Key Takeaways

California’s Transitional Kindergarten (TK) program provides an additional year of schooling aimed at preparing children for kindergarten within the K–12 system. TK was launched a decade ago with limited eligibility, but it will be open to all four-year-olds by 2025–26. While early research has shown that TK improved kindergarten readiness, little is known about its longer-term impact—especially among multilingual students and students with disabilities, who might benefit from early identification of special needs. To learn more, we followed initial TK cohorts through elementary school in five large school districts. Key findings include:

• Pre-expansion TK led to earlier identification of English Learner (EL) students and students with special education needs. Nearly four in ten TK students were identified as ELs in TK and kindergarten, compared to under three in ten kindergarteners who did not enroll in TK. Special education identification among TK students has primarily been driven by earlier identification of autism and speech/language impairments.→
• Evidence suggests positive social-emotional learning (SEL) outcomes, but only for English-only students. When we examine outcomes separately for English-only and multilingual students, TK appears to have large positive SEL impacts on the former group, and small or even slightly negative impacts on the latter group. This suggests a need for refinements to TK programing so that multilingual learners can benefit more fully.→
• TK does not appear to improve grade 3 and 4 test scores more than other pre-kindergarten options.  Previous research on pre-K programs shows additional academic improvement fading out in elementary school (but finds important positive outcomes in high school and beyond, such as improved graduation rates). It will be important to continue following trajectories to understand how TK’s impact may be evolving as it expands, and to see if initial TK cohorts see the long-run gains observed in previous research.→

As part of its overall effort to provide high-quality early childhood education to all three- and four-year-olds, California is expanding transitional kindergarten (TK). Research on the first California TK cohorts showed positive initial effects (Manship et al. 2017)—even relative to existing public universal pre-K programs—but little is known about the persistence of benefits in later grades and whether academic trajectories improve specifically for dual-language learners (a term often used for English Learners in early grades) and students with disabilities, who might benefit from earlier identification of their educational needs. Taking stock of such a sizeable—and promising—investment before, during, and after expansion can help ensure its success.

The share of four-year-olds eligible to attend TK held steady at 25 percent for many years, but all four-year-olds will be eligible as of the 2025–26 school year. The official expansion schedule adds two months of eligible birthdates each year between 2022–23 and 2024­–25, and in 2025–26, all children who turn four by September 1 will be eligible. The expectation is that all four-year-olds will be eligible for and have access to a TK program as a part of the K–12 system.

The Local Control Funding Formula will provide funding for additional TK students to school districts. The 2021–22 state budget committed to funding universal access for four-year-old children to TK by 2025–26. The cost of full implementation—estimated to be roughly $3 billion—includes $300 million to support district planning, teacher training, and expanding the teacher workforce (so that TK staffing ratios will be 10:1 by the end of the expansion period), as well as $490 million for constructing and modernizing facilities. While students of all income levels can attend TK, funding for expansion planning in the 2021–22 school year prioritized districts with more low-income families or EL students, especially at higher grade levels.

This report examines whether California’s pre-expansion TK program has affected dual-language learners’ educational trajectories. We also provide new research on the effects of TK for students with disabilities, for whom early identification and support may be beneficial. And finally, we document whether the benefits of TK have persisted into later elementary grades in the five districts we study.

How do we determine TK’s impact? Simply comparing outcomes between TK and non-TK students is not the best approach: there are differences in local availability of TK (Hill and Prunty 2022), as well as differences in the families that choose to opt into TK versus choosing another public, private, or home-care option for their four-year-old children. Instead, we rely on comparisons between students on either side of the December 2 birthdate cutoff for the school years 2013–14 through 2020–21. That is, we estimate the impacts of TK by comparing students born just days apart, and in effect differ only in their eligibility for TK. (We discuss the logic behind the approach, and its implications in full detail in Technical Appendix B.)

Our estimates are drawn from detailed administrative records from nearly 90,000 students who attended TK in five large districts across the state. Our study districts are all CORE districts, and data were obtained through the CORE-PACE research partnership. This is a larger sample than those in prior California TK studies. Because it includes multiple cohorts across hundreds of schools offering TK in different local contexts, we are able to examine how effects vary across student, school, and district characteristics. While the students in our study districts are not perfectly representative of students statewide, they provide a good representation of the state’s dual-language learners, Latinos, and children from low-income families.

