Ideally, results of a statistical analysis should not depend greatly on a single observation. To check this, it’s a good idea to conduct a sensitivity study. This entails redoing the analysis after deleting an outlier from the data set or changing its value to a more typical value and checking whether results change much. If results change little, this gives us more faith in the conclusions that the statistical analysis reports. For the weight changes in Table 9.4 from the anorexia study (shown again here and also in the Anorexia data file on the text CD), the greatest reported value of 20.9 pounds was a severe outlier. Suppose this observation was actually 2.9 pounds but was incorrectly recorded. Redo the two-sided test of that example, and summarize how the results differ. Does the ultimate conclusion depend on that single observation?

  • CreatedSeptember 11, 2015
  • Files Included
Post your question