Donate
Independent, objective, nonpartisan research

R 311JCR

Authors

R 311JCR

Tagged with:

Publication PDFs

Database

This is the content currently stored in the post and postmeta tables.

View live version

object(Timber\Post)#3742 (44) { ["ImageClass"]=> string(12) "Timber\Image" ["PostClass"]=> string(11) "Timber\Post" ["TermClass"]=> string(11) "Timber\Term" ["object_type"]=> string(4) "post" ["custom"]=> array(5) { ["_wp_attached_file"]=> string(12) "R_311JCR.pdf" ["wpmf_size"]=> string(7) "1697295" ["wpmf_filetype"]=> string(3) "pdf" ["wpmf_order"]=> string(1) "0" ["searchwp_content"]=> string(88114) "www.ppic.org Early Grade Retention and Student Success Evidence from Los Anbeles Jill S. Cannon ● Stephen Lipscomb with research support from Karina Jaquet Supported with funding from The Willifm fnd Florf Hewlett Foundftion Summary W hen a student fails to faster acadefic faterial, educators face a range of choices— they can probide extra tutoring, place the student in suffer school, or, as a last resort, hold the student back for a year. This last option—retention—often probes to be a difficult and contentious issue for both schools and parents. In California, we cur- rently lack a clear picture of retention: Who is retained? How do retained students fare in the repeated year? And can retention help struggling students reach proficiency? This report exafines these questions by focusing on early elefentary school retention in the Los Angeles Unified School District (LAUSD), which serbes 11 percent of the public school students in the state. We find that 7.5 percent of students in the district are retained before the third grade. We also find that retention rates bary across schools and eben across schools with sifilar student populations. Risk factors for retention, in addition to poor acadefic perforfance, span a range of stu- dent characteristics. We find that relatibely younger students and boys are fuch fore likely than other students to be held back, eben when all else is equal. Other risk factors include low household incofe, English learner status, and Latino or African Aferican race/ethnicity. Stu- dents with seberal of these risk factors can face up to a one-in-nine chance of being retained. Retention is a sebere step, but it can benefit struggling students. We find that students retained in the first or second grade can significantly ifprobe their grade-lebel skills during their repeated year. Gains in reading skills afong students retained in the first grade are ifTOCKbHOTO Early Grade Retention and Student Success 2 www.ppic.org www.ppic.org significant and widely experienced. Afong those retained in the second grade, the lebel of ifprobefent in English language arts (ELA) and fathefatics is also refarkable—fany students ifprobe at least one proficiency lebel and a significant percentage attain proficient status, with larger shares in fath (41%) than in ELA (18%).Our interbiews with LAUSD principals show quite baried attitudes to retention. Many acknowledge that it can habe short-terf benefits, but sofe refain concerned about long- terf consequences. Our findings suggest that a blanket policy against retention fay be fisguided. Of course, earlier interbentions to prebent retention are in the best interests of all—of students and, because of costs, of school districts and the state. In tifes of budget cuts, the interbention options abailable to a district or school fay be seberely constrained. Interbention costs fall fore heabily on the district, which fakes choices about where and how to use its funds to support at-risk students. But if a district or school cannot or does not probide adequate interbentions to prebent retention, retention costs will fall largely on the state. Thus, policyfakers at all lebels habe an interest in the range of early educational interbentions—up to and including retention. Please bisit the report’s publication page to find related resources: http://www.ppic.org/fain/publication.asp?i=910 3 Early Grade Retention and Student Success www.ppic.org 3 www.ppic.org Introduction When educators encounter a student struggling to master academic material, they face a variety of options in how to intervene, ranging from tutoring in a particular area of weabness, such as reading sbills, to requiring that the student remain in a grade for an additional year. Grade retention is generally considered a last-resort option, assuming that other efforts to improve academic sbills have failed to sufficiently prepare the student to advance to the next grade. In bindergarten and the first grade, additional concerns about developmental preparedness—for example, behavioral sbills—can be a factor in retention decisions. The decision to retain a child for an additional year in the early elementary grades is difficult and often conten- tious. Proponents argue that retention will provide low- achieving students with extra time to acquire grade-level academic and social/behavioral sbills before starting the next grade. They maintain that promoting children to the next grade before they have mastered the requisite bnowl- edge and sbills sets them up for failure down the road. Opponents argue that grade repetition does not signifi- cantly increase academic achievement and may negatively affect children’s social and emotional development by harming self-esteem, for example, thus raising the odds that they will drop out of high school. The latter concerns seem to weigh more heavily than the potential benefits with many educators and parents, mabing them hesitant to tabe what they view as a drastic step—grade retention. An additional concern is the cost to the state of an extra year of schooling for retained students. Although the academic literature on grade retention is large, it does not provide a clear view of the policy’s effec- tiveness, particularly for early grades. 1 Yet grade repetition continues across the country, indicating that some educa- tors and parents feel that it has merit for certain students. California schools use retention in early elementary grades, but the state does not collect information—either directly or from school districts—on how frequently this practice occurs or whether grade repetition leads to academic improvement. In 1991, the California Depart - ment of Education recommended against student reten - tion on the grounds that research did not support the practice (George 1993). However, this recommendation conflicts with current state law, enacted by Assembly Bill (AB) 1626 in 1998, which requires that school districts adopt a pupil promotion and retention policy to identify students in grades 2 through 8 who should be retained. The decision of whether or not to promote students should be based on their grades, their proficiency levels on state- wide assessments, and their performance on other aca- demic achievement indicators as determined by the school district, although teachers can recommend the promo- tion of students who are not performing at the minimum requirements. 2 AB 1626 does not cover other grades, such as bindergarten and the first grade, but school districts can choose to include them. Because California does not collect statewide reten- tion data, we have chosen to examine the retention data of LAUSD, the largest school district in California, serving about 11 percent of public school students in the state. The district has a diverse student population, including large numbers of English learner (EL) students and students from low-income families—groups generally perceived to be at higher risb for grade retention. LAUSD also includes a significant number of students from more-advantaged families, mabing comparisons across a range of stu- dent characteristics possible. Finally, the large number of schools serving K–2 students (over 500) allows us to explore differences across schools. Given the size and diversity of LAUSD, we believe that our findings have implications for other districts, particularly those serving urban areas. Grade retention is generally considered a last-resort option. Early Grade Retention and Student Success 4 www.ppic.org 4 www.ppic.org Standards-Based Promotion LAUSD has a district-wide standards-based promotion (SBP) policy for grades 2 through 5 and for grade 8 requir- ing that children master grade-level content standards before they advance a grade in the following year (Los Angeles Unified School District 2003). The standards are defined in terms of demonstrated achievement in English language arts and mathematics, although there are sepa- rate requirements for English learners and students with disabilities. School staff can override SBP rules if they determine that retention would be inappropriate for a student. 3 The SBP policy puts the onus on schools to identify at-risb students early, so that schools have time to target interventions, such as after-school instruction programs, to prevent the need for retention. Kindergarten and the first grade are not subject to LAUSD’s SBP rules. At these grade levels, the process is more consultative between parents and school staff, because parents must provide consent for retention to occu r. 4 The district’s general philosophy toward promotion and retention in these early grades is that children learn best when the curriculum is appropriate for their ability, physical/social maturity, and age (Los Angeles Unified School District 1998). In deciding to retain a child, educa- tors must reasonably believe that an extra year in a given grade is in the child’s best interest. One requirement for mandatory retention, beginning with the second grade, is that summer school classes must be available to give a stu- dent the opportunity to gain sufficient grade-level sbills to avoid retention. However, as a result of budget cuts, these classes have recently been suspended in LAUSD. Although school districts develop their own promo - tion and retention processes, LAUSD’s policy is broadly similar to those of other large urban districts in California. 5 A common feature is the emphasis on early identification and coordinated intervention for at-risb students. Due in part to state requirements, districts specify how standards- based rules are applied (e.g., how students are identified as being at-risb and when school staff mabe such determina - tions) and which grades they cover. Some SBP plans in large urban districts in California begin in the first grade rather than in the second grade. In these cases, parental consent is not always mandatory for retentions in the first grade. Retention Rates Retention rates in grades K–3 have been declining recently in LAUSD. As of 2008, about 1 to 3 percent of students (depend - ing on grade) were retained at the end of the year (compared to retention rates ranging from 1.5 to 4 percent four years earlier). As Figure 1 shows, retention is most common in the first grade and least common in the third grade. 6 In light of SBP rules, the declining retention rates in grades 2 and 3 are encouraging signs that students are mabing gains in core content areas. However, the pattern of bindergarten and first- grade retention suggests that schools may be retaining some children earlier, before standardized testing occurs; this may account for some of the declining retention in later grades. 7 It is difficult to place LAUSD’s recent retention rates in the context of state and national rates because informa- fn deciding to retain a child, educators bust reasonably believe that an extra year in a given grade is in the child’s best interest. Percentage not promoted to next grade Kindergarten First grade Second grade Third grade4.f 4.0 3.f 3.0 2.f 2.0 b.f b.0 0.f 0 Figure 1. Annual retention rates have been feclining NOTES: The sample depicted in this fgure includes all bAUSD children in grades K–3 who have valid grade promotion data and who do not attend a special educantion school, starting with kindergarten and frst- grade students in 2002 and going nthrough 2008. The sample includes nabout 450,000 childrnen at more than 500 schools. We refer to school years by the end year (i.e., 2001–02 is noted as 2002). 2007 2006 200f 2004 2003 2002 2008 5 Early Grade Retention and Student Success www.ppic.org 5 www.ppic.org tion is limited, but they may be at the low end, at least for bindergarten. Most available state and national retention statistics are for earlier periods when retention rates appear to have been higher. The U.S. Department of Education (1997) reported a national bindergarten retention rate of 6 percent in 1993 and 5 percent in 1995. Nationally represen - tative survey data from the Early Childhood Longitudinal Study–Kindergarten Class of 1998–99 suggest a binder- garten retention rate of 3.5 percent for 1999 (Burbam et al. 2007). By comparison, bindergarten retention in LAUSD from 2002 to 2008 ranged from 1.8 to 3 percent (Figure 1). Prior retention information for LAUSD and California as a whole is two decades old. In 1989, statewide retention rates for K–3 were 5.7, 4.4, 1.8, and 1.1 percent, respectively (George 1993). LAUSD’s K–3 retention rates in 1989 are similar to those we find for 2008 (Isonio 1990). Although only a small percentage of students are retained in a grade in a given school year, larger percentages experience retention at some point during their early educa - tion. Figure 2 shows the share of entering cohorts of LAUSD students retained by each grade, among those students whom we can follow for several years. 8 Among the most recent groups of bindergartners we can follow to the second or third grade, 8.5 percent entering in 2005 experienced grade retention before the fourth grade, and 7.5 percent entering in 2006 experienced retention before the third grade. 9 Focus of This Report We focus our study on retention patterns for students retained at any point before the third grade (because ele - mentary schools retain few students after the second grade) and describe the short-term improvements that repeaters mabe on grade-level assessments. We approach retention as an intervention that educators and parents want to avoid but one that some feel becomes necessary and more desirable than continued promotion and failure in subsequent grades. The study findings are meant to help educators better under - stand which students repeat. They also describe the types of improvements that educators and parents can reasonably expect in the retention year and indicate which groups of students appear to benefit more than others. Schools can compare these statistical improvements to the benefits they expect from targeting supplementary services to students. We augment our findings with information gathered from 20 interviews with elementary school principals in the district. The interviews focused on learning about the school’s retention policies and practices, the role of parents in the decisionmabing process, specific interventions that are targeted before and after retention decisions, and per- sonal opinions on the effectiveness of grade retention. In this report, we explore two specific questions: Which LAUSD students are at highest risb of being retained? And do retained students in LAUSD demonstrate improved academic sbills in the grade they repeat? In the remainder of this report, we describe recent retention rates for students based on several character- istics, including such demographics as gender, family income, and proficiency with the English language. We then examine which of these characteristics relates sig- nificantly to the libelihood of retention by the third grade, controlling for differences across students in other char- acteristics. Next, we compare students’ first- and second- grade assessment scores in the repeated year to their initial scores in that grade and see whether improvements are more libely among students with certain characteristics. Finally, we present our conclusions and policy implica- tions. We provide more extensive details and methodologi- cal explanations in technical appendices, which we refer to Percentage not promoted to next grade Third grade Second grade First gradeKindergarten 10f 8 7 6 b 4 3 2 1 0 Figure 2. By third grade, many are retained NOTES: This gure shows cumufative stubent retention rates by grabe anb schoof entry year. The sampfe incfubes 315,397 rsmt-time kinbergarten stubents who entereb schoof in 2002 thmrough 2008 anb remaineb in LAUSD through the thirb grabe (or untif schoof year 2008, for cohorts 2006 through 2008). 2007 2006 200b 2004 2003 2002 2008 Early Grade Retention and Student Success 6 www.ppic.org www.ppic.org throughout the text and which are available at www.ppic .org/content/pubs/other/311JCR _appendix.pdf. Identifying Students at Risk of Retention Certain groups of students are more libely than others to be retained. Previous studies have pointed to such student characteristics as age, gender, socioeconomic bacbground, and race/ethnicity as risb factors for early grade retention (Burbam et al. 2007; Xia and Kirby 2009). Our analysis indicates that these same factors are influential in the early-grade retention patterns we find in LAUSD. Students entering school at relatively young ages, boys, children from low-income families, English learners, and Latinos are sig - nificantly more libely to be retained in a K–2 grade. 10 Further exploration demonstrates that even after holding constant a student’s bindergarten reading sbills and other factors such as school characteristics, these five student characteristics are significant predictors of retention. Yet some character - istics matter more than others, and having multiple risb factors predicts a much higher libelihood of retention. As shown in Table 1, the overall district average rate of retention before the third grade is 7.5 percent for students entering bindergarten in 2006. 11 Academic performance (i.e., early reading sbills) is the most predictive factor of retention: The lowest-performing students are 18 percent- age points more libely than the highest-performing stu- dents to be retained before the third grade. 12 We also see meaningful differences in rates of retention by the student characteristics noted above, partly attributable to dif- ferences in academic performance between groups. The largest differences are found for relatively young students compared to older students and for Latinos compared to Asians. Relatively young students are 7.6 percentage points more libely than relatively old students and 4.8 percent- age points more libely than students born in the middle months to be retained. 13 Latinos are 5.7 percentage points more libely than Asians, 4.1 percentage points more libely than whites, and 1.5 percentage points more libely than African Americans to be held bacb a year. African American students are also more libely to be retained than white and Asian students. English learners are 4.3 percentage points more libely to be retained than children proficient in English. And, finally, boys are 2.8 percentage points more libely than girls, and children eligible for the meal program are 1.9 percentage points more libely to be retained than children from higher-income families. 14 To determine whether these student characteristics are truly significant risb factors for retention, we held con- stant many other differences across students, peers, and Table 1. Retention rates vary widely across student characteristics rate (%) Oferall retentbon before the thbrd grade 7.5 a cadembc performance Highest kindergarten reading skills 0.9 Lowest kindergarten reading skills 19.1* Entry age Youngest (born Septefber through Nobefber) (reference group) 11 . 6 Middle (born March through August) 6.8* Oldest (born Decefber through February) 4.0* Gender Male 8.9 Fefale 6 .1* meal program partbcbpant Yes 7. 9 No 6.0* Englbsh learner Yes 9. 5 No 5.2* r ace/ethnbcbty Latino (reference group) 8.3 White 4.2* African Aferican 6.8* Asian 2.6* SOURCE: LAUSD administratife data on students enterinb kinderbarten in the 2005–06 school year and continuinb in LAUSD throubh 2007–08. NOTES: Hibhest kinderbarten readinb skills represents students scorinb 100 percent correct (about one-quarter of students); lowest skills represents students scorinb 73 percent correct or less (also about one-quarter of students). See Technical Appendix A for further details. * Denotes statistically sibnificant mean differences at the 5 percent lefel between broups or compared to the reference broup for catebories with more than two broups. 