Institutional Research and Planning Room 306 1871 Old Main Drive Shippensburg, PA 17257 Phone: (717)-477-1154 Fax: (717)-477-4077 irp@ship.edu

Attrition Survey The topic of retention is one of the greatest areas of interest in research in higher education. Every year, institutions actively investigate retention through the use of progress records, surveys, interviews and other forms of assessment. With the increasing popularity of the Internet, colleges and universities across the United States are more eager than ever to share information regarding their student retention plans. They are less enthusiastic about discussing the other side of retention known as attrition. A review of thirty college web pages yielded only one mention of attrition. This mention came in the form of a no-longer active college exit survey administered to vacating students. The instrument was designed with the perspective that students leave because there is something inherently wrong with them. This individualistic perspective, along with the overall lack of Internet postings regarding attrition, illustrates the problem that exists: attrition is a difficult and complex issue. More specifically, institutions would rather focus attention on the seemingly positive topic of retention and away from the seemingly negative topic of attrition. At Shippensburg University, a variety of post-departure instruments have been contracted out or designed in-house by the Office of Institutional Research and Planning to study attrition. The instruments have evolved from a multiple sheet inventory, to a two-question survey, to most recently a 30 item rating scale with an option for open-ended responses. All of the instruments have followed a method of studying attrition known as the autopsy method (surveying an exiting student after s/he has departed) and produced similar and limited results. Due to low response rates, these results have not been representative of the departing populations. Autopsy studies deal with information gathered after the student has left; however, it is often difficult to track down students. Once located, it is difficult to ensure responses from the students. It may be interesting to review the responses received, but it is not very informative of the group as a whole and often provides a skewed or unrealistic picture. Attrition has been a popular topic among researchers for quite some time. Many theories exist regarding student development and with that, attrition. Most notably, there is Tinto’s model for student attrition. This model, which has been validated and replicated several times over, incorporates a variety of factors which researchers have found influence student development. It begins by taking into account a student’s background, individual attributes and pre-college schooling. From there, it suggests that in order for a student to be satisfied at college, they must have success in two areas: academic integration and social integration. Academic integration includes what Tinto refers to as goal commitments, grade performance, and intellectual development. Social integration refers to institutional commitments, peer-group interactions, and faculty interactions. Success in these areas leads to a re-evaluation of commitments (goal and institutional) and subsequently either the desire to persist or the desire to leave. Tinto stresses that it is necessary for institutions to understand that attrition may occur for a variety of reasons. While some of these reasons may stem from academic difficulty, dissatisfaction or adjustment issues, they could just as easily come from unique goals and commitments on the part of the student. A student who attends a four year institution for two years of that four, may do so because his/her goal was to accomplish particular requirements and then move on. If this student has successfully completed his/her objective, departure should be viewed in a positive rather than negative way. While the student did not graduate, the school did serve the student’s needs and therefore the interaction can be viewed as successful. Implementation Feasibility. With regard to time, the autopsy model produces results most quickly. Since the model only requires the production and administration of a single form, the time involved is minimal. On the other hand, while this form is simple to produce and send out, receiving feedback from it is not. The autopsy model is criticized for having low levels of representativeness due to low response rates. Higher response rates are attainable with other models. Current research cites that while autopsy studies yield a 15% to 40% response rate, cross-sectional studies may yield a 55% to 80% response rate and longitudinal studies a 40% to 60% response rate. With a more successful response rate than the autopsy model, the cross-sectional model also relies on a single form. However, it takes more time to receive and interpret results from the cross-sectional model. With the cross-sectional design, before reviewing the results, it is necessary to wait a year so that departure status can be confirmed (determine whom, out of those tested, is staying versus whom is leaving). Finally, the longitudinal models require the greatest amount of production, administration, and interpretation. The depth of data obtained and connectibility of information to other data sources remain significant advantages of the longitudinal method. Ultimately, the longitudinal study provides a greater quantity and quality of information if resources exist to conduct adequate longitudinal research. Cost. In most cases, the autopsy model is the least costly of the three models. In comparison, the autopsy model has a small sample size that translates into lower production costs. At the same time, the model examines students after they have left thus includes mailing costs. The cross-sectional model is the second least costly, with some additional costs as sample size increases. Administration occurs while the student is still enrolled and therefore does not incur any postage fees. Like the cross-sectional model, the longitudinal model employs larger sample sizes and consequentially costs more. While the use of multiple scales also adds to this cost, the scales are all administered while the student is still enrolled and therefore does not generate any additional postal fees (while insuring a higher response rate). It is important to bear in mind that a relationship exists between the cost and the effectiveness of the information generated; generally, as cost increases, so does the effectiveness of the information gathered. Potential Impact. Putting the issue of cost aside, there is the issue of impact. There are numerous considerations involved when looking for the most effective model. It is clear that there are advantages and disadvantages involved in tracking students for four years versus one year. It is also clear that it is desirable to have as much information about a student as possible, so that reasonable assessments can be made. With this in mind, the inherent weaknesses of the autopsy and cross-sectional models come to light. By limiting available information to data gathered from single assessments very little analysis can take place. Single assessments and low response rates do not create the quantity or quality of data necessary to calculate accurate pictures of attrition. At best, these measures scrape the surface; at worst they provide a picture which is unrealistic and unrepresentative. Over time, the more comprehensive data generated from a longitudinal model can provide more accurate information as well as trends that show existing institutional strengths and weaknesses. Recommended Model. With the merits for each model outlined above, it is clear that the more information gathered, the greater the ability to assess and potentially intervene in attrition. While it would be wise to continue with the autopsy model as a way to survey those students who leave the university, it is more important to augment the assessment plan with more current and comprehensive cross-sectional or longitudinal methods. For Shippensburg University, longitudinal models offer the most effective and least costly results. The only readily apparent drawback linked to these models is their potential drain on resources. To effectively implement the longitudinal method, there should be an addition of an instrument to the current assessment plan that incorporates the ASIS. This will be administered to new students near the end of their freshman year and again to juniors in the spring semester. Data generated from these scales will provide measures of integration from the first year that can then be compared with measures from the third year with specific attention being focussed on factors which exist for those who stay versus those who leave. Additionally, the data will be analyzed in conjunction with information generated from scales that are already a part of the current assessment plan. Scales such as the CIRP, NLSSI and CSEQ provide further measures that include the topics of personal background, social and academic support services, career and educational goals, all of which are known to impact attrition. By choosing to update current practices for studying attrition, Shippensburg University is clearly acknowledging the importance and complexity of this topic. Further, by augmenting the assessment plan, the university asserts not only knowledge of this topic’s importance but also a commitment to its study.