4. [13 points] a. Below is a snapshot of Figure 5 from Kearney and Levine (2014). How does this figure b. 16 and Pregnant/Teen Mom Ratings relate to the paper's identification strategy? B 5 + New Orleans + * # Birmingham 4 +* +* + + +* # lonesboro 3 * * * +* * : % . 2 # San Antonio * : s '0 P + . 4 Houston o \" . \""' * * & & # Austin 2 * - 0' * apt +% o + + + * * v & u * . ' \"4 - *, '~.. "' * & . * % R 'r" . U5 o", .o " o * - e o see s > +* +* + * 4t = * & oy 0e & 0 0.2 0.4 06 o8 i 12 14 i6 i8 MTV Shows, 2008/2009 In Table 3 of the paper, the authors estimate a regression where the dependent variable is the log of the tweet rate for the term \"birth control,\" and the key independent variable is the log of the tweet rate for the term \"16 and pregnant.\" The coefficient from this regression is 0.077. Interpret this coefficient. Can we reject the null hypothesis that the coefficient is zero at the 5% significance level? c. Now consider the following equation from the paper. The dependent variable is the log of the conception rate (births minus nine months) in market area j and quartert. The second through the eighth terms are interactions between 16 & Pregnant ratings and indicators for the quarters relative to the debut of the show (so for example, preQ3 is an indicator for the quarter three periods before the show debuts). In(B;) = fo + ai(RATE16P;*preQ3;) + ax(RATE16P;*preQ2;) + a;(RATE16P;*preQ1;) + BIRATEI6P, *postO1,) + PRATE16P,*postQ2;) + Ps(RATEI6P,*post(3,) + BRATE16PE*postQ4+;) + BUy + 31 + 0; + & Suppose we can reject the null hypothesis that a1 = a2 = a3 = 0. How would this affect your thinking about whether the paper has a credible identification strategy? d. Suppose that our estimate of B1=-0.05. How would you interpret this result? (Note: 16 & Pregnant ratings are measured as Nielsen points.) e. The term &; represents market-area fixed effects. Why is it important to include these fixed effects