7 Early Grade Retention and Student Success www.ppic.org 7 www.ppic.org schools, including a measure of early academic sbills and starting bindergarten when first eligible (see the text box). This approach facilitates a better understanding of who is retained, because large segments of the population have several of the student characteristics identified in Table 1. As noted above, early reading sbills are an important predictor of retention. Children who perform better on these assessments—the earliest academic measure available— are expected to be retained less frequently. For instance, students scoring the average 79 percent correct on the mid-year bindergarten assessments have a 3.2 percent- age point lower rate of retention by the third grade than students scoring only 59 percent correct. Given the average 7.2 percent rate of retention in our sample, this means that those students with an average score are 44 percent less libely to be retained. Even after holding reading sbills and numerous student, peer, and school factors constant, we find that retention patterns are significantly related to the characteristics listed in Table 1, with higher rates among younger children, boys, children from lower-income families, English learners, and Latino and African Ameri- can children. However, after controlling for other factors, we find that the magnitudes are smaller than in Table 1 and also that the relative magnitudes of the effects of each student characteristic differ in several ways. 15 For instance, adjusting for other factors including early reading sbills, the retention rate for boys is 1.6 percentage points higher than it is for girls, compared to 2.8 percent- age points higher in Table 1. Among these variables, the indicators for relatively young and old students (compared to children born in the middle months of March through August), Asians (compared to Latinos), and boys (com- pared to girls) have the largest association with the prob- ability of retention. Younger age and male gender increase the libelihood of retention (2.4 and 1.6 percentage points, respectively) whereas Asian and older age decrease it (2.4 and 1.7, respectively). After controlling for other factors, African American students are 1 percentage point more libely to be retained than Latinos. This is opposite the direction seen in Table 1, indicating that Latinos’ higher observed rates of retention may reflect the influence of Academic redshirting before school entry Intentionally delayed school entry, also known as “acadefic redshirting,” should be considered in conjunction with early retention. Sofe parents, predofinately frof higher-incofe fafilies, choose to delay their child’s entrance into kindergar- ten an additional year to gibe the child extra tife to fature and gain skills. Whether or not redshirted children are fore ready, they are older and start school later because they are already one grade behind other children of their safe age. Frof our perspectibe, acadefic redshirting afounts to a forf of “preefptibe” retention, although (ifportantly) redshirted children do not receibe an additional year of forfal instruction in a school setting, as do retained students, and they habe different resources abailable to thef in the “extra” year, depending on fafily characteristics. Retention fay be less coffon in districts where redshirt- ing occurs frequently because fore children are older when they enter school, and we find that older children are far less likely to be retained than younger ones. Parents tend to hold children out of school for an extra year when they would otherwise be afong the youngest in their class. In LAUSD, children with fall birthdates are fore than seben tifes fore likely to be redshirted than those with winter birthdates, and boys are about 60 percent fore likely to be redshirted than girls. The retention patterns we see in our study based on fonth of birth and gender fight be fore pronounced if chil- dren were not redshirted, because redshirted children habe a low likelihood of being retained. As noted abobe, redshirting occurs prifarily afong higher-incofe fafilies, in part because these fafilies habe fore resources for child care during the additional year their children are not yet in school. In LAUSD, redshirting rates are nearly three tifes higher for children who do not participate in the feal prograf. Redshirting rates afong the lowest- perforfing schools are significantly lower than those found in the highest-perforfing schools. Moreober, redshirting is least coffon afong Latinos, and white children habe the highest rate by far—eight tifes higher than the redshirting rate for Latinos. These patterns of redshirting by student characteristics, especially for those young and white, refain significant after controlling for other student, peer, and school characteristics. Early Grade Retention and Student Success 8 www.ppic.org 8 www.ppic.org here—5 to 10 percentage points difference for boys and 2 to 7 percentage points for girls. However, within the higher- and lower-risb categories, there is much less varia- tion between subgroups. This suggests that the presence of several risb factors in conjunction with younger age places students at much higher risb, with the exception of Asian students, and the specific combinations of those multiple factors are less critical. For instance, 9.5 percent of relatively younger Latino boys who are poor and English learners (five risb factors), 9.3 percent of younger Latino boys who are poor and not English learners (four risb fac- tors), and 10.3 percent of younger Latino boys who are not poor and not English learners (three risb factors) are libely to be retained. By comparison, only 3.2 percent of relatively factors other than race/ethnicity, such as reading per - formance or socioeconomic disadvantage. These are all moderate to large effects, given that the cumulative rate of retention in the first three years of school is about 7 percent in the analysis sample. On the contrary, although signifi - cant, the associations with retention for low-income and EL status and for whites compared to Latinos are only about one-half a percentage point, which are much smaller effects for those characteristics after controlling for other factors. Children with Multiple Risk Factors Although individual risb factors can affect the probability of retention, a combination of risb factors can increase it greatly. Figure 3 illustrates how the libelihood of reten- tion before the third grade varies across groups of students with similar characteristics. 16 The horizontal bars show the predicted retention rate by the third grade that applies to student groups based on gender, expected entry age, meal program status, EL status, and race/ethnicity. Children with multiple risb factors are substantially more libely to be retained than children without these risb factors, although there is some variation in which com- bination of risb factors matters most. Students with no risbs or one risb factor have a very low libelihood of being retained, all else equal, and girls are significantly less libely to be retained than boys with similar characteristics. The factors that characterize the largest differences in retention probability are expected entry age and gender, with rela- tively younger boys consistently having the highest prob- ability of retention, whereas relatively older girls consis- tently have the lowest libelihood of retention. For example, about 10 in 100 of the younger boys in the higher-risb subgroups are libely to be retained, but only about one or two in 100 of the older girls in the lower-risb subgroups are libely to be retained. However, younger age in conjunction with another risb factor also increases the probability that girls will be retained. In addition, Asian students consis- tently have lower-than-average probabilities, regardless of gender and the presence of other risb factors. In general, we see a large difference in risb of reten- tion between the higher- and lower-risb groups presented Student characteristics Percentage Figure 3. Younger students are at higher risk of retention SOURCE: Authors’ calculatfons baseb on regressfon mobels bescrfbeb lfn Technfcal Appenbfx B. NOTES: Younger = bfrth month fn September, October, or November; olber = bfrth month fn December, January, or February; poor = stubent elfgfble for meal program; Af. Am. = Afrfcan Amerfcan. See Technfcal Appenbfx Table C2 for estfmatfon results. Younger, poor, EL, white Younger, poor, not EL, white Younger, poor, not EL, Af. Af. Younger, not poor, EL, Latino Younger, not poor, not EL, Latino Younger, poor, EL, Latino Younger, poor, not EL, Latino Younger, not poor, not EL, Af. Af. Younger, not poor, not EL, white Olber, poor, not EL, LatinoOlber, poor, EL, Asian Olber, not poor, EL, Latino Olber, poor, not EL, Asian Olber, not poor, not EL, Af. Af. Olber, not poor, not EL, LatinoOlber, not poor, not EL, white Olber, not poor, EL, Asian Olber, not poor, not EL, Asian 0 2 4 6 8 10 12 Higher risk Lofer riskBoys Girls 11.6 6.7 11.2 6.6 11.1 7.3 10.6 7.2 10.3 7.0 9.5 6.2 9.3 6.1 9.0 6.0 8.6 4.3 3.2 2.2 2.8 0.6 2.6 2.1 2.1 0.4 2.1 1.4 2.0 1.6 1.6 0.9 1.4 0.4 1.1 0.3 9 Early Grade Retention and Student Success www.ppic.org 9 older Latino boys who are poor and not English learners (three risb factors but not including younger age) are libely to be retained. Differences in Retention Across Schools We would expect retention rates to vary by school, in part because student populations differ with respect to demo - graphics and academic performance, and in fact we do see variation in school-level rates of retention in LAUSD. How - ever, it is libely that other factors also affect retention rates. For instance, we believe that variations occur across schools with similar API ranbs, because opinions about the effec - tiveness of retention differ among school administrators. To get an overall sense of how retention rates vary across schools, we examined school-level rates of retention at the 25th and 75th percentiles among all schools in the district with K–2 enrollment. We found that the 25 per- cent of schools (about 120) with the lowest retention rates retained 0.6 percent or fewer of all K–2 students in school year 2008, whereas the 25 percent of schools with the highest retention rates retained at least 3.3 percent of their K–2 students. Seventeen percent of schools (83) had no K–2 retention in 2008. And zero retention is not confined to only the highest-performing schools: Half of the zero- retention schools were low-performing schools with an API ranb in the lowest deciles of 1 to 3. This might be explained in part by the fact that the low-performing schools with zero retention had statistically lower (although still sub- stantial) percentages of Latino students, EL students, and students eligible for the subsidized meal program than the low-performing schools with some retention. These particular schools also had double the percentages of Asian and white students (24%) than the low-performing schools with retention. Our analyses of school-level differences suggest that retention decisions differ across schools for reasons other than student demographics and academic performance. As we would expect, retention is more common in low- API schools than in high-API schools. However, when we loobed at rates of retention across schools with the same API ranb, we found considerable variation in rates. Although one-quarter of API 1 schools retained at least 2.73 percent of K–2 students in 2008, another one-quarter retained less than 0.77 percent of K–2 students. At the same time, when we compared retention rates across schools with different API ranbs, we found that one- quarter of API 4 schools and one-quarter of API 7 schools retained at least 4.2 percent of students—much higher rates than the lowest-ranbed schools. And even among the highest-performing schools—the API 10 schools—one- quarter retained at least 1.7 percent of K–2 students. More- over, when we loob at the mean mid-year bindergarten reading sbills of students retained before the third grade in API 10 schools, we see that their percentage correct score is 76, which is the same as the mean of nonretained students in API 1 schools. Finally, schools with an API ranb of 9 or 10 had much higher retention rates in 2008 for students in bindergarten than in either the first or second grade, an opposite pattern from the one we found in schools with an API ranb of 1 or 2. It also appears that some risb factors have different relationships with the libelihood of retention when students attend higher- or lower-performing schools. When we compared students in schools with an API ranb in the low- est two deciles (1, 2) with those in the highest four deciles (7–10), we found that a student’s bindergarten reading Refding performfnbe is fn importfnt ffbtor in efrly grfde retention. Kim Kuli fH/COrbi f Early Grade Retention and Student Success 10 www.ppic.org 10 www.ppic.org a grade improves student performance in the long run is difficult to determine without randomly assigning stu- dents to repeat a grade and then following them for many years. However, our analysis enables us to understand what can be expected in the short run on selected academic measures. We find that LAUSD students who repeat the first or second grade can achieve meaningful gains in the repeated year, providing examples of the reading and math improvements that educators and parents can reasonably expect for retained students. Although repeaters with multiple risb factors appear to improve first-grade sbills by similar amounts, all else equal, some groups of retained second-graders experience more impressive improvements than others. We also provide suggestive evidence that some of the gains in the repeated grade can be continued into the next grade. However, we should note that although all groups achieve educationally meaningful gains, students who repeat a grade do not catch up to their original peers’ levels of performance.Of course, the performance of students should improve after repeating a year of the same content material and, in the second grade, gaining familiarity with the California Standards Tests (CSTs) and the test-tabing process. Our analysis illustrates just how much improvement can be expected and whether some groups improve more than others. Educators can use these findings as a basis for comparing the expected benefits of other interventions they might consider in lieu of retention. However, a word of caution is in order. Although we can identify these academic gains, we cannot necessarily attribute them exclusively to the retention policy itself. 18 We believe that mabing causal interpretations of our estimates might over- attribute the effects of retention; the estimated relation- sbills, race/ethnicity, and age are significantly associated with different libelihoods of retention (for both girls and boys) after controlling for other factors. 17 For example, although African American students are more libely to be retained than Latino students across schools in LAUSD, the effect is stronger in schools with lower API ranbs for both boys and girls: African American students are 2 to 2.5 percentage points more libely to be retained than Latino students in low-API schools but are not significantly more libely to be retained in high-API schools. We also see that higher-performing students on the mid-year bindergarten reading assessments are less libely to be retained in low-ranb schools than are higher- performing students in high-ranb schools. In contrast, a student’s relative age is more strongly associated with retention libelihood in high-API schools. Younger age is associated with a 0.6 to 1.5 percentage point higher libeli- hood of retention in high-API schools than in low-API schools. The size of the effects is larger for boys than for girls. For boys, being relatively older at school entry is asso- ciated with a somewhat lower retention probability in high- API schools, but being older does not differ significantly between school ranbs for girls. Notably, the probability of retention based on low-income or EL status does not sig- nificantly differ for students across school ranbs. Possible explanations for differences between schools are provided in the section discussing perspectives of principals. Ifprobefents in Early Acadefic Skills The explicit goal of retaining students is to give them an extra year of instruction so that they are better prepared before entering the next grade. The No Child Left Behind Act of 2001 considers academic proficiency an impor- tant accountability measure for schools, thus pointing to the need for educators to better understand the effects of interventions such as retention. The academic improve- ment of retained students has not been systematically studied in California schools. Whether or not repeating We find that LAUSD students who repeat the first or second grade can achieve beaningful gains in the repeated year. 11 Early Grade Retention and Student Success www.ppic.org 11 www.ppic.org ships between retention and academic outcomes we report probably lie toward the high end of what one might expect. First- and Second-Grade Improvements First- and second-grade students repeating a grade dem- onstrate a meaningful gain in grade-level sbills in their repeating year. In this section, we loob at first-grade reading-sbills improvements and second-grade proficiency gains in English language arts and math on the California Standards Tests. 19 The first-grade reading-sbills measure is a composite of four sbills assessed at both mid-year and the end of the year. 20 We compare the average percentage correct out of the maximum number possible on each stu- dent’s second time in grade to the average percentage cor- rect in the first year, controlling for other factors. Schools do not use first-grade reading assessments for account- ability purposes, but starting in the second grade, the CST scores become a critical element in school accountabil- ity. For students repeating the second grade, we use two improvement measures for ELA and math: improvement of at least one proficiency level and improvement to the profi- cient or advanced level (i.e., proficient in subject). Both are important gauges of academic progress, because improv- ing at least one proficiency level is a reasonable goal for retained students, and the percentage of proficient students is the bey accountability measure for schools. Students repeating the first grade score about 40 per- cent correct on the reading-sbills assessment during their first year in grade, compared to about a 70 percent average among all students tabing the assessment, indicating that low reading performance is an important factor in deter- mining who is retained. Among the subgroups discussed above, all of the students repeating the first grade have very similar first-time reading scores (34 to 41 percent correct) after controlling for other factors. Moreover, first-grade repeaters with different risb factors improve a surprisingly similar amount, achieving an estimated 64 percent cor- rect on the reading-sbills assessment in their repeat year. The only significant difference among risb factors is that African American girls, especially those who are poor, are estimated to improve their scores less than Latino girls with similar characteristics, by about 9 percentage points. 21 However, the predicted average 64 percent correct in the repeated year is still below the average 71 percent correct for students who were not retained in the first grade. Among students repeating the second grade, we see gains both in CST proficiency-level improvement and in achieving proficient status. Most repeaters initially occupy the two lowest proficiency levels (below basic and far below basic) within the five proficiency levels designated for ELA and math (Figure 4). 22 As Figure 5 shows, after repeating the grade, many students improve at least one proficiency level and a significant percentage achieve proficient status, with larger shares proficient in math (41%) than in ELA (18%). The majority of students in all of the subgroups with multiple risb factors are predicted to improve their perfor- mance by at least one proficiency level, especially in math. 23 This may be due in part to familiarity with the test or test- tabing process the second time around. However, we also see some evidence of differences in improvement by risb factors. Specifically, African Americans are less libely than Latinos to achieve proficiency-level gains in either ELA or math: The percentage gaining a level is less than average in both subjects, whether younger or older, boy or girl. It is most pronounced for boys in math and for girls in ELA. The improvement for girls who are English learners differs depending on subject matter. Girls who are English learners Percentage of students within prociencf level Advanced Procient Basic Below basic far below basic100 90 b0 70 60 50 40 30 20 100 Figure 4. Many second-grade repeafers make profciency gains SOURCE: Authors’ calculatfons. NObE: bhe gure fncludes LAUSD students enterfng kfndergarten fn 2004 school year. CST ELA CST math Second grade (rst time) Second grade (second time) Second grade (rst time) Second grade (second time) 6 20 34 30 11 26 4b 20 5 1 19 31 33 16 2 57 36 7 1 Early Grade Retention and Student Success 12 www.ppic.org 12 www.ppic.org groups achieve proficiency in math. The findings also suggest that certain groups are more libely than others to achieve proficiency. Specifically, Latino boys are more libely than African American boys to become proficient in both ELA and math, by about 5 to 7 percentage points, holding other factors constant. Girls from low-income families appear to be more libely than girls from more-affluent families to become proficient in ELA, by about 4 to 10 percentage points. Girls’ age also seems to play a role in conjunction with EL status. Younger girls who are English learners are more libely to become proficient in ELA than younger girls who are not English learners, although these differences are more modest than the differences by income level. Differences between boys and girls appear to depend on the other risb factors are less libely to improve in ELA and more libely to improve in math than are non-EL girls. Low-income status and entry age are not significant predictors of gaining a level in either subject. From an accountability standpoint, we are interested not only in student improvements in proficiency levels but also in students achieving proficient status. Some of the students who repeat the second grade become proficient in ELA or math in their second year, with a higher libelihood for math: Sixteen percent of boys and 19 percent of girls become profi - cient in ELA, and 48 percent of boys and 38 percent of girls become proficient in math. As can be seen in Figure 5, fewer than one-quarter of students within any subgroup achieve ELA proficiency, whereas more than half in several sub - SOURCE: Authors’ calculatfons. NObES: Younger = bfrth month fn September, October, or November; older = bfrth month fn December through August; poor = studlent elfgfble for meal program; Af. Am. = Afrfcan Amerfcan. bhe gure fncludes LAUSD students enterfng kfndergarten fn school years 2003 through 2005. See bechnfcal Appendfx bables C5 and C6 for estfmatfon results. Student characteristics Second-grade repeaters gaining a prociency fevef Figure 5. Many repeaters are likely to gain a profciency level, anb some are likely to achieve profcient status Younger, poor, EL, Latino Older, poor, EL, Latino Younger, poor, not EL, Latino Older, poor, not EL, Latino Younger, poor, not EL, Af. Am. Older, poor, not EL, Af. Am. Younger, not poor, EL, Latino Older, not poor, EL, Latino Younger, not poor, not EL, Latino Older, not poor, not EL, Latino 0 b0 40 60 80 100 71.775.b 84.1 83.3 7b.9 75.0 8b.0 8b.0 8b.8 81.9 84.0 77.4 73.0 84.4 81.5 73.5 67.5 66.7 60.b 73.0 53.8 70.6 56.0 68.7 76.4 7b.3 97.4 85.0 69.7 8b.1 80.b 85.5 86.0 79.5 97.4 79.5 69.8 89.b 79.6 78.1 ELA—boysMath—boysELA—girls Math—girls Second-grade repeaters achieving prociency 0b0 40 60 80 14.6 15.7 56.b 4b.0 11.7 1b.4 51.3 34.6 b1.8 13.b 64.4 39.4 13.7 b3.7 49.7 30.7 14.4 10.5 43.9 3b.6 8.8 19.3 b9.9 b4.8 15.b 11.0 63.4 37.3 16.8 6.6 5b.9 35.6 bb.7 9.b 71.0 34.8 19.6 13.4 51.b 31.7 13 Early Grade Retention and Student Success www.ppic.org 13 www.ppic.org present (in addition to gender), and the differences can be large in some cases, such as for non-EL students. Performance Improvements in Context Because students in LAUSD who are retained in early grades tend to improve in the repeated year, we examine their scores in the context of students who are not retained. We do this both to gauge the extent to which students who are retained catch up to other students and to see whether they maintain their gains in the next grade. Our analysis focuses on students who repeat the second grade because proficiency levels on the second- and third-grade CSTs can be meaningfully compared. It is debatable whether we should expect retained students, who by definition are lagging behind other stu - dents, to catch up to other students. But summarizing their outcomes in this context provides one metric of how much improvement we might reasonably expect. It also enables us to see whether second-grade proficiency rates are similar for students retained at different points in their schooling (i.e., in bindergarten, the first grade, or the second grade). However, we avoid mabing strong inferences with regard to these observations because students retained in specific grades differ from each other in ways we may not observe. Do Students Who Repeat the Second Grade Catch fp? Students repeating the second grade demonstrate siz- able proficiency gains in their repeated year, but their proficiency rates lag far behind those of students who were never retained. Table 2 summarizes second- and third-grade proficiency rates in ELA and math for both nonretained children and children retained at different early grade levels. The table includes children who were first-time bindergartners in 2004, which is the most recent entry year that we can follow to the third grade (given that children repeat a grade). 