COVID-19-Induced Educational Disruptions and Children’s Executive Functioning: A Longitudinal Cohort Study

Abstract

Literature suggests that the COVID-19 pandemic may have disrupted children’s executive functioning (EF) development, but most studies rely on caregiver reports, cross-sectional data, or comparisons across cohorts. We build on the existing literature with repeated, direct assessments of EF from longitudinal pre-post COVID data on a race-ethnically diverse cohort of elementary-aged children (N = 667) from low-income families. Random-intercept models estimate children’s growth in two key EF skills between the fall of kindergarten (2018) and fifth grade (2023) as a function of school closures. We also test for moderation in children’s growth trajectories by teachers’ reports of children’s compliance with remote learning expectations. Results indicate that children’s EF growth stagnated during school closures, resulting in an estimated 11–12 months of lost growth compared to pre-pandemic trends. Post-reopening, EF growth continued but at a 65–74% slower rate than pre-closures. Children who completed insufficient remote work demonstrated less stagnation in their inhibitory control/attention growth, which may have been driven by selection. Changes otherwise did not vary according to children’s level of participation in remote learning during school closures. Findings underscore the need for interventions to support children’s recovery of EF growth, as well as more research on the roles of school closures versus other pandemic-related stressors in the observed patterns.

The COVID-19 pandemic profoundly disrupted the lives of millions of children in the U.S. (Benner & Mistry, 2020; Slavin & Storey, 2020). After the World Health Organization declared COVID-19 a global pandemic on March 11, 2020, most schools across the nation abruptly pivoted from in-person, face-to-face instruction to various forms of online remote learning (Cucinotta & Vanelli, 2020; Donohue & Miller, 2020). In many districts, like the one that is the focus of this study, this period of school closures lasted for a year or longer (National Center for Education statistics, n.d.). During this time, not only did the format of children’s schooling shift from in-person to remote, but children’s overall exposure to instruction decreased relative to before school closures. In the current sample, for example, district guidelines in December 2020 stated that children received 90+ minutes of synchronous instruction daily, which represented a decrease in instructional time relative to the standard 7-hour in-person school day (personal communication, TPS Executive Director of Elementary and Early Childhood Education, December 1, 2020). Parallel to these profound educational disruptions, many children experienced disruptions to their home lives such as changes in daily routines, decreased social interactions, increased family dysfunction, and other sources of pandemic-related stress (Eales et al., 2021; Cassinat et al., 2021; Korzeniowski, 2023).

The potential impact of these disruptions on children’s learning and development continues to concern education and developmental researchers, policymakers, teachers, and parents. While most research on COVID-related disruptions and child outcomes has focused on declines in academic achievement (see Betthäuser et al., 2023; Cortés-Albornoz et al., 2023), a smaller body of evidence suggests that such disruptions may have also impacted children’s executive functioning skills (Chichinina & Gavrilova, 2022; Colvin et al., 2023; Ghanamah et al., 2023; González et al., 2022; Hanno et al., 2022; Navarro-Soria et al., 2023; Perry et al., 2023). However, most studies have measured executive functioning using caregiver report measures, which tend to show weak convergence with direct assessments of these skills (Tamm & Peugh, 2019; Toplak et al., 2013) and may be confounded with caregiver stress during the pandemic. Additionally, most such research was initiated post-pandemic, relying on comparisons of different cross-sections of children before and after COVID.

The current study builds on prior literature by characterizing trajectories of executive functioning skill growth before and after the period of COVID-induced school closures among a single cohort of children who were in first grade at the onset of the pandemic—an age when such skills are rapidly developing and particularly sensitive to environmental influences (Anderson, 2002). We extend past work by using repeated direct assessments of children’s executive functioning skills from kindergarten through fifth grade to assess how their trajectories changed from before to after the protracted period of school closures and across the first two years following the return to “business as usual” schooling. We further test whether children’s individual level of participation in remote learning when schools were closed—based on teachers’ reports of whether children met their expectations for contact with the teacher and remote work completion—moderated any changes in their executive functioning trajectories from pre- to post-COVID.

COVID-Related Disruptions and Child Executive Functioning

Executive functions refer to the constellation of higher-order cognitive skills required to regulate and control one’s attention and behavior in a goal-directed manner, including inhibitory and attentional control, cognitive flexibility, and working memory (Happaney et al., 2004; Miyake et al., 2000). These skills underlie and are strongly predictive of children’s academic achievement (Best et al., 2011; Mann et al., 2017), health and well-being (Cassidy, 2016), and long-term life success (Ahmed et al., 2021; Korzeniowski et al., 2021). Executive functioning skills are also vulnerable to disruptions from stress (Liston et al., 2009; Pechtel & Pizzagalli, 2011), and in some contexts, fostered by school experiences (Ansari et al., 2021; Bardack & Obradović, 2019; Burrage et al., 2008; Nguyen et al., 2020). Furthermore, some evidence suggests that the use of digital technology and screen media—tools that were required for participation in remote learning during school closures—may be adversely associated with children’s executive functioning outcomes (Lakicevic et al., 2025; Likhitweerawong et al., 2024; Willingham, 2025). Thus, it is reasonable to expect that children’s development of executive functioning may have stagnated in the wake of COVID-related disruptions (Korzeniowski, 2023).

Indeed, a handful of studies have found that the children who experienced COVID-induced school closures and community lockdowns demonstrated poorer executive functioning outcomes or slower executive functioning growth in the first year of the pandemic than prior same-age cohorts (Chichinina & Gavrilova, 2022; Colvin et al., 2023; González et al., 2022), or an increase in executive functioning difficulties in the months following the pandemic outbreak (Navarro-Soria et al., 2023; Perry et al., 2023). While various pandemic-related factors may have contributed to these declines, there is evidence that school closures in particular played a key role. For example, in a study of U.S. elementary school children, an analysis of repeated measurements suggested that during the spring of 2021, children’s executive functioning tended to be worse during periods of remote instruction than during periods of in-person learning (Hanno et al., 2022).

While these studies suggest that the pandemic—and school closures in particular—may have negatively impacted children’s executive functioning development, the literature would benefit from additional studies that use alternate measurement approaches and prospective longitudinal data. Most studies rely on caregiver reports of children’s executive functioning. However, caregiver reports and direct assessments of children’s executive functioning tend to show weak convergence, and scholars have argued that both types of measures should be used in conjunction to provide complementary information about children’s skills (Tamm & Peaugh, 2019; Toplak et al., 2013). Furthermore, it is possible that caregiver reports of children’s executive functioning skills during the pandemic were confounded by increases in caregiver stress.

