Thread by @matt_motta

NEW at IJPOR - I show that aggregate change in anthropogenic climate change beliefs is less the result of attitude change ("changing minds") and more the result of demographic changes in sample composition ("changing samples.")

[THREAD]

academic.oup.com/ijpor/advance-…

Before summarizing the study, HUGE shoutout to the folks at C[C]ES, @pewresearch, and @YaleClimateComm who did the hard work of collecting the panel and cross-sec. data used in this study, and for making it free for all to use.

Next, some background: tracking polls have documented increases in the number of Americans who believe that climate change is human caused (anthropogenic: "ACC"), over the past decade. This could result from one of two sources.

OTOH: it could reflect attitude change at the individual level ("Changing Minds"). If true, we'd expect to see panelists in longitudinal studies "convert" their view in a pro-ACC direction, over time.

OTOH: it could reflect broader demo. changes in sample composition. If true, shifts in factors assoc. w/pro-ACC views (e.g., ideo. liberalism) should gradually make up a greater % of cross-sec. samples. ("Changing Samples").

In nationally representative panel data from CCES & Pew, I find little evidence of changing minds. Two key reasons why:

1. Few people revise their opinions on ACC *at all* over time (i.e., the dynamic reliability of ACC beliefs is quite strong)

2. The number of pro-ACC converts is similar in size to the number of anti-ACC converts; and typically not statistically different from the amount of movement we'd expect by random chance.

However, I find much more evidence of changing samples. In cross-sec. data from @YaleClimateComm, representative samples have changed in ways consistent with ACC acceptance.

For example, note the high degree of correspondence between the dashed blue line (% of liberals in samples) and the solid green line (% accept ACC). These findings hold in multilevel, multivariate models.

This work underscores the importance of continuing to track ACC beliefs longitudinally and cross-sectionally over time. This can help us disentangle the potential causes of change in climate attitudes.

It also has important #scicomm implications. Attitude change is rare, and difficult. We should make more effort to detect long-term changes in ACC belief in response to clim. comm. and other .