CoChat
Sign Up
Cited by 143Youth Substance Use and School AttendanceSchool Choice and PerformanceEducational Outcomes and Influences

The short- and long-run impacts of secondary school absences

Open access
Liu, Jing, Lee, Monica, Gershenson, Seth·2021·Journal of Public Economics
DOI 10.1016/j.jpubeco.2021.104441
Open this paper in CoChat — free account

Abstract

Abstract We provide novel evidence on the causal impacts of student absences in middle and high school on state test scores, course grades, and educational attainment using a rich administrative dataset that tracks the date and class period of each absence. We use two similar but distinct identification strategies that address potential endogeneity due to time-varying student-level shocks by exploiting within-student, between-subject variation in class-specific absences. We also leverage information on the timing of absences to show that absences that occur after the annual window for state standardized testing do not affect test scores, providing a further check of our identification strategy. Both approaches yield similar results. We find that absences in middle and high school harm contemporaneous student achievement and longer-term educational attainment: On average, missing 10 classes reduces math or English Language Arts test scores by 3–4% of a standard deviation and course grades by 17–18% of a standard deviation. 10 total absences across all subjects in 9th grade reduce both the probability of on-time graduation and ever enrolling in college by 2%. Learning loss due to school absences can have profound economic and social consequences.

Integrity signals

Registry

Crossref · 10.1016/j.jpubeco.2021.104441

Cites

49 of 49 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Claims the paper makes

0 of 24 cited sources checked · 0 supported · 0 flagged

  • Findingown data

    The effects of absences are approximately linear.

    “The effects of absences are approximately linear and similar in magnitude across student and school subgroups.”
    Quote in the paperNot a substantial claim
  • Findingown data

    Spring-semester absences are substantially more harmful to student achievement than fall-semester absences.

    “We find that absences in the spring semester are far more detrimental than those in the fall semester.”
    • Also stated in: § 1 Introduction · hedged
      “In contrast, absences in the fall semester appear to matter very little, if at all.”
    Quote in the paperSubstantial claim
  • Findingown data

    Missing 10 math classes in the spring semester lowers math standardized test scores by 7% of a standard deviation.

    “Our results show that on average, missing 10 math classes in the spring semester reduces math scores on state standardized tests by 7% of a standard deviation”
    Quote in the paperSubstantial claim
  • Findingown data

    Having 10 absences in the spring semester lowers course grades by 19% of a standard deviation.

    “Similarly, having 10 absences in the spring semester reduces course grades by 19% of a standard deviation.”
    Quote in the paperSubstantial claim
  • Findingown data

    The gaps in absences across subjects remain fairly stable as time passes.

    “Second, these between-subject differences are relatively stable over time.”
    Quote in the paperSubstantial claim
  • May combine two claims. 46
  • Classified as probably not a substantial claim. 17
  • The quote only partly states this claim. 1

  • May combine two claims.

    Findingown data

    The middle school math pattern extends to high school math and to ELA, where results are almost the same (Panel B of Table 4 and Appendix Table A2).

    “Panel B of Table 4 and Appendix Table A2 show nearly identical results in high school math and in ELA courses, respectively.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    Almost all grade-specific subject-absence coefficients are insignificant except for math versus science, which suggests student preferences do not shift systematically with grade progression.

    “Most of the coefficients are insignificant other than when comparing math with science, suggesting that students' preferences do not change systematically as they progress.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    Differences in absences across subjects are generally small and not statistically distinguishable from zero.

    “First, between-subject differences tend to be small in magnitude and statistically indistinguishable from zero.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    In Column 2, ten spring absences lower achievement by 9.5% of a standard deviation, a bit larger than the 7.0% preferred Table 4 estimate.

    “For example, as Column 2 shows, having 10 spring absences reduces student achievement by 9.5% of a standard deviation, a slightly bigger estimate than the coefficient from our preferred model in Table 4 (7.0%).”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    For middle school, fall absences show a small negative effect that is only marginally significant; in Column 1, ten absences reduce math scores by roughly 1% of a standard deviation.

    “The effect of fallsemester absences is negative and marginally significant but small in magnitude, suggesting that having 10 absences lowers test scores by about 1% of a standard deviation.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    Within-student-year variation in subject-specific absences is substantial; math and ELA absences correlate about .85 but not perfectly, with two-thirds of student-year observations having different counts.

