Table 1
Statistical approaches adopted for each research question.
| RESEARCH QUESTION | PRIMARY ANALYTIC MODEL | POST-HOC/EFFECT SIZE | RATIONALE |
|---|---|---|---|
| RQ1: Is there a gap between designers’ aspirations and their perceived reality of being able to encourage pro-environmental behaviours across different intervention types? | Paired t-test (aspiration versus reality) for each of the five interventions; Wilcoxon signed-rank used if Shapiro–Wilk p < 0.05 | Cohen’s d (paired) + 95% confidence interval (CI) | Tests the within-person aspiration–reality discrepancy |
| RQ2: Do aspirations, perceived realities and the gap between them differ across intervention types? | Three separate linear mixed-effects models (LMMs) with a random intercept for the participant:
| Kenward–Roger F-tests; Tukey-adjusted pairwise contrasts; partial η2 and partial ω2 (95% CI) | Retains all five repeated measures per person, accounts for within-subject correlation and quantifies the variance explained by delivery dependency |
| RQ3: To what extent are aspirations and perceived realities correlated across intervention types? | Pearson’s r for each intervention; Fisher’s r-to-z to compare coefficients | 95% CI for each r | Examines the strength of alignment within each intervention |
| RQ4: Is the variance in aspirations and perceived realities driven more by individual professional disposition or by intervention-specific characteristics? | Two-level random-intercept LMM (scores nested in persons) | Intraclass correlation coefficient ICC(1) + 95% CI | Partitions the score variance into between-person (dispositional) versus within-person (intervention-specific) components |
Figure 1
Violin plots of aspiration and perceived reality scores across intervention types.
Note: The horizontal width of each violin represents the probability density of responses; wider sections indicate a greater response frequency. Horizontal lines show the mean (thick) and 25th/75th percentiles (thin).
Figure 2
Correlations between aspiration and perceived reality by intervention type, showing a strong positive alignment across all domains, but a mixed dependency pattern.