For the 2014 GSS, Table 11.20 shows estimates

(with se values in parentheses) for four regression models for y = political party identification in the United States, scored from 1 = strong Democrat to 7 = strong Republican. The explanatory variables are number of years of education in model 1, also annual income last year (12 ordered categories, scored 1 to 12) in model 2, also religion (1 = fundamentalist, 2 = moderate, 3 =

liberal) in model 3, and also political views (scored from 1 = extremely liberal to 7 = extremely conservative) in model 4.

(a) Summarize what you learn from these four model fits.

(b) The effect of religion weakens considerably (and even changes direction) after adding political views to the model. Is it plausible that the relationship between y and religion is spurious? Explain.