    “there is considerable withinstudent-year variation in subject-specific absences: While highly correlated (with a .85correlation coefficient), they are not perfectly correlated, as two-thirds of student-year observations have a different number of math and ELA absences.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    Absences in math and ELA show almost the same association with prior math scores, which suggests their differences are not driven by math ability or preferences.

    “The two plots are nearly identical, suggesting that math and ELA absences differ for reasons unrelated to math ability or preferences (which are loosely captured by lagged math performance).”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    Using student-year fixed effects instead of separate student and year fixed effects reduces the coefficients somewhat, but they still match Table 4 closely; ten spring absences lower math scores by roughly 6% of a standard deviation.

    “The coefficients shrink a bit but are still closely aligned with those in Table 4: Having 10 spring absences reduces students' math test scores by about 6% of a standard deviation.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    Column 6's proxy model adjusts for ELA absences, and Appendix Table A4 indicates the estimate holds when other classes' absences, including physical education, are used.

    “The estimate reported in Column 6 controls for ELA absences, though Appendix Table A4 shows that this result is robust to using absences in other classes, including physical education.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    The preferred Column 6 proxy model estimates that ten spring absences reduce scores by 7% of a standard deviation, with strong statistical significance.

    “The point estimate in the preferred proxy-variable model of Column 6 is 7% of a standard deviation and is strongly statistically significant.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingown data

    The first stacked specification, with separate student and year fixed effects plus subject interactions, produces estimates close to Table 4.

    “The results are very similar to those in Table 4.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    Across specifications, ten fall absences shift math scores by 1 to 3% of a standard deviation, and in the preferred Column 6 proxy model the estimate is not statistically distinguishable from zero.

    “Depending on the model specification, the effect of having 10 fall-semester absences ranges from 1 to 3% of a standard deviation and is statistically indistinguishable from zero in the preferred proxy-variable model of Column 6.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    The two-year college enrollment estimate is double the four-year estimate and is not driven by unobservable selection, since its Oster bound nearly equals the point estimate.

    “Indeed, this is exactly what we see, as the point estimate in Column 5 is twice as large as that in Column 4 and is not driven by selection on unobservables: The Oster bound is nearly identical to the point estimate.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Findingown data

    The average student has 1.7 one-day absences and 0.7 two-day absence spells.

    “On average, a student has 1.7 absences that only last for one day and 0.7 absence spells that last for two days.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    Absence effects are comparable for boys and girls and across racial/ethnic groups, but black students show larger harm and white students' estimate is imprecise.

    “The effects are similar for boys and girls and for students from different racial and ethnic backgrounds, though the harm for black students is larger than for other groups, and the effect for white students is imprecisely estimated.”
    Quote in the paperNot a substantial claim

  • Classified as probably not a substantial claim. May combine two claims.

    Findingown data

    The linear, quadratic, and nonparametric fitted lines in Figure 8 largely coincide, although nonparametric estimates are noisy at high absence counts.

    “Figure 8 plots fitted regression lines from the linear, quadratic, and nonparametric models, which approximately overlap with each other, though the nonparametric estimates are fairly noisy for higher absence counts, as there are relatively few students in those bins.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    For college enrollment, the absolute effects are smaller but still statistically significant and proportionally similar given the lower base rate.

    “Columns 2 and 3 show that these absolute effects are smaller for college enrollment but remain statistically significant and in percentage terms are similarly sized effects, given the lower base rate of college enrollment.”
    Quote in the paperSubstantial claim

  • The quote only partly states this claim.

    Findingown data

    Across spells of different lengths, absence effects are roughly similar in size.

    “Still, the effects of absences are fairly similar in magnitude, especially when comparing across multi-way spells of different lengths.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    A single isolated absence hurts more than an additional absence within a longer spell, which fits fixed catch-up costs or scale economies in making up work.

    “The effect of a stand-alone absence is bigger than the marginal effect of an absence that is part of a longer absence spell; this finding is consistent with the notion of fixed costs to catching up following any absences and/or economies of scale in making up missed work.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingreasoning

    Week 21-22 absences have a near-zero, insignificant coefficient, which is reassuring since those absences likely happened after testing.

