Article #6
2000
 
 
 
Newsletter list 

Newsletter 

Previous Article 

Next Article 


 


Brunswik, the Berlin Way

Ulrich Hoffrage
Berlin, Germany

Almost all of my activities in 2000 that were related to Brunswik have been done together with my colleague Ralph Hertwig. Because he has already sent his report to the mailing list, I can keep my list short by just referring to his.

First, we organized this year's Society meeting - and let me once more use the opportunity to thank those of you who participated for making the effort to come and for your contributions.

Second, our Brunswikian report on the hindsight bias has been published this year in the JEP:LMC. The main idea is that a lack of direct memory on what we said or thought in the past can (and will) be compensated for by reconstruction based on the knowledge underlying our initial judgment.

Feedback on the correctness of this judgment leads to an update of the cue values we used for our original judgment. Note that in this process of updating, the cue-criterion relationship is reversed. The reconstructions from the updated knowledge base, in turn, lead to distorted "memories" - the so-called hindsight bias. (I wouldn't be too surprised if someone finds this short summary hard to understand, but I hope it gives you the feeling that this work is in the spirit of Brunswik and makes you curious enough to read the article.)

As Ralph already mentioned, we are now in the process of running simulations of this model - most of the work is done by Carola Fanselow, a graduate student in our group.

Third, Mandeep Dhami (and Ralph and I) have been - and still are - working on a review of the use of representative design in Social Judgment Theory research.

Fourth, we took the challenge of designing and testing a fast and frugal heuristic for a task at which regression is particularly strong: numerical estimation.

Last but not least, there is one thing that Ralph did not mention (at the time he sent his report, we didn't know yet). We are happy to announce that Science will publish in its Policy Forum Column an article by us (U. Hoffrage, S. Lindsey, R. Hertwig, G. Gigerenzer: "Communicating Statistical Information," scheduled for December 22nd, 2000). In this paper, we demonstrate that experts' and laypeople's difficulties in dealing with statistical information can be considerably reduced if the information is communicated in what we call "natural frequencies."

The link to Brunswik is that natural frequencies are the result of natural sampling, that is, of observing and counting cases as they occur in the environment, as one would do in a representative design. In the article, we describe what natural frequencies are in more detail, report two new studies in the domains of medicine and law, and discuss several applications and implications.

Contact Ulrich Hoffrage

Previous Article                Next Article

Newsletter list