Question 2. (Total: 30 points) You are a junior analyst for a political campaign

Question

Question 2. (Total: 30 points) You are a junior analyst for a political campaign, and you are asked to build a model to predict if an individual will turn out in the 2016 presidential election. You conduct a survey, and ask people how likely they are to vote in the 2016 election on a scale of 0-100. You use their name and address to append the publicly available turnout data for the previous three general elections from public voter files and append demographic data from a consumer file. This means you have the following information on an individual: if that individual voted2010, voted2012 and voted2014, their age, gender, race, and what party they are registered for (partyreg), and how likely they say they are to vote in the 2016 election (wotelik2016) (a) (5 points) Write out the regression model you will use to predict if someone will turn out in 2016.

Explanation / Answer

A-Regression model is

Votelik2016=beta0+beta1*voted2010+beta2*voted2012+ beta3*voted2014+ beta4*age+ beta5*gender+ beta6*race+ beta7*partyreg

Here gender, race and partyreg should be converted to dummy variables.

B-voted2012 is significantly related to voter’s stated likelihood of voting as the confidence interval does not includes zero and t-statistic is quite higher than 2. Corresponding to a unit increase in likelihood of voting in 2012 there is on an average an increase of 1.8 increase of likelihood of voting in 2016, holding other predictors fixed.

Voted2012 is significantly related to voter’s stated likelihood of voting as the confidence interval does not includes zero and t-statistic is quite higher than 2. This is causal as tendency of likelihood in 2016 related to earlier tendency in 2012.

C-the variables voted2010 and voted2014 should be deleted as they poses a problem of high multicollinearirty with voted 2012.

D- Existence of multicollinearity will inflate the standard error of estimate of voted2012 and hence have upward bias.

E-This is ratio scale as there are large values for ordinal variable 0-100

Related Questions

Navigate

• Browse (All)
• Browse Q
• Subjects

---

---

Integrity-first tutoring: explanations and feedback only — we do not complete graded work. Learn more.