preview

What Is The Credibility Of Republican Endorsements

Good Essays
Open Document

E(q | en = 0)= E
[
q | θn <

√ σ 2q +σ 2ε pn

]
=−βλr(pn), (8)

where the credibility of a Republican endorsement can be written as follows:

λr(pn)= φ (pn)
((pn)

. (9)

Similarly to the discussion of the credibility of Democratic endorsements, the credibility of
Republican endorsements is decreasing in the degree of a newspaper’s leaning to the right and such an endorsement from a left-leaning source provides more information to voters than does an endorsement from a right-leaning source.
Although we have taken editorial positions (pn) as exogenous here, in Appendix A, we ex-

plore two models with endogenous editorial positions, both of which follow the theoretical lit- erature on media bias. In a demand-side model, a monopoly …show more content…

Note that the voting equation depends upon the vector of

6. For the result regarding the monotonicity of the Mills ratio, see Heckman (1979).

“rdq037” — 2011/6/14 — 6:59 — page 802 — #8

802 REVIEW OF ECONOMIC STUDIES parameters from the endorsement equation (γ ), which is unobserved by the econometrician, and we thus estimate the model in two stages. In the first-stage, we estimate a Probit model in which newspaper endorsement decisions are related to newspaper characteristics (Zn). With the estimated parameters (γ̂ ) from this first stage, we can then compute the Mills ratios λd(γ̂ Zn) and λr(γ̂ Zn), which can be interpreted as generalized residuals from a Probit model (Gourier- oux et al., 1987). Finally, we then use these generalized residuals as generated regressors in the second-stage linear regression.
Given the presence of these generated regressors in the second stage, we compute the stan-

dard errors using bootstrapping techniques. In particular, we draw samples with replacement from the underlying set of newspapers and also independently draw samples with replacement from the underlying set of voters. The standard errors are based upon 200 replications. Note that this bootstrapping procedure assumes independence across voters and thus does not account for any potential clustering within newspapers. To address this issue, we also provide results below for a

Get Access
Related
Get Access