The cohort data for the 1994 school year yielded outcomes higher than those from the annual total data for all three outcomes. The cohort measure comparable to the graduation rate is percentage trained, and that was 70 percent, 5 percentage points higher than the annual total figure. Part of the reason the cohort figure is higher is that those who withdrew for related employment are included as completing their programs. The percentage placed in related employment in the cohort data is 82 percent, four points higher than the annual total results. Here again part of the higher figure is due to those who withdrew for related employment. The biggest difference between the annual total and cohort data is with regard to withdrawal. The annual total data estimated withdrawal at a fairly constant 20 to 22 percent. In the cohort data, the rate in 1994 was 30 percent. We think the higher cohort figure is due to a more careful tracking of those who entered programs in one school year but did not complete their programs that year. They informed their schools that they intended to return the next school year but did not. As a result, they were not counted as withdrawing in either school year. The cohort data, because it is longitudinal, counts these withdrawals more accurately than the crosssectional, annual total data. Both the annual total and cohort outcome measures were related to 39 measures of the characteristics of the students and the schools, using multiple regression analysis. This analysis determines the net, independent effect of each school characteristic on the outcomes, holding the effect of all the other characteristics constant. The school characteristics listed below were found to have consistent, statistically significant relationships with school performance in both the annual total and cohort data. Most of these relationships, however, were found for full-time enrollments. The outcomes for part-time enrollment have fewer systematic relationships with school characteristics. ■ Percentage of enrollment receiving Pell grants. This variable reflects the percentage of students from disadvantaged backgrounds served by a school. As the percentage of enrollment receiving Pells increase, graduation/percentage trained rates decline and withdrawal rates increase. The results for the Pell variable are consistent with other findings on the problems of educating students from disadvantaged backgrounds. Results from other studies suggest that schools accredited by the Commission may graduate more students from disadvantaged circumstances than other postsecondary institutions. Percentage of students classified as Ability to Benefit (ATB). This variable indicates the percentage of students served by a school who did not complete high school or obtain a General Educational Development certificate. As the percentage of ATB increases, withdrawal and default rates increase. The effect of the percentage of ATB was lower in 1993 and 1994 than in prior years. This may be because the schools and colleges are enrolling fewer ATB students. ■ Average program length in weeks. This variable indicates the average length of programs offered by a school. As program length increases, graduation/percentage trained rates decrease and withdrawal rates increase. In 1994, the difference in the graduation rates between the shortest and longest programs was 30 percentage points. As program length increases, however, default rates decrease. It may be that students who complete longer programs earn more money and are better able to pay off their loans. An alternative explanation is that longer programs are more costly and may draw their students from families with more financial resources. ■Main or branch campus. For four of the five years of annual total data and in the 1994 cohort data graduation/percentage trained rates were higher for schools that report they are the main campuses. We think it is likely that main campuses are likely to have more resources and more experience faculty that contribute to higher quality programs. ■Total enrollment. Graduation rates decrease and withdrawal rates increase as the total enrollment of schools increase. It appears that schools with enrollments over 600 have a more difficult time graduating as many of their students as smaller institutions, when all other factors are held constant. ■Faculty turnover. Turnover among faculty is associated with lower graduation rates and higher withdrawal rates. Turnover among part-time faculty is related to outcomes more than turnover among full-time faculty. It certainly is harder for teachers to establish supportive relationships with students when there are frequent changes. High turnover may also make it difficult to establish linkage with employers that often results in employment The cumulative effect of these characteristics can be quite substantial. A main campus school with an enrollment of 600 or less that had few Pell recipients and offered short programs could be expected to have a graduation rate 20 to 30 points higher than a branch campus school with an enrollment over 600, the average percentage of Pell recipients, and long programs. The percentage of Pell grant recipients, ATB students, and faculty turnover may all contribute to problems with retention and graduation. Large enrollments together with high percentages of Pell and ATB students, and high faculty turnover would be an especially worrisome combination, however, such a combination is rarely found in practice. The analyses in all three reports have identified those characteristics of schools that have a consistent relationship with school performance. The statistical method used to identify these characteristics was multiple regression analysis. This method relates several measures of school operations (inputs) to selected measures of school performance (outcomes). Schools want to graduate their students, i.e., retain them until their programs are completed, and place them in good jobs that are related to the fields in which they were trained. Schools also want students who borrowed money for their tuition to pay back their loans. These desired outcomes led to the four indicators of school performance examined in this study: graduation, withdrawal, trainingrelated placement, and default. The schools that produce these outcomes differ on many dimensions, such as the size of their enrollments, the prior education of their students, the turnover of faculty, whether they are main or branch campuses, and so