Predictive Modeling and ALEKS

In a recent article in Complete College America, it was reported that 50% of the incoming two year college students and nearly 20% of incoming four year college students are placed in remedial courses. The report suggest that too many students are needlessly placed in to remedial courses on the basis of flawed placement process. This has an adverse effect on student academic performance. It is an undeniable fact that not all students  are well prepared for college during their first semester. At the University of Connecticut, freshman courses in Calculus the DFW rates range from 25% to 50%. To analyze this issue of high DFW rates it is necessary to first establish some predictors (topics in Calculus I and Calculus II) that can give better understanding of variation in the grades on individual exams during the semester

Preliminary Analysis

Data on grades of students for midterm 1 and midterm 2 was analyzed using linear regression.

  • Predictors such as algebra skills were identified as the critical content related issues that have posed high level of difficulties for students.
  • Corrective measures to incorporate various pre-calculus materials in a more organized and effective manner were taken.
  • Pretests were used to assess the initial knowledge of students on pre-calculus topics.
  • Predictive models are underdevelopment to accurately forecast student grades on exams using indicator questions.

“ALEKS” is an acronym for Assessment and Learning in Knowledge Spaces. This is a web-based adaptive, artificially-intelligent education program. It is claimed that an ALEKS assessment results in

  •  The precise and comprehensive delineation of an individual’s competence in a subject in the form of his or her knowledge state describing all the types of problems mastered by that individual.
  • A comprehensive list of the topics the individual is ready to learn.

Although ALEKS has been successful in placing students in correct courses, which has directly resulted in the reduction of DFW rates at University of Austin, University of Illinois at Urban Champaign etc, it is important the success of ALEKS be majored in a careful and scientific way. For the purpose of providing a baseline of student performance and understanding the demographic population at the University of Connecticut following steps are being taken:

  •  A pretest was designed and administered to all the students in Math 1125 Calculus 1a and Math 1131 (regular calculus).
  • Data was requested from OIR for seven years starting from Fall 2005 to Fall 2011.
  • Demographic data on students with SAT scores, High school GPA, gender, AP score etc is being used to conduct predictive analysis.
  • Linear regression will be used to determine predictors that explain most variation in the data (grades in calculus courses).

The analysis will give a base line of student performance over the last three years. This will allow the University and the department of mathematics to evaluate the success of implementing a placement program such as ALEKS.