What I say depends on how you ask: Experimental evidence of the effect of framing on the measurement of attitudes

We use a survey experiment to document the presence of framing effects in the measurement of attitudes. Next, using standard techniques for generating aggregate indices, we find that statement framing can meaningfully influence the relationship of the index with relevant covariates—in some cases changing the magnitude, statistical significance, and even the sign of the estimated relationship. We conclude by discussing how randomizing statement framing across respondents can help address bias in the measurement of attitudes.