1.
- The mean value of the voting rate in the treatment group and control group in 2002
was 0.674 and 0.555, respectively, with a difference of -0.111, and the p-value was
0.0000, indicating that there is a big difference between. Age, new registered voter,
voted in 2000 and voted in1998 also differ significantly in the two groups, indicating
that the two groups are quite different. The control group will not provide a good
counterfactual for the treatment group’s voting potential outcomes. - Comparing to Assignment 1, covariates this time have large differences except for
the gender variable female. This is because whether the phone can be answered
completely is related to age, new registration, whether to vote in 2000 and whether to
vote in 1998.
4. - From the regression results of the, after gradually introducing other control
variables into the basic model, and comparing the regression results of each model,
there are slight changes in the coefficients of each variable. It can be concluded that
the added covariates have no relationship with whether listening to the call. - Comparing the regression results to Assignment 1, there is no significant change
between the two regression results. With the addition of control variables, the treat or
contact coefficients are both significant at the 1% significance level, but the
coefficients gradually become smaller. - From the regression results, with the addition of control variables, the R2 of the
model gradually increases, indicating that the fitting effect of the model is getting
better and better. Adding control variables can reduce the impact of missing variables
in the model on the regression of the model. vote00 and vote98 can reduce the
regression bias the most, mainly because R2 is significantly improved after adding
two variables. - No, in addition to the omitted variables, the regression error also affects the
measurement error of the variable observation value. Due to the accuracy of the
measurement tool and the incorrect measurement method, the observation value is
not completely consistent with the true value.
In order to simplify the model, the linear model is used to replace the non-linear
relationship, or the simple non-linear model is used to replace the complex non-linear
relationship, causing the error of inaccurate model relationship. Since people’s
understanding of economic laws is not completely consistent with the objective
economic laws themselves, it will also cause inaccurate errors in the model
relationship.