hey guys, I've developed a logit model to be applied to three different sets of cross-sectional data(the same model specification with differenet samples). What I'm trying to uncover is whether there are changes in the substantive effect of a given independent variable (IV) on the dependent variable (DV) controlling for other explanations at different times and across time. the first two models are logit models and third model is exlogistic model because the IV "interdependence" always associates with the same outcome when interdependence=!1
My questions are

1)How do I assess increased / decreased size in the association between the IV and DV? (2) Can I simply look at the different magnitudes of the coefficients across the logit and exlogistic models ?(3) why the the standard error of the IV "interdependence" is N/A
the first model
Iteration 0: log likelihood = -36.563979 |
Iteration 1: log likelihood = -35.084441 |
Iteration 2: log likelihood = -35.001635 |
Iteration 3: log likelihood = -35.001431 |
Iteration 4: log likelihood = -35.001431 |
Logistic regression |
Number of obs |
= 99 |
LR chi2(4) |
= 3.13 |
Prob > chi2 |
= 0.5371 |
Log likelihood = -35.001431 |
Pseudo R2 |
= 0.0427 |
|
|
collaboration Odds Ratio Std. Err. |
z P>z |
[95% Conf. Interval] |
|
interdependence1 2.21211 1.580962 |
1.11 0.267 |
.5450864 8.977344 |
novelty .9071175 .4687204 |
-0.19 0.850 |
.3294851 2.497419 |
incentive 1.200299 .4840858 |
0.45 0.651 |
.5444981 2.645957 |
incident1 .9977701 .0023237 |
-0.96 0.338 |
.993226 1.002335 |
_cons .178827 .2427856 |
-1.27 0.205 |
.0124967 2.55901 |
|
Note: _cons estimates baseline odds. |
the second model
. logit collaboration interdependence1 novelty incentive incident1 if period==2,or |
Iteration 0: log likelihood = -79.159074 |
Iteration 1: log likelihood = -52.106663 |
Iteration 2: log likelihood = -47.194014 |
Iteration 3: log likelihood = -46.644318 |
Iteration 4: log likelihood = -46.630959 |
Iteration 5: log likelihood = -46.630915 |
Iteration 6: log likelihood = -46.630915 |
Logistic regression Number of obs = 156 |
LR chi2(4) = 65.06 |
Prob > chi2 = 0.0000 |
Log likelihood = -46.630915 Pseudo R2 = 0.4109 |
|
collaboration Odds Ratio Std. Err. z P>z [95% Conf. Interval] |
interdependence1 70.25622 73.49304 4.06 0.000 9.042052 545.8868 |
novelty 1.77801 .5997315 1.71 0.088 .917952 3.443885 |
incentive 1.497619 .5009724 1.21 0.227 .7774339 2.884954 |
incident1 1.002408 .006256 0.39 0.700 .9902215 1.014745 |
_cons .0020589 .003575 -3.56 0.000 .0000685 .0618914 |
Note: _cons estimates baseline odds. |
the third model
observation 63: enumerations = 1411544 |
note: CMLE estimate for interdepen~1 is +inf; computing MUE |
Exact logistic regression Number of obs = |
63 |
Model score = |
23.69152 |
Pr >= score = |
0.0000 |
|
collaborat~n Odds Ratio Suff. 2*Pr(Suff.) [95% Conf. |
Interval] |
|
interdepen~1 32.16168* 21 0.0000 4.767067 |
+Inf |
novelty .8263797 30 0.7101 .3716871 |
1.627337 |
incentive 3.686022 35 0.0270 1.118811 |
24.93683 |
incident1 .9940054 4419 0.2823 .9857038 |
1.003811 |
|
(*) median unbiased estimates (MUE) |
|
. estat se |
|
collaborat~n Odds Ratio Std. Err. |
interdepen~1 32.16168* N/A |
novelty .8263797 .2806791 |
incentive 3.686022 2.326282 |
incident1 .9940054 .0046159 |
(*) median unbiased estimates(MUE) |
|
0 Response to Comparing logistic regression coefficients across models?
Post a Comment