Stat 412/512

Data Analysis 2

Due on canvas Feb 25th @ noon as a .pdf or Word

This is problem 14.20 in Sleuth and uses the data set ex1420.

These data are the weight losses of subjects randomly assigned to one of three diets, and additional covariates sex, initial age, and body mass index. Is there any evidence from these data that the mean weight loss differs for the different diets, after accounting for the effect of the covariates? How big are the differences?

Suggested strategy:

  • Start with exploring the dataset graphically, think boxplots, scatterplots, etc.
  • Fit a tentative model that answers the question of interest and allows for interactions.
  • Check regression diagnostics
  • Check for simplifications to the model and interpret the result. Do not use variable selection techniques Just specify a couple of levels of simplification and use F-tests to test them. For example, you might compare a model with all possible interactions to a model that has only up to two-way interactions, then a model with no interactions.

Other notes

Try to keep it to less than 2 pages You will need to be a bit picky about which plots to include, but make sure you comment on all the diagnositics even if you don’t include the plot.

You report should not include R code or raw R output. You may submit an R script in addition to your report on Blackboard.

Your report should include the following sections and be a pdf or Word doc:

  • Introduction Give a brief overview of the data, a little bit of background and the questions of interest. Keep this concise, understandable to someone outside of this class, free of statistical jargon and to the point. You should provide a summary graphic of the data involved or some basic summary statistics (it’s up to you how you do this).

  • Methods Your methods section should outline your analysis procedure. State your tentative model and any exploratory observations that lead you to it. Dicsuss the assumptions of your model and include relevant diagnostic plots. Describe any other models you considered and the results of F-tests. Present the model you will use for inference in the summary section, along with estimates and standard errors.

  • Summary Provide a brief non-technical summary of your findings that answers the questions of interest (like the statistical summaries we have been writing).