Students in an online undergraduate statistics class were asked to complete an online survey to provide data with which to work in class.  Fifteen of twenty students did, and each earned five course points for her or his contribution.  Five students declined to contribute, and earned zero class points.  A bit later a quiz was administered, testing the students on very basic concepts and definitions in statistics.  For pedagogical purposes, I conducted the simple statistical analysis that follows.  First, I sorted the students into two groups -- those who had completed the survey and those who had not.  Then I compared those two groups on their scores on the most recent quiz.  The analysis follows.



Group Statistics
  Survey_Credit N Mean
Week2 0 -- Survey Not Completed 5 3.60
5 -- Survey Completed 15 6.93

As you can see, those students who completed the survey did much better on the quiz than did those who did not.  Of course, this difference could be due to chance alone, even if in the larger population there were no relationship between survey participation and performance on a quiz. 


Independent Samples Test
  t-test for Equality of Means
t df Sig. (2-tailed) Mean Difference Std. Error Difference 95% Confidence Interval of the Difference
Lower Upper
    -3.368 18 .003 -3.333 .990 -5.413 -1.254
Were there no relationship between survey completion and quiz performance, the chances of getting, in the sample, a difference between groups as large as that observed here would be only 3 in 1,000.  This is strong evidence that those who take the time to contribute to class by completing a survey perform better on quizzes in the class than do those who do not take the time to complete a survey.

    So, why this association between completing the survey and doing well on the quiz?  It is, IMHO, unlikely that completing the survey has any causal relationship with quiz performance.  The association between taking the survey and doing well on the quiz is likely due to both of these being causally related to a third variable.  In this case, that third variable is what I call GaF.  Scoring high on Gaf causes one to complete the survey and also causes one to study the material and then do well on the quiz.  I am working on a device to measure GaF.  A prototype is shown below.