10,500 reputation
11441
bio website
location San Jose, CA
age 35
visits member for 3 years, 5 months
seen 10 hours ago

Mad data scientist.


21h
revised Post-estimation tests for ordinal probit
added 8 characters in body
21h
revised Post-estimation tests for ordinal probit
deleted 24 characters in body; edited tags; edited title
21h
comment Ordinal probit model tests
There's a user-written command omodel for testing the parallel regression assumption.
21h
answered Post-estimation tests for ordinal probit
1d
comment What is a data analysis book that is as math intense as Lattin's Analyzing Multivariate Data and Everitt's Applied Multivariate Data Analysis?
Neither of these seems particularly "mathy" to me. Can you clarify what you are looking for? Also, data analysis seems pretty broad. Are there particular areas, like multivariate data, that interest you?
1d
comment Ordinal probit model tests
See this answer and spost suite of command helpfile (particularly fitstat).
Apr
15
revised Significance of $1$ in the model: $Y_i=1[B_0+B_1X_i\geq \epsilon_i]$ in Binary Choice Model?
deleted 41 characters in body
Apr
15
revised Difference between different autoregressive models
added 85 characters in body
Apr
14
revised Multilevel regression modelling with replicate weights in stata
added 4 characters in body
Apr
14
answered Multilevel regression modelling with replicate weights in stata
Apr
14
revised After Monte Carlo simulations, should I do multiple test correction?
edited title
Apr
13
comment How to argue omitted variable problem is alleviated?
In the notation above, the first are the $x$s, the second group are the $z$s. The equation above is the data generating process, not an actual regression.
Apr
13
comment How to argue omitted variable problem is alleviated?
@AleksandrBlekh I think I finally understand your question. It is possible there are a set of unobserved controls which do not share covariance properties with the observed controls. This procedure will not deal with that sort of bias at all.
Apr
13
comment How to argue omitted variable problem is alleviated?
@AleksandrBlekh I am not sure what you mean. Once you make assumptions about $\delta$ and the hypothetical $R^2_{MAX},$ this offers a way to get at the magnitude of the OVB from how the $R^2$ and the coefficients change as you add observed covariates.
Apr
13
comment How to argue omitted variable problem is alleviated?
A $\delta$ of one indicates that the observed and unobserved variables have an equally important effect on the coefficient of interest. A $\delta$ greater than one indicates that the unobserved variables are relatively more important. The other ingredient for the bias is the hypothetical $R^2$ from that regression if you had access to the $z$s. I am not aware of an R implementation.
Apr
13
comment How to argue omitted variable problem is alleviated?
@AleksandrBlekh Consider a model $$ y=\alpha + \beta \cdot x + \Sigma_{j=1}^{C} \gamma_j \cdot x_j +\Sigma_{k=1}^{U} \eta_k \cdot z_k + \varepsilon = \alpha + \beta \cdot x + C + U + \varepsilon,$$ where $y$ is the outcome, $x$ is the variable of interest, $x_j$ are the controls and $z_k$ are the unobserved variables. The proportionality of selection relationship is defined as $$\delta = \frac{\frac{Cov(U,t)}{Var(U)}}{\frac{Cov(C,t)}{Var(C)}}.$$ Essentially, $\delta$ is the ratio of univariate regression coefficients of $x$ on $U$ and $x$ on $C$.
Apr
13
answered How to argue omitted variable problem is alleviated?
Apr
12
comment Discrete choice model
@marquidecarabas Nope, you're right. I must have run that right after I upgraded.
Apr
11
comment How to deal with 'cut-off' selection bias/sampling bias? (truncated distribution)
You might consider truncated regression, but it probably won't work very well with this degree of truncation.
Apr
11
comment How to deal with 'cut-off' selection bias/sampling bias? (truncated distribution)
@Leo The 2.73 is the overall mean for both rural and urban women in the sample. In the first regression, the constant 2.813102 is the mean among urban women, for rural women the mean is 2.813102 -.2936003 = 2.5195017. You can carry out a similar exercise for the interval regression to get 2.755219 and 2.408345. NB, you are not just using 54 uncensored observations. You are also using the 434 where you only know that log wage is in $(-\infty,3.367296]$, which is the known censoring point. A simpler version of this model is called a tobit and is more widely available in other software.