Biases in the production and reception of collective knowledge: the case of hindsight bias in Wikipedia

Abstract

The Web 2.0 enabled collaboration at an unprecedented level. In one of the flagships of mass collaboration-Wikipedia-a large number of authors socially negotiate the world's largest compendium of knowledge. Several guidelines in Wikipedia restrict contributions to verifiable information from reliable sources to ensure recognized knowledge. Much psychological research demonstrates, however, that individual information processing is biased. This poses the question whether individual biases translate to Wikipedia articles or whether they are prevented by its guidelines. The present research makes use of hindsight bias to examine this question. To this end, we analyzed foresight and hindsight versions of Wikipedia articles regarding a broad variety of events (Study 1). We found the majority of articles not to contain traces of hindsight bias-contrary to prior individual research. However, for a particular category of events-disasters-we found robust evidence for hindsight bias. In a lab experiment (Study 2), we then examined whether individuals' hindsight bias is translated into articles under controlled conditions and tested whether collaborative writing-as present in Wikipedia-affects the resultant bias (vs. individual writing). Finally, we investigated the impact of biased Wikipedia articles on readers (Study 3). As predicted, biased articles elicited a hindsight bias in readers, who had not known of the event previously. Moreover, biased articles also affected individuals who knew about the event already, and who had already developed a hindsight bias: biased articles further increased their hindsight.

Cultural Interpretations of Global Information? Hindsight Bias after Reading Wikipedia Articles across Cultures

“…Rather, it is possible that participants of our study also initially showed a reverse hindsight bias after learning about Fukushima (e.g., because they immediately experienced how surprised they were, Müller & Stahlberg, ), but that reading articles about the nuclear power plant, which were biased by hindsight still increased their perceptions of likelihood and foreseeability. In a related study with an unknown disaster, we found that perceptions of likelihood, inevitability, and foreseeability did not only increase after outcome knowledge (i.e., the ‘classic’ hindsight bias) but were additionally enhanced by reading a biased article about the event (Oeberst et al, ).…”

“…Interestingly, however, this was mostly information that had been known and was publicly available in advance, but had not been included in the foresight article version (e.g., risks of this type of reactor). Even more importantly, one previous study (Oeberst, Cress, Back, & Nestler, ; Oeberst, von der Beck, Back, Cress, & Nestler, ) demonstrated that the hindsight article version was significantly more suggestive of the nuclear disaster than the foresight article version. In other words, in this Wikipedia article, the nuclear disaster was presented as more likely inevitable and foreseeable — after it had taken place.…”

Collectively biased representations of the past: Ingroup Bias in Wikipedia articles about intergroup conflicts

“…Additional analyses suggested two reasons for this differential pattern: recency of the conflict and the proportion of ingroup members among the (top) editors. Both findings clearly point towards the transfer of individuals’ own biases into Wikipedia articles (see also Oeberst et al , ) as (1) an ingroup bias is much more likely for events that (still) pose a threat to the ingroup members (Ellemers et al , ; Hewstone et al , ; Stephan & Renfro, ) and (2) a smaller proportion of ingroup members (and a larger proportion of outgroup members) among the top authors indicates greater diversity and heterogeneity, which decreases bias (e.g., Schulz‐Hardt et al, ; Vinokur & Burnstein, ). It must be acknowledged, however, that there are also more sources available for more recent conflicts.…”

The Impact of Topic Characteristics and Threat on Willingness to Engage with Wikipedia Articles: Insights from Laboratory Experiments

“…Our studies are also limited in terms of overlooking the potential interactions between particular topics and participants’ personal opinions on those topics. Previous investigations showed that people could be affected by their own biases while working with Wikipedia articles ( Beck et al, 2017 ; Oeberst et al, 2017 ). An opinion bias might have played a role in participants’ motivation to engage with articles in our studies as well.…”