Stage	Step	Assumption
1	1	Fit a simple linear regression model to assess the relationship between the active social network size and physical activity enjoyment
	2	Note the standardised regression weight from Step 1
	3	Fit a multiple linear regression model in which all measured confounding variables are treated as predictors of the main independent variable, physical activity enjoyment
	4	Identify from Step 3 potential confounders that have a p-value ≥ 0.25
	5	Predictors from Step 4 that produced a p ≥ 0.25 should be removed from the analysis and the others kept for the next stage of the analysis
2	6	Adjust for each of the remaining confounding variables in the model fitted at Step 1
	7	Compute the per cent change between the standardised regression weight at Step 1 and the new weight resulting from Step 6
	8	All potential confounders that produce a change of 10% or more should be incorporated into the final analysis as the ultimate confounders
	9	Repeat the above eight steps, with the sedentary social network size now serving as the primary predictor