paierd t-test vs. two-sample t-test
统计真是一个复杂的东西。就比如最简单的 paierd t-test 和 two-sample t-test ，通常我们认为 paierd t-test 更容易得到显著的结果，因为他排除了被视间的效应。但是，如果考虑到增加被试间效应的代价，问题就变得很难说了，因为自由度也随着减小了。这种情况在被试少的情况下特别容易出现，我就碰到过一回了。可是，到底应该用sensitive的呢，还是应该用appropriate的呢？头疼。
The paired t-test can be a more appropriate model for a given data set, but more effects are modeled, i.e. there are less error degrees of freedom. This might come at the price of a decrease in sensitivity so that the two-sample t-test can be less appropriate, but more sensitive. This compromise is increasingly harder to make with a smaller number of scans J.
The paired t-test can be a more appropriate model for a given data set, but more effects are modeled, i.e. there are less error degrees of freedom. This might come at the price of a decrease in sensitivity so that the two-sample t-test can be less appropriate, but more sensitive. This compromise is increasingly harder to make with a smaller number of scans J.
评论
发表评论