Scoring the Monetary Choice Questionnaire using logistic regression 

Version 1.0b -------------------------------------------------

^mcqscore^ response_varlist, ^id(^id_variable^)^ ^saving(^out_filename^)^ [^nojk^]

Description -----------

MCQScore scores the Monetary Choice Questionnaire (questions in standard order), which uses a hyperbolic decay function to summarize the degree to which time discounts the value of a delayed reward. The scale parameter k is used as a measure of impulsivity. Conventional scoring requires a lot of hand labor, identifying indifference points where a subject's behavior changes from preferring an immediate to a delayed reward. Instead, we use logistic regression to identify a contour where the probability of choosing either reward is 0.5, identifying in turn the corresponding hyperbolic function. Logistic regression is applied to raw binary choices from the survey, modeling the probability of choosing the delayed reward in response to time and a transformation of the reward ratio. The estimate of k is obtained as a ratio of the two betas. Output will include log-k estimates for small, medium, and large delayed reward sizes, a pooled log-k estimate and the component betas, and optionally, a jackknife estimate of standard error for the pooled log-k estimate and a bias-reduced estimate.

Two other ado files must be included for this routine to run. The first is ^mcqlik.ado^, the penalized likelihood function that keeps the parameter estimates in bounds. The second is ^mcq2long.ado^, which transforms the data to long format, and associates question numbers with reward values.

Please note that this is a first release. Please contact me with errors.

^response_varlist^ is required. It lists the names of the variables containing the responses to the MCQ in standard order. Data must be 0s and 1s representing the choice of the delayed reward. Data must also be in wide format, with one line of data per subject. Variable naming should conform to the following convention: use a common root with the question number 1, 2, .. 27, on the end of the name.

Options ------------------------------------------------------------

^id^ is required. It contains the name of the subject identifier.

^saving^ is required. It is the name of an output file for saving results.

^nojk^ is optional. By default, the jackknife procedure is run. nojk prevents this.

Examples ---------

mcqscore choice1-choice27, id(subject) saving(d:\discounting\my_outfile) nojk

mcqscore c1 c2 c3 c4 c5 c6 c7 c8 c9 c10 c11 c12 c13 c14 c15 c16 c17 c18 c19 c20 c21 c22 c23 c24 c25 c26 c27, id(idname) saving(d:\mcq\outputfile)

References -----------

Wileyto, E.P., Audrain-McGovern, J., Epstein, L.H., & Lerman, C. (In Press) Using Logistic Regression to Estimate Delay- Discounting Functions. Behavior Research Methods, Instrumentation, & Computers.

Kirby, K.N., Petry, N.M., & Bickel, W.K. (1999) Heroin addicts have higher discount rates for delayed rewards than non- drug-using controls. Journal of Experimental Psychology: General, 128, 78-87.

Author ------

E. Paul Wileyto, Ph.D. Assistant Professor of Biostatistics Tobacco Use Research Center School of Medicine, U. of Pennsylvania 3535 Market Street, Suite 4100 Philadelphia, PA 19104-6068

215-746-7147 Fax: 215-746-7140 epw@mail.med.upenn.edu