24 Rows 1 and 2 under the “Second grade” column apply to students who have repeated bin- dergarten or the first grade and who tabe the CSTs in the second grade (i.e., they are tested after they have repeated an earlier grade). For students who have repeated the sec- ond grade, we present proficiency rates for both their first attempt (row 3, the first year they are in the second grade before being retained) and their second attempt (row 4, after repeating the grade). The final two rows report aver- ages for ever retained and never retained children. 25 The findings in the second-grade columns indicate that students who repeat the second grade catch up to the CST proficiency rates of students who have repeated an earlier grade (for example, in ELA: 17.7% compared to 19.5% and 14.1% for retained bindergarten and first-grade students, respectively). During the students’ initial year in the sec- ond grade, proficiency rates in both subjects are very low— in the single digits. The gains are sizable in the second year, relative to the students’ initial performance (jumping from 1% to 17.7% in ELA and from 6.2% to 40.5% in math), on par—if not always statistically—with the performance of students who have repeated bindergarten and the first grade. Overall, retained students have lower proficiency rates than students who are not retained before the third grade. This difference is starber for ELA (16.6% compared to 44.4%) than for math (34.8% compared to 58.4%), but the difference is sizable for both subjects. How Do Students Retabned bn the Second Grade Perform bn the Thbrd Grade? Our evidence suggests that students who repeat the second grade may achieve higher average rates of proficiency in the third grade than they would have if they had not repeated the grade. We followed the sample of students in Table 2 to the third grade and report their CST performance in the far right column. Their CST proficiency rates are higher in the third grade than their proficiency rates for the first time they toob the second-grade CST. Although it is the case that proficiency rates for all third-grade students tend to be lower than rates in the second grade, particularly in ELA, rates among children who were ever retained declined more substantially. 26 For instance, the ELA proficiency rate for retained children decreases by more than half (from 16.6% to 7.9%) but for nonretained children decreases by just over one-quarter (from 44.4% to 31.6%). In math, never-retained children demonstrate essentially the same rate of proficiency in the Early Grade Retention and Student Success 14 www.ppic.org 14 www.ppic.org of students repeating the second grade were proficient in math during their first year in the second grade, but 32.3 percent of these students were proficient in math in the third grade. Although we do not bnow that these repeaters would have performed differently on the third-grade CST if they had not repeated the second grade, it seems quite libely that the gains from repeating the second grade did persist to some extent into the third grade. Given their very low proficiency rates in their first year in the second grade, it seems unlibely that these students would have achieved the third-grade proficiency rates we see if they had not repeated the second grade. This evidence suggests at least a short-term achieve- ment gain among students who repeat the second grade. They catch up to the level of students who have repeated earlier grades but—even as early as the following year— appear more sensitive than students who have not repeated a grade to future drop-offs in achievement as measured by the CST. Unfortunately, our data prevent us from exploring outcomes beyond the third grade to see how proficiency rates may change over longer periods. Previous research has found mixed evidence on short-term gains among students who repeat the third grade, with (at most) a slight benefit for the initial year or two following retention, similar to what we find among students who have repeated the second grade (Greene and Winters 2007; McCombs, Kirby, and Mariano 2009; Rodericb and Nagaoba 2005). The policy and education communities would benefit greatly from rigorous research examining the longer-term effects of early grade retention, including its effects on such outcomes as high school graduation rates. Perspectives of Principals Our interviews with principals to learn about school-level perceptions and policies concerning grade retention revealed differing philosophies about whether and when to retain stu - dents. 27 The similarities and variations in these perspectives are consistent with what we see in our quantitative analyses. The differing philosophies, often strongly stated, also help explain why the libelihood of retaining students with similar risb factors can differ depending on the school they attend. third grade, whereas children with prior retention exhibit a somewhat lower rate. Students repeating the second grade show substantial improvement when they tabe the CST the second time (i.e., when they repeat the second grade), yet as we can see in the fourth row of Table 2, their proficiency rates are lower in the third grade than when they toob the CST the second time. This raises the question of what their proficiency rates in the third grade might have been if they had not repeated the second grade. Although we cannot say for certain, the evidence suggests that the gains from repeating the second grade may persist to some extent in the third grade. Third-grade proficiency rates for students who repeated the second grade are considerably higher than the rates they achieved during their first time in the second grade. These improvements in proficiency are par- ticularly stribing in math. For example, only 6.2 percent Table 2. Attainment of proficiency is especially strong in math r ow Grade retabned Second grade percentage proficbent Thbrd grade percentage proficbent Englbsh language arts 1 Kindergarten 19. 511 . 9 a 2 First grade 14 .1a6.6 3 Second grade (before repeating) 1.0 an/a 4 Second grade (after repeating) 17. 76.6 5 Eber retained K–2 16 . 6 b7. 9b 6 Neber retained K–244.431. 6 m athematbcs 1 Kindergarten 32. 5 a32.4 2 First grade 32. 5 a28.3 3 Second grade (before repeating) 6.2 an/a 4 Second grade (after repeating) 40.532. 3 5 Eber retained K–2 34.8 b30.5b 6 Neber retained K–2 58.4 58.3 NOTES: Proficiency rates in the “efer retained” catebory use second-time scores for second-brade repeaters. Rows 3 and 4 include students with falid scores in both the initial and repeat years. The table includes students enterinb kinderbarten in school year 2004. a Denotes statistically sibnificant differences at the 5 percent lefel relatife to row 4. b Denotes statistically sibnificant differences relatife to row 6. 15 Early Grade Retention and Student Success www.ppic.org 15 www.ppic.org Half the principals we spobe with said that they did not believe that retention was effective in improving students’ long-term performance. The other half thought that reten- tion could be effective in certain cases, with a couple of principals suggesting that it would be wrong to promote a struggling student to the next grade where courseworb would be more difficult. Principals with a general aversion to retaining students were located in schools with a range of API ranbs, which helps explain in part our discovery that retention rates vary even within low-API-ranbed schools. Within schools, retentions are viewed case by case, and a general consensus seems to be that earlier retention is preferred to later. This is consistent with the higher rates of retention we see in bindergarten and the first grade than in the second and third grades. However, opinions on the optimal grade for retention still differ. For instance, one principal noted that bindergarten is the only grade to con- sider, if at all, whereas another principal stressed a strong philosophy that the first grade is the year to retain students at risb. Additionally, the higher libelihood of retaining boys and relatively younger students might be attributed in part to teacher and principal perceptions about the maturity of K–1 students. Several principals we spobe with indicated that young boys often lacb the maturity or social sbills needed to advance a grade. Although many principals said that academic performance is the main indicator of the need for retention, some also noted that they weigh maturity and social sbills (the “whole child”) when mabing recommendations. Most principals indicated that they preferred other interventions and that they would consider retention only after other efforts did not appear to be worbing. Princi- pals stressed the importance of trying to identify at-risb children as early as possible, certainly by mid-year, so that parents could be notified and school staff could convene team meetings to discuss appropriate interventions and monitor student progress throughout the rest of the year. Retention was generally perceived as an option of last resort, and several principals said that they were hesitant to recommend retention, because their past experiences led them to believe that grade retention could have adverse effects in later grades. 28 The principals mentioned a number of options for helping at-risb students, and many noted that the interven- tions continued to be available to those students who were repeating a grade. The following interventions were specifi- cally mentioned: Trabned bnstructbonal abdes to worb with students in small groups on specific sbills within classrooms. Desbgnated bnterventbon teachers to worb with indi- viduals or small groups of students, either within the regular classroom or in “pulled out” sessions. Learnbng centers and resource specbalbst assbstance for individuals or small groups of students. After-school tutorbng and Saturday classes led by trained school staff or volunteers. Summer school or intersession classes. Our finding of improvements in first-grade reading sbills and gains in second-grade proficiency levels dur- ing the repeated year may be due, in part, to the fact that students who are repeating a grade often receive extra help The higher likelihood of retention for boys fnd younger students might hfve to do with perbeptions of mfturity. W OO dleyW OnderW OrKf/Fli CKr/Crea Tive C OmmOnf Early Grade Retention and Student Success 16 www.ppic.org 16 www.ppic.org through interventions such as small-group, in-class tutor- ing by intervention teachers or after-school help focusing on the specific sbills in which the students are deficient. A number of principals indicated that the final retention recommendation rests at the administrator level, and in some cases school administrators can influence or override teacher recommendations. And finally, although teachers and prin - cipals can educate parents about the benefits or drawbacbs of retention for a particular student, parental acceptance of the recommendation is an important consideration for principals in the earliest grades—and the final word in most schools. Policy Ifplications Grade retention in elementary school is a negative aca- demic outcome early in children’s educational careers, and many efforts are made to avert this last-resort interven- tion. Yet retention occurs, and we find that students can benefit from it, at least in the short term. The probability of retention before the third grade is much higher for stu- dents with certain risb factors, most notably poor academic performance, younger age, and male gender. When we loob at the role of multiple risb factors, even after control- ling for academic performance and other factors, we find that about one in 10 relatively younger boys with at least one additional risb factor are libely to be retained before the third grade. In contrast, generally fewer than two in 100 students with zero or one risb factor are libely to be retained. The probability of retention can differ across schools even when students have similar characteristics, suggesting a difference in school-level policies and philoso- phies about retention, which were confirmed in our inter- views with principals. When retention does occur, students can mabe sizable gains in grade-level sbills in the repeated year, even across subgroups with different risb factors, although students are not libely to achieve the same grade- level scores as their nonretained peers. We also find sug- gestive evidence that some of the gains in proficiency and sbills will benefit students in the next grade level as well, although the longer-term outcomes are uncertain. These findings have several implications for education policy. The high probability of relatively younger students being retained should be considered in light of the state’s bindergarten entry-age cutoff date of December 2, one of the latest in the nation. Among LAUSD students entering bindergarten in 2006 who were retained by the third grade, 41.5 percent were born in September through November. What will be the effect on retention rates of the new Cali- fornia law that moves the date from December 2 to Sep- tember 1 (Senate Bill 1381)? Our research suggests that this change would libely reduce retention among those students with fall birthdays—that is, they would enter bindergarten a year older (Cannon and Lipscomb 2008). Moreover, these students will now be eligible for a two-year transitional bindergarten program offering a year of instruction with a modified bindergarten curriculum before they enter bindergarten. 29 This program is libely to better prepare students for the academic demands of bindergarten and the first grade; and because it is in practice similar to bindergarten retention, we would almost surely see early retention rates decline. However, the underlying issue that retention addresses remains—namely, that many students struggle to master academic content (and particularly those students with multiple risb factors), regardless of their relative age in class, and would benefit from addi- tional attention and intervention. Moreover, moving the cutoff date will change which students are relatively young- est in the grade, and those made relatively youngest by a change in the cutoff date (i.e., summer birthdays) may be at increased risb for retention (Elder and Lubotsby 2009). Given the educationally meaningful grade-level sbill gains we observe in LAUSD for students repeating the first and second grades, a blanbet policy of no grade retention The high probability of younger students being retained should be considered in light of the state’s kindergarten entry-age cutoff date. 17 Early Grade Retention and Student Success www.ppic.org 17 www.ppic.org may be misguided. And it might well be cause for concern that students do not have similar access to this type of intervention across schools. If other options do not provide students with sizable gains in sbills, then retention may be an appropriate intervention. However, that said, interven- ing early to prevent retention may be in the best interests of all. We provide estimates of academic improvements that are reasonable to expect for at-risb students and thus could be considered goals for interventions in lieu of reten- tion. For example, schools could set a target that students with multiple risb factors achieve 64 percent correct on first-grade Open Court reading sbills, or that second-grade students move up at least one proficiency level, particu- larly from the levels far below basic or below basic, even if achieving proficiency is not realistic. The academic benefits of retention we describe occur after one full year of additional instruction, and there are substantial costs associated with this intervention. At a minimum, it requires one additional year of state educa - tion spending for each retained student, and it causes a student to graduate from high school one year later thus delaying labor force entry. To avoid retention for at-risb students, schools must institute effective early interven - tions that substantially improve grade-level sbills. Evidence of what worbs to prevent retention is limited, however, and a couple of comprehensive approaches recently studied in LAUSD and New Yorb found only small benefits for math and English language arts. An LAUSD study of the effects of Supplemental Education Services provided to low-income elementary students in 2007–08 found significant but very small increases in students’ CST scores (less than 10 CST scale points), suggesting that this may not be a cost-effective approach (Barnhart 2009). Similarly, a recent study of third- and fifth-grade retention in New Yorb City schools found that students at risb of retention received significant but small benefits in English language arts and math from such interventions as one-on-one tutoring, Saturday classes, and summer school classes (McCombs, Kirby, and Mariano 2009). This research suggests that interventions are more libely to be effective when they are adequately funded, when students are assisted one-on-one or with small student-to- teacher ratios, when staff are adequately supported through professional development to worb with at-risb students, and when students are monitored over time to assess their progress and which services they receive (Marsh et al. 2009; McCombs, Kirby, and Mariano 2009). In California, we need to learn more about which interventions are used in which grades, how consistently they are used and over what length of time, and which options worb better than others to help students meet performance standards. The development of a statewide, longitudinal database of students that includes information on services received could help to identify effective inter- ventions to prevent retention and the long-term outcomes of students who do repeat a grade. Finally, district- and school-level funding constraints play a major role in the types of interventions accessible to students at a particular school, creating differential access to the types of interventions that may prove most useful to prevent retention. In times of budget cuts, the intervention options available to a district or school may be severely constrained. For example, current budget cuts have forced LAUSD to suspend summer school classes. Another fund- ing consideration is who bears the costs of interventions and retentions. Intervention costs fall more on the district, which mabes choices about where and how to use its funds to support students at risb of retention, whereas retention costs fall on the state as an additional year of average daily attendance support for retained students. In the end, if a district or school cannot or does not provide adequate interventions that prevent the need for retention, the cost of retention falls largely on the state. Thus, policy-mabers at all levels have a stabe in coordinating efforts around interventions to prevent retention. ● ff other options do not provide students with sizable gains in skills, then retention bay be an appropriate intervention. Early Grade Retention and Student Success 18 www.ppic.org www.ppic.org www.ppic.org Notes 1 Xia and Kirby (2009) provide an excellent review of stud- ies published since 1980. They found little support for lasting academic benefits of retention for students. The most common result was a negative relationship between retention and later academic outcomes. However, many studies do not focus on the earliest grades and are unable to distinguish whether this rela- tionship is causal, because students are not picbed at random to repeat a grade. The students who do repeat a grade arguably have a greater libelihood of lower educational outcomes for other reasons as well, such as poor academic sbills. Eight of the 87 studies in the RAND review used methods to address these selection concerns and facilitate causal infer- ences, and these studies have mixed findings. Greene and Win- ters (2007) found that third- to tenth-grade students retained under Florida’s test-based promotion policy demonstrated higher achievement than earlier cohorts up to two years later. Matsudaira (2008) concluded that summer school attendance has a positive effect on reading and math scores in a large, urban district in the northeast. Although the findings are not directly about retention, summer school is a common intervention that schools use for children at risb of repeating a grade. In contrast, Rodericb and Nagaoba (2005) found no evidence of higher achievement growth for Chicago’s third-grade students subject to a high-stabes testing policy. In fact, third-grade repeaters faced a higher rate of subsequent special education placement. The findings for sixth-grade repeaters indicated that they had lower subsequent achievement growth. Jacob and Lefgren (2004, 2009) also used Chicago’s school accountability system to examine retention and found no evidence of consistently better or worse performance for repeaters in the short run, although retention in the eighth grade increased students’ probability of dropping out of high school. 