- Consideration of the most appropriate test statistic:
What aspect of the data am I interested in—mean, proportion,
relationship?- Identify the sampling distribution under the null hypothesis for the chosen test statistic:
This is shown in statistical tables.- Decide acceptable levels of probability for errors when drawing conclusions from a
statistical test of the null hypothesis:
Consider Type I and Type II errors.- Consider whether the statistical test has sufficient power:
Sample size, alpha, magnitude of effect that you want to detect, and
variability of the statistical variable all affect statistical power.In the next chapter we look more closely at choosing a statistical test and issues of
statistical power.
Statistical analysis for education and psychology researchers 116