Three studies have included direct assessments of children’s executive functioning skills, with mixed findings. Colvin et al (2023) found that average performance on a single working memory task among youth presenting at a psychiatric clinic was higher among those presenting at the clinic in the year prior to, as opposed to the year of, the pandemic onset. A study in Russia found that a cohort of children who entered school in 2017 demonstrated more kindergarten-year growth in their directly assessed cognitive flexibility and working memory—but not inhibitory control—than a cohort of children who entered kindergarten in 2019 and experienced COVID lockdowns that year (Chichinina & Gavrilova, 2022). However, a study that compared a cohort of children who experienced COVID-induced school closures during their preschool year (2019–2020) to a cohort who entered preschool after the onset of the pandemic (2020–2021) found that they demonstrated comparable improvement on a direct assessment of executive functioning between the fall and spring of preschool (Perry et al., 2023). Yet both cohorts in this study experienced COVID—albeit at different developmental stages—whereas the former two studies compared pre-COVID cohorts with post-COVID cohorts. Thus, the difference in findings may be a function of the differences in the comparison cohorts.

The current study provides a strong test of whether children’s trajectories changed from before to after COVID-induced school closures by measuring the same cohort of children repeatedly across the pre-post period. Further, we continue to follow children for two years after the return to full-time in-person schooling in September of 2021. Such evidence is important for understanding whether pandemic-related disruptions led to lasting impacts on children’s executive functioning development or if there has been evidence of recovery over time.

Variation in Children’s Remote Learning Participation

While school closures and the transition to remote learning appear to have played a role in pandemic-related executive functioning declines (Hanno et al., 2022), further research is needed to understand whether variation in children’s level of participation in remote learning during school closures moderated changes in their executive functioning. This line of inquiry is suggested by prior studies indicating that under ordinary circumstances, children’s level of participation in face-to-face learning affects their acquisition of executive functioning skills (Ansari et al., 2021; Fuhs et al., 2018). Thus, it is reasonable to expect that variation in children’s participation in remote learning similarly affected their executive functioning development. There is, in fact, evidence to suggest such variation occurred, as not all children complied with all requirements of remote instruction. For example, K-12 teachers in New York State reported that nearly 30% of their students did not complete assignments as requested during remote learning (Catalano et al., 2021).

Scholars have documented multiple barriers to participation in remote learning, particularly among children from low-resourced households, such as lack of reliable access to the internet and devices (Domina et al., 2021), as well as parents’ inability to keep up with the added demands of scaffolding and supporting their child’s engagement in remote learning activities (Garbe et al., 2020). Indeed, the shift to remote learning placed substantial added demands on parents’ time and cognitive resources. One study found that during the period of pandemic-related school closures, parents of elementary and middle school children spent more than double the usual time supporting their children’s learning (Balayar & Langlais, 2022). Notably, parents experienced these added demands alongside increases in other sources of stress related to the pandemic, such as loss of employment and income (Kalil et al., 2020), increases in household chaos and family dysfunction (Johnson et al., 2021a; Schmeer et al., 2023; Thomson et al., 2023), anxieties about oneself or loved ones contracting COVID (Peltz et al., 2021), and social isolation during community lockdowns (CTN; 2020a; Dawes et al., 2021).

These numerous challenges made it difficult—if not impossible—for many parents to provide the level of structure, scaffolding, and direct support required to ensure their child(ren) maintained consistent communication with their teachers, participated fully in remote learning activities, and regularly completed remote learning assignments (Garbe et al., 2020; Liu et al., 2022). For example, Singletary et al. (2022) found that household chaos and parental loneliness were negatively associated with parental time investment in children’s home-learning activities during the spring of 2020. Supporting children’s participation in distance learning may have been particularly challenging for parents in lower-income families, who faced disproportionate increases in material hardship (Karpman et al., 2020; CTN, 2022a) and mental health difficulties (CTN, 2020b) during the pandemic. This could explain why Catalano et al. (2021) found that teachers in high-needs districts reported lower levels of student participation during COVID-induced remote learning than teachers in other districts.

Of course, there was variation among low-resourced families in their experiences of health and material hardship and children’s concomitant participation in remote learning (Haskett et al., 2022; Johnson et al., 2021a), but the implications of this variation for children’s executive functioning trajectories after school closures are unknown. Perhaps children who demonstrated optimal participation in remote learning experienced less stagnation in their executive functioning growth during the period of school closures, rebounded to their pre-COVID rates of growth more quickly after schools re-opened, or both. It is also possible, however, that variation in remote learning participation is unrelated to the magnitude of stagnation or slowed rates of growth that children experienced. An empirical test of these possibilities is important for informing approaches to education in the event that future emergencies disrupt “business as usual” in U.S. schools.

The Current Study

The current study draws on a diverse longitudinal cohort of children from low-income families in Tulsa, Oklahoma who were in first grade when the pandemic hit and whose executive functioning skills were repeatedly assessed both before and after the period of COVID-induced school closures. With these data, we seek to address two questions. First, we test whether and how school closures during the COVID-19 pandemic changed children’s growth trajectories of executive functioning skill development. Second, we assess whether any change in children’s post-pandemic performance and growth rate varied according to their remote learning participation when schools were closed. To address limitations in the existing literature, we analyze repeated direct assessments of the same children’s executive functioning skills before and after the period of COVID-induced school closures. This allows us to test for both stagnation in growth during school closures—defined here as intercept changes reflecting lower scores following school reopening than would be expected based on those same children’s previous trends—as well as longer-term slowdown in executive functioning development—defined as slope changes reflecting growth in children’s scores occurring at a slower rate following school reopenings than that demonstrated before schools closed. Finally, we harness teacher reports of children’s level of participation during school closures to test whether having insufficient contact with one’s teacher or completing less remote work than requested moderated changes in children’s executive functioning growth trajectories. Importantly, the study includes a large portion of youth and families from minoritized and marginalized groups typically underrepresented in the developmental science literature yet reflective of the national landscape of public elementary school students in the United States (National Center for Education Statistics, 2023).

Method

Data Source and Sample

Data for the current study were drawn from an ongoing study of low-income children who have been followed since they were in preschool (2016). The Tulsa School Experiences and Early Development (SEED) Study initially recruited children enrolled in publicly funded preschool settings serving low-income children and families (household income was less than 185% of the federal poverty level or they received any public benefit within the last 12 months) in Tulsa, Oklahoma. When study children entered kindergarten in the fall of 2018, the Tulsa SEED Study recruited additional students from low-income families who had not attended preschool the prior year. All study children were in first grade at the onset of pandemic-related school closures, experienced second grade remotely (2020–2021) and returned to in-person school full-time in the fall of third grade (2021–2022), similar to a number of large urban school districts in the U.S. Children’s executive functioning skills were repeatedly assessed from the fall of kindergarten (2018) through the fall of fifth grade (2023).

