October 2022: Meta-Analysis of Relations Between Executive Function and Academic Outcomes

Spiegel, J. A., Goodrich, J. M., Morris, B. M., Osborne, C. M., & Lonigan, C. J. (2021). Relations between executive functions and academic outcomes in elementary school children: A meta-analysis. Psychological Bulletin, 147(4), 329-351. https://doi.org/10.1037/bul0000322

Summary by Paul T. Cirino

Overview and Purpose

Executive function (EF) is a cognitive resource that is well-known to be important for goal-directed behavior, and academic achievement is a key outcome that is a result of goal-directed behavior for children and adolescents. It is also well-known that EF is related to academic achievement. However, it is somewhat difficult to get a handle on this relationship because EF is defined and measured in so many different ways, and academic achievement has multiple components, with most studies looking at only one domain (e.g., reading or math), or only one type of reading or math (e.g., reading fluency or reading comprehension), and often in the context of a narrow age range. One way to improve understanding is to take a broader approach—that is, by collecting many individual studies that deal with EF and achievement and collapsing across their main results.

There are at least three advantages to such an approach. First, it can directly answer questions like what type of EF is most related to achievement (and what kind of achievement). Second, it can consider all types of EF at the same time. Third, it can evaluate whether the pattern of relationship changes over the course of a student’s development, which is important, given that academic demands change with development.

For these reasons, and perhaps others, Spiegel et al. used this approach (meta-analysis) to better understand the relations between dimensions of executive function and academic outcomes. Spiegel et al. formed their hypotheses about the relations between executive functions and academic outcomes from two important cognitive theories. First, Cognitive Load Theory (overly simplistically) implies that the more complex a task is, the more cognitive resources it will require. Therefore, it makes sense that EF, as a key cognitive resource that controls goal-directed behaviors, would be critically important for achievement. Second, Dual Process Theory draws a distinction between automatic and controlled processing. The former is fast and is more relevant for academic tasks that are well-learned, whereas the latter is needed for difficult tasks or tasks that have yet to be mastered.

Academic achievement skills vary in both complexity (Cognitive Load Theory) and difficulty (Dual Process Theory), with some academic achievement tasks easier to master because the range of outcomes is constrained (e.g., single-digit math facts, where the range of problems is finite) relative to those that are unconstrained (e.g., reading comprehension, which can be about anything). Pulling from such theories, Spiegel et al. proposed that:

  1. Because most prior studies, including prior meta-analyses, examine only general EF or a specific type of EF, the authors hypothesized that individual EF-achievement relations would be reduced when considering them all together versus individually, which is how most prior work has been done.
  2. Of the dominant ways of assessing EF (i.e., as working memory, inhibitory control, and shifting), the authors proposed that working memory (WM) and inhibition would increase in relevance throughout development, but that shifting would only emerge as important in adolescence because it is difficult to distinguish shifting from working memory and inhibition in the earlier grades.
  3. At earlier stages of learning, EF was hypothesized to be related to more basic (constrained) skills, but to decrease in relevance later in development, as these skills become mastered. But for more complex and difficult academic skills, which continue to be learned over time, EF would be similarly or more strongly related to achievement as development unfolds.

Study Methods

Spiegel et al. looked at 293 (!) studies, which had a total of over 65,000 students and over 2,600 “effect sizes” (a measure of the strength of relation between EF and achievement). In this study, as is many meta-analyses, this effect size was a correlation. Correlations can have values from -1 to +1, with numbers farther away from zero in either direction representing stronger relationships. The only difference between a positive and a negative correlation is that for positive correlations, both things being measured go in the same direction (e.g., like in this meta-analysis, where higher EF was associated with higher levels of achievement) whereas for a negative correlation, both things being measured go in opposite directions (e.g., if one were to correlate income with financial stress, then as income goes up, financial stress goes down).

The authors screened over 11,000 studies and kept over 2,000. With later screening for eligibility criteria, the authors retained 769 of those studies. They then coded 531 of these studies, although only 293 had effect size information and focused on the achievement domains studied here. It should be noted that this general process is common for good meta-analysis, and so the authors were doing all the right things to make sure that their data was comprehensive and relevant, so that their conclusions would be robust. It should be further noted that the authors focused only on students in kindergarten through Grade 6.