California’s transitional kindergarten program provides an extra year of schooling within the public school system. Starting in the 2012–13 school year, school districts were required to offer TK to four-year-olds whose birthdates were close to the kindergarten cutoff date. As of the 2022–23 school year, about 42 percent of four-year-olds were eligible.

TK is open to families of all income levels. TK students can be served in the same classrooms as kindergarten students (or in separate classrooms), in a California State Preschool Program, or in other blended settings. The TK curriculum is meant to be distinct from the kindergarten curriculum. Unlike kindergarten teachers, TK teachers must have a both Multiple Subject Credential and training in early childhood education. The TK school day must be at least three hours long, and districts may opt for full-day TK.

Currently, all school districts serving kindergarteners are required to offer TK, but an estimated 8 percent of eligible children live in districts that do not provide it (Hill and Prunty 2022). Moreover, there is no requirement that TK be offered at all elementary schools in each district. Earlier research estimated that 61 percent of eligible children who were not enrolled lived in school districts offering TK, but within elementary school zones where TK was not offered (Hill and Prunty 2022). As a result, some families with eligible four-year-olds may have opted out because the TK location in their district was inconvenient.

Moreover, since TK is optional, families may have chosen not to enroll their eligible children even if slots were available. The wide range of other options include public preschool (such as the California State Preschool Program or Head Start), private preschool, child care centers, private home-based child care, or care at home. The number of hours of care provided is also a critical factor. While schools are expanding hours of care for K–6 students (up to 9 hours), the some families may not be able to secure a full day of care at their neighborhood school or may require non-traditional hours of care.

Who Has Been Enrolling in TK?

In the 2021–22 school year, a quarter of all four-year-olds were birthdate-eligible to attend TK, as has been the case since 2014–15. Of those four-year-olds eligible to attend, the share that did so has ranged between 70 and 83 percent. Enrollment peaked just prior to the pandemic and fell to 70 percent during the worst of the pandemic (Figure 1). The most recent data suggests take-up rates have rebounded to 77 percent, or 75,000 students.

Despite the fact that not all eligible students enrolled in TK, we find that the demographic characteristics of TK students are quite similar to those of kindergarten students. In 2021–22, dual-language learners were slightly more represented among TK than among kindergarten students (27% vs. 25.8%, Table 1), but the two groups are currently fairly comparable in race/ethnicity and socioeconomic status. Prior to the pandemic, dual-language learners were most likely to enroll in TK if it was offered at their school; Black, Native American, and Pacific Islander students were least likely to enroll (Hill and Prunty 2022). This may be explained by several large districts’ robust early transitional kindergarten (ETK) offerings in which students too young for TK can participate in district instruction. Many of these districts with ETK programs enroll large numbers of dual-language learners, thus increasing the numbers of ELs in TK (Hill and Prunty 2022).

As noted earlier, TK is not the only early education or early care option available to four-year-olds in California: California has a mixed-delivery system for early learning and care. In the cohort of four-year-olds starting kindergarten in 2017, approximately 71 percent were in some form of non-parental care (62% in center-based care, 24% in relative care, and 11% in nonrelative care). Center-based care can be private or public. In the 2021–22 school year, an estimated 39 percent of all four-year-olds were enrolled in some form of public early childhood education. This includes TK (19%), California State Preschool Program (CSPP, 12%), Head Start (7%), and special education programs (1%) (Friedman-Krauss et al. 2023). School districts can be the location for all four programs.

TK eligibility is based on birthdate and is free, but many other early childhood care options are subsidized by state and federal governments based on income eligibility. And families that have participated in TK may differ in unobservable ways from families that have chosen other options for early education and care. Those other options could vary widely across the state. Center-based care (public or private) is currently chosen for over half of California’s four-year-olds (Master Plan for Early Learning and Care 2020), but we do not know what other options are chosen by families whose children do not enroll in TK, and we are not able to assess the quality of those experiences. This poses challenges to understanding the relative impact of TK on later outcomes.

Research has documented the benefits of early childhood education on school readiness, often finding that low-income and less-advantaged students see greater benefits (Currie 2001; Barnett 2011; Camilli et al. 2010; Magnuson and Shager 2010; Duncan and Magnuson 2013). Recent research in California found that preschool quality is particularly important for dual-language learners and children with disabilities (Sussman et al. 2023). In particular, early interventions for dual-language learners and special education students are thought to be especially important (Castro 2014): the critical period for development of a second language is in early childhood (Kuhl 2011), and earlier identification and supports for students with disabilities can improve outcomes.