Of the original study sample of children who had not withdrawn by the fall of kindergarten (N = 1,435), we selected the 833 children whose executive functioning skills were assessed both before and after COVID-induced school closures. (Typically, children stopped being assessed after they moved out of the district.) We then excluded 166 children whose second-grade teachers did not complete ratings of the child’s participation in remote learning. The analytic sample for the current study thus includes children with at least one executive functioning assessment before school closures (between the fall of 2018 and the fall of 2019), at least one executive functioning assessment after school closures (between the fall of 2021 and the fall of 2023), and teacher ratings of remote learning participation in second grade (N = 667). Compared to excluded cases, children who were included in the analytic sample were more likely to be Hispanic/Latinx (p < .001), less likely to be Black, non-Hispanic (p < .001) or White, non-Hispanic (p <.001), more likely to be dual language learners (DLLs; p < .001), less likely to have mothers who were single at the time of the child’s birth (p < .01), less likely to have a parent with more than a high school education (p < .01), and had mothers who were slightly older at the child’s birth (p < .05) and had slightly lower household incomes (p < .05). In the current sample, most children (59%) completed executive functioning assessments at all eight possible timepoints, and 88% were assessed at six or more timepoints.

As shown in Table 1, the sample was racially and ethnically diverse. Approximately 51% of children were Hispanic/Latinx¹, 20% were Black, 12% were White, 10% were Multiracial, 6% were American Indian or Alaskan Native, 1% were Asian American/Pacific Islander, and <1% were another race/ethnicity. Nearly half (49%) of children in the analytic sample were DLLs, meaning their parent reported that a language other than English was spoken inside the home. In this sample, approximately 94% of DLLs were Hispanic/Latinx, 2% were Asian American/Pacific Islander, 1% were Black, 1% were American Indian or Alaskan Native, 1% were White, and 1% were Multiracial. The average household income was $24,809, which is less than 100% of the federal poverty line for a family of four in 2018 ($25,100; ASPE, 2018). Mothers were on average 26 years old at the child’s birth, and just over half (52%) were single at the time of the child’s birth. On average, children were 5.53 years old at the fall of their kindergarten year, and 50% of children were female.

Table 1.

Sample Descriptive Statistics

	M/Prop.	SD	n
Child race/ethnicity
Hispanic/Latinx	.53		667
Black	.19		667
White	.11		667
Other race/ethnicity	.17		667
Child age in months at fall of preschool	66.38	3.71	737
Child is female	.50		667
Child is dual language learner	.50		667
Mother was single at child’s birth	.51		553
Mother’s age at child’s birth	26.26	6.15	548
Parent has more than a high school education	.37		464
Parent is employed (full- or part-time)	.67		443
Monthly household Income ($)	2050.45	1403.02	418
Child remote learning experiences
Less contact with teacher than requested	.45		666
Less work completed than teacher requested	.41		667

Note. Descriptive statistics are computed on unimputed data.

Procedures

Trained assessors collected repeated measurements of children’s executive functioning skills at the fall and spring of each year of the study, from kindergarten (2018–2019) through the fall of fifth grade (2023), except for the spring of first grade (2020) and the fall and spring of second grade (2020–2021), when school buildings were closed to in-person instruction. During the school day, each child was individually escorted out of their classroom to a separate location in the school where a trained research assistant administered a series of tasks. Each session lasted approximately 35 minutes. Information about family and household demographic characteristics used as covariates in the current analysis was drawn from a survey parents completed in the spring of the 3-year-old, preschool, or kindergarten year, depending on when the child enrolled in the study. Surveys were distributed via text message, email, and in children’s backpacks; parents received a $30 gift card for completing this survey. Between November of 2020 and January of 2021—during the period of COVID-induced remote learning—second grade teachers were sent a Qualtrics survey link for each participating study child in their classroom and asked to complete ratings of children’s level of participation in remote learning since the start of the school year (September 1, 2020). Each survey took approximately 6 minutes to complete, and teachers received $20-$60 in gift cards as compensation, depending on the number of participating children in their classroom. The University of Oklahoma-Tulsa and Georgetown University Institutional Review Boards reviewed and approved all study protocols.

Measures

Educational Disruptions Due to COVID-19

Pandemic-Related School Closures.

We constructed an indicator variable called “post-COVID” to capture the timing of child assessments relative to pandemic-related school closures. This variable was coded as a 0 for all timepoints before COVID-related school closures (i.e., prior to September 15, 2021) and a 1 for all timepoints following the return to full-time in-person learning (on or after September 15, 2021). Note that children were not assessed during the period of school closures; thus, all pre-COVID assessments occurred before March 15, 2020.

Insufficient Remote Learning Participation.

Children’s second grade teachers rated two dimensions of each child’s level of participation in remote learning during the fall 2020 semester: their contact with the child (1 = none at all, 2 = less than you requested, 3 = as much as you requested, 4 = more than you requested), and the amount of remote learning activities completed by the child (1 = none at all, 2 = less than you requested, 3 = a moderate amount, 4 = a lot, 5 = a great deal). We constructed two binary indicators of insufficient remote learning participation. First, we categorized children as having insufficient contact with their teacher if their teacher reported that they had less contact with the child than they requested or none at all (approximately 44% of children). Second, we categorized children as having insufficient remote work completion if their teacher reported that they completed less work than they requested or none at all (approximately 40% of children).

Child Executive Functioning

Children’s executive functioning skills were directly assessed using two tasks from the National Institute of Health Toolbox (NIHT) Cognition Battery (CB): The Flanker Inhibitory Control and Attention (FICA) test and the Dimensional Change Card Sort (DCCS) test (Zelazo et al., 2013). In the spring of third grade (2022), only the FICA test was administered; at all other timepoints, both tests were administered. Children were given an iPad and directed to complete a series of computerized tasks and games with stimuli designed to be engaging to young children. In kindergarten and first grade, children with Spanish as a home language received the Clinical Evaluation of Language Fundamentals-Fourth Edition (CELF-4) Sentence Structure subtest, a measure of receptive language, in both Spanish (Semel et al., 2006) and English (Semel et al., 2003). If the child failed the trial items in English or scored 6 or more points higher in Spanish than English, they completed the NIHT tasks in Spanish. If the child passed the English trial items and obtained a low raw score on the CELF-4 Sentence Structure subtest in both English and Spanish, the bilingual assessor evaluated their interactions with the child during rapport-building and used their discretion to decide whether to administer the NIHT tasks in Spanish. Among children in this sample who completed the NIHT at each timepoint, approximately 12% received the tasks in Spanish at the fall of kindergarten, 8% at the spring of kindergarten, and 8% at the fall of first grade. In the third and fourth grade, the NIHT tasks were administered in English with all children. For children with special needs, paraprofessionals or teaching assistants were invited to sit quietly with children to help them feel comfortable but were instructed not to provide any coaching to the child.

Inhibitory control and attention.

The FICA test measures children’s inhibitory control and attention. Children are asked to focus on a presented stimulus in the middle of the screen (namely, a fish) while inhibiting their attention to the stimuli flanking it. In a series of trials, children are asked to indicate the direction of the middle stimulus, which sometimes points in the same (congruent) direction as the stimuli flanking it, and sometimes points in the opposite (incongruent) direction of the flankers. Participants ages 3–7 who score greater than or equal to 90% on the first 20 trials are administered 20 additional trials with arrows instead of fish stimuli. Participants ages 8–85 proceed directly to the 20 trials with arrows. The FICA test included trial items at the fall of kindergarten; if children did not pass the trial items, their score was re-coded coded as 2 standard deviations below the sample mean (5% of children).

