GraphPad Statistics Guide

Interpreting results: Analyzing a stack of P values

Interpreting results: Analyzing a stack of P values

Previous topic Next topic No expanding text in this topic  

Interpreting results: Analyzing a stack of P values

Previous topic Next topic JavaScript is required for expanding text JavaScript is required for the print function Mail us feedback on this topic!  

Statistical significance approach

Adjusted P values

You entered each P value on a different row, and the results table has the same number of rows, each with three values:

The first column states whether the comparison associated with that P values is statistically significant or not, stated simply as yes or no.

The second column shows the P value you entered.

The third column shows the adjusted P value. If the alpha value you entered exactly equaled this adjusted P value, then this comparison would be right at the border of being defined as "statistically significant" or not.

Significant results

This table shows only the comparisons flagged as "statistically significant". The table is sorted, with the smallest P value on top. Each row shows the row title (or row number in the data table, if you didn't enter titles), the P value you entered and the adjusted P value.

Ranked P value

This table is only created if you checked the option to graph the P value distribution. The X values are the ranks of the P values (1 is smallest) and the Y values are the P values themselves. The table is not all that useful by itself, but is used to automatically create the graph.

Threshold P value

A floating note on the results page tells you the threshold P value. If a P value is less than this threshold, the result of the associated comparison is considered to be "significant". This is not the same as the value of alpha you entered. With the Bonferroni method, for example, the threshold equals alpha/K where K is the number of P values you entered.

 

False Discovery Rate method

q values

You enter each P value on a different row, and the results table has the same number of rows, each with three values:

The first column states whether the comparison associated with that P values is defined to be a "discovery" or not, stated simply as yes or no.

The second column shows the P value you entered.

The third column shows the q value. If the Q value you entered exactly equaled this q value, then this comparison would be right at the border of being defined a discovery.

Discoveries

This table shows only the comparisons flagged as "discoveries". The table is sorted, with the smallest P value on top. Each row shows the row title (or row number in the data table, if you didn't enter titles), the P value you entered and the q value

Ranked P value

This table is only created if you checked the option to graph the P value distribution. The X values are the ranks of the P values (1 is smallest) and the Y values are the P values themselves. The table is not all that useful by itself, but is used to automatically create the graph.

Threshold P value

A floating note on the results page tells you the threshold P value. If a P value is less than this threshold, the result of the associated comparison is considered to be a "discovery". This is not the same as the value of alpha you entered.  

Estimated number of true null hypotheses

If you chose the  adaptive method of Benjamini, Krieger and Yekutieli, Prism will report in a floating note the estimated number of true null hypotheses. The adaptive method works by first estimating this value, then using it when deciding which P values are small enough to be called "discoveries".