Internal validity. Answer each of the following questions. Recall that we have discussed five different "threats to internal validity": omitted variables bias, wrong functional form (nonlinearity), measurement error, sample selection, and simultaneity. a. We estimate a regression of average hours an individual works per week on the number of children under the age of 5 in their household. To estimate the regression I only use individuals who work full time, i.e., those who work at least 30 hours every week. Restricting the sample in this way biases the coefficient estimate because of which threat to internal validity? b. I use state level data (one observation for each of the 50 states) to estimate a regression of the unemployment rate in the state on the duration of unemployment benefits in the state. If state governments change the duration of unemployment benefits in response to changes in the unemployment rate then the estimated coefficient will be biased. Which threat to internal validity is this? c. In a study of Americans ages 50-70 I estimate the regression model ln(earningsi)=0+1agei+ui and I find that 1 is not statistically significant. I suspect that this is because earnings increase with age from 50-60 but start to decrease with age after 60 or 65 years of age. d. In (a)-(c) I have described a specific threat to internal validity. Can you identify a second potential threat to internal validity for each? Make sure to explain your reasoning. Internal validity. Answer each of the following questions. Recall that we have discussed five different "threats to internal validity": omitted variables bias, wrong functional form (nonlinearity), measurement error, sample selection, and simultaneity. a. We estimate a regression of average hours an individual works per week on the number of children under the age of 5 in their household. To estimate the regression I only use individuals who work full time, i.e., those who work at least 30 hours every week. Restricting the sample in this way biases the coefficient estimate because of which threat to internal validity? b. I use state level data (one observation for each of the 50 states) to estimate a regression of the unemployment rate in the state on the duration of unemployment benefits in the state. If state governments change the duration of unemployment benefits in response to changes in the unemployment rate then the estimated coefficient will be biased. Which threat to internal validity is this? c. In a study of Americans ages 50-70 I estimate the regression model ln(earningsi)=0+1agei+ui and I find that 1 is not statistically significant. I suspect that this is because earnings increase with age from 50-60 but start to decrease with age after 60 or 65 years of age. d. In (a)-(c) I have described a specific threat to internal validity. Can you identify a second potential threat to internal validity for each? Make sure to explain your reasoning