Abstract
Formative assessments and feedback are vital to enhancing learning outcomes but require that learners feel at ease identifying their errors, and receiving feedback from a trusted source – teachers. An experimental test of a new theoretical framework was conducted to cultivate a pedagogical alliance to enhance students’ (a) trust in the teacher, (b) well-being in the learning environment and (c) identification of confusion and errors for the purpose of learning, assessment and feedback. A sample of 101 undergraduate students was randomly assigned to either an intervention (n = 51) or control (n = 50) condition in Elementary Statistics. Results indicated that a pedagogical alliance could be created to enhance student trust in the instructor, leading students to report greater well-being and a higher number of potential areas of confusion in their understanding of new content material relative to a control group. These results have implications for formative feedback, assessments, and by extension learning outcomes.
Acknowledgement
Preparation of this chapter was supported by a grant to the first author from the Social Sciences and Humanities Research Council of Canada (SSHRC Grant No. 435-2016-0114). Grantees undertaking such projects are encouraged to express freely their professional judgment. This paper, therefore, does not necessarily represent the positions or the policies of the Canadian government, and no official endorsement should be inferred.
Notes
1. Cohen’s d was estimated using an effect size calculator for non-parametric tests (see Fritz et al., Citation2012; www.psychometrica.de/effect_size.html#nonparametric), which uses the Mann–Whitney U test value as input. Cohen’s d was then confirmed by estimating r, which is calculated by dividing the approximated Z value by the square root of N (see Fritz et al., Citation2012).
2. Observed power is not easily calculated for non-parametric tests. Thus, a Monte Carlo, simulation-based approach was performed in R with 10,000 data sets simulated based on the observed data distribution. Each data-set was analysed using a one-tailed Mann–Whitney U test in R (R Core Team, Citation2016). The proportion of times the mean difference was significant was observed to be 68.8%.
3. The literature indicates no straightforward or universally accepted way to evaluate the effect size for a Friedman’s Test (Mellinger & Hanson, Citation2017, p. 147). However, Kendall’s W is sometimes used as it is a measure of concordance or association among ranks, with 0 indicating no concordance and 1 indicating complete concordance (see Mellinger & Hanson, Citation2017, p. 147).