Cognitive flexibility.

The DCCS test measures children’s cognitive flexibility. Children are shown two target pictures that differ based on two dimensions (e.g., a yellow car and a blue boat) and are then asked to match a set of bivalent pictures to the target pictures, first according to one dimension (e.g., color), and then switching to the other (e.g., shape). Among children ages 3–7, those who get four out of five trials correct on both the pre- and post-switch trial blocks proceed to a block of 30 “mixed” trials in which they are required to actively change the dimension that is being matched. Participants ages 8–85 proceed directly into the mixed trials. The DCCS test included trial items at the fall of kindergarten; if children did not pass the trial items, their score was re-coded coded as 2 standard deviations below the sample mean (7% of children). Note that this measure was not administered in the Spring of third grade.

Scoring for both the FICA and DCCS tests are based on a combination of accuracy and reaction time. For each task, an accuracy score is computed that ranges in value from 0–5 (0.125 * number of correct responses out of 40). (Participants ages 8–85 automatically receive 20 correct response points on the FICA fish trials and 10 correct response points for the DCCS pre- and post-switch trial blocks; the measure developers determined that they typically score at the ceiling on these trials.) If the participant’s accuracy levels are greater than or equal to 80%, they also receive a reaction time score. (For the FICA, children ages 3–7 do not get a reaction time score unless they proceed to the “arrow” trials, and for the DCCS, children do not get a reaction time score unless they proceed to the “mixed” trials.) If a reaction time score is assigned, the accuracy and reaction time scores are combined and the computed score ranges in value from 0 to 10. These final computed scores are transformed into age-uncorrected standard scores (M = 100, SD = 15), which compare the performance of the test-taker to those in the entire NIHT nationally representative norming sample, regardless of age or any other variable. These scores provide a snapshot of the test-taker’s performance relative to the general U.S. population aged 3–85 and are recommended when monitoring performance over time (Slotkin et al, 2012; 2016). Both the NIHT Flanker and the DCCS demonstrate excellent test-retest reliability and convergent validity with young children (Read et al., 2022)

Covariates

Covariates were selected to increase the precision of our estimates and reduce omitted variable bias, on the basis of theory or empirical research suggesting that each covariate is associated with our dependent variables (executive functioning) and/or with both our independent variable (COVID disruption) and dependent variables (e.g., Barbee at al., 2025; Calvo & Bialystok; CTN, 2020; 2022a; Karpman et al., 2020; Klenberg et al., 2001; Sarsour et al., 2010). We coded child race/ethnicity as Hispanic/Latinx, Black, White, or Multiracial/another race (which, due to small sizes, combined children who were Asian American/Pacific Islander, Native American, or Multiracial). We coded children as DLLs if their parent reported that the household spoke a language other than or in addition to English at home. Parent education and employment were coded as binary variables capturing whether the parent had more than a high school education and whether they were employed full- or part-time as of the kindergarten year (2019), respectively. We drew household income from the kindergarten parent interview and log-transformed it in all analyses. Other covariates included mother’s age and whether she was unmarried at the child’s birth, child gender, and child age in months at the fall of the kindergarten year.

Analytic Strategy

To determine how time should be modeled, we first examined the shape of children’s pre-COVID growth trajectories by plotting the means of children’s scores on the FICA and DCCS between the fall of kindergarten (2018) and the fall of first grade (2019) and by comparing the results of models constrained to this time period that used linear versus quadratic specifications of time. Visual inspection of the plots and the non-significant quadratic time coefficients suggested that children’s growth in both executive functioning skills followed a linear trajectory prior to the onset of the pandemic. We thus proceeded under the analytic assumption that in the absence of COVID-induced disruptions, children’s growth in executive functioning skills during and after school closures would have continued to follow their prior linear trajectory.

We estimated two-level random intercept models with timepoints (Level 1) nested within children (Level 2) for each executive functioning outcome using the “mixed” command in Stata 18 (Robson & Pevalin, 2016; Singer & Willett, 2003). This approach allowed us to account for within-child correlations in assessment scores over time, while examining how children’s skills evolved before and after COVID-related school closures. In all models, we treated time as a continuous variable (months), centered at September 15, 2021, the date corresponding with the return to in-person learning.

Our first set of models examine children’s growth in executive functioning skills as a function of COVID-related school closures. Predictors at level 1 include time, the “post-COVID” indicator variable capturing whether the timepoint was before or after the period of COVID-related school closures, and the interaction between time and the post-COVID indicator. Level 2 predictors were time-invariant child and household characteristics, including child race/ethnicity, gender, DLL status, age at the fall of the preschool year, mother’s age and marital status at the child’s birth, parent education and employment, and the natural log of household income. All models included random intercepts to account for variability in children’s baseline skills, while treating the effects of time and COVID-related school closures as fixed predictors with consistent effects across individual children. These models can be summarized by the following equation:

$$Y_{ti}={\gamma }_{00}+{{\gamma }_{10}\left(\text{time}\right)}_{ti}+{{\gamma }_{20}\left(\text{postCOVID}\right)}_{ti}+{\gamma }_{30}({\text{time}\times \text{postCOVID})}_{ti}+{\sum \gamma }_{01-9}({\text{covariates})}_{ti}+{{\mu }_{0i}+r}_{ti}$$

In the above equation, $Y_{ti}$ represents the outcome score for child i at time t, and ${\gamma }_{00}$ is the overall model intercept (i.e., the mean value across all children when time = 0, or upon the return to in-person learning). The random portion of the intercept for child i is represented by ${\mu }_{0i}$. ${\gamma }_{10}$ represents the average monthly change in children’s scores prior to school closures. ${\gamma }_{20}$ represents the average change in children’s predicted scores from before to after the period of school closures (i.e., the shift in the intercept when time = 0). This is the term capturing suspected “stagnation” in executive functioning growth during the period of school closures. ${\gamma }_{30}$ represents the difference in children’s average monthly rate of growth from before school closures to after the return to in-person learning. This difference captures suspected “slowdown” occasioned by the pandemic. The simple slope following the period of school closures can be computed as the sum of ${\gamma }_{10}$ and ${\gamma }_{30}$. The term ${\sum \gamma }_{01-9}$ represents a vector of covariates on children’s average scores over time. Finally, $r_{ti}$ is the error term capturing the unexplained variation within children over time.

