Browse By Day
Browse By Time
Browse By Panel
Browse By Session Type
Browse By Topic Area
In Event: Special Poster Session 05 with Continental Breakfast Reception
In Poster Session: PS 05 - Policy Section
Reading comprehension (RC) is a complex accomplishment that underlies life outcomes, such as knowledge acquisition, educational attainment, and employment (Ricketts, Sperring, & Nation, 2014; Smart et al., 2017), yet many children have difficulty learning to comprehend text: only 37% of US 4th-graders score at or above proficient on the National Assessment of Educational Progress (NAEP, 2017). Substantial research points to the role of word reading skills in RC (García & Cain, 2014). However, some children demonstrate reading comprehension deficits (RCD), despite having age-appropriate word reading (Oakhill & Cain, 2017). Children with RCD have deficits in executive functions (EF), such as inhibition (Borella, Carretti, & Pelegrina, 2010), working memory (Cain, 2006), and cognitive flexibility (Cartwright et al., 2017). However, few studies have examined causal evidence regarding the relation between EF and RC, aside from a few intervention studies with small sample sizes, with ns ranging from 31 to 57 (e.g., Cartwright, 2002, Cartwright et al., 2017; Dahlin, 2011; García-Madruga et al., 2013). Thus, our purpose was to substantiate causal claims using innovative statistical techniques with a large sample size.
We assessed students in the first year of a three-year longitudinal study. The analytic sample included 672 1st to 4th-grade students from whom we collected data in fall and spring semesters. Students’ change in reading comprehension from fall to spring (on the Woodcock Johnson-IV passage comprehension subtest) was the outcome variable. EF was the primary variable of interest; students’ EF scores were a composite of inhibition (NEPSY inhibition subtest); working memory (TOMAL-2 letters backward); and cognitive flexibility (multiple classification, Cartwright, 2002). Students’ scores were standardized then averaged across the three assessments. We controlled students’ school, English learner status, vocabulary knowledge, homonym understanding, grade level, decoding skills, and free/reduced-price lunch status.
We used propensity score matching, which is a statistical technique utilizing covariates to predict the likelihood of students being a “success” on a dichotomous variable; with this method, assignment to treatment group becomes exogenous, meaning it mimics randomization in a randomized control trial (RCT) and reduces selection bias (Murnane & Willet, 2010). Specifically, we used covariates to predict whether a student had high EF skills, regardless of his/her actual EF score. Then, we matched students based on these probabilities, since these pairs were equally likely to have high EF skills, regardless of their actual EF values.
Students who started with high EF scores improved reading comprehension by 1.3 points (0.33 standard deviations [SD]) more than students with lower initial EF scores (p<.05). Using a traditional multiple linear regression model, we found a 1-SD change in EF skills led to a 0.203 SD increase in reading comprehension scores. The two models support each other, yet only propensity score matching permits causal claims. Specifically, it provides evidence that one’s executive functioning leads to changes in reading comprehension, which researchers could not test otherwise using traditional RCTs. Researchers and policymakers can use this finding to develop interventions that target EF skills in an effort to increase both reading comprehension and higher-order thinking skills.