There is evidence that early childhood education can also improve adult outcomes such as employment, earnings, educational attainment, health, and reduced criminal activity (Ludwig and Miller 2007; Heckman et al. 2010; Johnson and Jackson 2019). More generally, early childhood interventions have been shown to produce very high returns per dollar investment, particularly for programs serving low-income children (Hendren and Sprung-Keyser, 2020; Kline and Walters 2016).

However, less is known about the persistence of early childhood education effects on dual-language learners and students with disabilities, for whom early interventions and support might be more beneficial and persistent. Many studies show a “fade-out” of gains in elementary school (Gibbs et al. 2013; Li et al. 2020; Jenkins et al. 2018), though some research finds long-run effects on disciplinary and post-secondary outcomes even in the absence of test score effects (Gray-Lobe, Pathak, and Walters 2023).

California’s TK program launched in 2012–13, although most districts did not begin implementation until the following year. As a result, there is little research on the program’s effects beyond kindergarten. Manship et al. (2017) find that TK improved academic skills and engagement in kindergarten. Doss (2019) finds positive effects on kindergarten readiness among EL TK students in San Francisco. Holod et al. (2020) find positive TK effects upon kindergarten entry among Spanish-speaking EL students.

Notably, all of these studies examine only initial effects among the first TK cohorts; the extent to which effects persist through elementary school and result in earlier reclassification from English Learner status and improved English proficiency are unknown. This longer-term view may be important for dual-language learners as acquiring English proficiency may take years, especially when both home language and English are being learned, such as in a bilingual language program (Umansky and Reardon 2014) and as supported by the California’s English Learner Roadmap policy (CDE 2017). We are unaware of research specifically examining the effects of TK on the outcomes and identification of students with disabilities. And none jointly considers the relationship between EL and special education classification, a known challenge across grade levels in California (Umansky et al. 2015; Umansky et al. 2017; Hill et al. 2019).

How Can We Measure TK’s Impact?

Central to understanding impacts and improving TK expansion are questions of fade-out, the impact for dual-language learners, and the impact for students with disabilities. To understand whether students enrolled in TK are seeing benefits to their extra year of schooling in the K–12 system, we use student-level data linked from 2014–15 to 2020–21 in five large California school districts, which allows us to observe test scores through fourth grade for multiple cohorts, and up to sixth grade for other outcomes. Our sample is not perfectly representative of the state’s TK students (Figure 2). Notably, our TK sample has a higher share of Latino, dual-language-learner, and low-income children, and a lower share of white children, in part reflecting the fact that our study districts are larger and generally serve a higher share of nonwhite and high-need student groups. Thus, while not representative for the typical TK enrollee, our sample provides a good window into understanding impacts for Latino children, dual-language learners, and children from low-income families (i.e., eligible for free or reduced-price meals, FRPM). See Technical Appendix A for more detailed description of the data, sample selection, and the outcomes available for each TK cohort.

In the period we are studying, children born on or before December 2 were eligible for TK. Children born December 3 or later missed the cutoff and spent the year in some alternative to TK. While not all families chose to enroll their children in TK, the cutoff creates a discrete difference in the propensity to enroll in TK.

This birthday cutoff allows us to assess the impacts of TK by comparing otherwise similar children born on either side of the cutoff. Because we leverage these small birthdate differences, we can eliminate the impact of other factors that could confound attempts to assess the impacts of TK, such as race, socioeconomic status, and location. The students born just after December 2, who did not attend TK, started kindergarten alongside the cohort of TK students, who were born just days earlier.

In short, our empirical approach compares those born on both sides of the December 2 cutoff who enroll together in kindergarten in the same school the following year. This approach, often referred to as a regression discontinuity design (RDD), is commonly used to uncover causal impacts when there are discrete differences in program participation across a threshold.

Figure 3 shows how the birthdate cutoff influences TK participation. Each dot represents the share of kindergarteners in our sample with a birthdate up to 40 days before or after December 2 who enrolled in TK the prior year. We can see a large and discrete difference across the threshold: roughly 70 to 75 percent of students born earlier in November attend TK, compared to just under 20 percent of those born after December 2. Notably, for TK cohorts that entered kindergarten prior to 2017–18, take-up is under 3 percent for those born after December 2 (Technical Appendix Figure B4). For later cohorts, the introduction of “early TK” in some districts meant that students born after December 2 could enroll if the district opted to provide it; as a result, take-up among children born in December after the cutoff rose to roughly 30 percent (Technical Appendix Figure B4). Our final sample excludes students with missing demographic and background variables and totals nearly 70,000 kindergarten students within our baseline bandwidth of 40 days. (See Technical Appendix A and B for further details.)

