CAT in the classroom: A multilevel analysis of students’ experiences with instructor nonaccommodation

ABSTRACT

This study incorporates communication accommodation theory to investigate how student perceptions of instructor nonaccommodation influence affective and cognitive classroom outcomes. A series of two-level hierarchical linear models (students nested within instructors) revealed significant, negative associations between specific modes of instructor nonaccommodation (i.e., nonaccommodation related to nonverbal responsiveness, content knowledge, and student support) and students’ reported outcomes. Specifically, nonaccommodation related to nonverbal responsiveness and student support resulted in less communication satisfaction and instructor–student rapport when controlling for student sex and expected grade in a course. Contrarily, only nonaccommodation related to content knowledge predicted processing fluency. The research provides instructional communication researchers with a unique theoretical framework for conceptualizing and assessing student perceptions while also raising important questions regarding how students prioritize effective teaching behaviors in context. Practical implications are provided for how instructors better assess and enact behavior relative to individual student needs.

KEYWORDS:

• communication accommodation theory
• HLM
• nonaccommodation
• student perceptions

Acknowledgement

The authors would like to extend their gratitude to Drs. Marko Dragojevic, Xin Ma, and William Howe, as well as the anonymous reviewers, for their support in the construction of this manuscript. Their insights and feedback were critical to the research—thank you.

Notes

1 The original sample consisted of 554 participants. However, five participants were identified as multivariate outliers and were excluded from all analyses (Tabachnick & Fidell, 2013).

2 Guided by recommendations from Carpenter (2018), we also generated a single-level, randomized parallel analysis (n = 1000) that further identified latent factors. The results suggest that student perceptions of nonaccommodation may function as a single latent factor with the addition of a second factor (eigenvalue = 1.304) not larger than expected by chance. Thus, the parallel analysis suggests that future testing using confirmatory factor analysis (CFA) procedures may be needed to assess the possibility of a bifactor structure representing general perceptions of inappropriateness.

3 For the second phase (Models 2–9), the error terms for all predictors were treated as fixed, as the presence of random effects were not included as part of the study’s hypotheses (Ma et al., 2008; Raudenbush & Bryk, 2002). These error terms can be treated as random (i.e., assuming effects vary across level-two groupings) if (a) theory supports this decision or (b) the data reveal random effects to be present. As noted by Hayes (2006), the assumption of random effects “can always be tested, and if the data are inconsistent with this assumption, the model can be reestimated setting the effect to fixed” (p. 389).

4 As indicated by Tables 5–7, comparisons to grades other than A or B yielded nonsignificant results.

5 Notably, Tables 5–7 show that the effect of nonaccommodation related to verbal delivery on processing fluency disappeared when included alongside other variables; however, related to students’ satisfaction and rapport, the negative, individual effect reported in the parsimonious model changed direction entirely. This result was unexpected, and it is possible that the finding was merely a statistical artifact. Although the items used to assess this dimension of instructor nonaccommodation may be reflective of some instructor behaviors, the notion of instructor speech may be much more complex than what was captured in the current study. There may be intergroup differences or similarities in language use and expectations between instructors and students not represented in the research (Hosek & Soliz, 2016). For example, the general demographics of the instructors at the respective institution do not lend themselves to vast differences in verbal/linguistic behavior. The 17 instructors who were responsible for teaching the course were overwhelmingly white, female, and Caucasian. These characteristics are similar to the general demographic composition of the respondents for the survey. Thus, it seems reasonable that instructors’ use of slang, pronunciation, articulation, or language was already close to what the students might use themselves, and it is unclear whether this reversal in effect would be sustained when using a scale that better captures variability in perceived nonaccommodation related to verbal delivery.