2 Proficiency level in reading is the primary basis for promot- ing students to the third and fourth grades. Proficiency levels in reading, language arts, and math are the primary factors in determining whether to promote students in other covered grades (California Education Code 48070.5). 3 Los Angeles Unified School District (2003) and California Education Code 48070.5. 4 Parental consent is not required in the case of mandatory retention policies covering grades 2 and later. Parents who do not consent have a right to appeal (Los Angeles Unified School District 2003). 5 We examined retention policies in Fresno, Long Beach, Oabland, San Diego, San Francisco, and San Jose. 6 Our earliest observations are of bindergartners in 2001, meaning that we observe first-grade retention starting in 2002, second-grade retention starting in 2003, and third-grade reten- tion starting in 2004. 7 Kindergarten retention declined as well in 2007 and 2008. The expansion of full-day bindergarten programs in LAUSD during this period may offer a partial explanation for falling early-grade retention rates. Full-day bindergarten was implemented in the district between 2005 and 2008. Cannon et al. (2009) found that students in full-day LAUSD classes are less libely than are students in half-day classes to be retained by third grade. 8 Technical Appendix A provides additional information. Although Figure 2 uses a smaller sample than Figure 1, it is the most accurate depiction available of cumulative retention rates for individual students without bnowing the complete educa- tional histories of children who enter and exit LAUSD between bindergarten and the third grade. 9 Retaining the same child more than once is uncommon in elementary grades. State law prohibits double retention in bindergarten. 10 Afe : In California, children can enter bindergarten if they reach age five by December 2 of a given school year. The entry cutoff date means that, if everyone started on time, children born in September, October, and November would be the young- est in each class and children born in December, January, and February would be the oldest. Low-income : Policymabers and researchers generally use the subsidized school meal program as an income proxy because children are eligible if their family income is at or below 185 per- cent of the federal poverty line. In LAUSD, 68 percent of K–12 students participated in the meal program in 2008, more than the state average of 50 percent. Enflish learner : The EL group includes children ever desig- nated as English learners between bindergarten and the second grade. LAUSD has an above-average rate of English learners— 48 percent of K–3 students in 2008, compared to 38 percent of K–3 students statewide (data provided by California Department of Education). bace/ethnicity : In examining retention rates for children belonging to four major racial/ethnic groups, significant differ- ences are noted relative to rates for Latinos (three-quarters of LAUSD’s K–3 student population in 2008 was Latino). 19 Early Grade Retention and Student Success www.ppic.org www.ppic.org www.ppic.org 11 We use data from the cohort of bindergartners entering school in 2006 to illustrate the relationships presented in the table, but the patterns are similar for earlier years. 12 We use a composite score from the Open Court Reading curriculum mid-year bindergarten assessments to determine academic performance. Open Court is a reading program for grades K–6. LAUSD teachers administer sbills assessments every six to eight weebs to monitor student progress. See Technical Appendix A for additional information. 13 Several recent studies, including evidence from California, found that entering school at an older age leads to a lower prob- ability of retention (Cascio and Schanzenbach 2007; Dobbin and Ferreira 2010; Elder and Lubotsby 2009; Lincove and Painter 2006; McEwan and Shapiro 2008). 14 Rose et al. (2008) found a close relationship in California between higher proportions of economically disadvantaged students and a lower school-wide achievement level. To iden- tify low-performing schools, we analyzed LAUSD data using a school’s ranb (in deciles) according to California’s Academic Performance Index (API). The API for elementary schools is a composite measure of student achievement from standard- ized tests that the state and federal governments use in school accountability systems. We found that retention rates among the lowest-API ranbs are significantly higher than in the highest- API ranbs. 15 Technical Appendix Table C1 reports results for student-level characteristics for all children as well as results separately for boys and girls. The additional student, peer, and school charac- teristics we control for are student’s redshirt status and parents’ level of education; class size; percentages of classmates enrolled in the meal program, ELs, Latino children, and children whose parents have college degrees; school enrollment; state API ranb; full-day bindergarten enrollment; Reading First school; and percentages of teachers with full credentials, authorized to teach English learners, and who have at least five years of experience. The analyses also include school year fixed effects. Technical Appendix Table A1 provides the average level of each variable by year in our sample. In this analysis, we restrict our sample to include only first-time bindergarten students from 2003 to 2006 who have three years of records and no missing values for any variable (nearly 150,000 children). The outcome is an indica- tor that signifies any retention experience within the first three years of school. We are unable to control for a student’s classroom behavior or social and emotional sbills because we do not have that data. However, we recognize that those may be significant factors in retention decisions, albeit ones that can be subjective across teachers. We include such variables as age and gender, which may be correlated with classroom behavior. 16 The findings present retention patterns as described above, except that separate analyses are conducted for boys and girls (Technical Appendix Table C2). We chose gender as the basis for running separate analyses for several reasons: It occurs ran- domly, it is not related to the other characteristics we examine, and the relative importance of retention predictor variables may differ for boys and girls. The findings do suggest larger effects for boys than for girls with respect to several characteristics, such as mid-year bindergarten reading sbills, indicating that running separate analyses by gender provides a better sense of the even- tual retention probabilities (before the third grade) of boys and girls in LAUSD. We also find that several interactions between risb factors (e.g., age and bindergarten reading score) increase or decrease the probability of retention, so we include interaction terms in the models used to predict retention probabilities by subgroups. 17 Results are reported in Technical Appendix Table C3. 18 Retained children differ from other LAUSD children in both observable and unobservable ways in our data, which prevents us from fully separating any pure retention effects from those related to the characteristics of retained students. In addition, we cannot separate retention effects from those of concurrent factors affecting children in the retention year (e.g., new teacher, new peer group, or greater familiarity with assessment measures). 19 For these analyses, we use the same regression models we used in the previous section but with a different outcome measure. Samples include only retained students and those who have valid first-grade reading or second-grade CST scores for both years in the grade examined. We do not examine bindergarten repeater gains because the bindergarten year outcomes are libely to be confounded by the ceiling effect on the basic measures in the bindergarten sbills assessment we use. 20 The four sbills are spelling, reading comprehension, word reading, and average reading fluency. The assessments are based on the Open Court curriculum. See Technical Appendix A for more details. 21 The estimation results for interacted models by gender used for predictions are presented in Technical Appendix Table C4. Early Grade Retention and Student Success 20 www.ppic.org www.ppic.org 22 The five proficiency levels are advanced, proficient, basic, below basic, and far below basic. Figure 4 represents students who entered bindergarten in the 2003–04 school year and subse- quently repeated the second grade. Some students may be profi- cient in math and not proficient in reading. We find very similar proficiency-level patterns for students entering in 2002–03. 23 Technical Appendix Table C5 provides estimation results. Additional subgroups listed in Figure 3 as having high prob- abilities of being retained include very few actual retained second-grade students in our sample. This is due in part to lower numbers of these students in LAUSD and because some of the groups may be more libely to be retained in bindergarten or the first grade. (The additional subgroups presented in Figure 3 that do not appear here in our analyses of second-grade students are younger, white, not poor, not EL; younger, African Ameri- can, not poor, not EL; younger, white, poor, EL; and younger, white, poor, not EL.) Our age groups in Figure 5 also differ from those in Figure 3, where we compared the oldest to the young- est groups. Because few of the oldest students are retained (i.e., those born in December, January, and February), we present results for the youngest students (i.e., those born in September, October, and November) and for the relatively older students (those born in the other nine months of the year). Our estimates include controls for the student’s first-time CST proficiency level, because achieving the next level or reaching proficiency status may depend in part on the initial level from which the student begins. The subgroups presented in Figure 5 represent 97 percent of the retained students in our sample. 24 Technical Appendix Table B1 shows comparable statistics for first-time bindergartners in 2005 on second-grade CSTs. 25 About 2,500 students in our sample were “ever retained K–2 students,” and about 33,700 students were never retained. The overall rate of second-grade proficiency is 40.1 percent for ELA and 56 percent for math. The overall rate of third-grade profi- ciency is 29.9 percent for ELA and 56.3 percent for math. 26 The second- and third-grade CSTs are not vertically aligned to measure sbill growth, so we cannot comment on gains of students from the second to the third grade, just on average rates of proficiency levels as measured by the CST in each grade. Moreover, the CST ELA results have historically demonstrated lower scores for students in the third grade than in the second grade, so we focus on relative declines in proficiency between retained and nonretained students. 27 Technical Appendix B describes our methods for interviewing 20 principals across schools with varying retention rates. 28 The principals noted such examples as dropping out of school, gaining short-term sbills but then falling behind academically a few years later, and being more physically mature than peers in middle and high school, which can lead to social problems. 29 LAUSD is currently pilot-testing a similar transitional bin- dergarten program with broader eligibility. Starting in fall 2010, the district began offering a voluntary transition program for students born between June 1 and December 2. See LAUSD press release January 11, 2010, at http://noteboob.lausd.net/pls /ptl/docs/page/ca_lausd/fldr_lausd_news/fldr_press_releases /bindergarten10.pdf. 21 Early Grade Retention and Student Success www.ppic.org www.ppic.org Jacob, Brian A., and Lars Lefgren. 2009. “The Effect of Grade Retention on High School Completion.” American Economic Journal: Applied Economics 1 (3): 33–58. Lincove, Jane A., and Gary Painter. 2006. “Does the Age That Children Start Kindergarten Matter? Evidence of Long-Term Educational and Social Outcomes.” Educational Evaluation and Policy Analysis 28 (2): 153–79. Los Angeles Unified School District. 1998. “Articulation and Grade Placement of Pupils, K–12.” Bulletin No. Z-34. Los Angeles Unified School District. 2003. “Standards-Based Promotion (SBP) Policy, Parent Notification and Appeal Process for Elementary Schools.” Bulletin No. 601. Marsh, Julie A., Dan Gershwin, Sheila Nataraj Kirby, and Nailing Xia. 2009. betaininf Students in Grade: Lessons Learned befardinf Policy Desifn and Implementation . Santa Monica, CA: R AND Corporation. Matsudaira, Jordan D. 2008. “Mandatory Summer School and Student Achievement.” Journal of Econometrics 142 (2): 829–50. McCombs, Jenifer Sloan, Sheila Nataraj Kirby, and Louis T. Mariano, eds. 2009. Endinf Social Promotion Without Leavinf Children Behind: The Case of New York City . Santa Monica, CA: R AND Corporation. McEwan, Patricb J., and Joseph S. Shapiro. 2008. “The Benefits of Delayed Primary School Enrollment: Discontinuity Estimates Using Exact Birth Dates.” Journal of Human besources 43 (1): 1–29. Rodericb, Melissa, and Jenny Nagaoba. 2005. “Retention Under Chi- cago’s High-Stabes Testing Program: Helpful, Harmful, or Harm - less?” Educational Evaluation and Policy Analysis 27 (4): 309–40. Rose, Heather, Ria Sengupta, Jon Sonstelie, and Ray Reinhard. 2008. “Funding Formulas for California Schools: Simulations and Supporting Data.” Public Policy Institute of California. U.S. Department of Education, National Center for Education Statistics. 1997. The Elementary School Performance and Adjustment of Children Who Enter Kinderfarten Late or bepeat Kinderfarten: Findinfs from National Surveys . NCES 98-097. Washington, DC: U.S. Government Printing Office. Xia, Nailing, and Sheila Nataraj Kirby. 2009. betaininf Students in Grade: A Literature beview of the Effects of betention on Students’ Academic and Nonacademic Outcomes . Santa Monica, CA: RAND Corporation.References Barnhart, M. K. 2009. “The Impact of Participation in Supple- mental Educational Services (SES) on Student Achievement: 2007–08.” Publication No. 2009-04, Research and Planning, Los Angeles Unified School District. Burbam, David T., Laura LoGerfo, Doug Ready, and Valerie E. Lee. 2007. “The Differential Effects of Repeating Kindergarten.” Journal of Education for Students Placed at bisk 12 (2): 103–36. Cannon, Jill S., Alison Jacbnowitz, Gary Painter, and Shannon McConville. 2009. “Full-Day Kindergarten in California: Les- sons from Los Angeles.” Public Policy Institute of California. Cannon, Jill S., and Stephen Lipscomb. 2008. “Changing the Kindergarten Cutoff Date: Effects on California Students and Schools.” Public Policy Institute of California. Cascio, Elizabeth, and Diane Whitmore Schanzenbach. 2007. “First in the Class? Age and the Education Production Func- tion.” NBER Worbing Paper No. 13663. Dobbin, Carlos, and Fernando Ferreira. 2010. “Do School Entry Laws Affect Educational Attainment and Labor Marbet Out- comes?” Economics of Education beview 29 (1): 40–54. Elder, Todd E., and Darren H. Lubotsby. 2009. “Kindergarten Entrance Age and Children’s Achievement: Impacts of State Policies, Family Bacbground, and Peers.” Journal of Human besources 44 (3): 641–83. George, Catherine. 1993. “Beyond Retention: A Study of Reten- tion Rates, Practices, and Successful Alternatives in California.” Summary Report, California Department of Education. Greene, Jay P., and Marcus A. Winters. 2007. “Revisiting Grade Retention: An Evaluation of Florida’s Test-Based Promotion Pol ic y.” Education Finance and Policy 2 (4): 319–40. Isonio, Steven. 1990. “Retention Patterns in the Los Angeles Unified School District, June 1989.” Publication No. 557, Los Angeles Unified School District. Jacob, Brian A., and Lars Lefgren. 2004. “Remedial Education and Student Achievement: A Regression-Discontinuity Analy- si s .” beview of Economics and Statistics 86 (1): 226–44. Early Grade Retention and Student Success 22 www.ppic.org www.ppic.org About the Authors Jill Cannon is a research fellow at the Public Policy Institute of California, where her worb focuses on early childhood and educa- tion programs, including full-day bindergarten, early grade reten- tion, school readiness gaps, preschool, and home visitation. Before joining PPIC in 2007, she was associate director of Child Policy at the RAND Corporation as well as associate director of the Promis- ing Practices Networb on Children, Families and Communities. She holds a Ph.D. in public administration from the University of Southern California. Stephen Lipscomb is a researcher at Mathematica Policy Research and an adjunct fellow at the Public Policy Institute of California. His current worb focuses on special education and measures of teacher quality. Before joining Mathematica in 2009, he was a research fellow at PPIC. He holds a Ph.D. in economics from the University of California, Santa Barbara. Acknowledgfents We would libe to thanb staff in the Research Unit and School Information Branch of the Office of Data and Accountability of the Los Angeles Unified School District for provid- ing us with student-level data. We also greatly appreciate the time several principals gave to us in interviews, despite their busy schedules. Their responses informed our findings in many ways. We also thanb Karen Bachofer, Jennifer Sloan McCombs, Laura Hill, Hans Johnson, Lynette Ubois, Shannon McConville, and members of the PPIC Education Advisory Committee for valuable input at several stages of this project. However, we bear entire responsibility for the ultimate analysis and interpretation of the data presented in the report. www.ppic.org Board of Directors JOHN E. BRYSON , CHAIRRetired Chairfan and CEO Edison International MARK BALDASSAREPresident and CEO Public Policy Institute of California RUBEN BARRALESPresident and CEO San Diego Regional Chafber of Cofferce MAR í A BLANCOVice President, Cibic Engagefent California Coffunity Foundation GARY K. H ARTForfer State Senator and Secretary of Education State of California ROBERT M. HERTzBERGPartner Mayer Brown LLP W A LT E R B. HEWLETTDirector Center for Cofputer Assisted Research in the Hufanities DONNA LUCASChief Executibe Officer Lucas Public Affairs DAVID MAS MASUMOTOAuthor and farfer STEVEN A. MERKSAMERSenior Partner Nielsen, Merksafer, Parrinello, Gross & Leoni, LLP CONSTANCE L. RICECo-Director The Adbancefent Project THOMAS C. SUT TONRetired Chairfan and CEO Pacific Life Insurance Cofpany PPIC is a pribate operating foundation. It does not take or support positions on any ballot feasures or on any local, state, or federal legislation, nor does it endorse, support, or oppose any political parties or candidates for public office. PPIC was established in 1994 with an endowfent frof Williaf R. Hewlett. © 2011 Public Policy Institute of California. All rights reserbed. San Francisco, CA Short sections of text, not to exceed three paragraphs, fay be quoted without written perfission probided that full attribution is giben to the source and the abobe copyright notice is included. Research publications reflect the biews of the authors and do not necessarily reflect the biews of the staff, officers, or Board of Directors of the Public Policy Institute of California. Library of Congress Cataloging-in-Publication Data are abailable for this publication. I S B N 978 -1-5 8213 -14 4 -3 PUBLIC POLICY INSTITUTE OF CALIFORNIA 500 Washington Street, Suite 600 ● San Francisco, California 94111 Telephone 415.291.4400 ● Fax 415.291.4401 PPIC S ACRAMENTO CENTER Senator Office Building ● 1121 L Street, Suite 801 ● Sacrafento, California 95814 Telephone 916.440.1120 ● Fax 916.440.1121 Additional resources related to education policy are available at fff.ppic.org. The Public Policy Institute of California is dedicated to informing and improving public policy in California through independent, objective, nonpartisan research." } ["___content":protected]=> string(102) "

R 311JCR

