How to Change Significance Levels for Column Comparisons

From Q
Jump to: navigation, search

By default, the Column Comparisons in Q use:

  • Uppercase letters for results that are significant at the 0.001 level (99.9%) and lower-case at the 0.05 level (95%). That is, uppercase results are more significant.
  • The False Discovery Rate Correction when determining whether or not to show any letters (but not in determining which are lowercase and which are uppercase).

This is different to the norm within the industry, which is to test at the 0.05 and 0.1 levels (i.e., 95% and 90% confidence levels). If Q's settings are left at their default levels, the results reported in Q will tend to be more robust than those that are computed in many other programs, but fewer results will be reported as being significant.

Changing the significance levels

Settings governing how Q tests are set in Statistical Assumptions. For example, to change Q so that lowercase letters are shown at the 90% level of confidence and UPPERCASE at the 95% level:

  • Set the Overall significance level (top-left) to 0.10 (i.e., the 90% level of confidence).
  • Use the Significance levels and appearance grid to control the level of significance associated with different column letters. That is, copy one of the uppercase Column letters fields (i.e., A,B,C,...,) and paste it over the top of each of the lower-case fields (i.e., a,b,c,...,) for 0.005, 0.01 and 0.05.
  • Change both Multiple comparison correction settings to None.

Saving these changes

Use Project Templates to automatically apply these changes in future projects.

See also

Results Are Different to those from Another Program