Notable findings:
5 most controversial:
Estimates highly correlated with right wing views: ( p < .05 in italics, p < .01 in bold):
#code:
d <- public.personas.IQ.data
c <- rnorm(57)
e <- data.frame(c)
e$name = NA
e$median = NA
e$sd = NA
e$rwithrw = NA
e$pwithrw = NA
for(i in 14:70) {
d[, i][d[, i]<70] <- NA
print(colnames(d[i]))
print(median(d[, i], na.rm=T))
print(sd(d[, i], na.rm=T))
a <- cor.test(d[, 71], d[, i])
print(a$p.value)
print(a$estimate)
e$name[i-13] <- substring(colnames(d[i]),7)
e$median[i-13] <- median(d[, i], na.rm=T)
e$sd[i-13] <- sd(d[, i], na.rm=T)
e$rwithrw[i-13] <- a$estimate
e$pwithrw[i-13] <- a$p.value
}
>seb has paid subscriptions now
based. Hope you're getting sheckles.
You can estimate politically neutral estimates by regressing the value on the political views, and predicting the value at political centrism. I've used this approach in a number of analyses. This will probably change the rank-ordering quite a bit, primarily by fixing the issue with Noah Smith and Matt Yglesias.