COUNSELING CENTER

UNIVERSITY OF MARYLAND

COLLEGE PARK, MARYLAND

NON-COGNITIVE VARIABLES IN PREDICTING

ACADEMIC SUCCESS BY RACE

Terence J. Tracey and William E. Sedlacek

Research Report # 1-82

Paper presented at the annual meeting of the American Educational Research Association, New York, March, 1982.

COUNSELING CENTER

UNIVERSITY OF MARYLAND

COLLEGE PARK, MARYLAND

NON-COGNITIVE VARIABLES IN PREDICTING

ACADEMIC SUCCESS BY RACE

Terence J. Tracey and William E. Sedlacek

Research Report # 1-82

SUMMARY

A questionnaire designed to measure seven non-cognitive predictors of academic success was administered to two successive samples of incoming university freshmen. The responses were examined with regard to the reliability of the instrument, and three separate indicators of academic success—first semester college GPA, three semester cumulative GPA and persistence after three semesters. The results showed reliability and construct validity for the instrument. Further using this instrument added to the predictive validity of using traditional measures (SAT scores) on academic success. Also, different items were predictive of success for the different racial subsamples. The questionnaire was particularly predictive of the persistence of blacks. The implications of the results are discussed.

Non-Cognitive Variables in Predicting

Academic Success by Race

Since the 1970’s there has been evidence of the growing importance of retention in higher education for both human value reasons and for the continued existence of schools. A particularly vital aspect of this issue is minority student retention. The retention rate for minority students, particularly black, is lower than the rate for majority students (Astin, 1975; Sedlacek & Pelham, 1976;). The rate of minority retention is particularly low, and decreasing, in predominantly white institutions (Goodrich, 1978; Sedlacek and Webster, 1978). This is obviously a great loss of human potential and it is thus critical that steps be taken to understand and reduce attrition, particularly for minorities.

One means of increasing the retention rate is to do a better job of selection and admission (Ott, 1978). But most admissions criteria and procedures have been validated on typically white samples. Studies that have applied the usual college admissions criteria to blacks have tended to get lower validity than that obtained with the predominantly white samples (Baggeley, 1974; Borgen, 1972; Farver, Sedlacek, & Brooks, 1975; Pfeifer & Sedlacek, 1970, 1971, 1974). Among the possible explanations for this difference are cultural/racial biases in the traditional predictors (e.g., standardized tests, grades, etc.) and that minority applicants do not know how to play the admissions “game.” That is, white applicants tend to know what is viewed as desirable in college applications, but many blacks do not. Given these problems, steps must be taken to find alternative ways of obtaining valid information on minority applicants that are indicative of college success.

The purpose of this research was to design and test out a brief questionnaire for use at a predominantly white institution that might tap information related to retention not normally available. The questionnaire was specifically designed to assess the seven non-cognitive predictors of minority college success proposed by Sedlacek and Brooks (1976). Through research, they found seven variables that have been demonstrated to be related to college success, particularly for minorities. These seven variables are: positive self-concept, realistic self-appraisal, understanding of and ability to deal with racism, preference for long-range goals over short-term or immediate needs, availability of a strong support person, successful leadership experience, and demonstrated community service. While these variables have been studied individually, little work has been done on them collectively. So the focus of this study was the development of a quick, reliable and valid measure of these variables. This project was part of an ongoing research plan aimed at gaining a more complete, longitudinal picture of retention, particularly with regard to minorities.

Method

Sample

The two separate samples of incoming freshmen at the University of Maryland, College Park (1979 entering freshmen, N = 2137; 1980 entering freshmen, N = 573) were given the Non-Cognitive Questionnaire (NCQ) during summer orientation. Only those freshmen who had completed all the NCQ and whose SAT scores were able to be obtained from university records were included in this study. This resulted in final samples of 1644 for the 1979 freshmen and 478 for the 1980 freshmen. Of this final 1979 sample of 1644, 1339 identified themselves as white, 190 as black and 110 as being of other racial/ethnic backgrounds (predominately Asian-American). For the 478 freshmen in the 1980 sample, 355 were self-identified as white, 89 as black and 34 as other (again, predominantly Asian-American).

Instrument

The Non-Cognitive Questionnaire (NCQ) was designed with past research as a base and is intended to assess seven non-cognitive variables found to be related to minority retention (Sedlacek & Brooks, 1976). The NCQ consists of two nominal items relating to educational expectations, 18 Likert-type items relating to expectations about college and self-assessment, and three open-ended questions relating to present goals, past accomplishments and offices held/groups belonged to. All items, with the exception of the open-ended items, have been found to have adequate test-retest reliability. The two week correlations (N = 18) for the items range from .70 to .94. The open-ended items were included in the questionnaire as they may have been able to access dimensions not covered in the structured Likert-type item format. The responses to the question asking for one’s goals were rated for: 1) the amount of time required to complete the goals, i.e., how long-range they are (interrater r = .89) and 2) the degree to which the goals are related to academia (academic goals interrater r = .83). The open-ended item asking for which past accomplishments on is proudest of was rated for the degree of difficulty relative to all high school graduates (interrater r = .88). The final open-ended item asked the respondent to list all offices held and/or extracurricular activities. This item was rated on four dimensions: 1) number of activities (interrater r = 1.00), 2) degree of leadership exhibited (interrater r = .89), 3) degree list was related to academia (academic activities interrater r = .98) and 4) the degree to which community involvement was reflected (interrater r = .94). Lockett (1980) reported coefficient alpha reliabilities ranging from .54 to .73 for scales in a modified version of the NCQ employed in the present study.