    “Finally, we see an insignificant and close-to-zero coefficient estimate for week 21-22 absences, which is reassuring because those absences almost certainly occurred after the test was taken.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingreasoning

    Absences later in the year are more detrimental to course grades and exam scores.

    “Because absences later in the year are more detrimental to course grades and exam scores”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    The four-year college enrollment effect is smaller and probably reflects unobservable selection, because even 60% selection on unobservables can explain away the -0.02 estimate.

    “The effect of absences on four-year college enrollments is not only smaller but also likely due to selection on unobservables, as even a modest (60%) amount of selection on unobservables is enough to explain away the point estimate of -0.02.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    Most absences are single-day events, and the majority belong to spells lasting one or two days.

    “The modal absence is a one-off event (i.e., part of a one-day spell).Most absences are part of one-or two-day spells.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    For weeks 1-14 and 15-16, absence coefficients are about -0.05 and significant, but somewhat smaller than the baseline spring estimate.

    “Estimated coefficients for absences in weeks 1-14 and 15-16 are both around -0.05 and are statistically significant though slightly smaller than the baseline estimate that restricts the effects of all spring-semester absences to be the same.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    Absence harm does not differ systematically by school poverty level; it appears similar across school types.

    “We do not see any systematic differences between more-and less-advantaged schools; absences seem similarly harmful across school types.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    Although school-level harm varies somewhat, it does not follow student demographic or socioeconomic lines systematically.

    “Harm does vary somewhat across schools, but it does not vary systematically by student demographic or socioeconomic background.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    Absence coefficients for weeks 17-18 and 19-20 are nearly double, implying that missing class during the testing window hurts more than missing earlier in spring.

    “The coefficient estimates for absences in weeks 17-18 and 19-20 are almost twice as large, suggesting that absences during the testing window are more harmful than those that occur earlier in the spring semester.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    Overall, every absence is damaging, but an additional absence in a longer spell has a smaller marginal cost.

    “In sum, all absences are harmful, while the marginal cost of absences that are part of a longer absence spell is smaller.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    Over three-quarters of the school-level estimates are negative, and most lie within one standard deviation of the baseline.

    “More than 75% of the estimates are negative, and most are within one standard deviation of the baseline point estimate.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingnone given

    Patterns of absences are similar when using other school years.

    “Patterns are similar when using other school years.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S4 (based).

    Interpretationnone given

    The primary mechanism by which absences harm achievement is lost instructional time.

    “This assumption is plausible, as the primary mechanism through which absences harm achievement is via lost instructional time.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductnone given

    The studied district has a large and diverse student body that provides the power and variation needed for the analysis.

    “Lastly, the studied district has a large and diverse student body, providing the power and variation necessary to conduct the analysis.”
    Quote in the paperNot a substantial claim

  • Classified as probably not a substantial claim. May combine two claims.

    Premisereasoning

    Attendance is both an intermediate output and an input in the education production function, so it may be affected by the same shocks that affect test scores and other educational outcomes.

    “Because attendance is an intermediate output of the education production function as well as an input, it might be affected by the same shocks that affect test scores and other educational outcomes.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Study design and conductreasoning

    The lagged scores and classroom and tract-by-year fixed effects eliminate many threats to validity, but concerns about idiosyncratic student-level shocks remain.

    “The lagged scores and classroom and tract-by-year fixed effects included in X eliminate many of the threats to validity articulated above, but concerns about idiosyncratic student-level shocks to both A and y remain.”
    Quote in the paperSubstantial claim
    • The authors state this as an assumption: Absences in one subject do not affect achievement in another subject, conditional on other factors and the student-year shock.
    • The authors state this as an assumption: Within-student differences in absences between math and ELA are conditionally random, so absences in one subject serve as a good proxy for the student-year shock.

  • May combine two claims.

    Premisecitation

    Absence harm appears roughly similar across grade levels, with comparable North Carolina estimates for grades 4 and 5.

    “Moreover, the harm attributable to absences is fairly similar across all levels of schooling: Comparable estimates from North Carolina showed similar effects of absences in grades 4 and 5.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationown data

    With student fixed effects added in Columns 4 and 5, results remain robust, implying that stable unobserved student and family traits do not bias the simple lagged-score estimates in Column 1.

