The Influence of Early Respondents: Information Cascade Effects in Online Event Scheduling
https://deepblue.lib.umich.edu/bitstream/handle/2027.42/134703/Romero%20et%20al%202017%20%28WSDM%29.pdf?sequence=1&isAllowed=y
This essay was on the Influence of Early Respondents regarding Information Cascade Effects in Online Event Scheduling. It talks about when filling out a schedule, what effects are there that might affect a person decisions. It investigates whether this difference is due to information cascade effects in which later respondents adopt the decisions of earlier respondents. So essentially a person puts there time available and the person after gets to see the previous times and then put in there times. However, the fact that the people after the first have access to this information means that their choice is most likely biased to pick the best choice given the previous choices. This involves finding the bases probability or the probability that given the first person decision what is the second persons decision.
The work in the paper provides evidence of information cascade effects in online event scheduling, showing that the success of a poll depends on the availability of early respondents and that participants are particularly influenced by the availability of their directly preceding neighbor. In many cases, this can mitigate the success of a poll. It proposed a set of interventions that can help alleviate this bias and show that it can optimize the chances to find a mutually agreeable time. It hopes that its findings help inform the design of future event scheduling systems and enable groups to more effectively arrange a time to meet. Also after experiments exploring the role of early respondents on the overall success of the poll, and in particular the effect of early respondents with low availability, there was results for the effects. If respondents are simply reporting their availability, independently of the availability of other respondents, then we should observe no effect. However, if respondents are able to observe others’ responses and take these observations into account when reporting their own availability, then having an early respondent with low availability could have a negative effect on the overall poll. When a new respondent observes that earlier respondents reported low availability, this removes social pressure from the new respondent to report a wide availability. This effect could cascade through the sequence of respondents, making each new respondent more likely to report a limited availability. This aligns with the papers hypothesis that respondents are influenced by the level of availability of previous respondents.
This is quite similar to the marble problem we did in lecture that demonstrates the effects of previous decisions on future decisions. How if the person guess a blue bag then the next person should (if there isn’t a tie) should also pick blue since its the overall majority and leads to solid consensus.