Notably, the propensity to enroll in TK decreases slightly as birthdates get closer to December 2—from October to November to December (or 40 to 20 to 0 days before the cutoff). Our baseline models account for these trends separately for students born before or after December 2.

Enrollment in TK may affect educational trajectories through earlier identification of special educational needs and the provision of additional supports and services. Most notably, dual-language learners who enter TK or kindergarten speaking another language besides English at home may be more likely to be identified as English Learners (ELs)— in need of English language support services—in TK. We might expect this earlier additional support to influence later outcomes, perhaps through earlier reclassification from EL status to English-language proficient or improved outcomes based on earlier receipt of English language services. We examine reclassification through grade 5 for our earliest cohort and through kindergarten for the TK cohort starting in 2019–20.

TK Leads to Earlier EL Identification and Reclassification

We find stark differences in the timing of EL identification and reclassification due to TK. In our sample, nearly 40 percent of students who enrolled in TK were classified as ELs in TK and kindergarten, compared to under 30 percent of kindergarteners who did not enroll in TK. However, the difference is much smaller in first grade and is eliminated by second grade (Figure 4).

Using our model to compare students just across the birthdate cutoff, we can estimate the causal impact of TK on EL identification.

We find that students who enroll in TK are 10 percentage points more likely to ever be identified as ELs (Figure 5). When we break down the effect of TK on EL status by grade, we see that the TK leads to an increased likelihood of being classified as an EL in kindergarten. But starting in grade 1, TK students are no more likely to be classified as ELs. By grade 5, the impact of TK on EL status is very small and not statistically significant. Overall, TK students are classified as ELs for about 40 percent of an academic year longer than their non-TK counterparts (Appendix Table B2). This confirms the pattern observed in Figure 4: students who enroll in TK are both identified earlier and are more likely to ever be EL, with a higher likelihood of being EL in kindergarten for students who enrolled in TK versus those who did not.

Similarly, when we look at reclassification we see positive impacts in kindergarten and grade 1, and no difference due to TK in later grades (Figure 6). While kindergarten reclassification is relatively rare, it has been increasing statewide since the inception of the TK program, with the exception of the school years most impacted by COVID (CDE Dataquest).

These results suggest more students are identified for EL services because of TK, but that they do not receive those services for long. First, they are more likely to ever be identified as ELs, meaning that some students are identified in TK who would never have been identified had they started elementary school in kindergarten. Second, these students see very quick reclassification out of EL status in kindergarten and first grade. Thus, the average student only ends up seeing a modest increase in the years of EL services provided. Third, there appear to be no sizable net effects on EL status after grade 2: our point estimates are small and insignificant in grades 2 to 5.

TK May Lead to Earlier Identification of Special Education Students

Students with special education needs who enroll in TK may also get additional years of services that could affect their educational outcomes. Though most students will not be identified as having a special education need in early grades, they are not uncommon in the five districts in our study: 8 percent of students in TK have an identified special education need, rising to 9 percent in kindergarten, and then reaching roughly 13.4 percent by the end of elementary school (Figure 7). We are able to examine special education identification through grade 5. Specific learning disabilities (such as dyslexia) are the most common but are generally identified in later elementary school grades. In TK and kindergarten, speech/language impairments and autism are the most common.

Earlier (pre-kindergarten) enrollment in elementary school may mean earlier identification of autism and speech/language impairment, which make up roughly 80 percent of special education needs identified in TK. Indeed, Figure 8 suggests that this may be the case, as special education shares are initially higher in kindergarten through grade 2 for students who had TK than those who did not, with no or very small differences in later grades. This early identification pattern holds for both English-only and multilingual students (those speaking languages other than English at home) (Technical Appendix Figures C1 and C2). However, multilingual students have much higher rates of special education identification across all grades (10% in TK vs. 6% for English-only students). Moreover, the gap in special education identification between students who enrolled in TK and those who did not closes by grade 2 for English-only students; it shrinks but persists through grade 5 for multilingual students.

