0

I am interested in performing an ANOVA on the difference in two independent groups. I have 4 sets of data: the posttest and pretest scores of a treatment group, and the posttest and pretest scores of a control group. The numbers for the two groups are different, and I have the means and standard deviations of the 4 data sets (I do not have the data).

I would like to conduct an ANOVA on the difference in scores: is it possible to calculate a Fisher-Snedecor F statistic from the difference in scores? Do I need the correlation coefficients between the posttest scores (dependent data) and the pretest scores (independent data)?

I would like to do the calculations on a spreadsheet (and need therefore the formulas). I would also be very interested in solved spreadsheet examples.

chl
  • 53,725
Nathalie_RR
  • 71
  • This sounds more like an ANCOVA, not a simple ANOVA. – chl Nov 27 '20 at 19:40
  • Yes, an ANCOVA is indeed a solution. But I am looking for information on the analysis of variances of the differences in scores, even if this option is probably not the most relevant. – Nathalie_RR Nov 29 '20 at 09:08

1 Answers1

0

I found a pdf file on the internet which allowed me (I think) to find an answer to my question:

the pretest-poosttest x group design: how to analyze the data

who presents data for two independent groups of respondents twice, analyzing this data using a mixed-design ANOVA, then using a difference ANOVA, and concludes by saying that both types of analyzes lead to the same result (I have already read several times that these two analyzes are in fact the same). Since there are only two groups, the ANOVA of the differences is conducted by calculating Student's t. I have verified by calculation that, for each group, the variance of the differences is equal to the sum of the variances minus two times the product of the standard deviations by the correlation coefficient of the two data series. And it is therefore quite impossible to calculate F (or t) without this last result.

Nathalie_RR
  • 71