1

In order to estimate probability of default, banks and other financial institutions have often used logistic regression based on data involving credit history.

Suppose an entrepreneur or borrower with little credit or business history goes to a bank or some financial institution to borrow money.

Banks can use, as an alternative to a credit or business history, 'psychometric' or 'behavioural' tests. Hence, the bank will use some psychometric data, instead or with credit history data when using logistic regression to estimate probability of default.

How can Bayesian inference improve upon such logistic regression wherein psychometric data is used?

I previously asked about the Bayesian logit model and psychometrics.

So far, all I found are these EFL papers: 1 2

Anyone know any other literature out there?

Community
  • 1
BCLC
  • 707
  • 2
  • 9
  • 24
  • 2
    There's really nothing special about psychometric data in this regard, all of the usual advantages (and disadvantages) of Bayesian analyses, which have been well covered in your CrossValidated question, apply as well to psychometric data as to any other kind of analysis. What are you hoping to learn here? – Eoin Sep 07 '15 at 16:05
  • Thanks @Eoin ! Which part in the CV question gives an advantage of Bayesian analyses? It seems to explain it. I don't see any advantages involved – BCLC Sep 08 '15 at 00:12
  • I'm voting to close this question as off-topic because it's not clear how this question is specific to the topics of this SE. I think it should be asked at Cross Validated. –  Sep 10 '15 at 08:05
  • @MaríaAnt Edited. Still think it should go to CV? – BCLC Sep 10 '15 at 08:32
  • 1
    I think this is a suitable question. Bayesian analyses ask different questions than frequentist analyses, and that difference matters more in behavioral science than other areas. – Christian Hummeluhr Sep 10 '15 at 08:49
  • @ChristianHummeluhr That may not be before my last edit ^^; – BCLC Sep 10 '15 at 09:10
  • I am still skeptical. Bayesian statistics are used in many areas of science. As @Eoin has pointed out, the pros and cons of a Bayesian approach apply to all kinds of data, whether they are about loan decisions, processes in plant cells, or questions in engineering. In addition, I am not certain how the question goes beyond those already asked at CV about Bayesian vs. frequentist analyses. –  Sep 10 '15 at 12:12
  • I'm not sure if the question is really Bayesian specific. It seems like the basic question is whether incorporating psychometrics can improve a prediction of whether a person will default or not. – Josh Sep 10 '15 at 12:52
  • @Josh It's supposed to be how Bayesian can improve a prediction given that psychometric data is used – BCLC Sep 10 '15 at 13:07
  • 1
    I'm voting to close this question as off-topic because Should go to CrossValidate.SE – Robin Kramer-ten Have Jun 10 '16 at 13:32

0 Answers0