Recognizing the importance of instruction style to students' performance: some observations from laboratory research – a research note

Abstract

An experiment was undertaken to investigate the relationship between instruction style and student performance. Instruction style was modelled as a multidimensional approachability construct comprising degrees of immediacy and structure. The capacity for an instructor to manipulate perceived approachability is unequivocally demonstrated. The results also revealed an association between student learning outcomes and instruction style. Performance was positively associated with the presence of an instructor vis–à–vis no instructor, and also the degree of instructor approachability. Results also confirmed performance to be positively associated with students' ability, negatively associated with students' anxiety, but not associated with gender or several other control factors. The importance of task motivation as an important link between instruction style and students' performance was explored and some tentative propositions offered. Students' motivation varied with instruction style, with higher motivation observed with higher instructor approachability.

Keywords:

• instruction style
• approachability
• performance
• motivation
• anxiety

Notes

We wish to acknowledge the special contribution of our co-author Jennifer, who passed away in 2000. Her creative energy and special friendship are sadly missed.

¹For example, see Institute of Chartered Accountants in Australia (1994).

²‘The authoritarian leader was impersonal, the leader dictated the group's activities and eschewed participation by group members. The democratic leader by contrast, invoked group planning (structure), encouraged individual decision-making, and fostered an informal and friendly group climate. The laissez faire leader assumed a passive stance, left all decisions to the group and offered technical advice on request.’ (Scheidlinger, 1994, p:123).

³Although a degree of anxiety can be regarded as ‘facilitating’ or likely to enhance performance, excessive anxiety levels are ‘debilitating’ or impede performance (Clark and Schwartz, 1989). This threshold point will vary across individuals. However, there are state and trait anxiety ‘norms’ available for different groups of individuals (such as tertiary students) within the population (Spielberger, 1983a).

⁴The relationship between anxiety and performance is typically explained by reference to either the interference model (Wine, 1971, 1982; Hodapp and Henneberger, 1982) or the deficiency model (Paulman and Kennelly, 1984; Tobias, 1985). Proponents of the interference model maintain that high anxiety levels produce negative selfpreoccupying thoughts that interfere with information retrieval and processing in evaluative situations. The deficiency model posits that high levels of anxiety impede the development of the study skills necessary for the acquisition of knowledge required in evaluative situations.

⁵Analogous to the independent sample interpretations for the univariate tests, multiple regression assumptions of independent error terms are not violated by the fact that groups had been previously frequency-matched. Also, none of the correlations reported earlier in Table 3 indicated that multicollinearity was a problem. Further, the results of this higher-order performance analysis mirrored those of a multiple regression involving aggregate performance, with the same variables exhibiting significance. The conclusions were also identical for regressions of the performance score adjusted to its natural log, this adjustment strengthening assumptions regarding regression residual normality and homoscedasticity.

⁶The regression coefficients from the effect coding of discrete independent variables represent the deviation of each group's mean performance score from the overall mean of the dependent variable (i.e. from the ‘grand mean score’) and thus represent the effect of each treatment (see Pedhazur, 1982, pp. 290–92).

⁷Also, whilst it is possible that a highly-motivated student may work diligently but at fewer questions and thereby confound the proxy, a priori, there is no reason to expect that it would occur more readily in one student group than another. Indeed, no evidence of this situation was found. This possible confounding issue was investigated by comparing group homogeneity with respect to the ratio:

Higher-order performance score/Proportion of higher order questions attempted

If any student's ‘high’ motivation was manifested in such a way that he/she worked more diligently but on fewer questions, then, ceteris paribus, the above ‘diligence’ ratio would be higher for that student than for other students, and if any such behaviour was systematic to a particular group, the ‘diligence’ distribution for that group would be different to other groups. Groups were compared for the aggregate higher order questions and also each individual higher order question. In all cases, ratio distributions were quite similar such that the null hypothesis of group homogeneity was easily accepted using conventional ANOVA F-tests and nonparametric Kruskal–Wallis test equivalents (e.g. F = 0.4985, p = 0.6114 for group ‘diligence’ ratios re aggregate higher order questions). If anything, evidence of motivation being manifested as ‘greater diligence on fewer questions’ appeared to increase slightly from CTRL to LOW to HIGH groups (e.g. for CTRL, LOW and HIGH groups respectively, ‘diligence’ ratio means were 0.95, 0.95, 1.16; std devs were 0.55, 0.91, 0.50; medians were 0.93, 1.00, 1.21). Consequently, if any ‘diligence’ bias did confound the motivation proxy, it would have weakened the chances of observing the hypothesized relationship (that higher approachability enhances students' motivation as defined). However, the hypothesized relationship was still observed and thus can be considered a reasonably robust result given the above discussion.

⁸Note that in the absence of the CTRL group, instructor approachability was more finely measured in the regression as a continuous variable by simply using the approachability scores from each student.

Additional information

Notes on contributors

Jennifer Römcke†

We wish to acknowledge the special contribution of our co-author Jennifer, who passed away in 2000. Her creative energy and special friendship are sadly missed.