Servicenavigation

Project Details

Back

Projekt Print View

Sampling Effects through Self-Truncated Information Search

Subject Area Social Psychology, Industrial and Organisational Psychology
General, Cognitive and Mathematical Psychology
Term from 2018 to 2023
Project identifier Deutsche Forschungsgemeinschaft (DFG) - Project number 420128411
 
Final Report Year 2024

Final Report Abstract

One of the most well-established theoretical principles in psychological science is the amount-of-learning principle. Accordingly, learning increases with the number of trials. In accordance with Bernoulli’s (1713) law of large numbers, existing trends are more reliably discerned from large than from small samples of observations or learning trials. Applications of this principle can be found in various fields of behavioral science. However, a second, seemingly incompatible hypothesis predicts that trends can be more visible in smaller rather than in larger samples, when sample size is the result of self-truncated information search. We call this principle the self-truncation effect. While its underlying principles have been introduced in the established literature, it has only recently been investigated more thoroughly in experiments. Apparently, self-truncation seems to contradict the wisdom of crowds (Surowiecki, 2004), which seems to be well in line with law of large numbers. In fact, however, both principles support different aspects of Bernoulli’s law. It is easy to grasp that small sample size not only implies that sample statistics approximate to a lesser degree the population parameters. The same mathematical law also implies that the stronger dispersion small samples are more likely to produce very strong primacy effects. If the stopping rule that determined sample size is not predetermined experimentally but depends on self-truncation, the very inaccuracy of small samples allow the individual to stop sampling at the very moment when a trend is most visible. It can be shown, in simulation as in experimental research, that virtually all plausible stopping rules lead to negative correlations between sample size and extreme (sample-based) judgments. Self-truncation not only produces extreme decisions informed by small samples, but also accurate decisions under most reasonable conditions. Elaborating on the challenging competition between both principles, amount of information and self-truncation, was the focus of this project. As predicted on logical grounds, small samples led to less accurate decisions when sample size was an independent variable and to more accurate decisions when sample size became a dependent variable through self-truncation. Because democratic decisions are based on the majority principle, our findings are of general importance, granting that most committee protocols do not indicate whether the group deliberations were self-determined or pre-determined externally.

Publications

 
 

Additional Information

© 2026 DFG Disclaimer / Copyright / Privacy Policy

Textvergrößerung und Kontrastanpassung