15 Econometrics: Final

What is true about log-file?

• It is useless.

• Log-file can only be created from a do file.

• Log-file is Stata data format.

• It allows you to make a full record of your Stata session.

How can you combine if conditions in Stata?

• With & as "and" and | as "or".

• With & as "or" and | as "and".

• You cannot combine conditions, there can be only one per line.

• With () as "or" and [] as "and".

You are estimating 400 subsamples of a random variable x which is uniformly distributed between 3 and 5. What could you tell about the sampling distribution of the mean?

• It is impossible to predict.

• It will be chi-squared distributed with mean 3.

• It will be uniformly distributed with mean 4.

• It will be normally distributed with mean 4.

Is it true that there should be an additional column of ones in the X matrix in order to get reasonable estimates?

• Yes, always.

• No, never.

• Yes, if there are dummies in the regression.

• Yes, if the data is not demeaned.

After estimation you got y = 2.5 + 3.5*x, st.error of beta1 is 1, number of obs. is 421. Would you reject the null hypothesis of beta1 = 0.5 in favor of the alternative of not equal at 5% level?

• Yes, as 3 is greater than roughly 2.

• Yes, as 3.5 is greater than roughly 2.

• No, as 3 is greater than roughly 2.

• No, as 3.5 is greater than roughly 2.

What is the correct statement?

• t-statistic is F-statistic squared for simple beta1 = 0 test.

• F-statistic is t-statistic squared for simple beta1 = 0 test.

• F-statistic and t-statistic have no relation at all.

• t-statistic is F-statistic squared for joint significance of all coefficients test.

What is the correct statement?

• We can compare the models with different dependent variables using R-squared adjusted.

• We can compare the models with different dependent variables using R-squared.

• The greater the BIC, the better the model (given the same sample).

• None of the above.

In the presence of heteroscedasticity if White's robust standard errors are used, usually:

• Coefficients change, the standard errors are the same.

• Coefficients change, the standard errors become smaller.

• Coefficients do not change, standard errors become smaller.

• Coefficients do not change, standard errors become greater.

A person who transferred survey answers to computer was very tired and made some mistakes typing in the income section. Based on this dataset, you want to study the relationship of health (lhs) and income (rhs) estimate your model using OLS. Which problem would you face?

• Coefficients cannot be estimated consistently.

• Three is no problem at all.

• There will be more noise in the model.

• All the coefficients will have attenuation bias.