Conditional indirect effects of clarity on students’ information processing: disentangling sources of cognitive load

ABSTRACT

The effect of instructor clarity on student learning has been explained using cognitive load theory, which stipulates that students have limited mental resources to devote to activities pertaining to learning. To date, the effect of teacher clarity on students’ cognitive burden has been studied in reference to students’ extraneous cognitive load as clear teachers design their instruction to be straightforward and free from distraction. However, few instructional communication studies have examined how clarity’s impact on extraneous cognitive load interacts with the intrinsic difficulty of students’ course lessons to conditionally impact students’ learning processes. This study utilized data from 221 students to estimate a latent first-stage conditional process model with teacher clarity predicting students’ deep processing of course material through students’ receiver apprehension, dependent upon the intrinsic difficulty of the course material (despite teaching). Results indicated that as teachers were generally clearer during instruction, students more deeply processed their course material because they experienced less receiver apprehension. However, this indirect effect was greater when students had more working memory availability (i.e., lower intrinsic cognitive load controlling for clear teaching).

KEYWORDS:

• clarity
• cognitive load
• additivity

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 To demonstrate that items are tapping into the same latent construct in the bifactor measurement model, we also tested how the items from the six-item intrinsic load scale (which are based on the characteristics of the course only, and not the teacher) anchor to the latent residualized factor of working memory availability (i.e., allowing the residualized factor loadings of working memory availability purged of teacher clarity to load with the six items of intrinsic load). Results indicated that all items loaded highly (and significantly) on the residualized latent variable:

Table

This is evidence for construct validity (i.e., evidence based on internal structure; AERA, APA & NCME, 2014), as the same common factor strongly explains the item variance in residualized working memory availability (controlling for clarity) and intrinsic load (with only items related to the course difficulty, and not the instruction).

2 In further support of our argument that working memory availability serves as an appropriate proxy for intrinsic load, we re-examined our bifactor structural regression model after substituting the latent intrinsic load factor (from the six-item intrinsic load measure) with the residualized working memory availability factor, as a first-stage moderator in our mediation model. Results were largely unchanged the index of moderated mediation was significant with a 95% bootstrap confidence interval excluding zero (index = −.08 [−.17, −.01]), and similar to what we reported in the manuscript, we found that under conditions of high intrinsic load, the conditional indirect effect of general instructor clarity on students’ deep processing through students’ receiver apprehension was weaker (θab = .14 [.04, .35]) than when intrinsic load was average (θab = .22 [.07, .46]), or when intrinsic load was low (θab = .30 [.08, .60]). This is evidence for predictive validity (i.e., evidence based on relations to other variables; AERA, APA & NCME, 2014), as the structural regressions were replicated using either factor as a latent moderator. Considering the similarities in outcomes, we note that researchers could use a bifactor model to separate variance from working memory overload due to teaching clarity and variance leftover due to intrinsic difficulty, or they may opt to use the intrinsic-load measure reported here and include clarity in a regression model to control for each variable (or allow the two to interact). It may be more feasible for researchers to use two scales if they are not in the position to examine a bifactor latent variable model.