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 17.
The sign test is restricted to situations where one
group of subjects is tested twice.
The binomial test, like the sign test, has a null
hypothesis that contains the notion of "one-half."
Fisher's Exact Test is more accurate than either
the sign test or the binomial test.
All chi-square tests deal with frequencies.
In chi-square tests, it's the observed frequencies
(rather than the expected frequencies) that should not be too small.
In a one-sample chi-square test, the null hypothesis
must be set up to say that the population percentages are the same across
the various categories.
If 2 groups are compared on a dichotomous response
variable, the null hypothesis of a chi-square test says that each group's
population is split 50-50 across the two parts of the response variable.
A chi-square test is robust to the assumption of
McNemar's chi-square test is appropriate for comparing
2 groups on a dichotomous response variable.
If a 2x2 chi-square test is used to compare 2 independent
groups (each having 10 people) on a dichotomous response variable, the
df for the chi-square test will be equal to 18.