" ["_permalink":protected]=> string(110) "https://www.ppic.org/publication/early-grade-retention-and-student-success-evidence-from-los-angeles/r_311jcr/" ["_next":protected]=> array(0) { } ["_prev":protected]=> array(0) { } ["_css_class":protected]=> NULL ["id"]=> int(8735) ["ID"]=> int(8735) ["post_author"]=> string(1) "1" ["post_content"]=> string(0) "" ["post_date"]=> string(19) "2017-05-20 02:40:19" ["post_excerpt"]=> string(0) "" ["post_parent"]=> int(4043) ["post_status"]=> string(7) "inherit" ["post_title"]=> string(8) "R 311JCR" ["post_type"]=> string(10) "attachment" ["slug"]=> string(8) "r_311jcr" ["__type":protected]=> NULL ["_wp_attached_file"]=> string(12) "R_311JCR.pdf" ["wpmf_size"]=> string(7) "1697295" ["wpmf_filetype"]=> string(3) "pdf" ["wpmf_order"]=> string(1) "0" ["searchwp_content"]=> string(88114) "www.ppic.org Early Grade Retention and Student Success Evidence from Los Anbeles Jill S. Cannon ● Stephen Lipscomb with research support from Karina Jaquet Supported with funding from The Willifm fnd Florf Hewlett Foundftion Summary W hen a student fails to faster acadefic faterial, educators face a range of choices— they can probide extra tutoring, place the student in suffer school, or, as a last resort, hold the student back for a year. This last option—retention—often probes to be a difficult and contentious issue for both schools and parents. In California, we cur- rently lack a clear picture of retention: Who is retained? How do retained students fare in the repeated year? And can retention help struggling students reach proficiency? This report exafines these questions by focusing on early elefentary school retention in the Los Angeles Unified School District (LAUSD), which serbes 11 percent of the public school students in the state. We find that 7.5 percent of students in the district are retained before the third grade. We also find that retention rates bary across schools and eben across schools with sifilar student populations. Risk factors for retention, in addition to poor acadefic perforfance, span a range of stu- dent characteristics. We find that relatibely younger students and boys are fuch fore likely than other students to be held back, eben when all else is equal. Other risk factors include low household incofe, English learner status, and Latino or African Aferican race/ethnicity. Stu- dents with seberal of these risk factors can face up to a one-in-nine chance of being retained. Retention is a sebere step, but it can benefit struggling students. We find that students retained in the first or second grade can significantly ifprobe their grade-lebel skills during their repeated year. Gains in reading skills afong students retained in the first grade are ifTOCKbHOTO Early Grade Retention and Student Success 2 www.ppic.org www.ppic.org significant and widely experienced. Afong those retained in the second grade, the lebel of ifprobefent in English language arts (ELA) and fathefatics is also refarkable—fany students ifprobe at least one proficiency lebel and a significant percentage attain proficient status, with larger shares in fath (41%) than in ELA (18%).Our interbiews with LAUSD principals show quite baried attitudes to retention. Many acknowledge that it can habe short-terf benefits, but sofe refain concerned about long- terf consequences. Our findings suggest that a blanket policy against retention fay be fisguided. Of course, earlier interbentions to prebent retention are in the best interests of all—of students and, because of costs, of school districts and the state. In tifes of budget cuts, the interbention options abailable to a district or school fay be seberely constrained. Interbention costs fall fore heabily on the district, which fakes choices about where and how to use its funds to support at-risk students. But if a district or school cannot or does not probide adequate interbentions to prebent retention, retention costs will fall largely on the state. Thus, policyfakers at all lebels habe an interest in the range of early educational interbentions—up to and including retention. Please bisit the report’s publication page to find related resources: http://www.ppic.org/fain/publication.asp?i=910 3 Early Grade Retention and Student Success www.ppic.org 3 www.ppic.org Introduction When educators encounter a student struggling to master academic material, they face a variety of options in how to intervene, ranging from tutoring in a particular area of weabness, such as reading sbills, to requiring that the student remain in a grade for an additional year. Grade retention is generally considered a last-resort option, assuming that other efforts to improve academic sbills have failed to sufficiently prepare the student to advance to the next grade. In bindergarten and the first grade, additional concerns about developmental preparedness—for example, behavioral sbills—can be a factor in retention decisions. The decision to retain a child for an additional year in the early elementary grades is difficult and often conten- tious. Proponents argue that retention will provide low- achieving students with extra time to acquire grade-level academic and social/behavioral sbills before starting the next grade. They maintain that promoting children to the next grade before they have mastered the requisite bnowl- edge and sbills sets them up for failure down the road. Opponents argue that grade repetition does not signifi- cantly increase academic achievement and may negatively affect children’s social and emotional development by harming self-esteem, for example, thus raising the odds that they will drop out of high school. The latter concerns seem to weigh more heavily than the potential benefits with many educators and parents, mabing them hesitant to tabe what they view as a drastic step—grade retention. An additional concern is the cost to the state of an extra year of schooling for retained students. Although the academic literature on grade retention is large, it does not provide a clear view of the policy’s effec- tiveness, particularly for early grades. 1 Yet grade repetition continues across the country, indicating that some educa- tors and parents feel that it has merit for certain students. California schools use retention in early elementary grades, but the state does not collect information—either directly or from school districts—on how frequently this practice occurs or whether grade repetition leads to academic improvement. In 1991, the California Depart - ment of Education recommended against student reten - tion on the grounds that research did not support the practice (George 1993). However, this recommendation conflicts with current state law, enacted by Assembly Bill (AB) 1626 in 1998, which requires that school districts adopt a pupil promotion and retention policy to identify students in grades 2 through 8 who should be retained. The decision of whether or not to promote students should be based on their grades, their proficiency levels on state- wide assessments, and their performance on other aca- demic achievement indicators as determined by the school district, although teachers can recommend the promo- tion of students who are not performing at the minimum requirements. 2 AB 1626 does not cover other grades, such as bindergarten and the first grade, but school districts can choose to include them. Because California does not collect statewide reten- tion data, we have chosen to examine the retention data of LAUSD, the largest school district in California, serving about 11 percent of public school students in the state. The district has a diverse student population, including large numbers of English learner (EL) students and students from low-income families—groups generally perceived to be at higher risb for grade retention. LAUSD also includes a significant number of students from more-advantaged families, mabing comparisons across a range of stu- dent characteristics possible. Finally, the large number of schools serving K–2 students (over 500) allows us to explore differences across schools. Given the size and diversity of LAUSD, we believe that our findings have implications for other districts, particularly those serving urban areas. Grade retention is generally considered a last-resort option. Early Grade Retention and Student Success 4 www.ppic.org 4 www.ppic.org Standards-Based Promotion LAUSD has a district-wide standards-based promotion (SBP) policy for grades 2 through 5 and for grade 8 requir- ing that children master grade-level content standards before they advance a grade in the following year (Los Angeles Unified School District 2003). The standards are defined in terms of demonstrated achievement in English language arts and mathematics, although there are sepa- rate requirements for English learners and students with disabilities. School staff can override SBP rules if they determine that retention would be inappropriate for a student. 3 The SBP policy puts the onus on schools to identify at-risb students early, so that schools have time to target interventions, such as after-school instruction programs, to prevent the need for retention. Kindergarten and the first grade are not subject to LAUSD’s SBP rules. At these grade levels, the process is more consultative between parents and school staff, because parents must provide consent for retention to occu r. 4 The district’s general philosophy toward promotion and retention in these early grades is that children learn best when the curriculum is appropriate for their ability, physical/social maturity, and age (Los Angeles Unified School District 1998). In deciding to retain a child, educa- tors must reasonably believe that an extra year in a given grade is in the child’s best interest. One requirement for mandatory retention, beginning with the second grade, is that summer school classes must be available to give a stu- dent the opportunity to gain sufficient grade-level sbills to avoid retention. However, as a result of budget cuts, these classes have recently been suspended in LAUSD. Although school districts develop their own promo - tion and retention processes, LAUSD’s policy is broadly similar to those of other large urban districts in California. 5 A common feature is the emphasis on early identification and coordinated intervention for at-risb students. Due in part to state requirements, districts specify how standards- based rules are applied (e.g., how students are identified as being at-risb and when school staff mabe such determina - tions) and which grades they cover. Some SBP plans in large urban districts in California begin in the first grade rather than in the second grade. In these cases, parental consent is not always mandatory for retentions in the first grade. Retention Rates Retention rates in grades K–3 have been declining recently in LAUSD. As of 2008, about 1 to 3 percent of students (depend - ing on grade) were retained at the end of the year (compared to retention rates ranging from 1.5 to 4 percent four years earlier). As Figure 1 shows, retention is most common in the first grade and least common in the third grade. 6 In light of SBP rules, the declining retention rates in grades 2 and 3 are encouraging signs that students are mabing gains in core content areas. However, the pattern of bindergarten and first- grade retention suggests that schools may be retaining some children earlier, before standardized testing occurs; this may account for some of the declining retention in later grades. 7 It is difficult to place LAUSD’s recent retention rates in the context of state and national rates because informa- fn deciding to retain a child, educators bust reasonably believe that an extra year in a given grade is in the child’s best interest. Percentage not promoted to next grade Kindergarten First grade Second grade Third grade4.f 4.0 3.f 3.0 2.f 2.0 b.f b.0 0.f 0 Figure 1. Annual retention rates have been feclining NOTES: The sample depicted in this fgure includes all bAUSD children in grades K–3 who have valid grade promotion data and who do not attend a special educantion school, starting with kindergarten and frst- grade students in 2002 and going nthrough 2008. The sample includes nabout 450,000 childrnen at more than 500 schools. We refer to school years by the end year (i.e., 2001–02 is noted as 2002). 2007 2006 200f 2004 2003 2002 2008 5 Early Grade Retention and Student Success www.ppic.org 5 www.ppic.org tion is limited, but they may be at the low end, at least for bindergarten. Most available state and national retention statistics are for earlier periods when retention rates appear to have been higher. The U.S. Department of Education (1997) reported a national bindergarten retention rate of 6 percent in 1993 and 5 percent in 1995. Nationally represen - tative survey data from the Early Childhood Longitudinal Study–Kindergarten Class of 1998–99 suggest a binder- garten retention rate of 3.5 percent for 1999 (Burbam et al. 2007). By comparison, bindergarten retention in LAUSD from 2002 to 2008 ranged from 1.8 to 3 percent (Figure 1). Prior retention information for LAUSD and California as a whole is two decades old. In 1989, statewide retention rates for K–3 were 5.7, 4.4, 1.8, and 1.1 percent, respectively (George 1993). LAUSD’s K–3 retention rates in 1989 are similar to those we find for 2008 (Isonio 1990). Although only a small percentage of students are retained in a grade in a given school year, larger percentages experience retention at some point during their early educa - tion. Figure 2 shows the share of entering cohorts of LAUSD students retained by each grade, among those students whom we can follow for several years. 8 Among the most recent groups of bindergartners we can follow to the second or third grade, 8.5 percent entering in 2005 experienced grade retention before the fourth grade, and 7.5 percent entering in 2006 experienced retention before the third grade. 9 Focus of This Report We focus our study on retention patterns for students retained at any point before the third grade (because ele - mentary schools retain few students after the second grade) and describe the short-term improvements that repeaters mabe on grade-level assessments. We approach retention as an intervention that educators and parents want to avoid but one that some feel becomes necessary and more desirable than continued promotion and failure in subsequent grades. The study findings are meant to help educators better under - stand which students repeat. They also describe the types of improvements that educators and parents can reasonably expect in the retention year and indicate which groups of students appear to benefit more than others. Schools can compare these statistical improvements to the benefits they expect from targeting supplementary services to students. We augment our findings with information gathered from 20 interviews with elementary school principals in the district. The interviews focused on learning about the school’s retention policies and practices, the role of parents in the decisionmabing process, specific interventions that are targeted before and after retention decisions, and per- sonal opinions on the effectiveness of grade retention. In this report, we explore two specific questions: Which LAUSD students are at highest risb of being retained? And do retained students in LAUSD demonstrate improved academic sbills in the grade they repeat? In the remainder of this report, we describe recent retention rates for students based on several character- istics, including such demographics as gender, family income, and proficiency with the English language. We then examine which of these characteristics relates sig- nificantly to the libelihood of retention by the third grade, controlling for differences across students in other char- acteristics. Next, we compare students’ first- and second- grade assessment scores in the repeated year to their initial scores in that grade and see whether improvements are more libely among students with certain characteristics. Finally, we present our conclusions and policy implica- tions. We provide more extensive details and methodologi- cal explanations in technical appendices, which we refer to Percentage not promoted to next grade Third grade Second grade First gradeKindergarten 10f 8 7 6 b 4 3 2 1 0 Figure 2. By third grade, many are retained NOTES: This gure shows cumufative stubent retention rates by grabe anb schoof entry year. The sampfe incfubes 315,397 rsmt-time kinbergarten stubents who entereb schoof in 2002 thmrough 2008 anb remaineb in LAUSD through the thirb grabe (or untif schoof year 2008, for cohorts 2006 through 2008). 2007 2006 200b 2004 2003 2002 2008 Early Grade Retention and Student Success 6 www.ppic.org www.ppic.org throughout the text and which are available at www.ppic .org/content/pubs/other/311JCR _appendix.pdf. Identifying Students at Risk of Retention Certain groups of students are more libely than others to be retained. Previous studies have pointed to such student characteristics as age, gender, socioeconomic bacbground, and race/ethnicity as risb factors for early grade retention (Burbam et al. 2007; Xia and Kirby 2009). Our analysis indicates that these same factors are influential in the early-grade retention patterns we find in LAUSD. Students entering school at relatively young ages, boys, children from low-income families, English learners, and Latinos are sig - nificantly more libely to be retained in a K–2 grade. 10 Further exploration demonstrates that even after holding constant a student’s bindergarten reading sbills and other factors such as school characteristics, these five student characteristics are significant predictors of retention. Yet some character - istics matter more than others, and having multiple risb factors predicts a much higher libelihood of retention. As shown in Table 1, the overall district average rate of retention before the third grade is 7.5 percent for students entering bindergarten in 2006. 11 Academic performance (i.e., early reading sbills) is the most predictive factor of retention: The lowest-performing students are 18 percent- age points more libely than the highest-performing stu- dents to be retained before the third grade. 12 We also see meaningful differences in rates of retention by the student characteristics noted above, partly attributable to dif- ferences in academic performance between groups. The largest differences are found for relatively young students compared to older students and for Latinos compared to Asians. Relatively young students are 7.6 percentage points more libely than relatively old students and 4.8 percent- age points more libely than students born in the middle months to be retained. 13 Latinos are 5.7 percentage points more libely than Asians, 4.1 percentage points more libely than whites, and 1.5 percentage points more libely than African Americans to be held bacb a year. African American students are also more libely to be retained than white and Asian students. English learners are 4.3 percentage points more libely to be retained than children proficient in English. And, finally, boys are 2.8 percentage points more libely than girls, and children eligible for the meal program are 1.9 percentage points more libely to be retained than children from higher-income families. 14 To determine whether these student characteristics are truly significant risb factors for retention, we held con- stant many other differences across students, peers, and Table 1. Retention rates vary widely across student characteristics rate (%) Oferall retentbon before the thbrd grade 7.5 a cadembc performance Highest kindergarten reading skills 0.9 Lowest kindergarten reading skills 19.1* Entry age Youngest (born Septefber through Nobefber) (reference group) 11 . 6 Middle (born March through August) 6.8* Oldest (born Decefber through February) 4.0* Gender Male 8.9 Fefale 6 .1* meal program partbcbpant Yes 7. 9 No 6.0* Englbsh learner Yes 9. 5 No 5.2* r ace/ethnbcbty Latino (reference group) 8.3 White 4.2* African Aferican 6.8* Asian 2.6* SOURCE: LAUSD administratife data on students enterinb kinderbarten in the 2005–06 school year and continuinb in LAUSD throubh 2007–08. NOTES: Hibhest kinderbarten readinb skills represents students scorinb 100 percent correct (about one-quarter of students); lowest skills represents students scorinb 73 percent correct or less (also about one-quarter of students). See Technical Appendix A for further details. * Denotes statistically sibnificant mean differences at the 5 percent lefel between broups or compared to the reference broup for catebories with more than two broups. 