In the attachment is the dataset as described in the paper. This dataset provides 33 attributes, 3 of which are dependent measures:
1. "complete" indicates whether the student did complete the skill builder (1) or not (0)
2. "problem_count" is the number of problems the student has attempted. This is also known as Mastery Speed
3. "log(count)" is the logarithm based ten of "problem_count" if the students completed the assignments, "N/A" otherwise.
Most of other attributes are directly acquired from ASSISTments data, which are described here:
http://www.assistmentstestbed.org/the-data/ali-doc-request/interpreting-your-data-v-1-0
Some attributes are specific to this dataset, which are described here:
1. "ExperiencedCondition" indicates whether the student has experience the difference of the conditions.
In some experiments, some students may not experience the difference between the two conditions. For instance, if in the experiment condition students who answer incorrectly were given video hints, while in the control condition they were given text hints, then students who immediately answer 3 questions correctly do not experience the difference and should not be considered affected by the condition.
2. "Could See Video" indicates whether the student can see videos. In several RCE, we compare video hints vs text hints. We implemented a question that check whether the students can see videos or not. However, at the time of the experiments, the system was unable to direct students who could not see videos to non-video conditions. As a result, every student was randomly assigned to either video condition or text condition regardless of whether they could see videos or not.
Current Version:
- ThisOne.xlsx (V7) Updated 29 September 2016
- ThisOne.CSV (V3) Updated 29 September 2016
Change log:
29 September 2016
- added a few more problems to the "ignored problems" list so that the mastery speed is now more accurate.
- fixed a typo causing some problem sets to mark all students as "incomplete"
- use a new, more robust algorithm to infer conditions
23 September 2016
- rename column "Saw video" to "Could See Video"
- recalculate "problem_count" so that it is more accurate. problem_count now represents the total number of problems that are "actual" problems students have to solve (excluding all video checks, introduction messages, and such).
- recalculate "complete" so that it now excludes students who completed the assignments by exhausting the skill builders
- Added ThisOne.csv which is simple a ThisOne.xlsx in CSV format.