    “Once again, the results are robust, suggesting that relatively stable unobserved student and family characteristics are not biasing the simple lagged-score estimates reported in Column 1.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationcitation

    This spring effect is economically meaningful, comparable to moving from an average teacher to one at the 20th percentile of effectiveness (Herrmann and Rockoff, 2012).

    “This effect is economically significant, as it is roughly equivalent to replacing an average teacher with one from the 20th percentile of the effectiveness distribution (Herrmann and Rockoff, 2012).”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Herrmann and Rockoff, 2012

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S4 (based).

    Premisenone given

    Subject-specific absences in a given year are strongly correlated with the underlying process through which an idiosyncratic shock would affect attendance.

    “This too is plausible in the current setting, given the strong correlation between subject-specific absences in a given year and the underlying process through which an idiosyncratic shock would affect attendance.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Study design and conductreasoning

    The stacked approach with student-by-year fixed effects absorbs time-varying shocks specific to a student and assigns within-year math-versus-ELA performance gaps to subject-specific absences.

    “As shown in Equation 3, using student-byyear fixed effects, we control for time-varying student-specific shocks and directly attribute the difference between students' math and ELA performance to their subject-specific absences.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationown data

    To the authors' knowledge, this is the first credible evidence that absences reduce achievement in both middle and high school, implying absences matter throughout K-12.

    “To our knowledge, this is the first credible evidence that absences harm achievement in both middle and high schools, and this finding suggests that absences are a serious concern at every level of K-12 schooling.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Premisenone given

    Existing evidence on absences comes almost exclusively from primary school settings and uses full-day absences.

    “Existing evidence comes almost exclusively from primary school settings and considers absenteeism using full-day absences.”
    Quote in the paperNot a substantial claim

  • Classified as probably not a substantial claim.

    Study design and conductcitation

    The traditional paper scantron method of marking attendance has no systematic bias or error.

    “We have no reason to believe there would be any systematic bias or error in this traditional way of marking attendance (Whitney and Liu, 2017).”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Whitney and Liu, 2017

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S4 (based). May combine two claims.

    Premisereasoning

    Idiosyncratic time-varying shocks jointly affect current achievement and attendance but are not adequately controlled by lagged test scores or student or school fixed effects.

    “The concern is that idiosyncratic (timevarying) shocks jointly affect current achievement and attendance but are not adequately controlled for by lagged test scores or student or school fixed effects.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    Gottfried and Kirksey (2017) likewise reported that spring, not fall, absences correlate with end-of-year elementary test scores in a small California urban district.

    “An exception is a study by Gottfried and Kirksey (2017), who similarly found that spring, but not fall, absences are associated with end-of-year elementary school test scores in a small urban district in California.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Gottfried and Kirksey (2017)

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    The effects of annual primary school absences are approximately linear.

    “Gershenson et al. (2017) found the effects of annual primary school absences to be approximately linear”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Gershenson et al. (2017) source

      Not checked in the free run.

  • May combine two claims.

    Interpretationreasoning

    The pooled estimate misleads by giving too much weight to fall absences and too little to spring absences.

    “This weighted average of the fall and spring impacts uncovered in Table 4 is misleading, as it overstates the importance of fall absences and understates the importance of spring absences.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationown data

    Middle and high school absences harm both contemporaneous academic achievement and long-run educational attainment.

    “These results suggest that middle and high school absences are harmful for both contemporaneous academic achievement and long-run educational attainment.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S4 (based).

    Premisenone given

    Older students have more agency over their school attendance habits than younger students.

    “Absences in later grades are fundamentally different from those in earlier grades, as older students have more agency over their school attendance habits”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    The Column 1 estimate resembles prior estimates for North Carolina fourth and fifth graders (Gershenson et al., 2017).

    “This effect size is similar to that estimated using similar methods for 4th-and 5th-graders in North Carolina (Gershenson et al., 2017).”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Gershenson et al., 2017 source

      Not checked in the free run.

  • May combine two claims.

    Interpretationown data

    The Oster bounds indicate a causal effect on college enrollment, but sorting appears more influential than for high school graduation.

    “Once again, the Oster bounds suggest that there is a causal effect on college enrollment, though sorting seems to play a bigger role here than in the high school graduation analysis.”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim. May combine two claims.

