An Initial Analysis Of Freshman To Sophomore Retention In A New First Year Engineering Program

During the 2007-2008 academic year, the University of Arkansas (UofA) implemented the Freshman Engineering Program (FEP),a new first-year experience program for engineering students. The FEP was originally proposed to the UofA engineering faculty as an effort to improve the retention of new engineering students from their freshman to their sophomore years. As a result, the activities of the Academic and Student Services Sub-Programs executed by the faculty and staff of the FEP are all intended to improve students’ likelihood of academic success and/or to increase students’ desire to pursue an engineering degree. Since improving freshmanto-sophomore retention was a primary goal of the FEP, a significant amount of data has been collected on each of the 343 students enrolled in the first FEP cohort. This data includes demographic information, ACT (or similar) scores, high school GPA, Advances Placement (or similar) scores, scholarship data, fall 2007 class schedule and grades, spring 2008 enrollment data, spring 2008 class schedule and grades, fall 2008 enrollment data, and information related to the process used by students in selecting their engineering major for the sophomore year. Our primary objective in constructing this data set is to facilitate the completion of an exploratory data analysis to examine the interrelationships among the variables in hopes of identifying more effective methods for predicting student success in engineering. The long-term goal is to use the information and models obtained from this analysis to identify intervention programs that will promote increased retention rates for these students. In this paper, we present what we view to be the most interesting results of our initial analysis of this data. These results will range from tabulated counts from selected categories of the data to statistical models of relationships between these categories. We also present a brief synopsis of the activities associated with the executing of the Academic and Student Services Sub-Programs of the FEP. Review of the Literature A plethora of research has been generated regarding the prediction of success in college (Young and Korbin 11 ; Burton and Ramist 4 ; Ting 8 ; Pennock-Roman 7 ; Wilson 10 ; Bamforth et al. 1 ). However, a growing concern among researchers is the ability to retain students in the quantitative fields like math, science, and engineering. Retention of students is defined as either graduation or concurrent enrolment in a specific academic field. Without retention of students in mathand science-based fields, national and local economies suffer due to the increased demand for such research and development professionals (NARSET Report 6 ). Retaining students is a growing concern in many university departments, especially in the field of engineering. According to the National Access and Retention in Science, Engineering and Technology (NARSET) Report 6 , two factors determine the success of economic development in a country: 1) the amount and quality of human resources available, and 2) the extent of the research and development capacity. Without a retention and attraction program in place, the supply of graduates from the fields of science, engineering, and technology is unlikely to significantly grow. Identification and targeting of factors which influence retention is critical to the future growth of university engineering programs. Through the identification of prediction factors, P ge 14196.2 specific programs and models can be created which aid in the development of retention and success programs for current students. Predictive modeling can also aid in the identification, attraction, and support for future students. Research has shown that many factors such as gender, high school GPA, math and verbal SAT scores, and ethnicity contribute to college success of students and retention in science-based academic programs (Zhang et al. 12 , Bamforth et al. 1 , Budny et al. 3 , NARSET Report 6 , PennockRoman 7 , Wilson 10 ). Other associated factors known to influence the completion of a program include psychological, social, and cultural factors. Due to many associated factors, collecting and maintaining longitudinal tracking systems is often a complicated and expensive endeavor (Brainard and Carlin 2 ). According to the National Research Council in 1998, the inadequacies and inconsistencies of collection and maintenance of evaluation and retention data are major hindrances to projecting future manpower needs and identifying problems in the sciences field. Without access to consistent data which predicts success, engineering programs lack the ability to pinpoint deficiencies within their academic program and keep talented students. In addition to increasing attrition rates within engineering majors, another problem faced by departments is attracting talented high school applicants. Felder et al. 5 in their study on longitudinal engineering performance and retention found that both the increasing difficulty of attracting high school students into engineering and high attrition rates of enrolled engineering students have lead to the major decline in the graduation rates. With the need for math-based degrees on the rise, low graduation and attraction rates serve as a major detriment to local economies. Because of this fact, all prediction factors for college success, academic and non-academic, must be evaluated at the college as well as high school levels. Some common tools utilized by colleges in the admissions process are standardized test scores such as the SAT and ACT as well as high school performance in academic and community endeavors. For most colleges and universities, high school GPA and math SAT scores are positively correlated with graduation rates (Zhang et al. 12 ). However, the most frequently used criterion to assess the predictive validity of admissions based upon test scores is the freshman GPA (Wilson 10 ).However due to the rigors associated with the degrees, grades of students in the math and science fields of college tend to be lower than scores received in high school. These lower grades can lead to a lowered math confidence and increased risk of failure. Because the distribution of grades for physical science and engineering majors tend to more frequently be in the range of C or below (Pennock-Roman 7 ), students may be more inclined to seek a degree in the fields of arts and humanities where grade distributions are higher. Willingham’s 9 study with engineering students’ attrition rates found that students tended to migrate to majors where grading standards best fit their levels of preparation and lead to feelings of greater success. He found that the best-prepared students tend to major in more stringent disciplines whereas leastprepared students focus on more lenient disciplines. A more lenient degree where higher grades are possible could be viewed as appealing to an engineering student struggling in math. In preparation for college, many potential pre-engineering students engage in harder science and math classes in college as well as high school. However, many students find that for the field of engineering that they have gaps in their mathematical knowledge: gaps often due to the diversity of mathematical curricula taught in high schools. In most higher education institutions, a growing awareness exists in regards to the lack of mathematical preparedness exemplified in P ge 14196.3 freshman students for harder math and science courses (Bamforth et al. 1 ). This lack of knowledge of needed math skills and low levels of success can lead students to develop low self confidence in their ability to obtain a math-based degree during their freshman year; and unfortunately, research has shown that the freshman year can be critical to predicting the future success of an engineering student (Zhang et al. 12 ). Because of this mathematical diversity of the student intake, the Engineering Council in 2000 recommended that students embarking on a mathematics-based degree should have a diagnostic test before beginning the program and intervening support be provided immediately (Bamforth et al. 1 ). Through the provision of support to pre-engineering students lacking certain skills, mathematical competency can be reinforced and strengthened. This type of individualized support leads to the development of basic needed skills needed for the profession as well as increased student confidence in mathematics. Demographic factors which affect engineering students are gender and minority status. Due to the major decline in graduation rates, most engineering schools have undertaken major recruitment efforts directed at women and minorities (Felder et al. 5 ). Typically, more males than females major in engineering and physical sciences whereas more females major in the social sciences and humanities (Pennock-Roman 7 ). Many non-cognitive variables are associated with attraction and retention rates of women and minorities in the engineering field. In a study by Brainard and Carlin 2 , the most frequent non-cognitive barriers associated with women students in engineering are fear of losing interest, intimidation, lack of self-confidence, and poor advising. The researchers also found that young women who changed their major had experienced a loss of self-confidence prior to loss of performance in math and science classes; thus showing it was not a lack of ability that prompted the change. However, the successful establishment and accomplishment of women in many university engineering programs is an acknowledgement of the theory that given support and opportunity women can survive and thrive in a male-dominated field (Brainard and Carlin 2 ). The same theory could apply to minority students. Over the past several years, the retention of underrepresented groups has been a growing concern in the fields of science and engineering (Brainard and Carlin 2 ); therefore, universities have in