(Undergrad dissertation) I am running regressions on socio-economic factors of individuals 16-26 using UK data. When running an xtlogit regression (dependent variable being unemployment) on my control group (neurotypical- large sample) I manage to get decent results for both the individual specific and parental specific regressions (I'm assessing the impact individual vs parental parameters have on economic outcomes). However when I run the same models on the smaller, neurodiverse sample STATA does about 300 iterations ('not concave') and gives me a largely empty output table.
Given this I've also run LPM models (xtreg) on the same parameters and have obtained plausible and (hardly-as in very few variables) significant estimates. I know this isn't ideal and there are limitations given xtreg is mainly for continuous dependents- what caveats are worth mentioning or is it strongly discouraged?
Additionally- I've run all my models using eststo and have them all saved in my data- however when I try to tabulate I keep getting 'last estimates (.) not found' even though they are all in storage. I ask this because I've run 12 models on each group (eststo'd) I then appended these two datasets and wanted to only keep the stored estimates, tabulate and then repeat for each of my desired tables. Here is a snippet of the code:
//Log notEdu
keep _est_logit_NotEdu_MiND_FE _est_logit_NotEdu_MiNT_FE
esttab using Mlog_NotEdu.docx , se r2 ar2 title("Prob (%) of being out of Education") mtitles("ND" "NT")
//LPM notEdu
keep _est_lpm_NotEdu_MiND_FE _est_lpm_NotEdu_MiNT_FE _est_lpm_NoEdu_MpND_FE _est_lpm_NoEdu_Mpn1NT_FE
//Unemp
keep _est_lpm_unemp_MiND_FE _est_lpm_unemp_MiNT_FE _est_lpm_NotEdu_MpND_FE _est_lpm_unemp_Mpn1NT_FE
//wage
keep _est_reg_wage_MiND _est_reg_wage_MiNT _est_reg_wage_MpND _est_reg_wage_MpNT
//MH function
keep _est_reg_sf12_MiND _est_reg_sf12_MiNT
Given this I've also run LPM models (xtreg) on the same parameters and have obtained plausible and (hardly-as in very few variables) significant estimates. I know this isn't ideal and there are limitations given xtreg is mainly for continuous dependents- what caveats are worth mentioning or is it strongly discouraged?
Additionally- I've run all my models using eststo and have them all saved in my data- however when I try to tabulate I keep getting 'last estimates (.) not found' even though they are all in storage. I ask this because I've run 12 models on each group (eststo'd) I then appended these two datasets and wanted to only keep the stored estimates, tabulate and then repeat for each of my desired tables. Here is a snippet of the code:
//Log notEdu
keep _est_logit_NotEdu_MiND_FE _est_logit_NotEdu_MiNT_FE
esttab using Mlog_NotEdu.docx , se r2 ar2 title("Prob (%) of being out of Education") mtitles("ND" "NT")
//LPM notEdu
keep _est_lpm_NotEdu_MiND_FE _est_lpm_NotEdu_MiNT_FE _est_lpm_NoEdu_MpND_FE _est_lpm_NoEdu_Mpn1NT_FE
//Unemp
keep _est_lpm_unemp_MiND_FE _est_lpm_unemp_MiNT_FE _est_lpm_NotEdu_MpND_FE _est_lpm_unemp_Mpn1NT_FE
//wage
keep _est_reg_wage_MiND _est_reg_wage_MiNT _est_reg_wage_MpND _est_reg_wage_MpNT
//MH function
keep _est_reg_sf12_MiND _est_reg_sf12_MiNT
