T-test appropriate?

[zerostin]

Hello all,

I have a big dataset which I divided into different periods (years) and focused on one group. I calculated the means for the different time periods (for at least 3 variables) for this group, and now I want to verify if they are really different.

Is a T-test the most appropriate test if I assume that my data is normally distributed

[arpix]

You can use the t-test if you have two groups to compare, in your case it’s the years. If you have more than two groups, then you have to go for ANOVA. I assume the items were the same across the different years. If you have two groups, in this case the t test for two related samples (dependent t-test) would be the choice. You say you have at least three variables, in this case you might consider doing MANOVA (multivariate ANOVA).
Hope this helps.

[zerostin]

Yes, thank you very much.

[zerostin]

I still have a question about this, because it is a bit confusing.

Suppose I have the following data:

y2000	y2001	y2002	y2003	y2004	y2005	y2006
5603	5159	4661	4257	3973	3662	3279
0,38	-0,17	-0,28	0	-0,007	0,008	0,02
0,32	-0,021	-0,03	0	-0,008	0,005	0,016
471	-715	-845	-66	-319	372	1287
0,74	0,46	0,41	0,43	0,28	0,3	0,38
3,89	3,02	2,74	2,96	2,26	2,58	3,3
2,06	1,35	1,02	1	0,74	0,83	1,58
1,83	1,67	1,72	1,96	1,51	1,75	1,74

These are all means, and I want to know if they significantly differ from y2000. (e.g differs -0,17 significantly from 0,38, differs -0.021 significantly from 0,32, etc)
Is the Manova then still appropriate?

If I try a independent T-test with Stata, he takes all these observations together, but I want to take each observation separate