To examine whether children’s “stagnation” and/or “slowdown” in executive functioning skills varied according to their remote learning participation during school closures, our second and third set of models added a 3-way interaction between time, the post-COVID indicator, and each of our binary variables capturing incomplete remote learning participation (“disrupt”), respectively. These models can be summarized by the following equation:

$$Y_{ti}={\gamma }_{00}+{{\gamma }_{10}\left(\text{time}\right)}_{ti}+{{\gamma }_{20}\left(\text{postCOVID}\right)}_{ti}+{\gamma }_{30}({\text{time}\times \text{postCOVID})}_t+{{\gamma }_{01}\left(\text{disrupt}\right)}_{ti}+{{\gamma }_{11}\left(\text{time}\times \text{disrupt}\right)}_{ti}+{{\gamma }_{21}\left(\text{postCOVID}\times \text{disrupt}\right)}_{ti}+{{\gamma }_{31}\left(\text{time}\times \text{postCOVID}\times \text{disrupt}\right)}_{ti}+{\sum \gamma }_{02-010}({\text{covariates})}_{ti}{+{\mu }_0+r}_{ti}$$

In the above equation, ${\gamma }_{10}-{\gamma }_{30}$ are interpreted similarly to the first set of models, but specifically with respect to children whose level of contact with their teacher or completion of remote work during distance learning met their second-grade teachers’ expectations. ${\gamma }_{01}$ represents the average difference in the intercept associated with having insufficient student-teacher contact (Model 2) or insufficient work completion (Model 3) during school closures. ${\gamma }_{11}$ represents the average difference in children’s pre-COVID monthly rate of growth associated with the indicator of insufficient remote learning participation they experienced later. ${\gamma }_{21}$ represents the difference in the change in intercept from before to after the period of COVID-induced school closures associated with the indicator of insufficient remote learning participation they experienced during school closures. Finally, ${\gamma }_{31}$ represents the difference in the pre-school-closure to post-school-reopening change in children’s monthly rate of growth (i.e., the difference in the slope change) associated with the indicator of insufficient remote learning participation during school closures.

Rates of missing covariate data ranged from <1% for DLL status to 38% for household income, primarily due to parent survey non-response. Because the covariate missingness in this sample is likely non-random (e.g., children or households of parents who do not complete the survey may differ in systematic ways from those who do), and since non-random missing data can lead to biased parameter estimates, we chose to multiply impute missing covariate data (van Ginkel, 2020; Woods et al., 2023). Specifically, the ice command in Stata 18 created 25 multiply imputed data sets from a model that included all the child executive functioning assessment variables at each timepoint, the associated dates of assessment, indicators of insufficient remote learning participation, and child covariates. This technique—sometimes referred to as fully conditional specification—uses chained equations to fill in missing values by iteratively predicting them based on the observed values of other variables in the dataset (Royston, 2005). Missing values are initially replaced with starting values generated as random draws from the empirical distribution of the respective variable. For each variable with missing data, the algorithm fits a regression model using the other variables in the dataset as predictors and imputes (i.e., predicts) the missing values based on this model. These imputed values replace the initial starting values, and the process cycles through all variables with missing data, iteratively updating the imputed values a specified number of times (in this case, 10). In the current analysis, this entire iterative process was repeated 25 times to yield 25 imputed datasets. Model estimates were then obtained by analyzing each imputed dataset separately and combining the results using Rubin’s rules. We ran all models on the imputed data sets but used unimputed dependent variables in our analyses (Von Hippel, 2007).

Results

School Closures and Children’s Executive Functioning

Table 2 and Figure 1 display the results of multilevel models predicting children’s changes in executive functioning as a function of COVID-related school closures. In Table 2, the time coefficient represents children’s average pre-COVID monthly rate of growth in skills. The post-COVID coefficient in Table 2 represents the change in children’s predicted scores from before schools closed to after schools re-opened—what the current study refers to as “stagnation” during school closures. The coefficient for the interaction between time and post-COVID represents the change in children’s rate of growth from before schools closed to after they re-opened—what the current study refers to as “slowdown.”

Table 2.

Longitudinal growth models predicting children’s executive functioning outcomes from COVID-19-related school closures

	Inhibitory control and attention		Cognitive flexibility
	B	SE	B	SE
Time	1.02***	0.06	0.99***	0.06
Post-COVID	−11.49***	1.73	−11.76***	1.85
Time x Post-COVID	−0.66***	0.06	−0.73***	0.07
Constant	61.54***	8.22	61.62***	8.64
n	667		667

Notes. All models control for child race/ethnicity, gender, age at the fall of kindergarten, dual language learner status, mother’s age and marital status at child’s birth, parent education and employment, and the natural log of household income. Models include random intercepts to account for the nesting of timepoints in children. Missing covariates are multiply imputed.

^*p < .05;

^**p < .01;

^***p < .001

Figure 1.

Children’s executive functioning development before and after school closures.

Results suggest that children’s inhibitory control and attentional skills improved less during school closures than would be expected based on their pre-COVID trends (B = −11.49, SE = 1.73, p < .001). Furthermore, after schools re-opened, children’s inhibitory control and attentional skills continued to improve at a slower rate than prior to school closures (B = −0.66, SE = .06, p < .001). Specifically, before COVID, children’s rate of growth was equal to a standard score increase of approximately 1.02 points per months, but after COVID, it was equal to an increase of approximately 0.36 points per month. The results for cognitive flexibility are similar: children grew less during school closures than would be expected based on their pre-COVID trends (B = −11.76, SE = 1.85, p < .001), and after schools reopened, their rate of improvement was slower than before schools closed (B = −0.73, SE = .07, p < .001). Before COVID, children’s rate of growth in cognitive flexibility was equal to a standard score increase of approximately 0.99 points per months, but after COVID, it was equal to an increase of approximately 0.26 points per month. In sum, children experienced both learning stagnation and learning slowdown in both executive functioning skills.

Variation in Remote Learning Participation and Children’s Growth Trajectories

Tables 3 and 4 display the results of our second and third set of multilevel models predicting children’s growth in executive functioning as a function of each respective indicator of insufficient remote learning participation: less than requested contact with their teacher (Table 3) and less than requested remote work completion (Table 4). In Table 3, the non-significant coefficients on the interaction between the post-COVID indicator and the indicator for less contact than requested with the teacher, and the 3-way interaction between time, the post-COVID indicator, and the indicator for less contact than requested with the teacher, indicate that for both inhibitory control/attention and cognitive flexibility, whether children had sufficient contact with teachers during school closures was not associated with either a change in skill level following the period of school closures (stagnation), or with a change in the rate of growth in learning following the period of school closures (slowdown). Similarly, the insignificant coefficients on the interactions between time and the indicator for less contact than requested with the teacher suggest that there was no association between children’s pre-COVID growth rates and student-teacher contact during school closures. In other words, there is no evidence for selection into the indicator of insufficient remote contact by children’s pre-pandemic trajectories.

Table 3.

Longitudinal growth models predicting children’s executive functioning outcomes from contact with their teacher during COVID-19 school closures

	Inhibitory control and attention		Cognitive flexibility
	B	SE	B	SE
Time	1.05***	0.08	1.01***	0.08
Post-COVID	−13.42***	2.34	−13.43***	2.51
Time x Post-COVID	−0.68***	0.09	−0.77***	0.09
Less contact than teacher requested	−3.71	3.44	−4.30	3.64
Time x Less contact than teacher requested	−0.05	0.12	−0.05	0.12
Post-COVID x Less contact than teacher requested	4.16	3.47	3.88	3.71
Time x Post-COVID x Less contact than teacher requested	0.03	0.13	0.09	0.14
Constant	63.86***	8.42	64.61***	8.87
n	666		666

Notes. All models control for child race/ethnicity, gender, age at the fall of kindergarten, dual language learner status, mother’s age and marital status at child’s birth, parent education and employment, and the natural log of household income. Models include random intercepts to account for the nesting of timepoints in children. Missing covariates are multiply imputed.

