View All Case Studies

SAIT Polytechnic

Correlation of MyMathLab Results with Course Results in Calculus for Engineering Technology

Kerry E. Kijewski
Centre for Academic Learner Services
SAIT Polytechnic – Calgary, AB

Abstract

A pilot project was conducted in the fall semester (September-December) of 2008 to test the MyMathLab materials available for the U.S. 9th edition of Allyn J. Washington’s text “Basic Technical Mathematics with Calculus”, prior to the adoption of the SI version of the 9th edition of the text in Fall 2009, with the intent of reporting back to faculty about the new algorithmically-generated testbank and new utilities such as custom instructor-built algorithmically-generated questions.  A sample of 80 students in three different academic sections was required to complete on-line quizzes related to the material covered in class. An hypothesis test of linear correlation between the student grades obtained through MyMathLab and those obtained via testing in the course was conducted. A significant positive linear correlation was obtained (correlation coefficient r = 0.817) for a 0.01 level of significance. There is evidence to support the claim that a positive linear correlation exists between MyMathLab grades evaluated on the primary basis of final answer and course grades determined chiefly through adherence to solution methodology.

Introduction

Calculus is taught at SAIT Polytechnic as a core requirement for most engineering technology programs. New on-line resources created by book publishers has necessitated an examination of the relevance, applicability, and utility of these resources in motivating students and measuring their academic capability with course learning objectives. Our calculus teaching group decided to continue using Allyn J. Washington’s text “Basic Technical Mathematics with Calculus” as the primary resource for our courses. A new SI edition of the text (9th edition) will be available in Fall 2009 that should have a comprehensive testbank of algorithmically-generated questions. Many of these questions were already available in the U.S. version of the 9th edition of the textbook, so a pilot project was conducted to test the MyMathLab materials available with the intent of reporting back to faculty about the new algorithmically-generated testbank, the on-line textbook, the gradebook operation, and new utilities like custom-built algorithmically-generated questions.  The goal was to inform faculty how they could utilize MyMathLab in determining at-risk students, saving some marking time for assigned work, and acting as a motivational tool for the course. As part of this project, an hypothesis test of linear correlation between the student grades obtained through MyMathLab and those obtained via testing in the course was conducted. Since our engineering technology courses in mathematics are evaluated primarily on method of solution, and since MyMathLab evaluates on the basis of final answer only, there was an expectation that the results might demonstrate low correlation.

Methods

A sample of 80 students in three different academic sections was required to complete on-line examinations related to the learning objectives covered in the course forming 10% of their overall grade. Students were given three attempts on a quiz-type exercise, each attempt with different algorithmically-generated values and randomized question order using MyMathLab. The students could use the "Help Me Solve This" and "See an Example" utilities in Review Mode only after each attempt, and only the best mark counted. Since MyMathLab only marks the final answer for complete correctness and does not have the capability in the publisher-generated testbank questions to allow for rounding or other minor errors, a decision was made at the outset of the course to alter student grades from MyMathLab in the following ways:

• If a student submitted a correct answer but had a small rounding error (e.g. 5.43 kN submitted vs. 5.44 kN accepted answer) or number of decimals error (e.g. 8.993 cm submitted vs. 8.99 cm accepted) then the grade for that question would be altered to give part marks, penalizing only a small amount (approximately 10% of the marks for the question) for a rounding or decimal error.
• If a student submitted a functional solution and only had minor errors (e.g. variable substitutions, or a missing integration constant), or in cases where the answer was correct but merely unreduced, then the grade for that question would be altered to give part marks, penalizing only a small amount (approximately 10% to 20% of the marks for the question) for the committed errors.

In both situations, students communicated to the instructor which quiz attempt and question was affected in order to minimize time spent on this activity.

MyMathLab results were obtained from the average of 8 quizzes, each consisting of between 6 and 15 questions, with the intent that each quiz would take approximately one hour for a student to complete. Course grades were obtained from written exams marked primarily on the basis of methodology of solution (a midterm, final, and 6 handwritten quizzes throughout the term).

The data was analyzed with MS Excel, a linear regression line and linear correlation coefficient was calculated for the bivariate set. The data for the overall student course grade has had the 10% on-line testing component removed, so as to keep the variables independent and to not artificially inflate any resulting correlation.

Data & Analysis

A traditional two-tailed hypothesis test for linear correlation was conducted; the Pearson product moment correlation coefficient r was determined from the data, and compared to the critical values for this test statistic. The data is summarized in Figure 1 below.

What Students Are Saying

Lamatina has received the following comments from students using MyStatLab in Elementary Statistics.

MyStatLab helped me understand the material in the course. This was my second time taking the class.The first time, the notes were from the professor and we didn’t even have a book for the course.This time, the professor taught the material better and MyStatLab helped a lot.

I don’t know what the class would be like without MyStatLab. I like the features in MyStatLab. I particularly like that when I input an answer that isn’t correct, the program gives hints on how to work out the problem.