X hits on this document

PDF document

The Characteristics Associated with Student - page 28 / 94

286 views

0 shares

0 downloads

0 comments

28 / 94

Indicator of Secondary Major

Yes No All Undergraduates

N

% of row 100.0 95.3 95.3

N

% of row

100 12,076 12,176

0 600 600

0.0 4.7 4.7

No

Default

Yes

Total

100 12,676 12,776

Minor

The minor degrees obtained by borrowers are not significantly related to the probability of default. For a full list of minors and their default rates, see Appendix C.

Attendance Pattern

The variables that describe the manner in which borrowers are admitted to the university, the pattern of their attendance, the length of their tenure and the episodes of their withdrawal from college have a connection to the likelihood of their default. In fact, every Attendance Pattern variable has a statistically significant relationship to default. Moreover, some of the variables in this section have among the strongest relationships to default of any of the variables in this study. Some of these variables’ strong relationship to default is, in turn, probably due to their connection with whether or not borrowers graduate. For example, borrowers who take a number of course hours that would coincide with program completion (around 130 hours) or who attend the university for a length of time that is consistent with a traditional completion time (three to four years) have the lowest default rates. Conversely, borrowers who attend for very short periods of time and take very few hours probably do not graduate and have relatively high default rates. Nevertheless, these conclusions represent oversimplifications of the more complex and interesting relationships that follow.

Missing values do not pose a significant problem for variables in the Attendance Pattern section. In some cases, such as with the Number of Semesters Enrolled Less Than Full Time variable, it is clear that most missing values are actually zeroes, indicating, in this example, that some borrowers did not attend any semesters on a less than full time basis. In situations such as those, missing values were recoded to zeroes, even at the possible expense of designating some true missing values as zeroes. In the cases of the admissions variables, such as Admission Type and College of Admission, the default rate of the missing category (5.1 percent) was just slightly higher than the average for Texas A&M over the study period. This result is probably due in part to the fact that the missing values belonged to borrowers from earlier repayment cohorts, when default rates were higher.

Statistical Summary: Attendance Pattern

Variable Total Hours – Texas A&M Hours Plus Transfer Hours

Statistical Significance

Significant

Number of Years Between First

Significant

Uncertainty Coefficient

0.07 0.06

Cramer s V

0.18 0.16

Gamma

  • -

    0.42

0.27

Spearman Correlation

  • -

    0.13

0.09

25

Document info
Document views286
Page views286
Page last viewedTue Dec 06 08:31:03 UTC 2016
Pages94
Paragraphs7456
Words29247

Comments