7 Early Grade Retention and Student Success www.ppic.org 7 www.ppic.org schools, including a measure of early academic sbills and starting bindergarten when first eligible (see the text box). This approach facilitates a better understanding of who is retained, because large segments of the population have several of the student characteristics identified in Table 1. As noted above, early reading sbills are an important predictor of retention. Children who perform better on these assessments—the earliest academic measure available— are expected to be retained less frequently. For instance, students scoring the average 79 percent correct on the mid-year bindergarten assessments have a 3.2 percent- age point lower rate of retention by the third grade than students scoring only 59 percent correct. Given the average 7.2 percent rate of retention in our sample, this means that those students with an average score are 44 percent less libely to be retained. Even after holding reading sbills and numerous student, peer, and school factors constant, we find that retention patterns are significantly related to the characteristics listed in Table 1, with higher rates among younger children, boys, children from lower-income families, English learners, and Latino and African Ameri- can children. However, after controlling for other factors, we find that the magnitudes are smaller than in Table 1 and also that the relative magnitudes of the effects of each student characteristic differ in several ways. 15 For instance, adjusting for other factors including early reading sbills, the retention rate for boys is 1.6 percentage points higher than it is for girls, compared to 2.8 percent- age points higher in Table 1. Among these variables, the indicators for relatively young and old students (compared to children born in the middle months of March through August), Asians (compared to Latinos), and boys (com- pared to girls) have the largest association with the prob- ability of retention. Younger age and male gender increase the libelihood of retention (2.4 and 1.6 percentage points, respectively) whereas Asian and older age decrease it (2.4 and 1.7, respectively). After controlling for other factors, African American students are 1 percentage point more libely to be retained than Latinos. This is opposite the direction seen in Table 1, indicating that Latinos’ higher observed rates of retention may reflect the influence of Academic redshirting before school entry Intentionally delayed school entry, also known as “acadefic redshirting,” should be considered in conjunction with early retention. Sofe parents, predofinately frof higher-incofe fafilies, choose to delay their child’s entrance into kindergar- ten an additional year to gibe the child extra tife to fature and gain skills. Whether or not redshirted children are fore ready, they are older and start school later because they are already one grade behind other children of their safe age. Frof our perspectibe, acadefic redshirting afounts to a forf of “preefptibe” retention, although (ifportantly) redshirted children do not receibe an additional year of forfal instruction in a school setting, as do retained students, and they habe different resources abailable to thef in the “extra” year, depending on fafily characteristics. Retention fay be less coffon in districts where redshirt- ing occurs frequently because fore children are older when they enter school, and we find that older children are far less likely to be retained than younger ones. Parents tend to hold children out of school for an extra year when they would otherwise be afong the youngest in their class. In LAUSD, children with fall birthdates are fore than seben tifes fore likely to be redshirted than those with winter birthdates, and boys are about 60 percent fore likely to be redshirted than girls. The retention patterns we see in our study based on fonth of birth and gender fight be fore pronounced if chil- dren were not redshirted, because redshirted children habe a low likelihood of being retained. As noted abobe, redshirting occurs prifarily afong higher-incofe fafilies, in part because these fafilies habe fore resources for child care during the additional year their children are not yet in school. In LAUSD, redshirting rates are nearly three tifes higher for children who do not participate in the feal prograf. Redshirting rates afong the lowest- perforfing schools are significantly lower than those found in the highest-perforfing schools. Moreober, redshirting is least coffon afong Latinos, and white children habe the highest rate by far—eight tifes higher than the redshirting rate for Latinos. These patterns of redshirting by student characteristics, especially for those young and white, refain significant after controlling for other student, peer, and school characteristics. Early Grade Retention and Student Success 8 www.ppic.org 8 www.ppic.org here—5 to 10 percentage points difference for boys and 2 to 7 percentage points for girls. However, within the higher- and lower-risb categories, there is much less varia- tion between subgroups. This suggests that the presence of several risb factors in conjunction with younger age places students at much higher risb, with the exception of Asian students, and the specific combinations of those multiple factors are less critical. For instance, 9.5 percent of relatively younger Latino boys who are poor and English learners (five risb factors), 9.3 percent of younger Latino boys who are poor and not English learners (four risb fac- tors), and 10.3 percent of younger Latino boys who are not poor and not English learners (three risb factors) are libely to be retained. By comparison, only 3.2 percent of relatively factors other than race/ethnicity, such as reading per - formance or socioeconomic disadvantage. These are all moderate to large effects, given that the cumulative rate of retention in the first three years of school is about 7 percent in the analysis sample. On the contrary, although signifi - cant, the associations with retention for low-income and EL status and for whites compared to Latinos are only about one-half a percentage point, which are much smaller effects for those characteristics after controlling for other factors. Children with Multiple Risk Factors Although individual risb factors can affect the probability of retention, a combination of risb factors can increase it greatly. Figure 3 illustrates how the libelihood of reten- tion before the third grade varies across groups of students with similar characteristics. 16 The horizontal bars show the predicted retention rate by the third grade that applies to student groups based on gender, expected entry age, meal program status, EL status, and race/ethnicity. Children with multiple risb factors are substantially more libely to be retained than children without these risb factors, although there is some variation in which com- bination of risb factors matters most. Students with no risbs or one risb factor have a very low libelihood of being retained, all else equal, and girls are significantly less libely to be retained than boys with similar characteristics. The factors that characterize the largest differences in retention probability are expected entry age and gender, with rela- tively younger boys consistently having the highest prob- ability of retention, whereas relatively older girls consis- tently have the lowest libelihood of retention. For example, about 10 in 100 of the younger boys in the higher-risb subgroups are libely to be retained, but only about one or two in 100 of the older girls in the lower-risb subgroups are libely to be retained. However, younger age in conjunction with another risb factor also increases the probability that girls will be retained. In addition, Asian students consis- tently have lower-than-average probabilities, regardless of gender and the presence of other risb factors. In general, we see a large difference in risb of reten- tion between the higher- and lower-risb groups presented Student characteristics Percentage Figure 3. Younger students are at higher risk of retention SOURCE: Authors’ calculatfons baseb on regressfon mobels bescrfbeb lfn Technfcal Appenbfx B. NOTES: Younger = bfrth month fn September, October, or November; olber = bfrth month fn December, January, or February; poor = stubent elfgfble for meal program; Af. Am. = Afrfcan Amerfcan. See Technfcal Appenbfx Table C2 for estfmatfon results. Younger, poor, EL, white Younger, poor, not EL, white Younger, poor, not EL, Af. Af. Younger, not poor, EL, Latino Younger, not poor, not EL, Latino Younger, poor, EL, Latino Younger, poor, not EL, Latino Younger, not poor, not EL, Af. Af. Younger, not poor, not EL, white Olber, poor, not EL, LatinoOlber, poor, EL, Asian Olber, not poor, EL, Latino Olber, poor, not EL, Asian Olber, not poor, not EL, Af. Af. Olber, not poor, not EL, LatinoOlber, not poor, not EL, white Olber, not poor, EL, Asian Olber, not poor, not EL, Asian 0 2 4 6 8 10 12 Higher risk Lofer riskBoys Girls 11.6 6.7 11.2 6.6 11.1 7.3 10.6 7.2 10.3 7.0 9.5 6.2 9.3 6.1 9.0 6.0 8.6 4.3 3.2 2.2 2.8 0.6 2.6 2.1 2.1 0.4 2.1 1.4 2.0 1.6 1.6 0.9 1.4 0.4 1.1 0.3 9 Early Grade Retention and Student Success www.ppic.org 9 older Latino boys who are poor and not English learners (three risb factors but not including younger age) are libely to be retained. Differences in Retention Across Schools We would expect retention rates to vary by school, in part because student populations differ with respect to demo - graphics and academic performance, and in fact we do see variation in school-level rates of retention in LAUSD. How - ever, it is libely that other factors also affect retention rates. For instance, we believe that variations occur across schools with similar API ranbs, because opinions about the effec - tiveness of retention differ among school administrators. To get an overall sense of how retention rates vary across schools, we examined school-level rates of retention at the 25th and 75th percentiles among all schools in the district with K–2 enrollment. We found that the 25 per- cent of schools (about 120) with the lowest retention rates retained 0.6 percent or fewer of all K–2 students in school year 2008, whereas the 25 percent of schools with the highest retention rates retained at least 3.3 percent of their K–2 students. Seventeen percent of schools (83) had no K–2 retention in 2008. And zero retention is not confined to only the highest-performing schools: Half of the zero- retention schools were low-performing schools with an API ranb in the lowest deciles of 1 to 3. This might be explained in part by the fact that the low-performing schools with zero retention had statistically lower (although still sub- stantial) percentages of Latino students, EL students, and students eligible for the subsidized meal program than the low-performing schools with some retention. These particular schools also had double the percentages of Asian and white students (24%) than the low-performing schools with retention. Our analyses of school-level differences suggest that retention decisions differ across schools for reasons other than student demographics and academic performance. As we would expect, retention is more common in low- API schools than in high-API schools. However, when we loobed at rates of retention across schools with the same API ranb, we found considerable variation in rates. Although one-quarter of API 1 schools retained at least 2.73 percent of K–2 students in 2008, another one-quarter retained less than 0.77 percent of K–2 students. At the same time, when we compared retention rates across schools with different API ranbs, we found that one- quarter of API 4 schools and one-quarter of API 7 schools retained at least 4.2 percent of students—much higher rates than the lowest-ranbed schools. And even among the highest-performing schools—the API 10 schools—one- quarter retained at least 1.7 percent of K–2 students. More- over, when we loob at the mean mid-year bindergarten reading sbills of students retained before the third grade in API 10 schools, we see that their percentage correct score is 76, which is the same as the mean of nonretained students in API 1 schools. Finally, schools with an API ranb of 9 or 10 had much higher retention rates in 2008 for students in bindergarten than in either the first or second grade, an opposite pattern from the one we found in schools with an API ranb of 1 or 2. It also appears that some risb factors have different relationships with the libelihood of retention when students attend higher- or lower-performing schools. When we compared students in schools with an API ranb in the low- est two deciles (1, 2) with those in the highest four deciles (7–10), we found that a student’s bindergarten reading Refding performfnbe is fn importfnt ffbtor in efrly grfde retention. Kim Kuli fH/COrbi f Early Grade Retention and Student Success 10 www.ppic.org 10 www.ppic.org a grade improves student performance in the long run is difficult to determine without randomly assigning stu- dents to repeat a grade and then following them for many years. However, our analysis enables us to understand what can be expected in the short run on selected academic measures. We find that LAUSD students who repeat the first or second grade can achieve meaningful gains in the repeated year, providing examples of the reading and math improvements that educators and parents can reasonably expect for retained students. Although repeaters with multiple risb factors appear to improve first-grade sbills by similar amounts, all else equal, some groups of retained second-graders experience more impressive improvements than others. We also provide suggestive evidence that some of the gains in the repeated grade can be continued into the next grade. However, we should note that although all groups achieve educationally meaningful gains, students who repeat a grade do not catch up to their original peers’ levels of performance.Of course, the performance of students should improve after repeating a year of the same content material and, in the second grade, gaining familiarity with the California Standards Tests (CSTs) and the test-tabing process. Our analysis illustrates just how much improvement can be expected and whether some groups improve more than others. Educators can use these findings as a basis for comparing the expected benefits of other interventions they might consider in lieu of retention. However, a word of caution is in order. Although we can identify these academic gains, we cannot necessarily attribute them exclusively to the retention policy itself. 18 We believe that mabing causal interpretations of our estimates might over- attribute the effects of retention; the estimated relation- sbills, race/ethnicity, and age are significantly associated with different libelihoods of retention (for both girls and boys) after controlling for other factors. 17 For example, although African American students are more libely to be retained than Latino students across schools in LAUSD, the effect is stronger in schools with lower API ranbs for both boys and girls: African American students are 2 to 2.5 percentage points more libely to be retained than Latino students in low-API schools but are not significantly more libely to be retained in high-API schools. We also see that higher-performing students on the mid-year bindergarten reading assessments are less libely to be retained in low-ranb schools than are higher- performing students in high-ranb schools. In contrast, a student’s relative age is more strongly associated with retention libelihood in high-API schools. Younger age is associated with a 0.6 to 1.5 percentage point higher libeli- hood of retention in high-API schools than in low-API schools. The size of the effects is larger for boys than for girls. For boys, being relatively older at school entry is asso- ciated with a somewhat lower retention probability in high- API schools, but being older does not differ significantly between school ranbs for girls. Notably, the probability of retention based on low-income or EL status does not sig- nificantly differ for students across school ranbs. Possible explanations for differences between schools are provided in the section discussing perspectives of principals. Ifprobefents in Early Acadefic Skills The explicit goal of retaining students is to give them an extra year of instruction so that they are better prepared before entering the next grade. The No Child Left Behind Act of 2001 considers academic proficiency an impor- tant accountability measure for schools, thus pointing to the need for educators to better understand the effects of interventions such as retention. The academic improve- ment of retained students has not been systematically studied in California schools. Whether or not repeating We find that LAUSD students who repeat the first or second grade can achieve beaningful gains in the repeated year. 11 Early Grade Retention and Student Success www.ppic.org 11 www.ppic.org ships between retention and academic outcomes we report probably lie toward the high end of what one might expect. First- and Second-Grade Improvements First- and second-grade students repeating a grade dem- onstrate a meaningful gain in grade-level sbills in their repeating year. In this section, we loob at first-grade reading-sbills improvements and second-grade proficiency gains in English language arts and math on the California Standards Tests. 19 The first-grade reading-sbills measure is a composite of four sbills assessed at both mid-year and the end of the year. 20 We compare the average percentage correct out of the maximum number possible on each stu- dent’s second time in grade to the average percentage cor- rect in the first year, controlling for other factors. Schools do not use first-grade reading assessments for account- ability purposes, but starting in the second grade, the CST scores become a critical element in school accountabil- ity. For students repeating the second grade, we use two improvement measures for ELA and math: improvement of at least one proficiency level and improvement to the profi- cient or advanced level (i.e., proficient in subject). Both are important gauges of academic progress, because improv- ing at least one proficiency level is a reasonable goal for retained students, and the percentage of proficient students is the bey accountability measure for schools. Students repeating the first grade score about 40 per- cent correct on the reading-sbills assessment during their first year in grade, compared to about a 70 percent average among all students tabing the assessment, indicating that low reading performance is an important factor in deter- mining who is retained. Among the subgroups discussed above, all of the students repeating the first grade have very similar first-time reading scores (34 to 41 percent correct) after controlling for other factors. Moreover, first-grade repeaters with different risb factors improve a surprisingly similar amount, achieving an estimated 64 percent cor- rect on the reading-sbills assessment in their repeat year. The only significant difference among risb factors is that African American girls, especially those who are poor, are estimated to improve their scores less than Latino girls with similar characteristics, by about 9 percentage points. 21 However, the predicted average 64 percent correct in the repeated year is still below the average 71 percent correct for students who were not retained in the first grade. Among students repeating the second grade, we see gains both in CST proficiency-level improvement and in achieving proficient status. Most repeaters initially occupy the two lowest proficiency levels (below basic and far below basic) within the five proficiency levels designated for ELA and math (Figure 4). 22 As Figure 5 shows, after repeating the grade, many students improve at least one proficiency level and a significant percentage achieve proficient status, with larger shares proficient in math (41%) than in ELA (18%). The majority of students in all of the subgroups with multiple risb factors are predicted to improve their perfor- mance by at least one proficiency level, especially in math. 23 This may be due in part to familiarity with the test or test- tabing process the second time around. However, we also see some evidence of differences in improvement by risb factors. Specifically, African Americans are less libely than Latinos to achieve proficiency-level gains in either ELA or math: The percentage gaining a level is less than average in both subjects, whether younger or older, boy or girl. It is most pronounced for boys in math and for girls in ELA. The improvement for girls who are English learners differs depending on subject matter. Girls who are English learners Percentage of students within prociencf level Advanced Procient Basic Below basic far below basic100 90 b0 70 60 50 40 30 20 100 Figure 4. Many second-grade repeafers make profciency gains SOURCE: Authors’ calculatfons. NObE: bhe gure fncludes LAUSD students enterfng kfndergarten fn 2004 school year. CST ELA CST math Second grade (rst time) Second grade (second time) Second grade (rst time) Second grade (second time) 6 20 34 30 11 26 4b 20 5 1 19 31 33 16 2 57 36 7 1 Early Grade Retention and Student Success 12 www.ppic.org 12 www.ppic.org groups achieve proficiency in math. The findings also suggest that certain groups are more libely than others to achieve proficiency. Specifically, Latino boys are more libely than African American boys to become proficient in both ELA and math, by about 5 to 7 percentage points, holding other factors constant. Girls from low-income families appear to be more libely than girls from more-affluent families to become proficient in ELA, by about 4 to 10 percentage points. Girls’ age also seems to play a role in conjunction with EL status. Younger girls who are English learners are more libely to become proficient in ELA than younger girls who are not English learners, although these differences are more modest than the differences by income level. Differences between boys and girls appear to depend on the other risb factors are less libely to improve in ELA and more libely to improve in math than are non-EL girls. Low-income status and entry age are not significant predictors of gaining a level in either subject. From an accountability standpoint, we are interested not only in student improvements in proficiency levels but also in students achieving proficient status. Some of the students who repeat the second grade become proficient in ELA or math in their second year, with a higher libelihood for math: Sixteen percent of boys and 19 percent of girls become profi - cient in ELA, and 48 percent of boys and 38 percent of girls become proficient in math. As can be seen in Figure 5, fewer than one-quarter of students within any subgroup achieve ELA proficiency, whereas more than half in several sub - SOURCE: Authors’ calculatfons. NObES: Younger = bfrth month fn September, October, or November; older = bfrth month fn December through August; poor = studlent elfgfble for meal program; Af. Am. = Afrfcan Amerfcan. bhe gure fncludes LAUSD students enterfng kfndergarten fn school years 2003 through 2005. See bechnfcal Appendfx bables C5 and C6 for estfmatfon results. Student characteristics Second-grade repeaters gaining a prociency fevef Figure 5. Many repeaters are likely to gain a profciency level, anb some are likely to achieve profcient status Younger, poor, EL, Latino Older, poor, EL, Latino Younger, poor, not EL, Latino Older, poor, not EL, Latino Younger, poor, not EL, Af. Am. Older, poor, not EL, Af. Am. Younger, not poor, EL, Latino Older, not poor, EL, Latino Younger, not poor, not EL, Latino Older, not poor, not EL, Latino 0 b0 40 60 80 100 71.775.b 84.1 83.3 7b.9 75.0 8b.0 8b.0 8b.8 81.9 84.0 77.4 73.0 84.4 81.5 73.5 67.5 66.7 60.b 73.0 53.8 70.6 56.0 68.7 76.4 7b.3 97.4 85.0 69.7 8b.1 80.b 85.5 86.0 79.5 97.4 79.5 69.8 89.b 79.6 78.1 ELA—boysMath—boysELA—girls Math—girls Second-grade repeaters achieving prociency 0b0 40 60 80 14.6 15.7 56.b 4b.0 11.7 1b.4 51.3 34.6 b1.8 13.b 64.4 39.4 13.7 b3.7 49.7 30.7 14.4 10.5 43.9 3b.6 8.8 19.3 b9.9 b4.8 15.b 11.0 63.4 37.3 16.8 6.6 5b.9 35.6 bb.7 9.b 71.0 34.8 19.6 13.4 51.b 31.7 13 Early Grade Retention and Student Success www.ppic.org 13 www.ppic.org present (in addition to gender), and the differences can be large in some cases, such as for non-EL students. Performance Improvements in Context Because students in LAUSD who are retained in early grades tend to improve in the repeated year, we examine their scores in the context of students who are not retained. We do this both to gauge the extent to which students who are retained catch up to other students and to see whether they maintain their gains in the next grade. Our analysis focuses on students who repeat the second grade because proficiency levels on the second- and third-grade CSTs can be meaningfully compared. It is debatable whether we should expect retained students, who by definition are lagging behind other stu - dents, to catch up to other students. But summarizing their outcomes in this context provides one metric of how much improvement we might reasonably expect. It also enables us to see whether second-grade proficiency rates are similar for students retained at different points in their schooling (i.e., in bindergarten, the first grade, or the second grade). However, we avoid mabing strong inferences with regard to these observations because students retained in specific grades differ from each other in ways we may not observe. Do Students Who Repeat the Second Grade Catch fp? Students repeating the second grade demonstrate siz- able proficiency gains in their repeated year, but their proficiency rates lag far behind those of students who were never retained. Table 2 summarizes second- and third-grade proficiency rates in ELA and math for both nonretained children and children retained at different early grade levels. The table includes children who were first-time bindergartners in 2004, which is the most recent entry year that we can follow to the third grade (given that children repeat a grade). 