Try Multivariate approach. Explain further for better understanding of the nature of the research.  

Please provide more information about your study design, sample, and variables being measured. This will help us provide a more detailed response.

As you describe it, this is not really a "repeated measures" experiment in the classical term of repeated measures, which is usually defined as >2 observations. In your case, this is a "pre" and "post" measurement. Because you refer to this as an "experiment" I take this to mean that you have used random assignment to allocate units to treatment and control? If that is the case, AND your outcome is a continuous linear variable, then a t-test of the change score should suffice.

However, if there is more to your design than this simple framework, then a different approach may be required. I am also not sure why you are interested in comparing their variance? Variance of what, and why?

Ariel, I do not see where you read that Aucique has "pre-post measurements". I do not agree that "repeated measures is usually defined as >2". The criterion for repeated mesures is a dependency structure between the values. This means that groups (possibly simply pairs) of values share a common source of variance. This means that a part of the total variance can be explained as a systematic difference between such groups of values. This "common source of variance" is often the individual (when several measurements, like pre- and post treatment, or measurements at several time points) are taken from the same individuals. But this may also be some other factor like the experiment itself, the assay, the region of sampling (in geographic settings), the family, school, company, etc.

But however, Aucique's question can not be answered unless he gets more specific.

The ratio of two variances is asymptotically F-distributed. This may serve as a proxy. But to make sense of the effects I would suggest to inspect the likelihood function (that is, at best two-dimensional). Here one may see what for what particular combinations of variances the observed data is reasonably likely.

I agree with Jochen that to compare variances for two experiments an F-test could be performed. For three or more, Bartlett's test is often used.  Levene's test is a robust alternative.

Dear,

Awopeju, Ariel, Jochen and Stephen, I appreciate your comments. About my research is basically compare the variance of two experiments had the same objective and structure. At no time is a repeated measures design, simply want to compare the responses of repeated experiments.

Experiments were performed under a completely randomized design with 8 treatments and 6 repetitions, physiological variables were determined.

I had read that the test Fmax is one of the alternatives to determine the homogeneity of variances, however I want to know other approaches.

Thank you very much.

Thank you for clarifying your study design. Given your design, you may consider either a repeated measures ANOVA, or a regression model for longitudinal data. Depending on how the subjects were exposed to treatments (more than one treatment and more than one repetition?), you may need to use a mult-level approach that will involve nesting repetitions within treatments and treatments within subjects.