^*p < .05;

^**p < .01;

^***p < .001

Table 4.

Longitudinal growth models predicting children’s executive functioning outcomes from remote work completion during COVID-19 school closures

	Inhibitory control and attention		Cognitive flexibility
	B	SE	B	SE
Time	1.09***	0.08	0.93***	0.08
Post-COVID	−14.52***	2.25	−10.28***	2.41
Time x Post-COVID	−0.71***	0.08	−0.70***	0.09
Less work than teacher requested	−7.43*	3.47	2.73	3.69
Time x Less work than teacher requested	−0.16	0.12	0.13	0.12
Post-COVID x Less work than teacher requested	7.50*	3.51	−3.62	3.76
Time x Post-COVID x Less work than teacher requested	0.10	0.13	−0.09	0.14
Constant	65.63***	8.35	61.03***	8.83
n	667		667

Notes. All models control for child race/ethnicity, gender, age at the fall of kindergarten, dual language learner status, mother’s age and marital status at child’s birth, parent education and employment, and the natural log of household income. Models include random intercepts to account for the nesting of timepoints in children. Missing covariates are multiply imputed.

^*p < .05;

^**p < .01;

^***p < .001

In Table 4, the non-significant interaction terms similarly indicate that variation in children’s remote work completion during school closures was not associated with either a change in cognitive flexibility level following the period of school closures (stagnation), or with a change in the rate of growth in inhibitory control/attention or cognitive flexibility following the period of school closures (slowdown). However, one significant interaction emerged for inhibitory control/attention: The significant coefficient on the interaction between the post-COVID indicator and the indicator for less remote work than teacher requested (B = 7.50, SE = 3.51, p < .05) suggests that children who did not complete work sufficiently during school closures experienced less stagnation in their inhibitory control/attention growth than children who completed sufficient remote work during school closures. Yet there is also evidence for selection into the insufficient remote work completion by children’s pre-COVID children’s inhibitory control/attention (Figure 2). Specifically, children who would go on to have insufficient remote work completion had significantly lower intercepts in their inhibitory control/attention before COVID than children who would go on to have sufficient remote work completion (B = −7.43, SE = 3.47, p < .05). These children also had slower pre-COVID growth rates in their inhibitory control/attention, although this difference was not statistically significant (B = −.16, SE = .12, p = .16).

Figure 2.

Children’s inhibitory control/attention development before and after school closures by remote work completion during school closures.

Sensitivity Tests

Because we excluded a substantial number of children from our analytic sample due to missing data on remote learning participation, we re-estimated all models using imputed values on these two variables (n = 833). The results were substantively identical to our primary models (Appendix Tables 1–3). Additionally, because we excluded a large number of children whose executive functioning was assessed only before or after school closures, we re-estimated models with all children who had any executive functioning assessments between the fall of 2018 and the fall of 2023 using imputed values for all missing timepoints (n = 1,131). Again, the results were unchanged (Appendix Tables 2–6).

To account for the possibility that dichotomizing our remote learning participation variables may have obscured results by reducing variation in the moderators, we also ran a series of sensitivity tests with the original continuous measures in our models, and the results were substantively unchanged. Finally, to allow for the possibility that the effects of COVID-induced educational disruptions varied across individuals, we also tested models with random slopes for time, the post-COVID indicator, the insufficient remote learning disruption indicators, and the interaction terms. Again, the results did not change.

Discussion

This study was designed to examine how children’s executive functioning skills— key outcomes known to affect learning, benefit from being in a classroom setting (Ansari et al., 2021; Burrage et al., 2008), and be disrupted by stress (Blair et al., 2011; Johnson et al., 2021b; Shields et al., 2016)— were affected by the COVID pandemic. Scholars have speculated that the COVID pandemic and attendant transition to remote learning may have had negative implications for children’s executive functioning (Korzeniowski, 2023; Hanno et al., 2022; Perry et al., 2023). Our study makes an important contribution to the literature by testing this question empirically using multiple repeated direct assessments of children’s skills among a single cohort both before and after school closures. We modeled children’s growth in two key executive functioning skills that were measured repeatedly before and after the period of school closures, attending to changes in both the model intercept—representing children’s stagnation in the amount of growth during school closures—as well as changes in the slope—representing children’s slowdown in their rate of growth during the years after schools re-opened versus before they closed. We also explored whether pandemic-related stagnation and slowdown in children’s trajectories varied as a function of children’s level of participation in remote learning during school closures, defined as whether they met their teacher’s expectations for remote contact and remote work completion.

Our findings suggest that during school closures, children’s executive functioning development stagnated—that is, children improved less than would be expected according to their pre-COVID growth rates. Specifically, between March 2020 and September 2021 (first and second grade), children “lost” the equivalent of approximately 11–12 months of growth in inhibitory control and attention and cognitive flexibility that would have been expected had they continued to grow at their pre-COVID rates. Furthermore, during the two years after schools re-opened (third and fourth grade), children’s executive functioning skills continued to improve at less than half the rate – 65% to 74% slower – than prior to school closures. While it is possible that children’s executive functioning development normally slows down between kindergarten and fifth grade, there is reason to believe that the degree of growth deceleration suggested by our models is greater than what would be expected under normal conditions. For example, findings from a pediatric validation study of the FICA and DCCS with children ranging from 3–15 years of age suggest that children’s growth between ages 7 and 8—which correspond with our cohort’s remote learning year—should be, if anything, slightly faster than their growth between ages 5 and 6—which correspond with our pre-COVID assessments (Zelazo et al., 2013, Figure 3). Furthermore, findings from a meta-analysis suggest that children’s development of frontal lobe functions follows a linear trajectory between ages 5–8 and slows down only slightly—by about 25%—between ages 8 and 11 (Romine & Reynolds, 2005, Table 2; Figure 2). Thus, it is reasonable to speculate that the stagnation and sustained slowdown in executive functioning development observed in our sample were a function of COVID-related disruptions.

Our findings are largely consistent with prior studies that have employed parent or teacher report of child executive functioning (Ghanamah et al., 2023; González et al., 2022; Hanno et al., 2022; Navarro-Soria et al., 2023; Perry et al., 2023) or compared different cohorts before and after the pandemic (Chichinina and Gavrilova, 2022; Colvin et al., 2021; González et al., 2022). However, there are some discrepancies with prior literature. For example, Chichinina and Gavrilova (2022) found that children’s cognitive flexibility and working memory—but not inhibition—were negatively impacted by the pandemic, whereas the current study found that children’s inhibitory control and cognitive flexibility were both affected. However, that study was conducted in Russia, and the authors speculated that their findings for inhibitory control were culturally specific to caregiving norms and social expectations in the Russian context (Chichinina & Gavrilova, 2022). It’s possible that pandemic-related disruptions in the U.S. context—particularly for our predominantly minoritized, low-income sample—led to different impacts on children’s inhibition.

