How Should Negative Improvement Be Recorded in Pre-Post Test Data Analysis?

[craig100]

Hello There,

I'm after a little advice; I have data collected during experimentation that was designed to determine the improvement of two groups of subjects between a pre and a post-test. The results are as follows:
Pre & Post test numbers represent the number of errors made
Group 1:

Pre|Post|Improvement
0	0	0
1	0	1
5	4	1
12	0	12
2	0	2
7	6	1
0	0	0
13	0	13

Group 2:

Pre|Post|Improvement
0	1	-1
2	0	2
1	0	1
0	0	0
0	11	-11
5	23	-18

I'm applying the t-test to determine if there is a statistical difference between the means(of the improvement columns).

My question is really simple(hopefully)... in group 2, should I be recording improvement as a negative number? or simply zero?

Using zero will not give me a significant statistical difference between the means of group 1 and group 2... using the negative numbers does.

Also, if I'm keeping the negative improvement numbers, is it okay to represent improvement as a negative percentage?

I've not done much with statistics, so would appreciate your input.

Thanks,
Craig.

 

[statdad]

Hello There,

I'm after a little advice; I have data collected during experimentation that was designed to determine the improvement of two groups of subjects between a pre and a post-test. The results are as follows:
Pre & Post test numbers represent the number of errors made
Group 1:

Pre|Post|Improvement
0	0	0
1	0	1
5	4	1
12	0	12
2	0	2
7	6	1
0	0	0
13	0	13

Group 2:

Pre|Post|Improvement
0	1	-1
2	0	2
1	0	1
0	0	0
0	11	-11
5	23	-18

I'm applying the t-test to determine if there is a statistical difference between the means(of the improvement columns).

My question is really simple(hopefully)... in group 2, should I be recording improvement as a negative number? or simply zero?

Using zero will not give me a significant statistical difference between the means of group 1 and group 2... using the negative numbers does.

Also, if I'm keeping the negative improvement numbers, is it okay to represent improvement as a negative percentage?

I've not done much with statistics, so would appreciate your input.

Thanks,
Craig.

Since you are subtracting in the order Pre-Post, a negative difference indicates a higher value in the post data - that's what you are hoping to find, if I understand you correctly.

Note that you are free to subtract in either order: if the negative differences are giving you a conceptual problem, subtract in the other order.

I would be more concerned about doing t-tests with such small samples.