YourMorals.Org

The scale you just completed was a test of formal logical reasoning, and was based on a scale originally developed by Ara Norenzayan, Edward Smith, Beom Jun Kim, and Richard Nisbett. The scale measures the extent to which people use formal vs. intuitive reasoning styles. Past research has shown cultural differences in these reasoning styles. For example, Norenzayan and colleagues (2002) found that Asians and Asian Americans were more likely to demonstrate intuitive (rather than formal) reasoning styles compared to European Americans.

We are interested in examining ideological differences in logical reasoning styles. Specifically, we are examining group differences in formal vs. intuitive logical reasoning styles, as well as differences in susceptibility to a common psychological phenomenon called "belief bias".

The study involved three kinds of logic problems, but participants were randomly assigned to complete only one kind. You were randomly assigned to the "abstract logic" condition, and your results will appear in the 1st graph below. We have included results for the other two conditions (in the second and third graphs below) because they are interesting and informative, although you will not have scores on these two graphs.

The first graph below shows scores on a subset of the logic problems related to abstract logical reasoning. The scores range from 0% to 100% and represent the proportion of abstract logic problems correctly answered by logical reasoning. Thus, higher scores correspond with higher degrees of formal logical reasoning. The score of the average Liberal survey respondent is shown in blue (2nd bar), the average Conservative score is in red (3rd bar), and the average score for Libertarians is in orange.

The next graph, below, shows scores on a separate subset of the logic problems related to a common psychological phenomenon called "belief bias". Past research shows that people's judgment about whether or not an argument is logical is often biased by whether or not its conclusion seems believable.

For example, consider the following logic problem:

Premise 1: All animals which feed their young are mammals.
Premise 2: Birds are not mammals.
Conclusion: Birds feed their young.

This problem is actually invalid, although it is often judged as valid because its conclusion is believable.

In the graph below, scores range from 0% to 100%, with higher scores representing better performance (fewer errors) on questions related to belief bias. Thus, lower scores indicate a tendency to rate invalid arguments as valid when they contained believable conclusions, and to rate valid arguments as invalid when they contained less believable conclusions. The score of the average Liberal survey respondent is shown in blue (2nd bar), the average Conservative score is in red (3rd bar), and the average score for Libertarians is in orange.

The final graph, below, shows belief bias scores on a third subset of logic problems that involved partisan political issues. We were interested in investigating how pre-existing beliefs not only impair our ability to be logical (as shown above), but also how both liberals and conservatives are susceptible to this form of bias because of their unique sets of beliefs. For example, consider the following logic problem:

Premise 1: All things that add greenhouse gases into the atmosphere contribute to global warming.
Premise 2: Humans do not add greenhouse gases into the atmosphere.
Conclusion: Humans contribute to global warming.

This problem is actually invalid, although it is sometimes judged as valid by liberals because its conclusion is believable to them. Half of the arguments from this section tested this kind of "liberal" belief bias, and half tested a "conservative" belief bias.

In the graph below, you will see separate scores for these two forms of belief bias. Scores range from 0% to 100%, with higher scores representing better performance (fewer errors) on problems related to partisan belief bias. We expect liberals to be more resistant to "conservative" belief bias problems than "liberal" problems, and we expect the opposite pattern for conservatives. The scores of the average liberal respondent are shown in blue. The average conservative scores are shown in red.