Our findings are also partially inconsistent with findings from Perry and colleagues (2023). Although they found evidence for declines in teacher-rated executive functioning for children who experienced school closures during preschool—which is consistent with our findings of stagnation in directly assessed skill development—their sample demonstrated the opposite pattern in directly assessed executive functioning—one of continued growth that was on par with the preschool-year growth of a different cohort who entered preschool the following year (Perry et al., 2023). Yet the lack of multiple pre-COVID assessments precluded researchers from examining whether these gains were as large as would be expected based on these children’s pre-COVID learning trends. The current study—which makes use of multiple pre-COVID direct assessments—found that children’s executive functioning skills improved during the period of school closures, but to a much lesser extent than would be expected based on those same children’s pre-COVID trends. Furthermore, the comparison cohort in Perry et al. (2023) was assessed post-COVID—meaning they too likely experienced shocks associated with the pandemic onset, albeit prior to school entry.

It is possible that to some extent, COVID-related disruptions in children’s executive functioning development were not caused by school closures per se, but rather other sources of stress and lifestyle changes induced by COVID writ large. Indeed, studies show striking increases in maternal distress (Eales et al., 2021; CTN; 2021a, 2021b; 2022b) and household chaos (Cassinat et al., 2021) associated with the pandemic, both of which are known to be associated with children’s executive functioning (Andrews et al., 2021; Hughes et al., 2013; Vernon-Feagans et al., 2016). Studies have also identified pandemic-related changes in other factors linked to children’s EF, such as decreases in social interactions with peers (Larivière-Bastien et al., 2022), worsened sleep quality and high rates of sleep disturbances (Sharma et al., 2021), reductions in physical activity (Tulchin-Francis et al., 2021), and increases in screen media usage (Eales et al., 2021b; Tabullo et al., 2023). Nevertheless, the findings of Hanno and colleagues (2022)—that parents rated their child’s executive functioning as lower during periods of remote learning than during periods of in-person instruction—implicate educational disruptions in and of themselves in children’s pandemic-related executive functioning declines.

Some factors are difficult to tease apart from school closures and the resulting transition to remote learning. For example, virtual schooling inherently required children to engage with screens as a mean of receiving instruction, and prior literature suggests that screen time may be negatively associated with executive functioning (McMath et al., 2023; Vohr et al., 2021). Yet there is also evidence that children’s non-school-related screen media use increased from pre-pandemic to post-onset (Eales et al., 2021), and that children younger than school age—who had no online schooling requirements—were exposed to more screen time during COVID lockdowns than before (Bergmann et al., 2022). More research is needed to disentangle the impact of school closures from other pandemic-related stressors and lifestyle changes in order to direct pandemic recovery initiatives towards strategies and targets with the most promise for improvement. Additionally, future studies on the role of school closures should seek to isolate the importance of changes in instructional mode versus related changes in instructional time.

Moderation of Outcomes by Remote Learning Participation

Although, on average, children experienced post-COVID stagnation and slowdown in their executive functioning development, these findings may have varied by individual differences in children’s remote learning participation during school closures. We find that children who completed less remote work than their teacher requested during school closures demonstrated less stagnation in their inhibitory control and attention growth during school closures than children who completed sufficient remote work. However, this finding could be due at least in part to selection into insufficient remote work completion by pre-COVID inhibitory control/attention skills. Children who did not meet their teacher’s expectations for remote work completion began school closures with significantly lower levels of inhibitory/control attention and marginally slower pre-COVID growth rates. These children may have had “less to lose” in their inhibitory control/attention skills at the start of school closures than their peers who completed sufficient remote work. Yet even if children who complied with their teacher’s expectations for remote work completion were outperforming their peers before COVID, school closures may have eliminated this advantage.

Aside from this, children’s stagnation and slowdown in executive functioning did not vary according to whether they had sufficient student-teacher contact or sufficiently completed work during remote schooling. This is somewhat surprising considering the evidence that children’s in-school attendance and participation predict their acquisition of executive functioning skills (Ansari et al., 2021; Fuhs et al., 2018). The overall lack of moderation of children’s executive functioning trajectories by their remote learning participation suggests several possibilities. First, this may suggest—as alluded to earlier—that the stagnation and slowdown we observed in children’s executive functioning development were primarily a function of other pandemic-related lifestyle disruptions and sources of stress as opposed to educational disruptions specifically. Another possibility, however, is that school closures were so disruptive to children’s executive functioning development that not even maximal remote learning participation was protective. Indeed, even when children participated exactly as their teachers requested, they still were typically exposed to far less instruction during an average remote school day than they were during a pre-pandemic in-person school day. Finally, both of these possibilities could also be true: school closures themselves may have been harmful to children’s executive functioning primarily insofar as they produced cascading impacts on things like family stress, sleep, and screen time, which may have in turn served as the primary “active ingredients” in executive functioning declines, irrespective of remote learning participation (Beaugrand et al., 2023; Polizzi et al., 2021; Tabullo et al., 2023).

Limitations

The current study has several limitations that should be noted alongside its contributions. First, our measure of remote learning disruptions was captured in the fall of second grade, which may not accurately characterize children’s earlier engagement in remote learning when schools first closed during the spring of first grade, or their later engagement as remote learning continued into the spring of second grade. Teacher reports may also not be sufficiently sensitive to capture other types of remote learning disruptions that may matter for children’s outcomes, such as children’s level of attention and engagement during remote lessons. Future studies should test the role of remote learning participation in children’s executive functioning declines using alternative measures of remote learning participation. Finally, our sample was entirely low-income, homogenous in age, and drawn from a single school district in Tulsa where most schools remained closed for at least one full year. Future studies should seek to replicate these findings with other cohorts of children of varying socioeconomic status and grade level, and who experienced a variety of school closure lengths throughout the U.S.

Conclusion

This study provides new insights into the executive functioning growth trajectories of a single cohort of low-income children from pre- to post-pandemic as a function of school closures. By leveraging repeated direct assessments across kindergarten (2018) through fifth grade (2023), we identified marked stagnation in children’s inhibitory control and attention and cognitive flexibility development during school closures and sustained deceleration after schools reopened. However, contrary to our expectations, we found little evidence that children’s level of participation in remote learning moderated these changes. Our findings underscore the profound and lasting impacts of pandemic-related disruptions on executive functioning skill development and highlight the need for targeted interventions to support children’s recovery of skill growth (Korzeniowski, 2023). Further research is needed to disentangle the unique contributions of educational disruptions, family stress, and other lifestyle changes to the observed patterns, and to inform policies aimed at mitigating the impact of future crises on children’s executive functioning development.

