Good morning all,
I am trying to run a multinomial logit with endogenous variable using the gsem syntax. The challenge, however, is that each time I try running the model, the output replicates all the variables in the dataset and gives the error: If this is not your intention use the nocapslatent option, or identify the latent variable names in the latent() option.
Let me give a brief introduction to what am try to estimate:
I want to estimate the determinants of household poverty status and access to finance is envisaged to be endogenous to welfare.
- Household welfare poverty is either Never Poor (0), Chronically Poor (1), Exit Poverty (2) and Enter Poverty (3)
- Access to finance (AccFin) is 1 if household has accessed to finance, zero otherwise
Multinomial logit probabilities are functions of individual, household and locational factors: Hhd_Age Hhd_Age2 AccFin Marstat Urb
Access to finance (AccFin) is endogenous and is affected by individual, household and locational factors: Hhd_Age Marstat Internet Land Urb
Following the framework in Drukker (2014), I ran the following:
gsem (Hhd_POVCXT <- Hhd_Age Hhd_Age2 AccFin Marstat Urb, mlogit) (Hhd_ACCESSFIN <- Hhd_Age Marstat Internet Land Urb), nolog
The output was, however, confusing as it generated the following:
note: The following observed variable names will be treated as latent variables: Hhdhead_SpouseinHse, Hhdhead_SOCIALACTIVITY, Hhdhead_MARRIED, Hhdhead_MARSTAT, Hhdhead_DIVORCED, Hhdhead_SINGLE, Hhdhead_WIDOWED, Hhdhead_NOSCHOOL, Hhdhead_PRYSCHOOL, Hhdhead_SECSCHOOL, Hhdhead_POSTSECEDUC, Hhdhead_UNIEDUC, Hhdhead_OTHEREDUC, Hhdhead_EDUC, Hhd_EMPLOYER, Hhdhead_GOVEMPLOYED, Hhdhead_SELFEMPLOYED, Hhdhead_NGOCOOPEMPLOYED, Hhd_POVMATRIX, Hhd_CHRONICPOV, Hhd_ENTERPOV, Hhd_EXITPOV, Hhd_NEVERPOV, AveHhdSize_CHRONICPOV, AveHhdSize_EXITPOV, AveHhdSize_ENTERPOV, AveHhdSize_NEVERPOV, Hhd_POVCXT, N_Poor, C_Poor, X_Poor, E_Poor, Never_Poor, Chronically_Poor, Exit_Poverty, Enter_Poverty, Ave_CommEduc, Comm_FORMALBANKY1. If this is not your intention use the nocapslatent option, or identify the latent variable names in the latent() option.
note: Latent variable Hhd_POVCXT was specified with option family(multinomial), but family(gaussian) is the only option allowed. Assuming family(gaussian) for Hhd_POVCXT. note: Latent variable Hhd_POVCXT was specified with option link(logit),but link(identity) is the only option allowed. Assuming (identity) for Hhd_POVCXT.
model not identified;
no paths from latent variable Hhd_POVCXT to observed variables
It will be appreciated if anyone could offer help on this, please. Thanks in advance
I am trying to run a multinomial logit with endogenous variable using the gsem syntax. The challenge, however, is that each time I try running the model, the output replicates all the variables in the dataset and gives the error: If this is not your intention use the nocapslatent option, or identify the latent variable names in the latent() option.
Let me give a brief introduction to what am try to estimate:
I want to estimate the determinants of household poverty status and access to finance is envisaged to be endogenous to welfare.
- Household welfare poverty is either Never Poor (0), Chronically Poor (1), Exit Poverty (2) and Enter Poverty (3)
- Access to finance (AccFin) is 1 if household has accessed to finance, zero otherwise
Multinomial logit probabilities are functions of individual, household and locational factors: Hhd_Age Hhd_Age2 AccFin Marstat Urb
Access to finance (AccFin) is endogenous and is affected by individual, household and locational factors: Hhd_Age Marstat Internet Land Urb
Following the framework in Drukker (2014), I ran the following:
gsem (Hhd_POVCXT <- Hhd_Age Hhd_Age2 AccFin Marstat Urb, mlogit) (Hhd_ACCESSFIN <- Hhd_Age Marstat Internet Land Urb), nolog
The output was, however, confusing as it generated the following:
note: The following observed variable names will be treated as latent variables: Hhdhead_SpouseinHse, Hhdhead_SOCIALACTIVITY, Hhdhead_MARRIED, Hhdhead_MARSTAT, Hhdhead_DIVORCED, Hhdhead_SINGLE, Hhdhead_WIDOWED, Hhdhead_NOSCHOOL, Hhdhead_PRYSCHOOL, Hhdhead_SECSCHOOL, Hhdhead_POSTSECEDUC, Hhdhead_UNIEDUC, Hhdhead_OTHEREDUC, Hhdhead_EDUC, Hhd_EMPLOYER, Hhdhead_GOVEMPLOYED, Hhdhead_SELFEMPLOYED, Hhdhead_NGOCOOPEMPLOYED, Hhd_POVMATRIX, Hhd_CHRONICPOV, Hhd_ENTERPOV, Hhd_EXITPOV, Hhd_NEVERPOV, AveHhdSize_CHRONICPOV, AveHhdSize_EXITPOV, AveHhdSize_ENTERPOV, AveHhdSize_NEVERPOV, Hhd_POVCXT, N_Poor, C_Poor, X_Poor, E_Poor, Never_Poor, Chronically_Poor, Exit_Poverty, Enter_Poverty, Ave_CommEduc, Comm_FORMALBANKY1. If this is not your intention use the nocapslatent option, or identify the latent variable names in the latent() option.
note: Latent variable Hhd_POVCXT was specified with option family(multinomial), but family(gaussian) is the only option allowed. Assuming family(gaussian) for Hhd_POVCXT. note: Latent variable Hhd_POVCXT was specified with option link(logit),but link(identity) is the only option allowed. Assuming (identity) for Hhd_POVCXT.
model not identified;
no paths from latent variable Hhd_POVCXT to observed variables
It will be appreciated if anyone could offer help on this, please. Thanks in advance
