DEBIT stands for DEscriptive BInary Test.

Scientific papers sometimes report group membership as a binary variable (for instance, men=0  and women=1). This is usually in the context of running a regression analysis.

When papers describe a mean and standard deviation calculated in such a way, this gives us something to test -- because the mean and standard deviation are not independent, they are transforms of each other. If you know the  mean, you know the SD (and vice versa).

Some additional fun points:

• DEBIT isn't a frontline test. It's a little bit obscure, and less useful than other tests. It's great when it applies, but it doesn't apply often.

• DEBIT came about when an anonymous whistleblower (the uncreatively named 'John Smith') sent myself and Nick Brown some papers in criminology. We found this technique and sent it back. Later, two of authors, Pickett and Stewart, got involved -- Pickett went on to investigate his own papers that he wrote using data he got from Stewart. A few years later, Stewart was fired. Messy business.

• DEBIT is in the Scrutiny package as well. Type install.packages("scrutiny")from the prompt to install. More information is available on GitHub. 

• DEBIT also has Matlab code available in the OSF link above, but it's very simple and you won't need it.

• DEBIT can be calculated in your browser through Lukas Jung's online portal here.

Go back Home