10.2 Steps to hypothesis testing

Below is a list of all the steps required when conducting a hypothesis test.

Several of the steps we will learn about in later tutorials.

  • Identify the appropriate statistical test and thus null distribution for the test statistic
  • State the null (H0) and alternative (HA) hypotheses
  • Set an \(\alpha\) level (usually 0.05)
  • Determine whether a one-tailed or two-tailed test is appropriate (almost always the latter)
  • Check assumptions of the statistical test
    • state the assumptions of the test
    • use appropriate figures and / or tests to check whether the assumptions of the statistical test are met
    • transform data to meet assumptions if required
    • if assumptions can’t be met (e.g. after transformation), use non-parametric test and repeat the first four steps above
  • Provide an appropriate figure, including figure caption, to visualize the raw or transformed data
  • Provide a line or two interpreting your figure, and this may inform your concluding statement
  • Calculate the test statistic value using the observed data
  • Use the null distribution to determine the P-value associated with the observed test statistic
  • Draw the appropriate conclusion and communicate it clearly
  • Calculate and include a confidence interval (e.g. for t-tests) or R2 value (e.g. for ANOVA) when appropriate