Witness

A blog about forensic psychology.

by Karen Franklin, Ph.D.

Psychology Rife with Inaccurate Research Findings

By Karen Franklin, Ph.D. on November 15, 2011 in Witness

The case of a Dutch psychologist who fabricated experiments out of whole cloth for at least a decade is shining a spotlight on systemic flaws in the reporting of psychological research.

Diederik Stapel, a well-known and widely published psychologist in the Netherlands, routinely falsified data and made up entire experiments, according to an investigative committee. Read More

Diederik Stapel is dutch, not

Submitted by Anonymous on November 16, 2011 - 7:00am.

Diederik Stapel is dutch, not danish....please correct!

reply
quote

Thanks!

Submitted by Karen Franklin, Ph.D. on November 16, 2011 - 8:12am.

Oops! Thanks for catching that error.

reply
quote

What's wrong with AUC stat?

Submitted by Anonymous on November 16, 2011 - 8:58am.

What exactly is the problem with the Area Under the Curve statistic? I'm not aware of any problems with the stat itself except for the sometimes erroneously way people COMPARE two incompatible ROC curves.

reply
quote

AUC statistic

Submitted by Karen Franklin, Ph.D. on November 16, 2011 - 10:09am.

Anonymous,
Jay Singh and colleagues have been conducting a line of research into measurements of predictive accuracy. They find that the AUC statistic does poorly as a method for measuring the predictive accuracy of risk assessment instruments, when compared with other methods such a the Diagnostic Odds Ratio (DOR). I have more information on that research, as well as contact information for Dr. Singh in case you would like to request his article(s), at my professional blog (http://bit.ly/niKm0j). An upcoming issue of Behavioral Sciences and the Law will also address problems with the AUC statistic (see my blog announcement at http://bit.ly/tA6OXE).

reply
quote

honestly when it comes to

Submitted by Anonymous on November 16, 2011 - 10:10am.

honestly when it comes to studies that can actually have a decent impact on what people believe, they should not be allowed to be published unless peer reviewed extensively.

reply
quote

It would also be nice if

Submitted by Anna on November 16, 2011 - 12:00pm.

It would also be nice if during a time of reform they addressed the file drawer problem that leads many researchers to lie about their results because otherwise they can't publish and can't keep their jobs.