    Study design and conductown data

    Because exact test dates are unknown and can differ among students in the same school, the testing windows are imputed.

    “As discussed in Section 2, we can only impute testing windows, as the exact timing of testing is unknown and may vary across students within a school.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationown data

    The timing findings support that missed instruction from absences lowers performance and that selection does not bias the baseline estimate (Herrmann and Rockoff, 2012).

    “These results further confirm that lost instructional time induced by absences harms student performance and that the baseline estimate is not biased by selection (Herrmann and Rockoff, 2012).”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Herrmann and Rockoff, 2012

      Not checked in the free run.

  • May combine two claims.

    Interpretationown data

    Table 7 adds evidence that absences reduce middle and high school achievement, that the link is causal, that later absences hurt more, and that lost instruction is a likely mechanism.

    “Overall, the results in Table 7 provide additional evidence that middle and high school student absences harm achievement, that this relationship is causal, that absences later in the year are more harmful than those earlier in the year, and that a likely mechanism is simply lost instructional time.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationown data

    The effects of absences on test scores are economically significant and comparable to the impacts of other popular interventions.

    “These are economically significant effects on par with the impacts of other popular interventions.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationown data

    The overlap among models indicates that absence effects are roughly linear and show no threshold nonlinearity at chronic absenteeism.

    “The similarity between the three models suggests that the effect of absences is approximately linear and that there is no nonlinearity in achievement at the threshold of chronic absenteeism (though chronically absent students certainly perform significantly worse than students with just one or two absences).”
    Quote in the paperSubstantial claim
    • The authors state this as an assumption: The effects of absences are linear, with no diminishing or increasing costs and no discrete effect at the chronic absenteeism threshold.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Interpretationreasoning

    This replication is unsurprising given the high correlation between math and ELA absences.

    “Again, this is to be expected because math absences and ELA absences are highly correlated.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationown data

    Table 8 supports a causal reading of the main absence-harm finding and offers arguably novel evidence that ninth- and tenth-grade absences causally affect later attainment, suggesting high school absence interventions may be valuable.

    “In sum, the results of Table 8 reaffirm a causal interpretation of the main finding that absences harm achievement and provide arguably novel evidence that 9th-and 10th-grade absences have a causal effect on subsequent educational attainment, which motivates the potential value of absence-related interventions in high school.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Interpretationreasoning

    It is unlikely that all absences are discretionary.

    “it is unlikely that all absences are discretionary”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationreasoning

    Late-year absences are harder to make up, either because there is less time to do so or because students are busier and encounter more difficult material later in the year.

    “it appears that late-year absences are simply harder to make up, either because there is less time to do so or because students are busier and encountering more difficult material later in the year”
    Quote in the paperSubstantial claim
    • Rests on an assumption: Grades in the district are based on academic performance and cannot be based on nonacademic factors such as absences.
59 of 59

  • The authors state this as an assumption.

    Absences in one subject do not affect achievement in another subject, conditional on other factors and the student-year shock.

    “second, that any spillover effects of absences in one subject on academic mastery in another are relatively trivial.”
    “Conditional on math absences, X, and the year-specific student shock, ELA absences do not affect math achievement.”

  • The authors state this as an assumption.

    Within-student differences in absences between math and ELA are conditionally random, so absences in one subject serve as a good proxy for the student-year shock.

    “first, that in a given school year, within-student differences in absences in math and English language arts (ELA) are conditionally random”
    “A -j needs to be a good proxy in the sense that conditional on A -j , A j is no longer correlated with the idiosyncratic shock (v it in Equation 2).”

  • The authors state this as an assumption.

    Within-student differences in absences between subjects are conditionally random and not due to year-specific subject preferences.

    “both the proxy plug-in and stacked student-year fixed-effects strategies assume that such differences are conditionally random (e.g., because of the timing of a dentist appointment or leaving school early due to illness) and not the result of a year-specific preference for one subject over the other.”

  • The authors state this as an assumption.

    Non-math absences do not directly affect math performance, ruling out cross-subject spillover effects.

    “the redundancy assumption made by the proxy plug-in solution, which is that non-math (e.g., ELA) absences do not directly affect math performance.”

  • The authors state this as an assumption.

    The effects of absences are linear, with no diminishing or increasing costs and no discrete effect at the chronic absenteeism threshold.

