The null hypothesis H0 set in the Chi-square test
assumes that there are no differences in passing the
course between different categories of Forum members
and non-members. The working hypothesis H1 assumes
the dependence between these values and categories.
The level of significance is set to P=0.05 with Yates's
correction, due to the small number of posts in some
categories, i.e. less than 10.
According to the results shown in Table IV, the null
hypothesis is not validated in five out of six monitored
cases. To sum up, there is a statistically significant
correlation between the course pass or fail grade and the
activity in forum discussions.
The working hypothesis is not validated only in the
case of Forum members with student tutorship in
generation 2010/11. The reason for this may lie in the
purpose of the forum for student tutorship; it is not clear
whether the students attend tutoring because they need
supplemental tuition or advanced tuition.
Generation 2010/11 has an additional motivating
factor: they are invited to post the errors identified in the
script, and to propose corrections. Thereby, the number of
students actively participating in forum discussions is
higher in that generation. Text mining in that group might
identify and differentiate students who provide assistance
and solutions from those who seek help.