SPSS Findings Chapter — Regression and Assumption Testing

Excerpt from the sample

Assumption testing

Prior to conducting the multiple regression, the assumptions of the model were assessed. Linearity between each predictor and the outcome was confirmed by inspection of partial regression plots. Independence of residuals was supported by a Durbin-Watson statistic of 1.94, within the conventionally acceptable range of 1.5 to 2.5.

Homoscedasticity was assessed by plotting studentised residuals against unstandardised predicted values; the distribution showed no discernible funnelling. Multicollinearity was not indicated, with all variance inflation factors below 2.1 and tolerance values above 0.47. Two cases exceeded ±3 standard deviations on the studentised residuals; both were examined individually, found to be legitimate responses rather than data-entry errors, and retained, with the analysis re-run excluding them to confirm that the pattern of results was unchanged.

Regression results

The model significantly predicted academic self-efficacy, F(3, 186) = 24.71, p < .001, accounting for 28.5% of the variance (adjusted R² = .285). Perceived tutor support emerged as the strongest unique predictor (β = .38, p < .001), followed by prior attainment (β = .24, p < .001). Weekly contact hours did not contribute significantly to the model (β = .07, p = .284).

The non-significant contribution of contact hours is reported here without interpretation; its implications are considered in Chapter 5 in relation to the literature reviewed in Chapter 2.

Marker’s notes

This is the part worth reading twice. These are the specific reasons the piece scored what it did — the decisions you can apply to your own work.

• Assumptions are tested and reported before results — routinely skipped, and routinely penalised.
• The outlier decision is stated, justified and sensitivity-checked.
• The final sentence explicitly defers interpretation to the discussion chapter — the split markers look for.
• Statistics reported in full APA format with exact p values.