Stat 301

1
Stat 301 Day 38

• Putting it all together!

2
Last Time Bootstrapping with 2 samples

• With two independent random samples, can
  bootstrap each sample, calculate the difference
  in the statistics, and build a bootstrap
  distribution to get standard error of statistic

Observed difference in medians 186
3
Results may vary!

• Randomization distribution
• Some issues with medians
• Type of study design mostly affects scope of
  conclusions

4
Investigation The Biggest Loser

• Stat 301 Other data files page
• Choice of procedures

5
Investigation The Biggest Loser

• (a) one-sample z-test (H0 p .5)
• (b) one-sample z-interval
• (c) two-sample t-test
• (d) two-sample z-test for proportions
• (e) one-sample t-interval (matched pairs)
• (f) one proportion test on whether lost less
• (g) one-sample t-interval for a mean
• (h) one-sample t-prediction interval
• (i) randomization test?

6
Investigation The Biggest Loser

• (j) Each test has a .05 probability of a Type I
  Error.
• The chances of making at least one Type I Error
  across 6 tests rises to .265
• Would prefer a procedure that allows you to
  compare all 4 means at once
• with a follow-up for adjusting the overall error
  rate if do find is at least one difference

7
Investigation The Biggest Loser

• (k) Where EV diet and statistically significant
• Overweight individuals in Boston area who would
  volunteer for such a study?

8
Investigation The Biggest Loser

• Stat 301 Other data files page
• Choice of procedures
• Prepare a mini-analysis for the research
  question, be ready to present your analysis
• (a)/(b), (c), (d), (e)/(f), (g)/(h), (i)
• Solutions online

9
For Friday

• HW 8 due
• Review!
• Draft handout online
• Send suggestions for handout
• Course evaluation
• Final exam sign up done!