Y(Croprotation)=a=ba*time + b2*grpmem + b3*(time* grpmem)
Where,
Croprotation= 1 if HHs practice crop rotation and 0 otherwise
time takes 0 for baseline and 1 for the endline
grpmem is whether or not HH has membership into an agriculture group.
I ran this regression to evaluate the impact of membership into agriculture group on adoption of improved agriculture practices (Above example is for crop rotation). Command was logistic agpcrpr i.time##i.grpmem and I got following results
Table 1:
Odds Ratio | Std. Err. | z | P>z | [95% Conf. | Interval] | |
time | ||||||
endline | 3.1205 | 0.264247 | 13.44 | 0 | 2.643282 | 3.683874 |
grpmem | ||||||
Has membership | 1.549721 | 0.227346 | 2.99 | 0.003 | 1.162468 | 2.06598 |
time#grpmem | ||||||
endline*Has membership | 1.158914 | 0.236407 | 0.72 | 0.47 | 0.776988 | 1.728574 |
After this I also ran post estimation command
margins time#grpmem
margins grpmem, dydx(time)
and results were
Table 2:
Delta-method | ||||||
Margin | Std. Err. | z | P>z | [95% Conf. | Interval] | |
time#grpmem | ||||||
Baseline#No membership | 0.50034 | 0.0124978 | 40.03 | 0 | 0.475845 | 0.524835 |
Baseline#Has membership | 0.595741 | 0.0287983 | 20.69 | 0 | 0.539297 | 0.652184 |
Endline#No membership | 0.73365 | 0.0106975 | 68.58 | 0 | 0.712684 | 0.754617 |
Endline#Has membership | 0.824793 | 0.0177039 | 46.59 | 0 | 0.790094 | 0.859492 |
Table 3:
Delta-method | ||||||
dy/dx Std. Err. | z | P>z | [95% Conf. | Interval] | ||
0.time | (base outcome) | |||||
1.time | ||||||
grpmem | ||||||
No membership | group | .2333107 .0164791 | 14.16 | 0 | 0.201012 | 0.265609 |
Has membership | group | .2290519 .0334853 | 6.84 | 0 | 0.163422 | 0.294682 |
My questions:
- Can I say that (from Table 1) ; Having membership in agriculture group increases the probability of adopting crop rotation by 15% (though the result is not significant, I am just wondering about the interpretation). In that case what would be my reference group.
- How do I interpret the results from post estimation, Table 2 & Table 3? And how are they calculated?
0 Response to How do I interpret the results of Difference in Difference model
Post a Comment