It is important to note that these differences by grade are only suggestive. Students with special education needs may be more likely to take up TK, and there are differences in special education identification by age—and relative age within the classroom—that may influence comparisons between TK and non-TK students when measured in kindergarten and later grades. When we estimate the causal impact of TK on special education identification by comparing only students in the same cohorts whose birthdates differ in a narrow window around the cutoff—our regression discontinuity design outlined above—we find differences in kindergarten and grade 1 that are still positive but slightly smaller than the differences shown in Figure 8, and statistically insignificant (Technical Appendix Figure B20). We also find no effects on special education rates in later grades, and no significant impact on whether a student is ever identified with a special education need.

We do find notable impacts of TK on the grade of identification, however (Figure 9). On average, TK leads to identification of special education needs roughly a half a grade earlier. When we look at the three most common primary disability categories, we see that that the main drivers are autism (one grade earlier identification) and speech/language impairments (half a grade earlier identification). We see no impact on the timing of identification for specific learning disabilities. This is unsurprising, as these disabilities are most often identified later in elementary school.

Taken together, these special education findings provide new evidence that TK affects the timing of special education identification for the most common early disabilities. To the extent that earlier identification is used to provide additional years of services that benefit students with disabilities, this is a promising effect of the TK program.

Earlier identification of autism speech/language occurs mainly among multilingual students

We also estimate the impact of TK on special education identification separately for English-only and multilingual students. On average, multilingual students who enroll in TK see higher rates of special education identification through grade 5 (Technical Appendix Figure C2), but these differences are smaller (and not statistically significant) when we look at the difference just across the December 2 cutoff. TK does not appear to affect the likelihood of either English-only or multilingual students ever being identified as in need of special education. However, identification of autism and speech and language impairment are identified earlier among multilingual students than among English-only students (Technical Appendix Figure B25). In particular, multilingual students with autism are identified nearly 1.5 grades earlier, on average, due to TK enrollment. More research is needed to understand the mechanisms driving earlier identification multilingual students and determine whether earlier identification improves student outcomes in the long term.

To better understand TK’s longer-term impact, we need to look at differences in later elementary school outcomes, including standardized test scores in math and English Language Arts (ELA) and social-emotional learning (SEL). We also explore whether there are differences in TK’s longer-term impact across demographic groups, including gender, socioeconomic status, and race/ethnicity.

TK Does Not Appear to Boost Test Scores in Grades 3 and 4

The first cohort in our sample participated in TK in 2013–14 and reached grade 4 by 2018–19. Due to the COVID-19 pandemic, testing was suspended in 2019–20 and optional in 2020–21. Our data do not cover 2021–22, so here we examine grade 3 and 4 test scores for the first two TK cohorts in our sample.

There are very small differences in the mean test scores in math and ELA in grades 3 and 4 for students near the cutoff who enrolled in TK and those who did not (Technical Appendix Figure C3). We confirm this lack of difference with a comparison of students in the same cohorts with birthdates within a narrow window before and after the cutoff (Figure 10).

Our comparison finds small impacts—roughly 1–2 percent of a standard deviation, which is smaller than most educational interventions. Looking at the upper bound of our confidence intervals for pooled test scores, we can rule out large positive (or negative) impacts (i.e., more than 9% of a standard deviation). However, we cannot statistically rule out more modest test score improvements (of less than 9% of a standard deviation), which are within our 95 percent confidence intervals. We see no difference between impacts across ELA and math subjects.

The lack of statistical differences in grade 3 or 4 that we find is not surprising, given prior research on the effects of pre-kindergarten programs (Gibbs et al. 2013; Li et al. 2020; Jenkins et al. 2018). As noted earlier, TK has been shown to increase kindergarten readiness in California, but other research shows academic impacts often fade out shortly after pre-K interventions, even in studies that find positive benefits later in school and adulthood (such as reduced disciplinary issues and improved postsecondary attendance) of a specific pre-K program (e.g., Gray-Lobe et al. 2023).

Moreover, our findings do not indicate that TK has no positive impact on test scores of early cohorts; rather, we find that it produces no additional positive benefits beyond those produced by the other programs students would have attended had they not enrolled in TK. While we cannot observe these alternative programs, the availability of high-quality public and/or private options could mean that even a very high-quality TK program produces no relative positive academic impact, even if it may be preferable for other reasons (e.g., cost, non-academic benefits, etc.). In this context, it is worth noting that in 2016–17, a greater share of four-year-olds enrolled in subsidized public early childhood education programs than in TK: 17 percent of four-year-olds statewide attended California State Preschool (CSPP) and 8 percent attended Head Start, while 18 percent attended TK (Stipek 2018).

