Thread: t test with diversity index

1. t test with diversity index

So, I am trying to perform a t-test using diversity indexes, but the biostat book I am using is a little bit confusing.

the calculated t value = -1.911
calculated v value = 196

then they say something about t0.05(2), 196 = 1.972
They seem to have gotten this value from the critical t value table.

They then say that 0.05 < P < 0.10 and that you do not reject the null hypothesis. This is where I am getting confused. Shouldn't I be finding a p-value? How do they know that this 1.972 means I can reject the null?

Please help. I am doing a scientific presentation using this data as well as publishing a paper (hopefully) and need this answered ASAP!

Thanks!

2. Originally Posted by agarcin
So, I am trying to perform a t-test using diversity indexes, but the biostat book I am using is a little bit confusing.

the calculated t value = -1.911
calculated v value = 196

then they say something about t0.05(2), 196 = 1.972
They seem to have gotten this value from the critical t value table.

They then say that 0.05 < P < 0.10 and that you do not reject the null hypothesis. This is where I am getting confused. Shouldn't I be finding a p-value? How do they know that this 1.972 means I can reject the null?

Please help. I am doing a scientific presentation using this data as well as publishing a paper (hopefully) and need this answered ASAP!

Thanks!
This is an hypothesis test, you decide on the significance level that is appropriate. In this case it appears to be a two sided 5% significance level test. That means that for the number of degrees of freedom that you have (196) you find the critical values of the test statistic (t) that will define an interval that contains 95% of all results if the null hypothesis is in fact true.

If your observed test statistic lies outside this interval you reject the null hypothesis at this level of significance.

So we see that for such a test we do not need to compute the p value corresponding to the test statistic.

CB

3. I got it.

I figured it out. If the t value found from the v value on the chart is less than the calculated t value then you may reject the null hypothesis.

4. Originally Posted by agarcin
I figured it out. If the t value found from the v value on the chart is less than the calculated t value then you may reject the null hypothesis.
Not quite the two sided test gives an interval (-1.972, 1.972) that the test statistic will lie in with probability 0.95, if the test statistic is outside this interval you reject the null hypothesis.

CB