Abstract:
How do cross-cutting social divisions affect public opinion? The COVID-19 pandemic generated cleavages based on compliance with public health suggestions related to masking and vaccination that sometimes aligned with partisan identity and sometimes cut across it. But distinguishing ambivalence from avoidance – survey satisficing – among responses of cross-pressured individuals is difficult using traditional, forced-choice survey items. By combining forced-choice and open-ended responses analyzed using human-in-the-loop large language model–assisted coding, we are able to show the the middling forced choice responses of cross-pressured individuals reflects ambivalence rather than avoidance. The responses are noisier, but in ways that suggest internal conflict rather than disengagement. Substantively, our findings highlight the importance of integrating open- and closed-ended data to address methodological issues in the interpretation of survey responses from cross-pressured individuals that prove essential for demonstrating how individuals respond when cross-pressured.