A paper by Gregory Martin and Ali Yurukoglu (2017) studies the effects of exposure to cable news bias on voting patterns. In particular, they wish to run the following regression VoteShare = Bo + B1 Minutes Watched + (1) where the dependent variable is the Republican party's vote share in zip code z during the 2008 presidential election and the independent variable is the average number of minutes of Fox News watched (Fox News is understood to be a conservative-leaning cable news network). The error term is Ez. (a) In words (no math) state the zero conditional mean assumption which must hold in order for B to have a causal interpretation. Why is there reason to believe that the assumption fails in this case? (b) The authors propose an instrument for exposure to media bias: the channel positions of news channels in cable television lineups. The channel position is the ordinal position of news. channels in the cable lineup, which varies by zip code. What are two conditions that must hold in order for this to be deemed a "good" instrument? Which of these conditions can be tested and which cannot? (c) Consider the condition in (b) that you determined cannot explicitly be tested. Do you believe that this condition holds? Explain why or why not. (d) The table below details the results of the first stage regression (with standard errors reported below the point estimates, in parenthesis). Interpret the sign of the first stage regression coefficient and discuss whether the instrument is weak. Fox News Channel position Observations R Fox News Channel Minutes per Week -0.174 (0.028) 59,541 0.377 Table 1: First Stage Regression (e) Suppose that you add an exogenous variable to regression (1) denoting the share of women in zip code z, (FemaleShare). Write out the reduced form equation for this regression model.