    “The main results presented in Section 4.1 assume that the effects of absences are linear -that is, that there are neither diminishing nor increasing costs to student absences, nor is there a discrete effect of crossing the threshold for being considered chronically absent, which in this district is defined as missing at least 10% of total school days (accruing about 18 absences).”

  • The authors state this as an assumption.

    The harm of absences is the same across all student and school types.

    “The models estimated to this point also assume that the harm of absences is the same across all student and school types.”

  • The paper relies on this without support.

    Grades in the district are based on academic performance and cannot be based on nonacademic factors such as absences.

    “Grades in the district are explicitly based on academic performance and are not allowed to be based on nonacademic factors such as absences.”
7 of 7

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

  • Both correlational and quasi-experimental studies find that student absences have negative effects on several academic and longer-run economic outcomes.

    “Both correlational and quasi-experimental studies find that student absences have negative effects on several academic and longer-run economic outcomes (5–7).”

    From Higher chronic absenteeism threatens academic recovery from the COVID-19 pandemic (2024) · cites this paper as [5]

    Sections 4 and 5.2 report harmful effects of absences on test scores, high school graduation, and college enrollment. But Section 2.4 identifies the measured long-term outcomes as graduation and enrollment; it does not report effects on longer-run economic outcomes such as earnings or employment. (LLM, conf=0.86)

    Experimental — accuracy not yet measured

  • Consistent school attendance is an educationally consequential behavior.

    “More generally, consistent school attendance is an educationally consequential behavior.”

    From Higher chronic absenteeism threatens academic recovery from the COVID-19 pandemic (2024) · cites this paper as [5]

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

  • Four-year enrollment exceeds any-college enrollment in Table A6

    Major

    Table A6 reports a four-year college enrollment average of 0.580 but an ever-enrolled-in-college average of only 0.481 for the same math-classroom sample. Four-year enrollees must be included among students ever enrolled in college. The ELA-classroom averages have the same problem (0.553 versus 0.442).

  • The stated full sample excludes a grade used in the middle-school analysis

    Minor

    Table 2 describes its observations as the full analytic sample but says they include grades 7 through 11. Section 4.1 describes the middle-school analysis sample as grades 6 through 8, which includes sixth-graders.

2 of 2

Cites

Flagged first, then by citations

9 of the 58 references couldn’t be matched to a known paper.

Research works printed but not linked: 2

  1. Dundar, 2016, The National Student Clearinghouse as an integral part of the national postsecondary data infrastructure, National Student Clearinghouse Research Center, 20, 1
  2. Gottfried, Michael A. and E. Hutt, Addressing Absenteeism, Harvard Education Press, 2019.

Cited by

Sorted by citations

Related

Authors
Liu, J., Lee, M., & Gershenson, S.
Affiliations
  • University of Maryland, College Park (US)
  • Brown University (US)
  • American University (US)
Corresponding author
Liu, Jing
ORCID
Venue
Journal of Public Economics
Publisher
Elsevier BV
Published
2021-07
Volume
199
Pages
104441
ISSN
0047-2727 (ISSN-L)
DOI
10.1016/j.jpubeco.2021.104441
OpenAlex
W2972629176
Semantic Scholar
1565bc898d49f78abe44a5b35d24ec9dc5743a4c
Cited by
143
References
58
Citations by year
2026: 29 · 2025: 20 · 2024: 41 · 2023: 25 · 2022: 19 · 2021: 6 · 2020: 3
FWCI
18.87
Type
journal article
Language
en
Open access
Licence
Locations
  • Journal of Public Economics (publishedVersion)
  • SSRN Electronic Journal (publishedVersion)
  • RePEc: Research Papers in Economics (submittedVersion)
  • Econstor (Econstor) (submittedVersion)
Funders
  • Institute of Education Sciences
  • Spencer Foundation
Topics
  • Social Sciences › Education › Youth Substance Use and School Attendance
  • Social Sciences › Education › School Choice and Performance
  • Social Sciences › Education › Educational Outcomes and Influences
Keywords
school absenteeism, course grades, educational attainment, high school graduation, college enrollment, learning loss, junior high school, high schools, endogeneity

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Claims read by deepseek-v4.1-flash under definition version 3 (extract-v3) on 2026-10-07.

Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-10-07.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review