Okay, pressed the send accidentally. I've completed the
post below. Still trying to get this internet thing down. ;-)
-MP
----- Original Message -----
Robyn Dawes has argued that a simple linear regression-type
equation is better at predicting a person's behavior or achievement
in some process than subjective judgments (e.g., impressions from
an interview) in the following wonderfully titled article:
Dawes, R. M. (1979). The robust beauty of improper linear models
in decision making. American psychologist, 34(7), 571.
Dawes' claims that such linear decision models were more fair
that subjective judgments (along with their implicit biases) were
but this was not accepted by many because people, especially
clinical psychologists (or, if you're a sports fan, sports management;
read the book or see the movie "Moneyball" for a comparable
example) who felt their intuitions and experience were better
guides to judging a person than some equation, even if the
equation gave more successful predictions. Dawes has
said such experts may know what variables are important
but they can't "add them together" (i.e., integrate the information
into an appropriate predication without bias and error).
People in economics and finance, however, took Dawes advice
to heart and realized that if they were going to lend people money,
they'd better be able to predict whether they pay back that money
along with interest. Computer algorithms now go over a person's
credit history, assess the risk of lending money to such a person,
and related information.
But what about a person who went to college right from high school,
graduates from college, has a job, and now wants a loan. Such
a person has a limited credit history so the traditional algorithms
really don't apply or, if used, will excessively deny good risks.
But what makes a person a good risk for a loan if they have
a very limited credit history? Some people have suggested
using indicators of a person's "character" to assess their riskiness.
The NY Times has a article in its "Bits" section on a couple of
companies that have developed such algorithmic systems,
systems that indicate "good" or "bad" or neutral character.
See:
http://bits.blogs.nytimes.com/2015/07/26/using-algorithms-to-determine-character/ Quoting from the article:|"If
you take two people with the same job and circumstances,|like whether they have kids, five years later the one who
had|the higher G.P.A. is more likely to pay a debt," said Paul Gu,|Upstart's co-founder and head of product.
"It's not whether you|can pay. It's a question of how important you see your obligation."||The idea,
validated by data, is that people who did things like|double-checking the homework or studying extra in case there|was
a pop quiz are thorough and likely to honor their debts.||Analytics, meet judgment of people. "I guess you could
call it|character, though we haven't used that label," said Mr. Gu,|who is 24.So, students who are conscientious
in their school work and infulfilling course requirements -- fullfillng their educationalobligations --may be better
credit risks, even if they are a "subprime"candidate given traditional algorithmic evaluation.This
seems to make sense and I think Robyn Dawes would bepleased with this but it
will be interesting to see what getspublished on this. I think, however, I'll
bet my money oncharacter. ;-)-Mike PalijNew York universitym...@nyu.edu
---
You are currently subscribed to tips as: arch...@mail-archive.com.
To unsubscribe click here:
http://fsulist.frostburg.edu/u?id=13090.68da6e6e5325aa33287ff385b70df5d5&n=T&l=tips&o=46161
or send a blank email to
leave-46161-13090.68da6e6e5325aa33287ff385b70df...@fsulist.frostburg.edu