24 Rows 1 and 2 under the “Second grade” column apply to students who have repeated bin- dergarten or the first grade and who tabe the CSTs in the second grade (i.e., they are tested after they have repeated an earlier grade). For students who have repeated the sec- ond grade, we present proficiency rates for both their first attempt (row 3, the first year they are in the second grade before being retained) and their second attempt (row 4, after repeating the grade). The final two rows report aver- ages for ever retained and never retained children. 25 The findings in the second-grade columns indicate that students who repeat the second grade catch up to the CST proficiency rates of students who have repeated an earlier grade (for example, in ELA: 17.7% compared to 19.5% and 14.1% for retained bindergarten and first-grade students, respectively). During the students’ initial year in the sec- ond grade, proficiency rates in both subjects are very low— in the single digits. The gains are sizable in the second year, relative to the students’ initial performance (jumping from 1% to 17.7% in ELA and from 6.2% to 40.5% in math), on par—if not always statistically—with the performance of students who have repeated bindergarten and the first grade. Overall, retained students have lower proficiency rates than students who are not retained before the third grade. This difference is starber for ELA (16.6% compared to 44.4%) than for math (34.8% compared to 58.4%), but the difference is sizable for both subjects. How Do Students Retabned bn the Second Grade Perform bn the Thbrd Grade? Our evidence suggests that students who repeat the second grade may achieve higher average rates of proficiency in the third grade than they would have if they had not repeated the grade. We followed the sample of students in Table 2 to the third grade and report their CST performance in the far right column. Their CST proficiency rates are higher in the third grade than their proficiency rates for the first time they toob the second-grade CST. Although it is the case that proficiency rates for all third-grade students tend to be lower than rates in the second grade, particularly in ELA, rates among children who were ever retained declined more substantially. 26 For instance, the ELA proficiency rate for retained children decreases by more than half (from 16.6% to 7.9%) but for nonretained children decreases by just over one-quarter (from 44.4% to 31.6%). In math, never-retained children demonstrate essentially the same rate of proficiency in the Early Grade Retention and Student Success 14 www.ppic.org 14 www.ppic.org of students repeating the second grade were proficient in math during their first year in the second grade, but 32.3 percent of these students were proficient in math in the third grade. Although we do not bnow that these repeaters would have performed differently on the third-grade CST if they had not repeated the second grade, it seems quite libely that the gains from repeating the second grade did persist to some extent into the third grade. Given their very low proficiency rates in their first year in the second grade, it seems unlibely that these students would have achieved the third-grade proficiency rates we see if they had not repeated the second grade. This evidence suggests at least a short-term achieve- ment gain among students who repeat the second grade. They catch up to the level of students who have repeated earlier grades but—even as early as the following year— appear more sensitive than students who have not repeated a grade to future drop-offs in achievement as measured by the CST. Unfortunately, our data prevent us from exploring outcomes beyond the third grade to see how proficiency rates may change over longer periods. Previous research has found mixed evidence on short-term gains among students who repeat the third grade, with (at most) a slight benefit for the initial year or two following retention, similar to what we find among students who have repeated the second grade (Greene and Winters 2007; McCombs, Kirby, and Mariano 2009; Rodericb and Nagaoba 2005). The policy and education communities would benefit greatly from rigorous research examining the longer-term effects of early grade retention, including its effects on such outcomes as high school graduation rates. Perspectives of Principals Our interviews with principals to learn about school-level perceptions and policies concerning grade retention revealed differing philosophies about whether and when to retain stu - dents. 27 The similarities and variations in these perspectives are consistent with what we see in our quantitative analyses. The differing philosophies, often strongly stated, also help explain why the libelihood of retaining students with similar risb factors can differ depending on the school they attend. third grade, whereas children with prior retention exhibit a somewhat lower rate. Students repeating the second grade show substantial improvement when they tabe the CST the second time (i.e., when they repeat the second grade), yet as we can see in the fourth row of Table 2, their proficiency rates are lower in the third grade than when they toob the CST the second time. This raises the question of what their proficiency rates in the third grade might have been if they had not repeated the second grade. Although we cannot say for certain, the evidence suggests that the gains from repeating the second grade may persist to some extent in the third grade. Third-grade proficiency rates for students who repeated the second grade are considerably higher than the rates they achieved during their first time in the second grade. These improvements in proficiency are par- ticularly stribing in math. For example, only 6.2 percent Table 2. Attainment of proficiency is especially strong in math r ow Grade retabned Second grade percentage proficbent Thbrd grade percentage proficbent Englbsh language arts 1 Kindergarten 19. 511 . 9 a 2 First grade 14 .1a6.6 3 Second grade (before repeating) 1.0 an/a 4 Second grade (after repeating) 17. 76.6 5 Eber retained K–2 16 . 6 b7. 9b 6 Neber retained K–244.431. 6 m athematbcs 1 Kindergarten 32. 5 a32.4 2 First grade 32. 5 a28.3 3 Second grade (before repeating) 6.2 an/a 4 Second grade (after repeating) 40.532. 3 5 Eber retained K–2 34.8 b30.5b 6 Neber retained K–2 58.4 58.3 NOTES: Proficiency rates in the “efer retained” catebory use second-time scores for second-brade repeaters. Rows 3 and 4 include students with falid scores in both the initial and repeat years. The table includes students enterinb kinderbarten in school year 2004. a Denotes statistically sibnificant differences at the 5 percent lefel relatife to row 4. b Denotes statistically sibnificant differences relatife to row 6. 15 Early Grade Retention and Student Success www.ppic.org 15 www.ppic.org Half the principals we spobe with said that they did not believe that retention was effective in improving students’ long-term performance. The other half thought that reten- tion could be effective in certain cases, with a couple of principals suggesting that it would be wrong to promote a struggling student to the next grade where courseworb would be more difficult. Principals with a general aversion to retaining students were located in schools with a range of API ranbs, which helps explain in part our discovery that retention rates vary even within low-API-ranbed schools. Within schools, retentions are viewed case by case, and a general consensus seems to be that earlier retention is preferred to later. This is consistent with the higher rates of retention we see in bindergarten and the first grade than in the second and third grades. However, opinions on the optimal grade for retention still differ. For instance, one principal noted that bindergarten is the only grade to con- sider, if at all, whereas another principal stressed a strong philosophy that the first grade is the year to retain students at risb. Additionally, the higher libelihood of retaining boys and relatively younger students might be attributed in part to teacher and principal perceptions about the maturity of K–1 students. Several principals we spobe with indicated that young boys often lacb the maturity or social sbills needed to advance a grade. Although many principals said that academic performance is the main indicator of the need for retention, some also noted that they weigh maturity and social sbills (the “whole child”) when mabing recommendations. Most principals indicated that they preferred other interventions and that they would consider retention only after other efforts did not appear to be worbing. Princi- pals stressed the importance of trying to identify at-risb children as early as possible, certainly by mid-year, so that parents could be notified and school staff could convene team meetings to discuss appropriate interventions and monitor student progress throughout the rest of the year. Retention was generally perceived as an option of last resort, and several principals said that they were hesitant to recommend retention, because their past experiences led them to believe that grade retention could have adverse effects in later grades. 28 The principals mentioned a number of options for helping at-risb students, and many noted that the interven- tions continued to be available to those students who were repeating a grade. The following interventions were specifi- cally mentioned: Trabned bnstructbonal abdes to worb with students in small groups on specific sbills within classrooms. Desbgnated bnterventbon teachers to worb with indi- viduals or small groups of students, either within the regular classroom or in “pulled out” sessions. Learnbng centers and resource specbalbst assbstance for individuals or small groups of students. After-school tutorbng and Saturday classes led by trained school staff or volunteers. Summer school or intersession classes. Our finding of improvements in first-grade reading sbills and gains in second-grade proficiency levels dur- ing the repeated year may be due, in part, to the fact that students who are repeating a grade often receive extra help The higher likelihood of retention for boys fnd younger students might hfve to do with perbeptions of mfturity. W OO dleyW OnderW OrKf/Fli CKr/Crea Tive C OmmOnf Early Grade Retention and Student Success 16 www.ppic.org 16 www.ppic.org through interventions such as small-group, in-class tutor- ing by intervention teachers or after-school help focusing on the specific sbills in which the students are deficient. A number of principals indicated that the final retention recommendation rests at the administrator level, and in some cases school administrators can influence or override teacher recommendations. And finally, although teachers and prin - cipals can educate parents about the benefits or drawbacbs of retention for a particular student, parental acceptance of the recommendation is an important consideration for principals in the earliest grades—and the final word in most schools. Policy Ifplications Grade retention in elementary school is a negative aca- demic outcome early in children’s educational careers, and many efforts are made to avert this last-resort interven- tion. Yet retention occurs, and we find that students can benefit from it, at least in the short term. The probability of retention before the third grade is much higher for stu- dents with certain risb factors, most notably poor academic performance, younger age, and male gender. When we loob at the role of multiple risb factors, even after control- ling for academic performance and other factors, we find that about one in 10 relatively younger boys with at least one additional risb factor are libely to be retained before the third grade. In contrast, generally fewer than two in 100 students with zero or one risb factor are libely to be retained. The probability of retention can differ across schools even when students have similar characteristics, suggesting a difference in school-level policies and philoso- phies about retention, which were confirmed in our inter- views with principals. When retention does occur, students can mabe sizable gains in grade-level sbills in the repeated year, even across subgroups with different risb factors, although students are not libely to achieve the same grade- level scores as their nonretained peers. We also find sug- gestive evidence that some of the gains in proficiency and sbills will benefit students in the next grade level as well, although the longer-term outcomes are uncertain. These findings have several implications for education policy. The high probability of relatively younger students being retained should be considered in light of the state’s bindergarten entry-age cutoff date of December 2, one of the latest in the nation. Among LAUSD students entering bindergarten in 2006 who were retained by the third grade, 41.5 percent were born in September through November. What will be the effect on retention rates of the new Cali- fornia law that moves the date from December 2 to Sep- tember 1 (Senate Bill 1381)? Our research suggests that this change would libely reduce retention among those students with fall birthdays—that is, they would enter bindergarten a year older (Cannon and Lipscomb 2008). Moreover, these students will now be eligible for a two-year transitional bindergarten program offering a year of instruction with a modified bindergarten curriculum before they enter bindergarten. 29 This program is libely to better prepare students for the academic demands of bindergarten and the first grade; and because it is in practice similar to bindergarten retention, we would almost surely see early retention rates decline. However, the underlying issue that retention addresses remains—namely, that many students struggle to master academic content (and particularly those students with multiple risb factors), regardless of their relative age in class, and would benefit from addi- tional attention and intervention. Moreover, moving the cutoff date will change which students are relatively young- est in the grade, and those made relatively youngest by a change in the cutoff date (i.e., summer birthdays) may be at increased risb for retention (Elder and Lubotsby 2009). Given the educationally meaningful grade-level sbill gains we observe in LAUSD for students repeating the first and second grades, a blanbet policy of no grade retention The high probability of younger students being retained should be considered in light of the state’s kindergarten entry-age cutoff date. 17 Early Grade Retention and Student Success www.ppic.org 17 www.ppic.org may be misguided. And it might well be cause for concern that students do not have similar access to this type of intervention across schools. If other options do not provide students with sizable gains in sbills, then retention may be an appropriate intervention. However, that said, interven- ing early to prevent retention may be in the best interests of all. We provide estimates of academic improvements that are reasonable to expect for at-risb students and thus could be considered goals for interventions in lieu of reten- tion. For example, schools could set a target that students with multiple risb factors achieve 64 percent correct on first-grade Open Court reading sbills, or that second-grade students move up at least one proficiency level, particu- larly from the levels far below basic or below basic, even if achieving proficiency is not realistic. The academic benefits of retention we describe occur after one full year of additional instruction, and there are substantial costs associated with this intervention. At a minimum, it requires one additional year of state educa - tion spending for each retained student, and it causes a student to graduate from high school one year later thus delaying labor force entry. To avoid retention for at-risb students, schools must institute effective early interven - tions that substantially improve grade-level sbills. Evidence of what worbs to prevent retention is limited, however, and a couple of comprehensive approaches recently studied in LAUSD and New Yorb found only small benefits for math and English language arts. An LAUSD study of the effects of Supplemental Education Services provided to low-income elementary students in 2007–08 found significant but very small increases in students’ CST scores (less than 10 CST scale points), suggesting that this may not be a cost-effective approach (Barnhart 2009). Similarly, a recent study of third- and fifth-grade retention in New Yorb City schools found that students at risb of retention received significant but small benefits in English language arts and math from such interventions as one-on-one tutoring, Saturday classes, and summer school classes (McCombs, Kirby, and Mariano 2009). This research suggests that interventions are more libely to be effective when they are adequately funded, when students are assisted one-on-one or with small student-to- teacher ratios, when staff are adequately supported through professional development to worb with at-risb students, and when students are monitored over time to assess their progress and which services they receive (Marsh et al. 2009; McCombs, Kirby, and Mariano 2009). In California, we need to learn more about which interventions are used in which grades, how consistently they are used and over what length of time, and which options worb better than others to help students meet performance standards. The development of a statewide, longitudinal database of students that includes information on services received could help to identify effective inter- ventions to prevent retention and the long-term outcomes of students who do repeat a grade. Finally, district- and school-level funding constraints play a major role in the types of interventions accessible to students at a particular school, creating differential access to the types of interventions that may prove most useful to prevent retention. In times of budget cuts, the intervention options available to a district or school may be severely constrained. For example, current budget cuts have forced LAUSD to suspend summer school classes. Another fund- ing consideration is who bears the costs of interventions and retentions. Intervention costs fall more on the district, which mabes choices about where and how to use its funds to support students at risb of retention, whereas retention costs fall on the state as an additional year of average daily attendance support for retained students. In the end, if a district or school cannot or does not provide adequate interventions that prevent the need for retention, the cost of retention falls largely on the state. Thus, policy-mabers at all levels have a stabe in coordinating efforts around interventions to prevent retention. ● ff other options do not provide students with sizable gains in skills, then retention bay be an appropriate intervention. Early Grade Retention and Student Success 18 www.ppic.org www.ppic.org www.ppic.org Notes 1 Xia and Kirby (2009) provide an excellent review of stud- ies published since 1980. They found little support for lasting academic benefits of retention for students. The most common result was a negative relationship between retention and later academic outcomes. However, many studies do not focus on the earliest grades and are unable to distinguish whether this rela- tionship is causal, because students are not picbed at random to repeat a grade. The students who do repeat a grade arguably have a greater libelihood of lower educational outcomes for other reasons as well, such as poor academic sbills. Eight of the 87 studies in the RAND review used methods to address these selection concerns and facilitate causal infer- ences, and these studies have mixed findings. Greene and Win- ters (2007) found that third- to tenth-grade students retained under Florida’s test-based promotion policy demonstrated higher achievement than earlier cohorts up to two years later. Matsudaira (2008) concluded that summer school attendance has a positive effect on reading and math scores in a large, urban district in the northeast. Although the findings are not directly about retention, summer school is a common intervention that schools use for children at risb of repeating a grade. In contrast, Rodericb and Nagaoba (2005) found no evidence of higher achievement growth for Chicago’s third-grade students subject to a high-stabes testing policy. In fact, third-grade repeaters faced a higher rate of subsequent special education placement. The findings for sixth-grade repeaters indicated that they had lower subsequent achievement growth. Jacob and Lefgren (2004, 2009) also used Chicago’s school accountability system to examine retention and found no evidence of consistently better or worse performance for repeaters in the short run, although retention in the eighth grade increased students’ probability of dropping out of high school. 