For the purposes of this report, it is not important to detail the specific analytic approach and techniques utilized, but it should be noted that the authors’ approach was what could be considered state-of-the-science, and so their results should be quite trustworthy. Also, for purposes of this report, the results below focus only on reading and math skills.

Key Findings

Overall, the average relationship between EF and academic skill was .33, which is a moderate correlation. The authors considered the extent to which the different EF measures related to one another as well, and so somewhat weaker effects were expected relative to past meta-analyses. Also, of the different types of EF assessed, it was apparent that working memory (WM) had stronger relations (.33) with academic achievement relative to either inhibitory control (.26) or shifting (.26). The overall effect was higher for math (.36) than for reading (.32).

These values were also examined for younger (GK-G2) and older (G3-G6) students. A main finding was that for older students, inhibitory control no longer had a significant relationship with achievement. In addition, the role of inhibitory control decreased for general reading over time, and the role of shifting decreased for general math over time.

The authors then examined only the reading studies, so they could compare different kinds of reading (i.e., decoding, fluency, and comprehension). WM had the largest effect and was significant for all reading subskills, including comprehension (.26), decoding (.27), and fluency (.17), and this was true for both younger and older students. For younger students, both inhibitory control and shifting were related to comprehension (.24 and .19, respectively) and to decoding (.18 and .12), but not fluency, whereas for older students the opposite pattern held: inhibitory control and shifting were only significant for fluency (.21 and .15) but not comprehension and decoding.

When the authors examined only the math studies, they were able to compare the relation of EF to word problem solving, computation, and math fluency. In general, the pattern of relations of EF and math did not vary when younger and older samples were compared, except that WM was more relevant for word problems in older samples than younger samples. Again, WM had the most consistent effects, including for word problem solving (.32), for calculation (.28), and for fluency (.23). However, the only nonsignificant effect was the effect of shifting on math fluency.

Conclusions, Implications, and Future Directions

Regarding the hypotheses, the Spiegel et al. found that:

  1. The relation between EF and achievement was was reduced when all the EF types were considered together. This is in part a by-product of the authors using more rigorous analytic methods than prior studies. However, it also supports the “unity” idea of EF, because the reduction is due to what is common to all the EF types. On the other hand, the individual EF types each related to achievement, and sometimes differentially so, supporting the “diversity” idea of EF.
  2. The authors expected the effects of WM and inhibitory control to increase with development and the role of shifting to be relevant only at older grades. Instead, they found that the effect of WM was stable and relevant for both reading and math. The effects of inhibitory control and shifting varied for reading but not for math; specifically, these EF types were important for comprehension and decoding early and for reading fluency later.
  3. The authors expected that EF would be more related to constrained skills early and unconstrained skills later. This was only partially supported. That is, for the skills that are most unconstrained (i.e., word problem solving, reading comprehension), EF was consistently, or more, related later in development; however, there was more variability for constrained skills.

The authors conclude that relations between EF and achievement may be more nuanced than expected. For example, rather than the contribution of EF to achievement increasing with development, it is necessary to consider the type of EF and the particular achievement subskill being considered. Further, development of EF and achievement are likely to occur in tandem. Given that this meta-analysis found consistent support for the role of EF for achievement, considering individual differences in EF, particularly WM, early in school may be useful in identifying those at academic risk.

Further Reading

Karr, J. E., Areshenkoff, C. N., Rast, P., Hofer, S. M., Iverson, G. L., & Garcia-Barrera, M. A. (2018). The unity and diversity of executive functions: A systematic review and re-analysis of latent variable studies. Psychological Bulletin144(11), 1147-1185. http://dx.doi.org/10.1037/bul0000160

This is a review of the structure of executive function, showing that the extent to which working memory, inhibition, and shifting are separable varies across development.

Takacs, Z. K., & Kassai, R. (2019). The efficacy of different interventions to foster children’s executive function skills: A series of meta-analyses. Psychological Bulletin145(7), 653-698. http://dx.doi.org/10.1037/bul0000195

This is a group of studies, similar to the Spiegel et al. study reviewed here, except instead of focusing on the relation of EF to achievement, this study focuses on the extent to which EF can be improved.