Dear All,
I use Graphpad Prism 6 to analyze my 6 sample group's data (N=3) by using one-way ANOVA with Turkey comparison rest. The multiple comparison result doesn't show any significant between any pair of group. However, when I delete some of sample group for example delete group: 4, 5, 6 and leave only group: 1, 2, and 3. It shows significantly difference between 1&2 and 1&3.
Thank you
You need to set up the hypothesis you are testing in advance. If you set out to discover if there were significant differences between any pairs of groups among the six groups, you cannot arbitrarily choose which groups to include in the comparison after the fact. Note also that the first stage of an ANOVA is to determine if there are any differences between any two means. This is determined by the F-statistic, which is the ratio of the mean variability among groups divided by the mean variability within groups. If the F-statistic is significant, the next stage is to determine which pairs of means were significantly different from each other. If you want to compare all possible pairs of means, then the Tukey test is an appropriate test for this. However, if you want to compare each group mean with a control mean, then the Dunnett test is appropriate.
That's a pretty scary Turkey comparison, Jochen!
A very happy Christmas to you, I should add
- Badges
- Science topic
- Similar topics
- Biological Science
- Bioecology
- Systems Ecology
- Landscape Ecology
- Connectivity