A woodshop that refinishes old furniture decides to run an experiment on the impact of sand paper brand on the time to strip off paint. They just received an order to redo 600 identical chairs from a local event center and randomly assign 200 to each of three brands A, B, and C (keeping all else equal, such as grit). Unfortunately, for 50 of the chairs from brand B, the notepad that had the recorded times was ruined by a coffee spill so that these 50 data points are missing. Which answers are true?

Answers

1. The 550 remaining observations can still be analyzed using a General Linear Model.

2. To increase power, 50 chairs from each of the other brands should be deleted before analysis.

3. Before the missing data, the experiment had 200 replicates.

4. Since the design is no longer balanced, the data is useless and the experiment must be restarted with new chairs.