Evidence of Social-Emotional Learning Gains from TK Is Suggestive but Imprecise

Some previous research has found that pre-kindergarten programs have positive impacts on behavioral and non-academic outcomes (e.g., Gray-Lobe, Pathak, and Walters 2023). To better understand TK’s non-cognitive impact, we examined student responses on social-emotional learning (SEL) surveys that are available from four of the five districts in our sample. The surveys cover grades 3 to 12; we looked at results for grades 3–6.

The surveys measure four SEL skills (Meyer, Wang, and Rice 2018):

• Self-management: regulation of emotions, thoughts, and behaviors; e.g., managing stress, delaying gratification, motivating oneself, and goal-setting (CASEL 2005).
• Growth mindset: belief that one’s abilities can improve with effort, practice, and/or perseverance.
• Self-efficacy: belief in one’s ability to succeed at achieving a specific outcome, e.g., reaching a goal.
• Social awareness: ability to recognize others’ perspectives, with empathy, recognition, and understanding of social norms, diverse backgrounds, and community supports (CASEL 2005).

We aggregated responses across questions about each skill to develop a single “score”; we also aggregated an average overall “SEL score.” See Technical Appendix A for details on the construction of these scores from the survey data.

On average, we see small differences in SEL outcomes between students who did and did not enroll in TK in grades 3 to 6 (Technical Appendix Figure C3). Figure 11 reports estimates from our model comparing students just across the December 2 cutoff on these SEL measures. Here, estimated impacts of TK are generally positive, though none are statistically significant. This analysis yields suggestive evidence of SEL improvements due to TK, though the estimates are imprecise. Future research examining additional cohorts may be able to provide a more precise estimate of TK’s impact on SEL.

Evidence of Differences in TK’s Impact on Test Scores and SEL across Socioeconomic Groups Is Inconclusive

Prior research has found that early childhood education has larger impacts on lower-income students and identified some gender differences—especially in non-academic outcomes (Duncan and Magnuson 2013). We tested for differences across genders and socioeconomic status. To identify socioeconomic status (SES) differences, we relied on two factors: (1) whether or not a student is eligible for free and/or reduced-price meals (FRPM) and (2) whether or not a student has a parent with any college education.

In all cases, we cannot rule out the hypothesis that TK’s effects are zero. We do see some evidence of large positive SEL impacts for non-FRPM students, though our results are imprecise. We also find that test score impacts are more positive for students with parents who have attended college and are small and negative for those with parents who did not attend college. We find smaller differences by gender.

Overall, we take this as inconclusive but suggestive evidence that the higher-SES students in the initial cohort benefitted more from TK than the lower-SES students. Future research that examines more TK cohorts across a larger set of districts will be needed to more precisely understand the extent to which impacts vary across socioeconomic statuses.

Black and Asian Students May Have Seen SEL Benefits from TK

We examined TK’s academic and SEL impacts among Latino, white, Black, and Asian students (Technical Appendix Figures B18–19). Test score impacts are greater for Black students than for Asian, Latino, and white students, though the results are not statistically significant. SEL impacts appear to have been greater for Asian students, who saw significant positive improvements overall as well as in self-management and self-efficacy. Black students also saw some positive SEL impacts, though the only significant change was in self-efficacy. SEL impacts for Latino and white students are not statistically distinguishable from zero. Overall, we take this as suggestive evidence that TK may have more positive impacts for Asian and Black students, but we cannot make firm conclusions.

TK Appears to Have Boosted SEL Outcomes among English-only Students but Not Multilingual Students

Our earlier examination of EL classification among our TK cohorts indicated that TK’s impact on academic and SEL outcomes may differ for multilingual students and English-only students. We examined test score and SEL outcomes separately for these two groups.

As Figure 12 shows, there are no significant differences in test score outcomes: TK impacts were statistically insignificant for both English-only and multilingual students in ELA, while English-only students may have seen larger impacts in math. However, there are differences for SEL outcomes. Impacts are small and slightly negative (but insignificant) for multilingual students. For English-only students, overall SEL impacts are positive and significant. We find an especially significant impact on social awareness, and impacts are large across all SEL areas (representing sizable improvements ranging from roughly 25% to 35% of a standard deviation).

