# SCS 2014: Visualizing Regression/list of questions

• You are studying observational data on the relationship between Health and Coffee. Suppose you want to control for a possible confounding factor 'Stress'. In this kind of study it is more important to make sure that you measure coffee consumption accurately than it is to make sure that you measure 'stress' accurately? What are the consequences of measurement error in Coffee? What are the consequences of measurement error in Stress? Which consequences are more consequential?
• Is it ever important to include a variable that is not significant? Is it ever important to exclude a variable that is highly significant? When are fitting criteria (e.g. AIC) suitable for model selection and when are they not?
• In a model with two predictors X and Z, and an interaction between them, how can we interpret the main effects? Is it okay to interpret the main effects if the interaction is not significant?
• In a model with three predictors X, W and Z (no interactions), if neither W nor Z are significant, can we just drop them and refit with only X?
• In a model with two predictors X, Z (no interactions), if Z is not significant, can we expect that dropping it will have little effect on the estimate and on the p-value for the effect of X?
• If X is not significant alone in a regression, is it unlikely to be significant if you add it to another variable, Z?
• Daniel Kahneman is a Nobel prize winning psychologist. Early in his career he was training air force instructors in psychological methods to improve their own training methods for flight personnel. Kahneman told his class that praise was much more effective than criticism to encourage better performance among students. The members of Kahneman’s class strongly disagreed and told Kahneman that, in their experience, student performance tended to improve after criticism and deteriorate after praise. Can you reconcile Kahneman’s claim that praise is better than criticism with the experience of his students that suggests the contrary?
• A news item on the radio says that new research shows that people who use sunscreen are at a higher risk of developing skin cancer than people who don’t. A friend who heard the item tells you that they plan to stop using sunscreen when they go out into the sun. What advice do you give your friend? What would you do to determine whether you should stop using sunscreen?
• Have your heard that 'suppression' and 'Simpson's Paradox' are both example of the same phenomenon: sign reversal? Is this correct, what does it mean, and why would it matter?
• The relationship between Y and X is highly significant in a sample of men but not in a sample of women. Is it safe to conclude that the relationship between Y and X is different in men from what it is in women?