CHI-SQUARE GOODNESS OF FIT TEST
• Statistical tests such as the t-test and z-test
are parametric because they test a hypothesis
about a particular population value
• Nonparametrics such as chi-square test a
hypothesis, but not about one particular
• Parametric statistics require assumptions that
are often not satisfied (e.g., shape of the
population distribution, interval/ratio data).
• Nonparametric statistics require assumptions,
but they are easier to meet.
What is the Purpose of the Chi-Square
Goodness of Fit?
• Test whether an observed frequency
distribution differs from a Null Hypothesis
• Use for a design in which individuals
categorized into two or more groups.
What are the Assumptions?
• mutually exclusive groups
• expected frequencies at least 5 per cell
How Does it Work?
• Determine the frequencies you expect if
the Ho is true.
• Compare the observed frequencies to the
Ho expected frequencies.
• Large differences between observed and
expected give a large value of chi-square,
likely to be significant.