I have the folllowing models:
fmlogit dropout change stay, eta(perc_female perc_Parent_HE matric1op ent_int ent_rga ent_m23 i.sector i.field_study i.estab)
margins, dydx(*)
and
fracreg logit delayed perc_female perc_Parent_HE matric1op ent_int ent_rga ent_m23 i.sector i.field_study i.estab
margins, dydx(*)
However, I want to examine the impact of the covid years through a panel. I have done interactions with all the variables and a dummy for covid years. Yet the results are not great, my coefficients don't change much:
Labels: Models (1) Multinominal Fractional Logit Model with control variables, (2) Multinominal Fractional Logit Model with control variables interacted with COVID-19 years (2019 – 2021)
Notes: t statistics in parentheses. Significance levels: * p<0.05, ** p<0.01, *** p<0.001
Source: Created by the authors.
Labels: Models (1) Fractional Logit Model with control variables, (2) Fractional Logit Model with control variables interacted with COVID-19 years (2019 – 2021)
Notes: t statistics in parentheses. Significance levels: * p<0.1, ** p<0.01, *** p<0.001
Source: Created by the authors.
Now I would like to try the work achieved by Papke, L.E. and J.M. Wooldridge (2008). Except I am not sure I could adapt the code provided in this paper.
Can you please help me how to get the marginal average effects from the adapted GLM model, as studied in the same work.
Also if you have any insights on how to improve my work, would be highly appreciated.
Thanks a lot
fmlogit dropout change stay, eta(perc_female perc_Parent_HE matric1op ent_int ent_rga ent_m23 i.sector i.field_study i.estab)
margins, dydx(*)
and
fracreg logit delayed perc_female perc_Parent_HE matric1op ent_int ent_rga ent_m23 i.sector i.field_study i.estab
margins, dydx(*)
However, I want to examine the impact of the covid years through a panel. I have done interactions with all the variables and a dummy for covid years. Yet the results are not great, my coefficients don't change much:
| Multinominal Fractional Logit (1) |
Multinominal Fractional Logit (2) |
|||||
| Dependent variable: | Dropout | Transfer | Stay | Dropout | Transfer | Stay |
| Prop Female | - .0475*** (.01) |
- .0339*** (.00) |
.0814*** (.01) |
- .0469*** (.01) |
- .0345*** (.01) |
.0814*** (.01) |
| Prop Parent HE | - .0826*** (.01) |
- .0011 (.01) |
.0814*** (.01) |
- .0860*** (.01) |
.0066 (.01) |
.0793*** (.01) |
| Enrolled 1st op | - .0214*** (.00) |
- .1108*** (.01) |
.1322*** (.01) |
- .0208*** (.00) |
- .1133*** (.01) |
.1341*** (.01) |
| Entry Internationals | .1406*** (.01) |
.0185 (.02) |
- .1591*** (.02) |
.1112*** (.01) |
.0182 (.02) |
- .1302*** (.02) |
| Entry General | - .1099*** (.01) |
.0555*** (.01) |
.0544*** (.02) |
- .1055*** (.01) |
.0525*** (.01) |
.0530*** (.02) |
| Entry >23 | .0751*** (.01) |
- .0071 (.02) |
- .0679*** (.03) |
.0720*** (.01) |
- .0038 (.02) |
- .0683** (.03) |
| Sector | .0016 (.00) |
- .0176* (.01) |
.0160 (.01) |
.0011 (.00) |
- .0174* (.01) |
.0163 (.01) |
| Covid | .0135*** (.00) |
- .0042* (.00) |
- .0092** (.00) |
|||
| N | 8.054 | 8.054 Yes Yes |
||||
| Institution Fixed Effects | Yes | |||||
| Field of Study Fixed Effects | Yes | |||||
Notes: t statistics in parentheses. Significance levels: * p<0.05, ** p<0.01, *** p<0.001
Source: Created by the authors.
| Fractional Logit (1) |
Fractional Logit (2) |
|
| Dependent variable: | Delayed Completion | Delayed Completion |
| Prop Female | - .2816*** (.02) |
- 0.2800*** (.02) |
| Prop Parent HE | - .1973*** (.02) |
- .1868*** (.02) |
| Enrolled 1st op | - .1633*** (.01) |
- .1685*** (.02) |
| Entry Internationals | .0940** (.03) |
.0881* (.04) |
| Entry General | - .1269*** (.02) |
- .1253*** (.02) |
| Entry >23 | .0765** (.04) |
.0768* (.04) |
| Sector | - .0138 (.02) |
- .0135 (.02) |
| Covid | - .0350*** (.01) |
|
| N | 6.078 | 6.078 |
| Institution Fixed Effects | Yes | Yes |
| Field of Study Fixed Effects | Yes | Yes |
Notes: t statistics in parentheses. Significance levels: * p<0.1, ** p<0.01, *** p<0.001
Source: Created by the authors.
Now I would like to try the work achieved by Papke, L.E. and J.M. Wooldridge (2008). Except I am not sure I could adapt the code provided in this paper.
Can you please help me how to get the marginal average effects from the adapted GLM model, as studied in the same work.
Also if you have any insights on how to improve my work, would be highly appreciated.
Thanks a lot
