All potential discounting predictors were also investigated using step-wise linear regression with Bayesian Information Criterion (BIC) analysis — these BIC models revealed that education level and UMCE accounted for large portions of the variance. We conclude that education level and UMCE were the most consistent predictors of discounting. 

Discussion

We examined whether those who continued to participate in a MTruk multisession study differed from those who previously were enrolled but dropped out. 
[Interpretation of choosing medium magnitude]
We further explore the subset of the MCQ based on magnitude in order to predict the dropout. We found that DD rate at medium magnitude predict dropout robustly, with one measure of increase in ln(K) predict
 
[limitation]
·      Our study was the first to investigate this issue directly, and this is an important finding. However, Given the constraint of our moderately small sample size, this ove-ride effect of DD rate on education was not significant (t= p=). Further investigation might be needed with larger sample size. The logistic model compare
·      Other limitation includes:
·       
[future suggestion]
This study provide evidence DD is a reliable proxy to predict subject return to subsequent sessions in the multisession studies. It provide important insigt in engaging participants in the multisession behavior change programs.
No previous studies have tracked changes in performance on behavioural impulsivity tasks in relation to alcohol involvement over such an extended period of time, and we believe that the current study makes a very important contribution in this regard.
 
·      The mechanism of this predictive power need to be examined – fMRI study on excusion in mendeley
·      Suggesting incooperate DD rate measurement in design of  behavior change program.