Hi All,
I am running a mixed logit model regression on my discrete choice experiment results. We have two scenarios, and I want to see how preferences vary in each scenario. For scenario 1, I have assigned "full analysis 1" and "full analysis 2" to have random distributions, as I believe peoples preferences will vary across these. I have attached my code and output below. Is it normal that the co-efficient signs would change from negative to positive, when they have fixed vs random distributions? I'm confused on how to interpret this. Many thanks for any advice!
mixlogit choice simplifiedanalysis_1 simplifiedanalysis_2 time_1 interest_holder_1 interest_holder_2 if scenario==1, group
> (ncs) id(record_id) rand(full_analysis_1 full_analysis_2) nrep(1000)
Iteration 0: Log likelihood = -234.92997 (not concave)
Iteration 1: Log likelihood = -223.98086
Iteration 2: Log likelihood = -223.01103
Iteration 3: Log likelihood = -222.94619
Iteration 4: Log likelihood = -222.94612
Iteration 5: Log likelihood = -222.94612
Mixed logit model Number of obs = 840
LR chi2(2) = 24.65
Log likelihood = -222.94612 Prob > chi2 = 0.0000
--------------------------------------------------------------------------------------
choice | Coefficient Std. err. z P>|z| [95% conf. interval]
---------------------+----------------------------------------------------------------
Mean |
simplifiedanalysis_1 | -1.042023 .2115606 -4.93 0.000 -1.456674 -.6273721
simplifiedanalysis_2 | -.8097927 .2130863 -3.80 0.000 -1.227434 -.3921512
time_1 | -.186743 .0441925 -4.23 0.000 -.2733588 -.1001272
interest_holder_1 | -.8321485 .2281618 -3.65 0.000 -1.279337 -.3849596
interest_holder_2 | -.6209333 .212516 -2.92 0.003 -1.037457 -.2044095
full_analysis_1 | -2.388495 .40041 -5.97 0.000 -3.173284 -1.603706
full_analysis_2 | -1.832243 .3741539 -4.90 0.000 -2.565571 -1.098915
---------------------+----------------------------------------------------------------
SD |
full_analysis_1 | 1.430649 .3894929 3.67 0.000 .6672571 2.194041
full_analysis_2 | 1.38886 .3490175 3.98 0.000 .7047987 2.072922
--------------------------------------------------------------------------------------
The sign of the estimated standard deviations is irrelevant: interpret them as
being positive
I am running a mixed logit model regression on my discrete choice experiment results. We have two scenarios, and I want to see how preferences vary in each scenario. For scenario 1, I have assigned "full analysis 1" and "full analysis 2" to have random distributions, as I believe peoples preferences will vary across these. I have attached my code and output below. Is it normal that the co-efficient signs would change from negative to positive, when they have fixed vs random distributions? I'm confused on how to interpret this. Many thanks for any advice!
mixlogit choice simplifiedanalysis_1 simplifiedanalysis_2 time_1 interest_holder_1 interest_holder_2 if scenario==1, group
> (ncs) id(record_id) rand(full_analysis_1 full_analysis_2) nrep(1000)
Iteration 0: Log likelihood = -234.92997 (not concave)
Iteration 1: Log likelihood = -223.98086
Iteration 2: Log likelihood = -223.01103
Iteration 3: Log likelihood = -222.94619
Iteration 4: Log likelihood = -222.94612
Iteration 5: Log likelihood = -222.94612
Mixed logit model Number of obs = 840
LR chi2(2) = 24.65
Log likelihood = -222.94612 Prob > chi2 = 0.0000
--------------------------------------------------------------------------------------
choice | Coefficient Std. err. z P>|z| [95% conf. interval]
---------------------+----------------------------------------------------------------
Mean |
simplifiedanalysis_1 | -1.042023 .2115606 -4.93 0.000 -1.456674 -.6273721
simplifiedanalysis_2 | -.8097927 .2130863 -3.80 0.000 -1.227434 -.3921512
time_1 | -.186743 .0441925 -4.23 0.000 -.2733588 -.1001272
interest_holder_1 | -.8321485 .2281618 -3.65 0.000 -1.279337 -.3849596
interest_holder_2 | -.6209333 .212516 -2.92 0.003 -1.037457 -.2044095
full_analysis_1 | -2.388495 .40041 -5.97 0.000 -3.173284 -1.603706
full_analysis_2 | -1.832243 .3741539 -4.90 0.000 -2.565571 -1.098915
---------------------+----------------------------------------------------------------
SD |
full_analysis_1 | 1.430649 .3894929 3.67 0.000 .6672571 2.194041
full_analysis_2 | 1.38886 .3490175 3.98 0.000 .7047987 2.072922
--------------------------------------------------------------------------------------
The sign of the estimated standard deviations is irrelevant: interpret them as
being positive
