Background: The past half-century has resulted in a substantial expansion of early childhood center-based care in the United States. Today, over 74% of American children aged 3-5 attend center-based care (U.S. Department of Education, National Center for Education Statistics, 2021). Although public pre-k is considered socially favorable, evidence-informed, and essential for a robust economy, policymakers are still uncovering how to make contemporary pre-k programs effective in the long-term (Duncan & Magnuson, 2013; Farran & Lipsey, 2016). The evidence on effectiveness is mixed; some programs have produced meaningful, long-term effects (Gray-Lobe, 2023) while others show null, medium-term effects (Puma et al., 2010). The most concerning is evidence from Tennessee, where participants randomly assigned to the Voluntary Pre-K (VPK) program are scoring lower on state achievement tests, and have higher disciplinary incidents, poorer attendance, and higher rates of special education placement than children not assigned to the program (Durkin et al., 2022). However, the current VPK estimates do not account for the type of alternative care that the control group received, and this may be masking important variation in effects (Weiss et al., 2014). For example, if children assigned to the control group attend similar types of center-based care, then the contrasting experiences (and the corresponding effect sizes) between the treatment and control group might be small or null. Similar studies have found that failing to account for alternative care options underestimated the effects of the Head Start Impact Study (Feller et al., 2016; Kline & Walters, 2016). Research Questions: In this study, we aim to provide exploratory evidence that speaks to the counterfactual experiences that children would have received in absence of VPK. We ask the following research questions: 1. What is the effect of VPK for students who would have otherwise stayed home? 2. What is the effect of VPK for students who would have otherwise attended another type of center-based care? 3. Do these effects sustain or fade out between the end of pre-k through third grade, and does this vary by each subgroup of children? Setting: This study involves a secondary analysis of the original Tennessee VPK study which was implemented across Tennessee between 2009-2011 in 79 state-funded centers. The VPK centers in the study were in urban, rural, and suburban settings throughout the state. Population/Participants/Subjects: We use the VPK intensive sub-study sample (ISS) (N = 1,076 students; 36% of the full sample of N = 2,990 from 58 VPK centers) which is a subsample of children selected at the time of the pre-k lottery for lengthier, in-person assessments and measures. Intervention/Program/Practice: The original study consisted of a waitlist-lottery randomized trial of oversubscribed VPK centers throughout Tennessee. The study was implemented in two cohorts (2009-2010 and 2010-2011). The VPK program was a full-day, 9-month program that met quality standards (i.e., it met 9 of 10 2009 NIEER benchmark standards; Barnet et al., 2014). Research Design: Because children were not randomized to alternative childcare settings, we employ a principal stratification analysis (Frangakis & Rubin, 2002) which yields principal causal effects for two subgroups of children: those who enrolled in VPK but otherwise would have stayed home with a caregiver (home compliers) and those who enrolled in VPK but otherwise would have attended an alternative center-based care (center-based compliers). We observe each child i's treatment assignment, Z[subscript i] = 1 or Z[subscript i] = 0, and which care arrangement a child attends; home-based care S[subscript i] = hc, alternative center-based care S[subscript i] = cc, or VPK S[subscript i] = vpk. The combination of two treatment assignments and three possible care arrangements provides us with six strata. Although we cannot point-identify the effects for the home compliers and alternative center-care compliers, invoking several assumptions allows us to calculate the upper and lower bounds for each principal causal effect (Miratrix et al., 2018). These assumptions include SUTVA, the exclusion restriction, and no irrelevant alternatives (Feller et al., 2016). Data Collection and Analysis: Information about alternative childcare arrangements was collected through a parent survey administered during the Fall of the pre-k year. We combine survey responses with administrative data from the state to create three categories: VPK attenders, alternative center-based care attenders (Head Start center, private child center, home- based childcare center), and home-based care attenders (stayed at home with a caregiver). We explore effects on subtests of the Woodcock-Johnson III which was administered at several time points between pre-k entry through third grade. Estimation with principal stratification is simplified with the use of outcomes with sharp cut points. We standardize outcome measures by cohort using the mean and standard deviation of the control group, and then truncate the outcomes to include 3 standard deviations above and below the mean. Findings/Results: We find initial benefits of VPK for home compliers on most outcomes, but these effects fade rapidly and almost entirely disappear by first grade. Results for center compliers are wide with some indication of adverse effects. Conclusions: The intent-to-treat impacts of the original VPK study were puzzling. This study provides some evidence that the adverse impacts were experienced by the children who would have attended other types of center-based care, and that the VPK offer diverted children away from other types of center-based care that were higher quality than what VPK offered. However, children who would have stayed home still benefitted in the short-term, with no clear adverse effects in through third grade. These findings suggest that shifting children into a state-run program depends on the quality of the "business-as-usual" care.