{[ promptMessage ]}

Bookmark it

{[ promptMessage ]}


2 over application of available information diagnoses

Info iconThis preview shows pages 8–10. Sign up to view the full content.

View Full Document Right Arrow Icon
2. Over-application of available information (diagnoses based on recency). 3. Bad feedback (lie detection; complex business decisions). 4. Biased feedback; motivated reasoning (political experts; sports gamblers). 5. Learned associations are overgeneralized (stereotypes). Alternatives to intuitive decision making: 1. Experimentation / Pilot Testing 2. Base Rates / Find Similar Cases 3. Statistical Models / Consistently-Applied Decision Rules 4. Aggregating (Independent) Opinions Lecture 11 Why is statistical judgment so consistently better than human judgment? Reliability: How much do two judgments of the same thing correlate with each other? Validity: How much does a judgment correlate with the variable we are trying to predict? It is difficult to have validity without reliability. Create a random model by: 1. Determine the sign of each regression coefficient. 2. Standardize all the variables. 3. Choose a coefficient at random over the range 0 to 1. 4. Use those coefficients to make predictions. Human judges often use information that is not predictive. Human judges often assign non-optimal weights to the criteria. Optimizing is not something that humans are good at – especially when two criteria are somewhat redundant.
Background image of page 8

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
Caveats 1. The data that is used to create this model must reflect the range of possibilities in the real world. 2. You must have a quantifiable criterion, using quantifiable attribute values. 3. Statistical models don’t often account for “soft”, qualitative attributes that, in some cases, should matter. 4. Diversity is difficult; what if you care about more than one criterion? Ways to Improve Judgment: 1. Improve Reliability 2. Discard Useless Information Takeaways 1. When accurate prediction, diagnosis, and judgment is important, a statistical model will outperform expert judges. 2. Human judges are unreliable, susceptible to influence by factors that should not matter, prone to ignoring factors that should matter, and incapable of optimally weighting the relevant factors. SECOND HALF OF THE SEMESTER Lecture 12 Problems with Group Deliberations 1. Groups are re-affirming 2. Some opinions are not voices 3. Opinions are not independent Averaging opinions often out-performs the vast majority of individual estimates. Reason 1: Judgments are often unreliable, and susceptible to random errors. Averaging often helps to cancel out those random errors. Reason 2: Individual judgments are often biased by one’s own unique perspective or tendencies. Whereas group discussion can amplify such biases, averaging can cancel them out. Potential Problems with Averaging Opinions: Problem 1: In some cases, judges are not independent or diverse. Even if they are physically separate, people may rely on the same sources (media, rumors, beliefs) to make estimates.
Background image of page 9
Image of page 10
This is the end of the preview. Sign up to access the rest of the document.

{[ snackBarMessage ]}

Page8 / 17

2 Over application of available information diagnoses based...

This preview shows document pages 8 - 10. Sign up to view the full document.

View Full Document Right Arrow Icon bookmark
Ask a homework question - tutors are online