Student academic achievement
After reading the article carefully and taking notes, answer the following three questions in your initial response (all in one post; do not do separate posts for each question so that the discussion stays organized!):
1. What do the overall results (of the various models) tell us about how school gardens affect student academic achievement? (Summarize and synthesize the findings clearly IN YOUR OWN WORDS!)
2. Looking specifically at the regression analyses that involve step-wise regression (same DV but different combinations of IVs for different models, e.g., Model 1, Model 2, etc.), what do you learn from these analyses about the importance of using step-wise regression for examining the effects of different IVs/sets of IVs? In other words, what do these analyses reveal about why we don’t just include all of the variables into a single model from the start?
3. Thinking sociologically, what other variables might it have been good to include as controls in the regression analyses?
When including control variables, it is important to avoid redundancy since that leads to collinearity (multiple variables measuring or “tapping into” the same underlying concept). For example, we would NOT suggest average income of parents of children at the school since the authors already included a measure for participation in free/reduced price lunch. That variable serves as a “proxy” for the SES of families at the school. Including a measure of average parental income would actually be counterproductive since having both means that each essentially would “wash out” the effect of the other. Keep this in mind as you think through and propose additional control variables you think the authors could/should have included. (NOTE: Your goal here is not to undermine or “trash” this study; your goal is to think critically about the models used and also to think about how we might elaborate these models further.)
