Appendix A: Treatment of the student grades data
Treatment 1: Removal of duplicate entries due to multiple professors
The initial dataset contained one row per term, student, course, course section and professor (where the professor also had a new unique identifier assigned). For course sections that were taught by more than one professor, duplicate identical grades entries, one for each professor, were listed. These duplicates were removed.
Treatment 2: Removal of empty (null) grades
The initial dataset included entries missing a grade value entirely (a null or empty value). Any row with an empty grade was removed.
Treatment 3: Restriction to valid grades
Algonquin College’s policy AA14 enumerates valid values for grades. The dataset was subsequently restricted to the following valid grades to enable analysis. A non-exhaustive list of other possible statuses has been included for reference.
| Grade type | Non-fail values | Fail values | Included |
|---|---|---|---|
| Numeric grades, including incomplete variants | A+ to D-, IA+ to ID- | F, IF, FSP | Yes |
| Pass/fail grades, including incomplete variants | P, IP | F, IF, FSP | Yes |
| Other statuses | Aegrotat (AG), audit (AU), credit (CR), satisfactorily/not satisfactorily completed (SC/NSC), not available (N), withdrawn (W) | Aegrotat (AG), audit (AU), credit (CR), satisfactorily/not satisfactorily completed (SC/NSC), not available (N), withdrawn (W) | No |
Treatment 4: Removal of Centre for Continuing and Online Learning, Jazan, and Kuwait
To restrict the dataset as much as possible to grades earned in a similar academic context, the Centre for Continuing and Online Learning (CCOL), which handles continuing and distance education, was removed as well as the College’s two international campuses, Jazan and Kuwait.
Treatment 5: Limitation to full-time programs using section numbering
Using information on section numbering practices at the College, the dataset was further restricted to include, as best as possible, only sections denoted as full-time. Owing to a section numbering format change between academic years 2013/2014 and 2014/2015, two separate formatting processes were consulted. As with treatment 4, the goal was to restrict the dataset, as much as possible, to grades earned in a similar academic context.
| Academic year(s) | Sections included |
|---|---|
| 2013/2014 | 010 to 489 |
| 2014/2015, 2015/2016, 2016/2017 | 010 to 279, 300 to 399 |
Treatment 6: Addition of a Pass/Fail course flag and an eText/IPM flag
Two variables were added for each row of grades data:
First, a pass/fail flag was added to denote courses whose grading system was pass/fail. This was important to be able to properly attribute failing grades (i.e. F, IF, FSP) to pass/fail courses and not have them be included in the body of numeric grades.
Second, an eText/IPM flag was computed as follows:
| Academic year(s) | Method of computation |
|---|---|
| 2013/2014, 2014/2015 | Based on historical documents from academic years 2013/2014 and 2014/2015, a list of eText/IPM courses per term was derived and used for the eText/IPM flag |
| 2015/2016, 2016/2017 | The Registrar’s Office data contained a flag for a given grade based on current data from the platform managing IPM resource distribution. A new flag was calculated such that if at least one student in a given term, course, and section was flagged by the Registrar’s office as eText/IPM, the entire section was flagged as IPM. |
The computed eText/IPM flag for academic years 2015/2016 and 2016/2017 was added after it was noticed that students who were repeating an IPM course and already had the resource from the previous time they took the course were not being flagged by the Registrar’s Office as IPM.
