Chapter 18: Misconceptions
When people read, hear, or prepare research summaries,
they sometimes have misconceptions about what is or isn't "sound
practice" regarding the collection, analysis, and interpretation
of data. Here are some of these common (and dangerous) misconceptions
associated with the content of Chapter 18.
- If, at the end of the baseball season, the top 5
batters have these batting averages [.357, .351, .350, .350, and .348],
their ranks would be 1, 2, 3, 3, 4.
- Nonparametric tests were invented specifically for
situations where samples are nonrandom.
- Nonparametric tests have lower power than parametric
tests.
- Since they are based on ranks, the results of nonparametric
tests are easy to interpret.
- To compare 2 groups with a median test, it's necessary
to compute each group's median.
- If the Wilcoxon matched-pairs signed-ranks test is
used to compare a group's pretest and posttest performance, and if those
people with identical pre and post scores are dropped from the analysis,
this will cause the power of the Wilcoxon test to decrease.
- A Kruskal-Wallis one-way analysis of variance of
ranks leads to an F-test that's usually presented in an ANOVA summary
table.
- There are 2 independent variables in Friedman's two-way
analysis of variance of ranks.
- Large-sample approximations of nonparametric tests
should not be used unless the sample size(s) are equal to or greater
than 30.
- Being "distribution-free," nonparametric
tests have no underlying assumptions that deal with the distributional
shape of population(s).
|