Public Significance Statement:

This study suggests that the COVID-19 pandemic and associated period of school closures led to disruptions in children’s development of executive functioning, the cognitive skills that enable goal-directed behavior. These changes showed minimal variation according to children’s level of participation in remote learning during school closures.

Appendix Tables

Appendix Table 1.

Longitudinal growth models predicting children’s executive functioning outcomes from COVID-19-related school closures (sample expanded to include children missing teacher reports of remote learning participation)

	Inhibitory control and attention		Cognitive flexibility
	B	SE	B	SE
Time	1.03***	0.05	1.01***	0.06
Post-COVID	−11.31***	1.55	−12.16***	1.67
Time x Post-COVID	−0.67***	0.06	−0.76***	0.06
Constant	61.47***	7.71	64.91***	7.93
n	833		833

Notes. Sample includes children with at least one executive functioning assessment both before and after school closures. All models control for child race/ethnicity, gender, age at the fall of kindergarten, dual language learner status, mother’s age and marital status at child’s birth, parent education and employment, and the natural log of household income. Models include random intercepts to account for the nesting of timepoints in children. Missing covariates are multiply imputed.

^*p < .05;

^**p < .01;

^***p < .001

Appendix Table 2.

Longitudinal growth models predicting children’s executive functioning outcomes from imputed contact with their teacher during COVID-19 school closures

	Inhibitory control and attention		Cognitive flexibility
	B	SE	B	SE
Time	1.05***	0.07	1.02***	0.08
Post-COVID	−12.91***	2.2	−13.52***	2.42
Time x Post-COVID	−0.68***	0.08	−0.78***	0.09
Less contact than teacher requested	−3.33	3.34	−3.63	3.73
Time x Less contact than teacher requested	−0.04	0.11	−0.02	0.12
Post-COVID x Less contact than teacher requested	3.5	3.37	3.06	3.76
Time x Post-COVID x Less contact than teacher requested	0.03	0.12	0.05	0.14
Constant	63.60***	7.85	67.62***	8.09
n	833		833

Notes. Sample includes children with at least one executive functioning assessment both before and after school closures. All models control for child race/ethnicity, gender, age at the fall of kindergarten, dual language learner status, mother’s age and marital status at child’s birth, parent education and employment, and the natural log of household income. Models include random intercepts to account for the nesting of timepoints in children. Missing covariates and remote learning participation indicators are multiply imputed.

^*p < .05;

^**p < .01;

^***p < .001

Appendix Table 3.

Longitudinal growth models predicting children’s executive functioning outcomes from imputed remote work completion during COVID-19 school closures

	Inhibitory control and attention		Cognitive flexibility
	B	SE	B	SE
Time	1.09***	0.07	0.96***	0.08
Post-COVID	−14.22***	2.07	−10.88***	2.3
Time x Post-COVID	−0.71***	0.08	−0.73***	0.08
Less work than teacher requested	−7.14*	3.27	2.14	3.7
Time x Less work than teacher requested	−0.16	0.11	0.12	0.12
Post-COVID x Less work than teacher requested	7.07*	3.28	−3.05	3.76
Time x Post-COVID x Less work than teacher requested	0.1	0.12	−0.08	0.13
Constant	65.36***	7.88	64.54***	8.1
n	833		833

Notes. Sample includes children with at least one executive functioning assessment both before and after school closures. All models control for child race/ethnicity, gender, age at the fall of kindergarten, dual language learner status, mother’s age and marital status at child’s birth, parent education and employment, and the natural log of household income. Models include random intercepts to account for the nesting of timepoints in children. Missing covariates and remote learning participation indicators are multiply imputed.

^*p < .05;

^**p < .01;

^***p < .001

Appendix Table 4.

Longitudinal growth models predicting children’s imputed executive functioning outcomes from COVID-19-related school closures (sample further expanded to include all children with at least one assessment before or after school closures)

	Inhibitory control and attention		Cognitive flexibility
	B	SE	B	SE
Time	1.21***	0.05	1.25***	0.05
Post-COVID	−15.17***	1.57	−17.79***	1.64
Time x Post-COVID	−0.85***	0.05	−1.00***	0.06
Constant	58.62***	7.32	65.85***	8.13
n	1131		1131

Notes. Sample includes children with at least one executive functioning assessment before or after school closures. All models control for child race/ethnicity, gender, age at the fall of kindergarten, dual language learner status, mother’s age and marital status at child’s birth, parent education and employment, and the natural log of household income. Models include random intercepts to account for the nesting of timepoints in children. Missing covariates and dependent variables are multiply imputed.

^*p < .05;

^**p < .01;

^***p < .001

Appendix Table 5.

Longitudinal growth models predicting children’s imputed executive functioning outcomes from imputed contact with their teacher during COVID-19 school closures

	Inhibitory control and attention		Cognitive flexibility
	B	SE	B	SE
Time	1.21***	0.07	1.31***	0.08
Post-COVID	−16.07***	2.21	−20.39***	2.49
Time x Post-COVID	−0.86***	0.08	−1.07***	0.09
Less contact than teacher requested	−1.97	3.47	−6.03	4.16
Time x Less contact than teacher requested	−0.01	0.12	−0.12	0.14
Post-COVID x Less contact than teacher requested	1.99	3.4	5.71	4.25
Time x Post-COVID x Less contact than teacher requested	0.01	0.13	0.14	0.14
Constant	60.05***	7.54	69.65***	8.04
n	1131		1131

Notes. Sample includes children with at least one executive functioning assessment before or after school closures. All models control for child race/ethnicity, gender, age at the fall of kindergarten, dual language learner status, mother’s age and marital status at child’s birth, parent education and employment, and the natural log of household income. Models include random intercepts to account for the nesting of timepoints in children. Missing covariates, remote learning participation indicators, and dependent variables are multiply imputed.

^*p < .05;

^**p < .01;

^***p < .001

Appendix Table 6.

Longitudinal growth models predicting children’s imputed executive functioning outcomes from imputed remote work completion during COVID-19 school closures

	Inhibitory control and attention		Cognitive flexibility
	B	SE	B	SE
Time	1.28***	0.07	1.19***	0.07
Post-COVID	−18.12***	2.11	−16.22***	2.29
Time x Post-COVID	−0.91***	0.08	−0.95***	0.08
Less work than teacher requested	−7.60*	3.73	2.56	3.79
Time x Less work than teacher requested	−0.18	0.13	0.15	0.12
Post-COVID x Less work than teacher requested	7.07	3.79	−3.66	3.9
Time x Post-COVID x Less work than teacher requested	0.14	0.14	−0.13	0.13
Constant	63.04***	7.6	65.78***	8.17
n	1131		1131

Notes. Sample includes children with at least one executive functioning assessment before or after school closures. All models control for child race/ethnicity, gender, age at the fall of kindergarten, dual language learner status, mother’s age and marital status at child’s birth, parent education and employment, and the natural log of household income. Models include random intercepts to account for the nesting of timepoints in children. Missing covariates, remote learning participation indicators, and dependent variables are multiply imputed.

^*p < .05;

^**p < .01;

^***p < .001