Dear all,
I have conducted a discrete choice experiment. The alternatives are three policies and there is a status-quo alternative (no policy). I use Stata17. For my basic analysis, I use cmxtmixlogit. As a robustness check (and to include individual characteristics, since it is an unlabeled experiment), I use nlogit. Now I struggle with the interpretation of the nested logit, since adding an alternative-specific constant (ASC) turns the attributes insignificant. The ASC is coded as follows: 1 for all policies, 0 for the status-quo alternative (no policy). For your information: the results from cmxtmixlogit are significant, also with the ASC.
Here are the two outputs:
Without ASC:
With ASC:
Can anyone help me in interpreting what's going on here?
Thank you in advance!
I have conducted a discrete choice experiment. The alternatives are three policies and there is a status-quo alternative (no policy). I use Stata17. For my basic analysis, I use cmxtmixlogit. As a robustness check (and to include individual characteristics, since it is an unlabeled experiment), I use nlogit. Now I struggle with the interpretation of the nested logit, since adding an alternative-specific constant (ASC) turns the attributes insignificant. The ASC is coded as follows: 1 for all policies, 0 for the status-quo alternative (no policy). For your information: the results from cmxtmixlogit are significant, also with the ASC.
Here are the two outputs:
Without ASC:
HTML Code:
RUM-consistent nested logit regression Number of obs = 7,536
Case variable: id Number of cases = 1884
Alternative variable: Alternative Alts per case: min = 4
avg = 4.0
max = 4
Wald chi2(4) = 742.85
Log pseudolikelihood = -1685.8381 Prob > chi2 = 0.0000
(Std. err. adjusted for clustering on id)
------------------------------------------------------------------------------
| Robust
Choice | Coefficient std. err. z P>|z| [95% conf. interval]
-------------+----------------------------------------------------------------
Alternative |
Costs | -.0645111 .0049329 -13.08 0.000 -.0741795 -.0548427
RecMat | .0187528 .0014709 12.75 0.000 .0158698 .0216358
GHG | .0205729 .0009496 21.66 0.000 .0187117 .0224342
Reuse | .0125093 .0011585 10.80 0.000 .0102388 .0147799
------------------------------------------------------------------------------
dissimilarity parameters
------------------------------------------------------------------------------
/nests |
policies_tau | .5548734 .0305705 .4949563 .6147904
none_tau | 1 .0008902 .9982553 1.001745
------------------------------------------------------------------------------
HTML Code:
RUM-consistent nested logit regression Number of obs = 7,536
Case variable: id Number of cases = 1884
Alternative variable: Alternative Alts per case: min = 4
avg = 4.0
max = 4
Wald chi2(5) = 796.41
Log pseudolikelihood = -1683.5947 Prob > chi2 = 0.0000
(Std. err. adjusted for clustering on id)
------------------------------------------------------------------------------
| Robust
Choice | Coefficient std. err. z P>|z| [95% conf. interval]
-------------+----------------------------------------------------------------
Alternative |
Costs | -.0231164 .0214353 -1.08 0.281 -.0651289 .018896
RecMat | .0066912 .0061856 1.08 0.279 -.0054324 .0188148
GHG | .007314 .0067917 1.08 0.282 -.0059975 .0206256
Reuse | .0044247 .0041392 1.07 0.285 -.003688 .0125373
ASC | 1.785674 .913887 1.95 0.051 -.0055119 3.576859
------------------------------------------------------------------------------
dissimilarity parameters
------------------------------------------------------------------------------
/nests |
policies_tau | .1970386 .1830114 -.1616571 .5557344
none_tau | 1 .4572899 .1037283 1.896272
------------------------------------------------------------------------------
Can anyone help me in interpreting what's going on here?
Thank you in advance!
