Which statement best distinguishes planned contrasts from post hoc tests?

Which statement best distinguishes planned contrasts from post hoc tests?

• A. Planned contrasts are theory-led and pre-specified with greater power; post hoc are exploratory with less power.
• B. Post hoc tests are theory-led and pre-specified.
• C. Planned contrasts have less power than post hoc.
• D. Post hoc tests partition variance into smaller portions.

Answer: (A)

Planned contrasts are designed to test specific, theory-driven hypotheses about how the group means should relate to one another, and they are specified before looking at the data. Because you focus on a few meaningful comparisons that you’ve pre-specified, you reduce the number of tests, which gives you greater statistical power to detect the hypothesized differences while maintaining appropriate error control. Post hoc tests, in contrast, are used after you’ve observed the data and are exploring which group means differ. They involve many possible comparisons (often all pairs), so you need adjustments to control the overall Type I error rate. Those adjustments come at the cost of reduced power for each individual test, and the analyses are considered exploratory rather than theory-driven. The other statements mix up these ideas or make mischaracterizations. Planned contrasts aren’t the ones with less power, and post hoc tests aren’t typically theory-led or pre-specified, nor are they primarily about partitioning variance in the way that phrase suggests.

Planned contrasts are designed to test specific, theory-driven hypotheses about how the group means should relate to one another, and they are specified before looking at the data. Because you focus on a few meaningful comparisons that you’ve pre-specified, you reduce the number of tests, which gives you greater statistical power to detect the hypothesized differences while maintaining appropriate error control.

Post hoc tests, in contrast, are used after you’ve observed the data and are exploring which group means differ. They involve many possible comparisons (often all pairs), so you need adjustments to control the overall Type I error rate. Those adjustments come at the cost of reduced power for each individual test, and the analyses are considered exploratory rather than theory-driven.

The other statements mix up these ideas or make mischaracterizations. Planned contrasts aren’t the ones with less power, and post hoc tests aren’t typically theory-led or pre-specified, nor are they primarily about partitioning variance in the way that phrase suggests.