Provided that no outside information is given and the choice is not changed, a cascade will continue indefinitely. One different choice can disrupt a cascade, making it fragile as stated in class. The model used to introduce cascades was built so people would make the same choice beyond a string of two of the same choices because no one can infer about how the person before them made their choice.

When more factors are analyzed, the length of the cascade factors into the strength of the cascade. Longer cascades are stronger than shorter cascades, however, they will both end once a different choice is made. If there are four players, is player 4 more likely to continue the cascade than player 3? This is possible because Player 4 can think that player 3 did not learn from the previous players. If A was chosen up to player 3, player 4 can think player 3’s private information was a, and this would make player 4 pick A also. This category of players making assumptions about previous player’s choices comes from Quantal Response Equilibrium. In this Kübler and Weizsäcker, a poll of each player was taken and the further along the cascade the player was, the more confident the player was in their choice.

The strength of the length of the cascade is also shown by a decrease in using the option to buy outside information further into the cascade. Also some people were given information that contradicts the cascade and in experiments run by Kübler and Weizsäcker for these cases, there were results where these people continued the cascade, once again attesting to the strength of the length of a cascade. This paper by Kübler and Weizsäcker offers more detail than we discussed in class when cascades were covered at an introductory level. This paper discussed the different experiments set and the prior experiments cited that some of their ideas drew from.

The conclusion states parallels between the multiple experiments conducted on cascades. “We therefore argue that while some of the cascade papers seem to contradict each other, one should also take notice that all of the data sets are consistent in a central feature of the aggregate cascade development, the correlation of length and strength.” The set up of the experiments is also discussed in the conclusion.

This is the link to the paper http://www.mitpressjournals.org/doi/pdf/10.1162/jeea.2005.3.2-3.330?cookieSet=1