English Deutsch Français Italiano Español Português 繁體中文 Bahasa Indonesia Tiếng Việt ภาษาไทย
All categories

For a t-test is one for paired data and one for independent samples?

2006-10-25 08:04:16 · 6 answers · asked by sticky 7 in Science & Mathematics Mathematics

6 answers

The way you would structure a t-test is different for paired data and independent samples. Recall that a t-test is useful if you don't know the population variance and the sample size is not large. You must also assume that the random samples were selected from a normal population.

For independent samples, one often is testing a hypothesis of whether the means of the two samples are stastically different. We structure this as x1-x2 (or x1=x2), and then use one of several available statistical methods to test if this hypothesis is true. You can use a t-test for this with n1+n2-2 degrees of freedom.

For paired observations, the two means are not independent. Unlike the situation described above, the conditions are the two populations are connected and the pair have something in common. Therefore, one typically wants to know if there is a meaningful difference in the paried observations. We structure this as x1-x2=xd (the difference) and then test the hypothesis of whether this difference is statistically meaningful. You can use a t-test for this as well, with n-1 degrees of freedom.

I realize this is subtle, but hope this helps!

2006-10-25 09:28:49 · answer #1 · answered by nice2mice 2 · 0 0

The way you wrote your question is confusing and I nearly gave up on it
.
If x1 is the mean of a set of values and x2 is the mean of another set of values, and the two sets are not related, you apply the t-test to find out whether x1 - x2 is significantly different from zero, ie whether t1 - t2 = 0.
This is called the unrelated (or independent) t- test

If the two sets of values are related (for example the subjects under test were in matched pairs), then you apply the t-test to find out whether the mean difference x is significantly different from zero., ie whether x = 0
This is called the related (or dependent) t-test

2006-10-25 10:29:52 · answer #2 · answered by Anonymous · 1 0

if x1=x2 then x1-x2=0 in maths

2006-10-25 08:07:17 · answer #3 · answered by raj 7 · 0 1

absolutely nothing, what happens when u add x2 to both sides.

since u do it to both sides it preserves the equality!

2006-10-25 08:29:21 · answer #4 · answered by Johnny T 2 · 0 1

No difference.

2006-10-25 08:06:11 · answer #5 · answered by campbelp2002 7 · 0 1

It seems like they would be the same to me.

2006-10-25 08:06:20 · answer #6 · answered by I ♥ AUG 6 · 0 1

fedest.com, questions and answers