The Items in the CSFI Were Written to Describe Theoretically Derived Factors Which Contribute

VALIDITY

The items in the CSFI were written to describe theoretically derived factors which contribute to college success. Between versions LP1 and LP5, individual item analysis led to the rewriting of many items. The criteria used for predictive validation were student retention and grade point average. Internal consistency within the scales (alpha) was also examined.

Construct Validity

Between-scale correlations ranged from +.68 (Time management with Task Precision), to +.27 (Competition with Time Management), suggesting that while some scales are only moderately related, a common theme underlies several behavioral scales.

Several factor analyses were run to check correspondence between the theoretically developed scales and statistically derived factors. The derived Factor One accounted for 21.4% of the variance. This factor corresponded very well with the Time Management Scale plus the Task Precision Scale—items which described behavioral skills relating to study habits, good use of time, and doing a good job on assignments. Another factor was composed exclusively of Family Involvement items (10.2% of the variance was accounted for by this factor). A third empirically derived factor was composed of competition items (9.8%) of the variance). A factor composed of College Involvement items contributed 10.2% of the variance. A composite factor included a number of Expectations items (19% of the variance).

Criterion Tests

Table 3 (see Norming) shows substantial differences between categorized sample types. There is an inverse progression of scores—at one end of the continuum are students who have acknowledged difficulties. These are represented by regular community college and university students. At the other end are regular entry students who have not been grouped into special help classes, including college bound high school seniors.

Results obtained from several key criterion groups were in the expected direction. In an upper middle-class high school, college bound seniors with a grade point average above 3.5, had a CSFI total mean score of 165 in the direction of greater success.

Though not the intended end-users of the CSFI, graduate students, through a process of selection and attrition, should embody the requisite skills for college success. The graduate students’ CSFI aver-age total score was 157. The CSFI scales confirmed that their scores were substantially toward the “success direction.” Another group of high achieving students was represented by a group of freshmen at a prestigious university with competitive entry. These scores, also, were skewed toward the success end of the spectrum.

In contrast, several study skills groups and probation groups in community colleges had high scores (low success potential). Their scores ranged from 183 to 20.

The 2.0 version of the CSFI was tested for concur-rent validity with 100 college freshmen. The resultant coefficient between overall test (mean) scores and students’ present GPA was r.=.41.

Continuing validation studies for Version 2.0 of the CSFI in 2009 included two new variables, and one task planning as a stand-alone variable. A correlation of r.-40 was calculated between the 2.0 version of the CSFI.

Predictive Tests

Correlational studies with Form LP5 have resulted in a correlation of -.35 between CSFI Total Score and GPA. Tests with Form LP5-R (Wellness scale deleted) resulted in a correlation of -.50 between CSFI Total Score second year GPA. Individual scales were between -.20 and -.45.

In one attrition study of 865 freshmen, the CSFI significantly distinguished between dropouts and students registering for their second year. After three academic quarters, the mean scores were significantly different in the two groups in Total (p=.004), Control (p=.02), Precision (p=.01), College Involvement (p=.0001), and Family Involvement (p=.006). All mean differences showed the dropouts had higher means (lower success).