Demographic characteristics of learner population
In total, 335 first-year learners (M age = 23.6 years, SD = 4.44) responded to the pre- and post-survey (n = 76 in 2019; n = 154 in 2020; n = 105 in 2022). As seen in Table 2, the majority of learners identified as women and as non-Indigenous. Political views were only measured in the 2020 and 2022 cohorts, with the majority expressing left-leaning views (the New Democratic Party). A series of chi-square independence tests indicated that health professional program, χ2 (6, N = 304) = 74.33, p < 0.001, gender, χ2 (4, N = 304) = 110.86, p < 0.001, Indigenous identity χ2 (2, N = 299) = 7.19, p = 0.028, and political views, χ2 (6, N = 238) = 21.48, p = 0.002), varied across the three cohorts (see Table 2), as did age, F(1, 301) = 4.74, p = 0.030. The 2020 cohort had the highest representation of Indigenous students, and expressed more liberal views (35.8%) compared to the 2022 cohort which was strongest in support of the New Democratic Party (37.5%). The 2019 cohort was slightly older than the 2022 cohort, but neither differed in mean age from the 2020 cohort.
To determine whether causal beliefs, blaming attitudes, professional responsibility to address inequities and support for government action or policies changed as a function of the educational intervention, a series of 2 (pre- and post-course) X 3 (cohort) mixed measures analyses of variance (ANOVAs) was conducted. There were significant interactions suggesting that the intervention differentially affected the three cohorts in terms of learners’ causal beliefs regarding historical factors of colonialism, F (2,273) = 2.79, p < 0.001, η2 = 0.020, blaming attitudes, F (2, 276) = 102.09, p < 0.001, η2= 0.425, and support for government action and policies to address inequities, F (2,273) = 83.79, p < 0.001, η2 = 0.380. Simple main effects comparing the pre-post responses for each of the three cohorts showed that, following the intervention, learners in the 2020 cohort were more likely to believe historical factors contributed to present day inequities (Fig. 1a) and to express blaming attitudes (Fig. 1b) and were less likely to support government action and policies to address inequities (Fig. 1c); these differences were not significant among learners in the 2019 and 2022 cohorts, although they demonstrated a similar trend to the 2020 cohort. This interaction was not significant in relation to causal beliefs regarding ongoing factors, F (2, 273) = 1.43, p = 0.240, η2 = 0.020, or professional responsibility to address inequities, F (1, 223) = 3.07, p = 0.081,η2 = 0.014, nor were there significant main effects for the intervention itself on professional responsibility to address inequities, F (1, 223) = 0.001, p = 0.971, η2 = 0.000. However, there was a main effect for cohort for ongoing factors, F (2, 274) = 47.14, p < 0.001,η2 = 0.256. As seen in Fig. 1d, beliefs that ongoing colonialism contributed to health inequities for Indigenous Peoples were less likely to be endorsed by learners in the 2020 cohort compared to learners in the 2019 and 2022 cohorts, which did not differ from each other (with Tukey adjustment for familywise error at p < 0.05). There was no significant main effect for cohort on taking professional responsibility to address inequities, F (1, 223) = 0.001, p = 0.971.
Correlations among outcome variables
Pearson correlation coefficients were conducted to assess the linear relationships among causal beliefs regarding the role of historical factors, blaming attitudes and support for social action or policies. There was a negative correlation between historical factors and blaming attitudes, r (283) = -0.36, p < 0.001, suggesting that learners who were less willing to recognize the role of historical factors on health inequities were more likely to express blaming attitudes. Moreover, stronger support for government action or policies to address such inequities was associated with greater recognition of the causal effects of historical factors, r (282) = 0.52, p < 0.001, and lower inclination to express blaming attitudes, r (282) = -0.88, p < 0.001.
Demographic predictors of responses to the intervention
Multiple regression analyses were conducted to assess the demographic features that predict learners’ responsiveness to the educational intervention. Based on the above reported interaction effects, each outcome variable at Time 2 (Causal Beliefs – Historical Factors, Blaming Attitudes, Support for Government Action and Policies) was regressed onto levels at time 1 on the first step, followed by age (continuous), gender (male coded 0 vs. female coded 1), and Indigenous identity (non-Indigenous coded 0 vs. Indigenous coded 1) on the second step, the null variables representing health professional program on the third, and finally the null variables representing political views on the last step. By entering the variables in blocks, we were able to evaluate the added variance accounted for in the outcome variables. Regression coefficients in the final (4th) step were the basis of our interpretations of the patterns of relationships.
The multiple linear regression model demonstrated that learners who had awareness of historical factors prior to the intervention were more likely to recognize historical factors as being an important causal belief as to why Indigenous Peoples are more likely to experience inequities following the intervention, R2 = 0.464, F(1, 214) = 184.88, p < 0.001. However, none of the demographic variables predicted such beliefs (see Table 2).
The extent to which learners expressed blaming attitudes following the educational intervention was, not surprisingly, greater among those who held such attitudes at the outset, R2 = 0.146, F(1, 219) = 38.16, p < 0.001. As a whole, demographic variables accounted for an additional 35.2% of the variability of post-intervention blaming attitudes. In particular, as seen in Table 3, males (Madj = 3.65, se = 0.069) were more likely than females (Madj = 2.53, se = 0.065) to continue to express blaming attitudes following the intervention (means are adjusted for baseline scores). In addition, the intervention appeared to be differentially effective depending on the program learners were registered in, R2change = 0.012, F(3, 210) = 6.11, p = 0.001 (Table 3). Post hoc pairwise comparisons (with Tukey adjusted p-values to maintain family-wise α less than 0.05) indicated, following the intervention, the dentistry and dental hygiene learners (Madj = 3.58, se = 0.090) were significantly more likely to express blaming attitudes than those in medicine (Madj = 2.71, se = 0.111), nursing (Madj = 2.87, se = 0.083), or pharmacy (Madj = 2.19, se = 0.252). None of the other demographic variables was a significant predictor of changes in blaming attitudes.
Similarly, support for government action and policies was greater among those who expressed more supportive views at the onset of the course, R2 = 0.179, F(1, 215) = 46.93, p < 0.001. Once again, as seen in Table 4, participant gender was significant with females (M = 5.68, se = 0.062) being more likely to express greater support government social action and policies following the intervention (i.e., controlling for pre-intervention levels) than males (M = 4.74, se = 0.065). The intervention also appeared to be more effective for learners in different programs, R2change = 0.027, F(3, 209) = 27.435, p = 0.001 (Table 4). Post hoc pairwise comparisons indicated support for government policies was lower among dentistry and dental hygiene learners (M = 4.79, se = 0.083) than among medicine (M = 5.57, se = 0.102), nursing (M = 5.35, se = 0.079), and pharmacy (M = 6.04, se = 0.235) at p < 0.001. None of the other demographic variables was a significant predictor of changes in blaming attitudes.