Main content

Home

Menu

Loading wiki pages...

View
Wiki Version:
**Original citation.** Vul, E., & Pashler, H. (2008). Measuring the crowd within: Probabilistic representations within individuals. *Psychological Science, 19*, 645-647. **Target of replication.** We attempted to replicate the crowd within effect, which is the effect that the average of two guesses within one person provides a more accurate answer than either guess alone. Specifically, we tried to replicate the effect that the error of the average of two guesses is lower than both the error of guess 1, t(255) = 4.41, p < .001, and the error of guess 2, t(255) = 9.90, p < .001. **A priori replication criteria.** A successful replication includes a higher accuracy of the aggregated guess compared to the individual guesses. In order to test this, we compared the error of the average of both guesses with the error of guess 1 on the one hand, and with the error of guess 2 on the other hand, by performing two t-tests for paired observations. We excluded data from those participants which defocused the browser window running the study, as the latter may be an indication of participants looking up the answers to the questions. On the same line of reasoning, we planned to exclude data from participants who indicated that they had looked up the answers to the questions at the end of the experiment. **The experiment component** contains instructions in the immediate and delayed condition and some screen shots from the experiment. **The analysis code component** contains matlab code used for post-processing the data, matlab code used for the confirmatory analyses and matlab code used for the post-hoc analyses. **The data component** contains the raw data from the immediate and delayed condition and the post-processed data. **The sample plan component** contains matlab code used for sample size calculation to achieve a 95% power level. **The slides on science 2.0 component** contains a presentation explaining the what, how and why of some of the features of science 2.0 (replication, registration, high power, bayesian statistics, estimation, co-pilot multi-software approach, distinction between confirmatory and exploratory analyses, and open science) using this study as a running example. **Conclusions.** Our results successfully replicated the original finding by Vul and Pashler (2008) that averaging two guesses within one person provides a more accurate answer than either guess alone. This effect was found when the second guess was made immediately after the first guess (immediate condition), as well as when the second guess was made three weeks later (delayed condition). The three-week delay between the two guesses improved the accuracy gain of averaging compared to guess 2, but not compared to guess 1. These results are comparable to those in Vul and Pashler (2008), where an increase in accuracy gain was also observed with the comparison of one of both guesses only. <a href=" http://journal.frontiersin.org/Journal/58626">Download the full report</a>