Attitude Polarisation

What it is

In the 1979 paper, the authors proposed that people who hold strong opinions on a complex social issue judge research on that issue unevenly. They accept findings that support their views with little scrutiny and pick apart findings that challenge them. If that is how the reading goes, a mixed body of evidence will not pull two opposing groups together. Each group will find the supportive part more persuasive, and the gap between them can grow.

The paper's title names two linked ideas. Biased assimilation is the uneven treatment of the evidence itself: the same method or result is judged sound in one case and flawed in the other, depending on whether it fits the reader's prior view. Attitude polarisation is the outcome, a widening of the distance between the groups. The first is a process and the second is a result, and the experiment was designed to show them together.

In effect

The experiment used the death penalty because opinion on it was strong on both sides. Participants who favoured it and participants who opposed it each read descriptions of two studies on whether it deters crime. One study appeared to support the participant's view and the other appeared to contradict it. Participants rated how convincing each study was and how well it had been conducted, and they reported how their attitudes shifted as they read.

Both groups judged the study that matched their prior beliefs as more convincing and better designed, and both shifted in the direction of that study. Because each group moved towards its original position, the net effect was greater polarisation. Everyone had read identical material, so the difference lay in the reading and not in the evidence supplied.

The finding fits the earlier picture from Albert Hastorf and Hadley Cantril, whose 1954 football study showed students at two universities counting more infractions by the opposing team on the same film. That study concerned perception of events, and the 1979 experiment concerned judgement of research, so the two address different stages of the same broad habit. Ray Nickerson's 1998 review of confirmation bias places both within a wider family of related findings, including motivated forms, in which people defend beliefs they value, and unmotivated forms, in which they are partial even on questions that matter little to them.

What it does not say

Attitude polarisation is a claim about particular conditions: strong prior views, material that bears directly on them, and readers who are asked to evaluate it. It does not claim that all exposure to opposing information hardens opinion.

That broader claim, often called backlash, was tested by Andrew Guess and Alexander Coppock. They defined backlash as people who meet evidence against their view holding it more strongly afterwards, and ran three survey experiments, two using online worker samples and one using a nationally representative sample. They found no sign of backlash, even in conditions set up to make it most likely, and concluded that backlash is much less common than the information-processing literature often implies.

The two sets of findings measure different things. The 1979 experiment recorded how convincing participants found each study and how their attitudes moved while they read mixed evidence. The survey experiments recorded whether counter-attitudinal information moved views in the wrong direction. Both can be reported accurately. Uneven judgement of evidence has a long record, and the stronger claim that opposing evidence makes opinions more entrenched has found less support in these experiments.

For anyone applying the idea, the practical reading is modest. Readers with a stake tend to weigh the evidence unevenly, so it helps to have evidence assessed by someone without a stake. Matthew Syed's account of aviation in Black Box Thinking describes this kind of arrangement, with independent investigation and shared findings, and sets it against settings where failures are explained away.

---