Look for significant difference between two groups

[Doris Schmidt]

Hello,

I run a survey in which the participants have to decide how they would allocate money between an ingroup member and an outgroup member.

I have two groups. The first group will answer the questions hypothetically and the second group will do it incentivized (so their allocation is really implemented)

Now I have to look if there is a significant difference in what they give to the ingroup between the hypothetical and the incentivized group. (Does the incentivized group give significantly more or less to the ingroups than the hypothetical group?)

I was told to do it with a dummy, where the incentivized group is like a treatment group. But I do not really get how I can implement it in Stata.

Many thanks in advance

[Carlo Lazzaro]

Doris:
an idea might be:

svy: logit <depvar> i.group <otherregressors> <controls>

[Doris Schmidt]

Thanks for your reply, but I don't get it yet.

Here is my pre-test data:
1st column: for each participant in the incentivized group the allocation to the ingroup
2nd column: for each participant in the hypothetical group the allocation to the ingroup

Now I have to look if there is a significant difference between what those groups give to the ingroup (I hope that there is no significant difference, so I can put the observation of both groups together in one variable)

inc_|
|----------|
1. | . |
2. | . |
3. | . |
4. | . |
5. | 0 |
|----------|
6. | 70 |
7. | . |
8. | . |
9. | 100 |
10. | . |
|----------|
11. | 30 |
12. | . |
13. | . |
14. | 60 |
15. | 60 |
|----------|
16. | . |
17. | 50 |
18. | . |
19. | 50 |
20. | . |
|----------|
21. | 40 |
22. | . |
+----------+





hyp |
|----------|
1. | . |
2. | 40 |
3. | 60 |
4. | . |
5. | . |
|----------|
6. | . |
7. | . |
8. | 50 |
9. | . |
10. | 30 |
|----------|
11. | . |
12. | 60 |
13. | 70 |
14. | . |
15. | . |
|----------|
16. | 50 |
17. | . |
18. | 80 |
19. | . |
20. | 0 |
|----------|
21. | . |
22. | 40 |

[Carlo Lazzaro]

Doris:
my previous reply heavily drew upon your description.
1) what is the metric of the question (regressand)? I thought it was a yes/no answer, whereas it seems a share out of 100;
2) did your data come from a survey with PSU and related stuff or by survey you meant, in general term, an inquiry?

[Doris Schmidt]

1) the participants have 100€ and decide how they allocate it between an ingroup and an outgroup member (it is a slider that goes from 0 to 100, on the left side is the outgroup on the right side the ingroup)
2) it is a survey that I created on Qualtrics for my thesis

[Carlo Lazzaro]

Doris:
as per you last clarification, I would go:

. label define group 0 "incentivized" 1 "hypothetical"

. label val group group

. regress ingroup i.group

      Source |       SS           df       MS      Number of obs   =        19
-------------+----------------------------------   F(1, 17)        =      0.07
       Model |  45.8479532         1  45.8479532   Prob > F        =    0.7900
    Residual |  10648.8889        17  626.405229   R-squared       =    0.0043
-------------+----------------------------------   Adj R-squared   =   -0.0543
       Total |  10694.7368        18  594.152047   Root MSE        =    25.028

-------------------------------------------------------------------------------
      ingroup | Coefficient  Std. err.      t    P>|t|     [95% conf. interval]
--------------+----------------------------------------------------------------
        group |
hypothetical  |  -3.111111   11.49961    -0.27   0.790    -27.37317    21.15095
        _cons |   51.11111   8.342696     6.13   0.000     33.50956    68.71266
-------------------------------------------------------------------------------
.

[Doris Schmidt]

Ok many thanks, I think this works