Why are t-tests preferred over other statistical tests for the analysis of fMRI data?
Hello,
T-test can show you accurately the correlation coefficient in the stage of validating your results.
I think one of the reasons is that t-test is easy to understand and widely used in the academic community.
Because t-contrasts allow testing of directional hypotheses (cond. A evokes greater activation than condition B or vice versa, not just condition A different from condition B).
I also like the discussion on t-contrasts and F-contrasts here:
http://mindhive.mit.edu/node/60
I think the t-rest shows the increasing or decreasing in the activation between two conditions ( or as you said it has directional) while F-test dose not has this advantage.
Is there a reason which is specific to the type of data under consideration. In this case the data being fMRI
From statistics point of view, t-test will be used for comparing 'continuous' data.
t-tests are simple and easy to understand. you have to pay attention to issues of normality, IID. and in an fMRI context, it's terribly tempting to just run t-tests for each voxel (between conditions) - you need to correct for multiple comparisons then. I've been out of the fMRI game for years, but even 15 years ago, everyone had moved on to more statistically sophisticated approaches than t-tests...
Hello, I think the t-student is a valid tool to compare fMRI. But varabilidad of fMRI signals are large and may be recommended to seek another more powerful statistic. Multivariate statistics could try.
- Badges
- Science topic