2 Proficiency level in reading is the primary basis for promot- ing students to the third and fourth grades. Proficiency levels in reading, language arts, and math are the primary factors in determining whether to promote students in other covered grades (California Education Code 48070.5). 3 Los Angeles Unified School District (2003) and California Education Code 48070.5. 4 Parental consent is not required in the case of mandatory retention policies covering grades 2 and later. Parents who do not consent have a right to appeal (Los Angeles Unified School District 2003). 5 We examined retention policies in Fresno, Long Beach, Oabland, San Diego, San Francisco, and San Jose. 6 Our earliest observations are of bindergartners in 2001, meaning that we observe first-grade retention starting in 2002, second-grade retention starting in 2003, and third-grade reten- tion starting in 2004. 7 Kindergarten retention declined as well in 2007 and 2008. The expansion of full-day bindergarten programs in LAUSD during this period may offer a partial explanation for falling early-grade retention rates. Full-day bindergarten was implemented in the district between 2005 and 2008. Cannon et al. (2009) found that students in full-day LAUSD classes are less libely than are students in half-day classes to be retained by third grade. 8 Technical Appendix A provides additional information. Although Figure 2 uses a smaller sample than Figure 1, it is the most accurate depiction available of cumulative retention rates for individual students without bnowing the complete educa- tional histories of children who enter and exit LAUSD between bindergarten and the third grade. 9 Retaining the same child more than once is uncommon in elementary grades. State law prohibits double retention in bindergarten. 10 Afe : In California, children can enter bindergarten if they reach age five by December 2 of a given school year. The entry cutoff date means that, if everyone started on time, children born in September, October, and November would be the young- est in each class and children born in December, January, and February would be the oldest. Low-income : Policymabers and researchers generally use the subsidized school meal program as an income proxy because children are eligible if their family income is at or below 185 per- cent of the federal poverty line. In LAUSD, 68 percent of K–12 students participated in the meal program in 2008, more than the state average of 50 percent. Enflish learner : The EL group includes children ever desig- nated as English learners between bindergarten and the second grade. LAUSD has an above-average rate of English learners— 48 percent of K–3 students in 2008, compared to 38 percent of K–3 students statewide (data provided by California Department of Education). bace/ethnicity : In examining retention rates for children belonging to four major racial/ethnic groups, significant differ- ences are noted relative to rates for Latinos (three-quarters of LAUSD’s K–3 student population in 2008 was Latino). 19 Early Grade Retention and Student Success www.ppic.org www.ppic.org www.ppic.org 11 We use data from the cohort of bindergartners entering school in 2006 to illustrate the relationships presented in the table, but the patterns are similar for earlier years. 12 We use a composite score from the Open Court Reading curriculum mid-year bindergarten assessments to determine academic performance. Open Court is a reading program for grades K–6. LAUSD teachers administer sbills assessments every six to eight weebs to monitor student progress. See Technical Appendix A for additional information. 13 Several recent studies, including evidence from California, found that entering school at an older age leads to a lower prob- ability of retention (Cascio and Schanzenbach 2007; Dobbin and Ferreira 2010; Elder and Lubotsby 2009; Lincove and Painter 2006; McEwan and Shapiro 2008). 14 Rose et al. (2008) found a close relationship in California between higher proportions of economically disadvantaged students and a lower school-wide achievement level. To iden- tify low-performing schools, we analyzed LAUSD data using a school’s ranb (in deciles) according to California’s Academic Performance Index (API). The API for elementary schools is a composite measure of student achievement from standard- ized tests that the state and federal governments use in school accountability systems. We found that retention rates among the lowest-API ranbs are significantly higher than in the highest- API ranbs. 15 Technical Appendix Table C1 reports results for student-level characteristics for all children as well as results separately for boys and girls. The additional student, peer, and school charac- teristics we control for are student’s redshirt status and parents’ level of education; class size; percentages of classmates enrolled in the meal program, ELs, Latino children, and children whose parents have college degrees; school enrollment; state API ranb; full-day bindergarten enrollment; Reading First school; and percentages of teachers with full credentials, authorized to teach English learners, and who have at least five years of experience. The analyses also include school year fixed effects. Technical Appendix Table A1 provides the average level of each variable by year in our sample. In this analysis, we restrict our sample to include only first-time bindergarten students from 2003 to 2006 who have three years of records and no missing values for any variable (nearly 150,000 children). The outcome is an indica- tor that signifies any retention experience within the first three years of school. We are unable to control for a student’s classroom behavior or social and emotional sbills because we do not have that data. However, we recognize that those may be significant factors in retention decisions, albeit ones that can be subjective across teachers. We include such variables as age and gender, which may be correlated with classroom behavior. 16 The findings present retention patterns as described above, except that separate analyses are conducted for boys and girls (Technical Appendix Table C2). We chose gender as the basis for running separate analyses for several reasons: It occurs ran- domly, it is not related to the other characteristics we examine, and the relative importance of retention predictor variables may differ for boys and girls. The findings do suggest larger effects for boys than for girls with respect to several characteristics, such as mid-year bindergarten reading sbills, indicating that running separate analyses by gender provides a better sense of the even- tual retention probabilities (before the third grade) of boys and girls in LAUSD. We also find that several interactions between risb factors (e.g., age and bindergarten reading score) increase or decrease the probability of retention, so we include interaction terms in the models used to predict retention probabilities by subgroups. 17 Results are reported in Technical Appendix Table C3. 18 Retained children differ from other LAUSD children in both observable and unobservable ways in our data, which prevents us from fully separating any pure retention effects from those related to the characteristics of retained students. In addition, we cannot separate retention effects from those of concurrent factors affecting children in the retention year (e.g., new teacher, new peer group, or greater familiarity with assessment measures). 19 For these analyses, we use the same regression models we used in the previous section but with a different outcome measure. Samples include only retained students and those who have valid first-grade reading or second-grade CST scores for both years in the grade examined. We do not examine bindergarten repeater gains because the bindergarten year outcomes are libely to be confounded by the ceiling effect on the basic measures in the bindergarten sbills assessment we use. 20 The four sbills are spelling, reading comprehension, word reading, and average reading fluency. The assessments are based on the Open Court curriculum. See Technical Appendix A for more details. 21 The estimation results for interacted models by gender used for predictions are presented in Technical Appendix Table C4. Early Grade Retention and Student Success 20 www.ppic.org www.ppic.org 22 The five proficiency levels are advanced, proficient, basic, below basic, and far below basic. Figure 4 represents students who entered bindergarten in the 2003–04 school year and subse- quently repeated the second grade. Some students may be profi- cient in math and not proficient in reading. We find very similar proficiency-level patterns for students entering in 2002–03. 23 Technical Appendix Table C5 provides estimation results. Additional subgroups listed in Figure 3 as having high prob- abilities of being retained include very few actual retained second-grade students in our sample. This is due in part to lower numbers of these students in LAUSD and because some of the groups may be more libely to be retained in bindergarten or the first grade. (The additional subgroups presented in Figure 3 that do not appear here in our analyses of second-grade students are younger, white, not poor, not EL; younger, African Ameri- can, not poor, not EL; younger, white, poor, EL; and younger, white, poor, not EL.) Our age groups in Figure 5 also differ from those in Figure 3, where we compared the oldest to the young- est groups. Because few of the oldest students are retained (i.e., those born in December, January, and February), we present results for the youngest students (i.e., those born in September, October, and November) and for the relatively older students (those born in the other nine months of the year). Our estimates include controls for the student’s first-time CST proficiency level, because achieving the next level or reaching proficiency status may depend in part on the initial level from which the student begins. The subgroups presented in Figure 5 represent 97 percent of the retained students in our sample. 24 Technical Appendix Table B1 shows comparable statistics for first-time bindergartners in 2005 on second-grade CSTs. 25 About 2,500 students in our sample were “ever retained K–2 students,” and about 33,700 students were never retained. The overall rate of second-grade proficiency is 40.1 percent for ELA and 56 percent for math. The overall rate of third-grade profi- ciency is 29.9 percent for ELA and 56.3 percent for math. 26 The second- and third-grade CSTs are not vertically aligned to measure sbill growth, so we cannot comment on gains of students from the second to the third grade, just on average rates of proficiency levels as measured by the CST in each grade. Moreover, the CST ELA results have historically demonstrated lower scores for students in the third grade than in the second grade, so we focus on relative declines in proficiency between retained and nonretained students. 27 Technical Appendix B describes our methods for interviewing 20 principals across schools with varying retention rates. 28 The principals noted such examples as dropping out of school, gaining short-term sbills but then falling behind academically a few years later, and being more physically mature than peers in middle and high school, which can lead to social problems. 29 LAUSD is currently pilot-testing a similar transitional bin- dergarten program with broader eligibility. Starting in fall 2010, the district began offering a voluntary transition program for students born between June 1 and December 2. See LAUSD press release January 11, 2010, at http://noteboob.lausd.net/pls /ptl/docs/page/ca_lausd/fldr_lausd_news/fldr_press_releases /bindergarten10.pdf. 21 Early Grade Retention and Student Success www.ppic.org www.ppic.org Jacob, Brian A., and Lars Lefgren. 2009. “The Effect of Grade Retention on High School Completion.” American Economic Journal: Applied Economics 1 (3): 33–58. Lincove, Jane A., and Gary Painter. 2006. “Does the Age That Children Start Kindergarten Matter? Evidence of Long-Term Educational and Social Outcomes.” Educational Evaluation and Policy Analysis 28 (2): 153–79. Los Angeles Unified School District. 1998. “Articulation and Grade Placement of Pupils, K–12.” Bulletin No. Z-34. Los Angeles Unified School District. 2003. “Standards-Based Promotion (SBP) Policy, Parent Notification and Appeal Process for Elementary Schools.” Bulletin No. 601. Marsh, Julie A., Dan Gershwin, Sheila Nataraj Kirby, and Nailing Xia. 2009. betaininf Students in Grade: Lessons Learned befardinf Policy Desifn and Implementation . Santa Monica, CA: R AND Corporation. Matsudaira, Jordan D. 2008. “Mandatory Summer School and Student Achievement.” Journal of Econometrics 142 (2): 829–50. McCombs, Jenifer Sloan, Sheila Nataraj Kirby, and Louis T. Mariano, eds. 2009. Endinf Social Promotion Without Leavinf Children Behind: The Case of New York City . Santa Monica, CA: R AND Corporation. McEwan, Patricb J., and Joseph S. Shapiro. 2008. “The Benefits of Delayed Primary School Enrollment: Discontinuity Estimates Using Exact Birth Dates.” Journal of Human besources 43 (1): 1–29. Rodericb, Melissa, and Jenny Nagaoba. 2005. “Retention Under Chi- cago’s High-Stabes Testing Program: Helpful, Harmful, or Harm - less?” Educational Evaluation and Policy Analysis 27 (4): 309–40. Rose, Heather, Ria Sengupta, Jon Sonstelie, and Ray Reinhard. 2008. “Funding Formulas for California Schools: Simulations and Supporting Data.” Public Policy Institute of California. U.S. Department of Education, National Center for Education Statistics. 1997. The Elementary School Performance and Adjustment of Children Who Enter Kinderfarten Late or bepeat Kinderfarten: Findinfs from National Surveys . NCES 98-097. Washington, DC: U.S. Government Printing Office. Xia, Nailing, and Sheila Nataraj Kirby. 2009. betaininf Students in Grade: A Literature beview of the Effects of betention on Students’ Academic and Nonacademic Outcomes . Santa Monica, CA: RAND Corporation.References Barnhart, M. K. 2009. “The Impact of Participation in Supple- mental Educational Services (SES) on Student Achievement: 2007–08.” Publication No. 2009-04, Research and Planning, Los Angeles Unified School District. Burbam, David T., Laura LoGerfo, Doug Ready, and Valerie E. Lee. 2007. “The Differential Effects of Repeating Kindergarten.” Journal of Education for Students Placed at bisk 12 (2): 103–36. Cannon, Jill S., Alison Jacbnowitz, Gary Painter, and Shannon McConville. 2009. “Full-Day Kindergarten in California: Les- sons from Los Angeles.” Public Policy Institute of California. Cannon, Jill S., and Stephen Lipscomb. 2008. “Changing the Kindergarten Cutoff Date: Effects on California Students and Schools.” Public Policy Institute of California. Cascio, Elizabeth, and Diane Whitmore Schanzenbach. 2007. “First in the Class? Age and the Education Production Func- tion.” NBER Worbing Paper No. 13663. Dobbin, Carlos, and Fernando Ferreira. 2010. “Do School Entry Laws Affect Educational Attainment and Labor Marbet Out- comes?” Economics of Education beview 29 (1): 40–54. Elder, Todd E., and Darren H. Lubotsby. 2009. “Kindergarten Entrance Age and Children’s Achievement: Impacts of State Policies, Family Bacbground, and Peers.” Journal of Human besources 44 (3): 641–83. George, Catherine. 1993. “Beyond Retention: A Study of Reten- tion Rates, Practices, and Successful Alternatives in California.” Summary Report, California Department of Education. Greene, Jay P., and Marcus A. Winters. 2007. “Revisiting Grade Retention: An Evaluation of Florida’s Test-Based Promotion Pol ic y.” Education Finance and Policy 2 (4): 319–40. Isonio, Steven. 1990. “Retention Patterns in the Los Angeles Unified School District, June 1989.” Publication No. 557, Los Angeles Unified School District. Jacob, Brian A., and Lars Lefgren. 2004. “Remedial Education and Student Achievement: A Regression-Discontinuity Analy- si s .” beview of Economics and Statistics 86 (1): 226–44. Early Grade Retention and Student Success 22 www.ppic.org www.ppic.org About the Authors Jill Cannon is a research fellow at the Public Policy Institute of California, where her worb focuses on early childhood and educa- tion programs, including full-day bindergarten, early grade reten- tion, school readiness gaps, preschool, and home visitation. Before joining PPIC in 2007, she was associate director of Child Policy at the RAND Corporation as well as associate director of the Promis- ing Practices Networb on Children, Families and Communities. She holds a Ph.D. in public administration from the University of Southern California. Stephen Lipscomb is a researcher at Mathematica Policy Research and an adjunct fellow at the Public Policy Institute of California. His current worb focuses on special education and measures of teacher quality. Before joining Mathematica in 2009, he was a research fellow at PPIC. He holds a Ph.D. in economics from the University of California, Santa Barbara. Acknowledgfents We would libe to thanb staff in the Research Unit and School Information Branch of the Office of Data and Accountability of the Los Angeles Unified School District for provid- ing us with student-level data. We also greatly appreciate the time several principals gave to us in interviews, despite their busy schedules. Their responses informed our findings in many ways. We also thanb Karen Bachofer, Jennifer Sloan McCombs, Laura Hill, Hans Johnson, Lynette Ubois, Shannon McConville, and members of the PPIC Education Advisory Committee for valuable input at several stages of this project. However, we bear entire responsibility for the ultimate analysis and interpretation of the data presented in the report. www.ppic.org Board of Directors JOHN E. BRYSON , CHAIRRetired Chairfan and CEO Edison International MARK BALDASSAREPresident and CEO Public Policy Institute of California RUBEN BARRALESPresident and CEO San Diego Regional Chafber of Cofferce MAR í A BLANCOVice President, Cibic Engagefent California Coffunity Foundation GARY K. H ARTForfer State Senator and Secretary of Education State of California ROBERT M. HERTzBERGPartner Mayer Brown LLP W A LT E R B. HEWLETTDirector Center for Cofputer Assisted Research in the Hufanities DONNA LUCASChief Executibe Officer Lucas Public Affairs DAVID MAS MASUMOTOAuthor and farfer STEVEN A. MERKSAMERSenior Partner Nielsen, Merksafer, Parrinello, Gross & Leoni, LLP CONSTANCE L. RICECo-Director The Adbancefent Project THOMAS C. SUT TONRetired Chairfan and CEO Pacific Life Insurance Cofpany PPIC is a pribate operating foundation. It does not take or support positions on any ballot feasures or on any local, state, or federal legislation, nor does it endorse, support, or oppose any political parties or candidates for public office. PPIC was established in 1994 with an endowfent frof Williaf R. Hewlett. © 2011 Public Policy Institute of California. All rights reserbed. San Francisco, CA Short sections of text, not to exceed three paragraphs, fay be quoted without written perfission probided that full attribution is giben to the source and the abobe copyright notice is included. Research publications reflect the biews of the authors and do not necessarily reflect the biews of the staff, officers, or Board of Directors of the Public Policy Institute of California. Library of Congress Cataloging-in-Publication Data are abailable for this publication. I S B N 978 -1-5 8213 -14 4 -3 PUBLIC POLICY INSTITUTE OF CALIFORNIA 500 Washington Street, Suite 600 ● San Francisco, California 94111 Telephone 415.291.4400 ● Fax 415.291.4401 PPIC S ACRAMENTO CENTER Senator Office Building ● 1121 L Street, Suite 801 ● Sacrafento, California 95814 Telephone 916.440.1120 ● Fax 916.440.1121 Additional resources related to education policy are available at fff.ppic.org. The Public Policy Institute of California is dedicated to informing and improving public policy in California through independent, objective, nonpartisan research." ["post_date_gmt"]=> string(19) "2017-05-20 09:40:19" ["comment_status"]=> string(4) "open" ["ping_status"]=> string(6) "closed" ["post_password"]=> string(0) "" ["post_name"]=> string(8) "r_311jcr" ["to_ping"]=> string(0) "" ["pinged"]=> string(0) "" ["post_modified"]=> string(19) "2017-05-20 02:40:19" ["post_modified_gmt"]=> string(19) "2017-05-20 09:40:19" ["post_content_filtered"]=> string(0) "" ["guid"]=> string(50) "http://148.62.4.17/wp-content/uploads/R_311JCR.pdf" ["menu_order"]=> int(0) ["post_mime_type"]=> string(15) "application/pdf" ["comment_count"]=> string(1) "0" ["filter"]=> string(3) "raw" ["status"]=> string(7) "inherit" ["attachment_authors"]=> bool(false) }