One thing that I’m interested in is how people with strong political beliefs – so-called extremists – differ from people with more moderate political views. With Anthony Evans and Jarret Crawford, I’ve been fortunate to have a paper accepted for publication at Psychological Science (pre-print). In this post I’m going to briefly note some interesting things about the paper besides the actual findings. First, the abstract and a figure:
People with extreme political opinions are alternatively characterized as relatively unthinking or as confident consumers and practitioners of politics. We test these competing hypotheses using cognitive anchoring tasks in three studies (Total N = 6767). Using two different measures of political extremity, we found that extremists were less influenced by two types of experimenter-generated anchors compared to political moderates (Studies 1 – 3) and this result was mediated by extremists’ belief superiority (Study 2). Extremists and moderates, however, were not differentially influenced by self-generated anchors (Study 2), suggesting that extremists differentiate between externally- and internally-generated anchors. These results are consistent with the confident extremist perspective and contradict the unthinking extremist perspective. The present studies demonstrate the utility of adopting basic cognitive tasks for understanding the ideology-cognitive style relationship and suggest that extremity does not necessarily beget irrationality.
We think this is pretty interesting data (which is here), which is why we published it. And we are interested to see if the ideas in this paper can be extended to other cognitive tasks. But to get the gist of the paper, please just read it. It isn’t too long.
I want to use this space to quickly highlight some other aspects of the paper that I think deserve some notice.
One interesting thing theoretically is that we do not find consistent differences between liberals and conservatives. The only consistent effects we find across our two measures of political beliefs and three studies are the effects of extremity. This is in contrast with recent (and popular) work linking more conservative (and typically socially conservative) beliefs with a negativity bias (.pdf). Like our anchoring tasks, negativity bias is often assessed with more purely cognitive or psychophysiological tasks (e.g., gated pdf). So, it may not be that people on the left and the right are always different, but rather they are more likely to be different when the task has some degree of arousal (e.g., .html).
There are also a few interesting things from a methodological point of view, especially if you are interested in the whole open science and replication business.
- Data for Study 1 is from the first version of the Many Labs Project (.osf, .html). This project collected data from labs around the world, analyzed it, wrote it up, and then posted the data for anyone to use. We analyzed the data they collected in the United States on four anchoring tasks. This was extremely helpful for our project because it gave us another source of data and it gave us data that was (relatively) geographically diverse.
- There were a lot of replications in this paper. Ostensibly, our paper is about a novel effect (or rather, a novel test of an existing hypothesis). Yet, we worked in replications of several different effects which gives us more confidence in those effects, but also in our own data. Some of the replications were more or less close/direct and some of the effects that we replicated were more or less established. Besides our effects, here are the effects we replicated and a link to a relevant previous study:
- Experimenter-generated anchoring effects were replicated in all of our studies (.pdf).
- Extremists’ perception that their beliefs are superior was replicated in our Study 2 (gated pdf).
- People are more certain about their judgments on experimenter-generated compared to self-generated anchoring tasks was replicated in our Study 2 (.pdf).
- Political conservatives tend to score higher on individual difference measures of the need for cognitive closure was replicated in our Study 2 (.pdf).
- But not every effect we found replicated some previous known effect. For example, some previous work suggests that high need for closure (operationalized as cognitive load) makes people more susceptible to anchoring-type tasks (gated pdf). This may be true for a cognitive load manipulation with their specific anchoring tasks, but in our data with measured individual differences in need for closure we did not find any evidence for this idea.
A couple of takeaways. 1. It is possible for researchers to replicate effects and ideas without running studies specifically to replicate studies. 2. Extremists are not necessarily unthinking and they are not necessarily irrational (but that doesn’t mean they are clear headed thinkers making only rational decisions!).