on. These characteristics of schools can be considered indicators of the inputs used to produce the desired outcomes. Multiple regression indicates which characteristics of schools have statistically significant relationships with the four outcomes. In the following sections we explain how we defined the measures of school performance and characteristices used in this report. Measures of School Performance As noted above, annual reports are submitted to the Accrediting Commission on a school year basis, defined as July 1 through June 30. The information in the reports is verified annually by site visits to a sample of reporting schools. The numbers reported in the various categories— new enrollments, continuing students, re-entries, graduation, and withdrawals—are the totals for the year. The cohort information that traces a defined group of students from initial entry until exit, either through graduation or permanent withdrawal, was collected for the first time for the 1994 school year. For the annual total data, graduation rates, withdrawal rates, and training-related placement rates were calculated for each school by aggregating information provided for each program offered by a school and calculating overall school rates. Separate rates were calculated for full-time and part-time enrollments. The default rates used in the 1990 and 1991 analyses were obtained from reports prepared by the U.S. Department of Education. The rates used in the 1992, 1993, and 1994 analyses were obtained from the annual reports submitted by the schools for those year. It should be noted that these rates are always for the students who left school two years prior to the year in which they are reported. Outcome Definitions Used with Annual Total Data Graduation rate. The first of the three reports prepared by the Center on Education and Training for Employment (CETE) discussed how we tested various ways of calculating graduation rates and selected the method we used. The primary problem in defining graduation, lies with how to deal with continuing students, those who neither graduate nor withdraw in a given school year. This is the definition we decided to use: Number graduating Number graduating + Number withdrawing This definition disregards the problem of continuing students for the reporting year. Some of these students will graduate and some will withdraw in the next school year and will be included in the rate when they do so. Withdrawal rates. Withdrawal rates present less of an analytic problem: a withdrawal is a withdrawal regardless of when it occurs. Withdrawal rates were calculated by dividing the total number withdrawing during a school year by the total number enrolled during that school year. This definition, however, like that for graduation, is not based on a defined group of entering students. Because of continuing students and the definition of graduation rates adopted to allow for them, for the annual total data, withdrawal rates are not simply the reverse of graduation rates. Since continuing students are not included in the calculation of graduation rates, graduation plus withdrawal rates do not sum to 100 percent. In each school year we have analyzed, about 40-45 percent of full-time and part-time students neither graduate nor withdraw. Instead they start during one school year and continue their studies into the next year. Consequently, not all the variables found to have a significant relationship with graduation have a similar reverse relationship with withdrawal. Training-related placement. Training-related placement (TRP) is a critical measure of the performance of schools whose primary mission is to teach specific occupation skills. In the past decade, research has established that obtaining employment related to training is essential if program completers are to realize increased earnings in the labor market. The final report of the National Assessment of Vocational Education (Boesel et al. 1994) summarized the importance of obtaining related employment as follows: The strongest, most consistent finding throughout the literature is that improved To calculate TRP, we adjusted the number graduating by eliminating those who were unavailable for employment because they were continuing their education, entering the military, or had other documented reasons why they were not seeking employment, such as illness or pregnancy. The number excluded for these reasons has been consistent over the three years, averaging 8 to 9 percent of the total number of graduates. The mean numbers excluded from the calculation of TRP are shown in Table 1.1. The mean is the average value for all reporting schools. The standard deviation is an indicator of the variability, or how the separate schools are distributed from lowest to highest value, for each of the variables. The large standard deviations reflect the wide variation in the number of graduates, which in 1994 varied from 0 to 1,664. The due date for the 1994 annual reports was three months later than the reports from prior years. These three months were added to allow a more accurate estimate of the students who obtained related employment following completion of their programs. Definitions for Cohort Data Percentage trained. Because 1994 was the first year for which cohort data were collected, only three totals from the completion and placement data were coded and provided to CETE: the number that started a program, the percentage trained, and the percentage placed. Percentage trained is not entirely the same as graduation because it includes a small number of students who withdrew from the program prior to graduation because they obtained employment in the fields in which they were being trained. We know from the annual total data that those who withdraw for related employment are typically only 2 or 3 percent of the number who graduate in a given year. The definition used to calculate percentage trained in the cohort data was as follows: Grads within 150% of program length + Withdrawn employed in field Percentage placed. The percentage placed in related employment calculated for the cohort data also includes those students who withdrew for related employment. In the percentage placed, however, the number who withdrew appears in both the numerator and denominator of the definition: Withdrawn employed in field + Grads employed in field Withdrawn employed in field + [Grad within 150% of program length - (Further education + Military service + Other unavailable for employment)] |