Belief system networks can be used to predict where to expect dynamic constraint

Felicity Turner-Zwinkels*, Mark Brandt

*Corresponding author for this work

Research output: Contribution to journalArticleScientificpeer-review

7 Citations (Scopus)
114 Downloads (Pure)


We test if a change in an attitude affects other related attitudes (i.e., dynamic constraint), a core prediction of belief systems theory. We use psychological network methods to represent the belief system and make preregistered predictions about which attitudes should change and to what extent. We collected data in two longitudinal experiments (N = 3004; N = 2999) and three pilot studies (combined N = 2788) from community samples of US Americans. We use data from T1 as pretest measures of attitudes and to estimate the structure of the sample’s belief system from which to generate and preregister predictions. At T2 participants were randomly assigned to one of three conditions: a control condition (no manipulation), a terrorism attitude manipulation (Study 1), a crime attitude manipulation (Study 2) attitude manipulation, or a banking attitude manipulation (Studies 1 & 2). We successfully manipulated the targeted attitude and also observed changes in non-targeted attitudes in the belief system. Multilevel models provided evidence that changes in non-targeted attitudes were moderated by their distance from the targeted attitude within the belief system: Non-targeted attitudes closer to the experimentally targeted attitude typically changed more. Changes in non-targeted attitudes were generally related to (and mediated by) changes in the targeted attitude. We discuss the implications of our findings for belief systems theory and the value of network methods in studying attitude change.
Original languageEnglish
Article number104279
JournalJournal of Experimental Social Psychology
Publication statusPublished - 2022


  • Attitude change
  • Belief systems
  • Dynamic constraint
  • Psychological networks


Dive into the research topics of 'Belief system networks can be used to predict where to expect dynamic constraint'. Together they form a unique fingerprint.

Cite this