Equity & Fairness of Multiple Provincial Exam Forms
On March 13th to 21st 2006 the individual student provincial exam results will be released. This year when students in the Nanaimo school district walked out of their provincial exam and compared notes on how they did, some discovered that they wrote several different exam versions for a given subject. In the 2005-2006 hand book of procedures, under the Security section, the following is listed:
Beginning in 2004-2005 with Grade 10 Graduation Program examinations, the Ministry began using enhanced statistical methods to equate tests and improve fairness and equity for students from one examination session to another. In 2005-2006 these statistical methods will be applied to exams in courses that meet the 2004 Graduation Program social studies requirement and to selected Grade 12 examinations. The goal of test equity can only be met if selected test questions, called “anchor items” are re-used.
The reason the Ministry chose to do this are listed:
• ensure equivalency of scores from administration to administration and across years
• permit more frequent administration while limiting development cost
• permit rapid response to emergent situations
• provide a basis for multiple forms of examinations to support electronic-examination delivery.
A mark is generally understood as a measure of knowledge gained. It indicates that if a student receives 60% then they should have an amount of knowledge equivalent to 60% of the learning outcomes. Let’s look at the example of two students with identical capabilities. One student writes version A—a version measuring the appropriate level of knowledge—and the other student writes version B—a version measuring different learning outcomes than expected. The marks for the student who wrote version B are then statistically post processed to compensate for the discrepancy in learning outcomes being measured. For the student writing exam version B, is this an accurate assessment of the knowledge they have acquired? Does statistical exam mark post-processing decrease the importance placed on exam design quality and exam validation? Is it fair to students when they don’t know whether they wrote exam A or exam B, whether their mark will be post-processed or not, how it will be processed, and by how much?
The Ministry of Education has indicated that exams can be equated to one another, within and across exam sessions, through the linking of a relatively small proportion of common questions and through looking at the relative difficulty of those questions, based on students' responses. This does not account for the situation where “the small proportion of common questions” may be completely missed by students who decide to skip or speed right through them if they are writing a difficult version of the exam. In this case, their comparative base score on the common questions will be lower. This can be further aggravated by the psychological impact of watching one’s classmates leave early, wonder why they are having more difficulties, and get flustered.
Also there is the issue of exam content bias. Recently an English exam had a section that used specialized terminology. For example, let’s assume a section was about basketball and that it used basketball specific jargon. In this example, one would expect basketball players and fans to do better than others. There is bias in the exam that favours a group of students. This bias cannot be corrected by statistical methods unless one knows which students are familiar with the basketball jargon.
The mark post-processing and the exam multiple versions are intended to ensure fairness in evaluation and treat students the same over time, but let's imagine that the first round of grade 10 exams or this year's grade 11 socials exams were unfair to some students because some teachers were not prepared to teach the provincial prescribed learning outcomes being evaluated by the provincial exams. Will the Ministry retro-actively correct the marks for an unfair exam implementation? I doubt it; this hardly seems fair or equitable.
Labels: Provincial Exams