What do we make of these findings? They seem to suggest, at least for the initial cohorts, that TK shows positive impacts on students’ self-reported SEL later in elementary school for many students—but that multilingual students are not seeing these SEL benefits. Unfortunately, we lack data on the TK instruction that multilingual students received; this could be an important factor driving the differences in SEL benefits. In an attempt to account for differences in the language environment for multilingual learners in TK, we broke down outcomes for multilingual students according to the share of the kindergarten cohort receiving EL services. These estimates are imprecise, but we find positive SEL outcomes among only the highest EL shares, which suggests that classes with a higher share of multilingual students might have a language environment that is more conducive to improved SEL outcomes for this group.

TK Does Not Seem to Affect Attendance and Discipline Outcomes

We also examined the impact of TK on attendance and disciplinary issues. When we looked at attendance records, we found very small and statistically insignificant effects on attendance rates and chronic absenteeism (10% or more missed school days). Similarly, we find near-zero disciplinary effects, such as suspensions—though disciplinary incidents are very uncommon among elementary students in our dataset (less than 1%–2% of students per year, depending on the outcome). See Technical Appendix Figure B21.

The absence of effects on these outcomes in elementary grades is unsurprising, given the prior literature. Insofar as TK has positive impacts on noncognitive skills to the extent demonstrated in prior studies of early childhood education programs, we might expect these effects to accrue in middle and high school, when disciplinary incidents are more common and attendance is driven more by a student choices and circumstances, as opposed to parental factors.

Critical to improving TK expansion in California are the questions of its impact on dual-language learners and students with disabilities, and its effects on outcomes on student academic and social-emotional outcomes in later grades. There is a substantial body of research examining the short-, medium-, and long-term impacts of early childhood education programs across the nation—but the available evidence on TK’s effects past kindergarten is very limited in California. We have found that TK leads to earlier identification of EL and special education needs—and quicker reclassification of EL students. However, we found that TK also increases the likelihood of a student ever being identified as ELs (leaving open the possibility that TK students are over-identified as ELs) and has no net impact on EL status past grade 2. The positive test score impacts observed upon kindergarten entry for early TK cohorts in prior studies are not found in this study by grade 3 and 4, consistent with what most prior research on early childhood education programs in other settings has found.

Our analysis also yielded some promising suggestive evidence of positive impacts on SEL outcomes, which may mediate the longer-term gains that often appear, even in longitudinal studies showing academic score fade out. However, this impact seems to be limited to English-only students: we found no positive SEL impacts of TK on multilingual students. This suggests the potential for improvements to the TK program to better serve multilingual students and improve non-TK offerings available to them.

Our findings yield several policy considerations and recommendations:

The absence of social-emotional learning benefits for multilingual students suggests a need for a deeper examination of program offerings for multilingual students. As TK expands to all four-year-olds, state and district policymakers should devote additional attention to how to best serve dual-language learners to fully maximize the potential of TK. We find suggestive evidence that multilingual students see greater SEL benefit in settings with high shares of students from non-English speaking households; this may indicate that a more supportive language environment would be beneficial for these students.

Earlier special education identification suggests a promising area for improving services and outcomes. Special education needs—particularly for autism and speech/language impairments—are identified earlier for students who enroll in TK. To the extent additional services can be provided and tailored to younger children, TK may provide an opportunity to improve these students’ outcomes through earlier interventions.

The lack of positive test score effects in elementary school for TK participants is consistent with prior research and may reflect high-quality alternatives for four-year-olds in California. That we find neither positive nor negative effects is not surprising given prior research findings—nor is it an indication that TK is not providing value to students and their families. Instead, our findings may suggest TK’s impact is comparable to that of current alternative preschool programs, such as the CSPP and Head Start. Notably, CSPP, Head Start, and other public preschool programs have been shown to have benefits beyond test scores (e.g., Barr and Gibbs 2022). Furthermore, our test score findings are limited to early TK cohorts in five large districts, before many programmatic changes and improvements in quality, like teacher aides, class size reductions, and curriculum changes.

Future research will continue to track outcomes for these initial cohorts as they progress through middle school. And as expansion continues, data will become available for later cohorts, allowing researchers and policymakers to assess changes and improvements to the TK program and compare outcomes to those of the initial TK cohorts. A continued focus on TK program offerings, quality, and eventual impacts will be key to ensure that California’s TK expansion continues to improve and reaches its full potential of providing high-quality universal early childhood education to all the state’s four-year-olds.