Missing values with ordinary (not repeated measures) ANOVA

Note that one value is blank. It is fine to have some missing values, but you must have at least one value in each row for each data set in order to fit a full model (column effect, row effect, and column/row interaction).

The following table cannot be analyzed by two-way ANOVA using a full model because there are no data for treated women. Every cell must have at least one value. However, if you choose to fit a main effects only model, Prism will still be able to analyze this data (no interaction term will be included).

If you are entering mean, SD (or SEM) and n, it is fine if n is not always the same, but ANOVA won't work if you leave n blank or enter zero.

Missing values with repeated measures ANOVA

There are two different situation regarding missing values and repeated measures two-way ANOVA:

•Prism can compute repeated measures two-way ANOVA fine if treatment groups have different numbers of participants, but each participant (experiment, litter, ...) has data at each repeat.

•Prism (starting with Prism 8) can  also do the equivalent of repeated measures two-way ANOVA if values at some repeats are missing, so long as not too many points are missing and they are missing completely at random. If some values are missing because they would have been too large to measure, then any results would be meaningless because the missing values actually represent the largest values in the table. But if a few values are missing completely at random, Prism uses a  can fit a mixed effects model.