GraphPad Statistics Guide

Table of means (or predicted means)

Table of means (or predicted means)

Previous topic Next topic No expanding text in this topic  

Table of means (or predicted means)

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

Starting with Prism 8, Prism can report a table of means as part of two-way ANOVA. Choose to create this table in the Options tab of the two-way ANOVA dialog.

If there are no missing values

If the data are balanced (no missing values) you'll see the label "Means" at the top of this table which shows the cell means, the row means, the  column means, and the grand mean. You could easily compute the values yourself.

If there are any missing values

If the data are unbalanced (there are any missing values), this table is labeled "Predicted means (LS Means)"  for the cell, row, column and grand means. These predicted means, also known as "LSmeans", may not match the ones you compute by hand.

For ANOVA without repeated measures, LS means are easy to compute. Say you have a row with these values and want to compute the row mean.

There are two ways to compute the row mean. One way would be to add up all five values and divide by five. That value (4.4) is not what Prism reports.  Prism averages the three values in the first column (4), then averages the two values in the second column (5), and finally averages those two values, reporting 4.5. This is the LS mean.

Cell, row, column and grand SD

Prism also reports the SD of each cell, which is straightforward. It also reports the SD for each row and column, and SD for the whole table (the grand SD). These are computed by first averaging the replicates in each cell and then computing the SD of those cell means.  If a row has three data set columns, each with five subcolumns for replicates, the n for computing the SD is 3, not 15.