Analyses

There were two basic types of analyses performed on the data, each reflecting the major purposes of this study. First, the properties of the instrument itself were examined to see if the responses did vary across the races and if the items were content valid in their ability to tap the seven non-cognitive dimensions posited by Sedlacek & Brooks (1976). To accomplish this, the relationships among the Likert-type items were examined using separate factor analyses for each race. A principal components factor analysis, using squared multiple correlations as commonality estimates and varimax rotation was done on the Likert-type items for the entire sample, the white sample, and the black sample. These factor analyses would yield information on the degree to which the items clustered along the posited seven non-cognitive dimensions and how this varied by race.

The second set of analyses was designed to establish the external validity of the NCQ as a predictor of collegiate success. College success can be defined in many different ways, i.e., grade point average, continued enrollment, etc. Examining retention using only one of these definitions can lead to an invalid or biased picture of what contributes to retention (Tracey & Sedlacek, 1981). As such, this study used two separate, albeit not mutually exclusive, measures of collegiate success, grade point average (GPA) and enrollment status, to move toward gaining a complete understanding of this issue.

Separate step-wise regressions were performed on each sample (1979 freshmen and 1980 freshmen) examining the relationship of the NCQ items and SAT scores to GPA (one semester GPA and three semester GPA for the 1979 sample and one semester GPA for the 1980 sample). In addition, as traditional cognitive data (i.e., SAT scores) often dominate regression equations which include noncognitive variables, separate regressions were done using only the NCQ responses as predictors. The above regressions were performed on: 1) the entire sample for each year, 2) whites only subsample for each year, and 3) blacks only subsample for each year.

To examine the relationship of the NCQ responses and SAT scores to persistence, stepwise discriminant analyses were used. As an accurate determination of persistence (enrolled vs. not enrolled) could only be obtained after several semesters, this analysis was done using only the 1979 sample, as enrollment status was determined over three semesters not just one. As with the regression analyses, stepwise discriminant analyses were done using NCQ responses plus SAT scores, and NCQ responses alone as predictors to determine the extent of overlap between the data seta. Also, as with the regressions above, separate discriminant analyses were done for the entire sample for each year, the white samples for each year and the black subsample for each year.

Results

This section will be divided into two parts, the first describing those results which examine the differences and similarities in the responses to the instrument across races. In a sense, this part concerns itself with measurement properties of the questionnaire. The second part of the results will be concerned with describing the analysis done relating questionnaire responses to success in college. The section emphasizes the application of the data. As the questionnaire as a whole was developed with minority selection in mind, most of the subsequent writeup will center on the minority data, particularly black, as this group had sufficient numbers for all the analyses, which most of the other minority groups did not. All differences noted below are statistically significant at the .05 level.

Internal Questionnaire Results

The results of the separate factor analyses conducted showed fairly similar structures for each racial group. Because this study was most concerned with minority students and because of space limitations, only the factor analysis on the black sample will be presented. Table 1 is a summary of the factors obtained and the items that loaded from this factor analysis on the black sample. As can be seen from Table 1, the results of the factor analysis demonstrate support of six of the seven non-cognitive variables suggested by Sedlacek and Brooks (1976). The six variables that were supported by the factor analysis were: Leadership (Factor I), recognizing racism (Factor II), preference for long-range goals (Factor III), realistic self-appraisal (Factor IV), support for college plans (Factor V), self-confidence (Factors VI and VII). Factor VIII seemed to be assessing general familiarity with academia unrelated to academic self-confidence. So the items used do appear to cluster along the seven variables as designed.

Predicting Collegiate Success

Given the number of items and analyses done on the different samples, only those items that significantly added to the prediction of any of the criteria (first semester GPA, three semester cum or enrollment status) will be presented. The specific items that significantly added to prediction in each analysis and the overall multiple correlation coefficients are summarized in Table 2.

In all critical analyses, the NCQ items were at least as highly predictive of the criteria examined as SAT scores alone. Combining the NCQ items with SAT scores resulted in significant increases in prediction in each of the eight separate analyses performed. So for all criteria, the usage of the NCQ items added to the ability to predict collegiate success for black and whites.

When the criterion that was examined was first semester grades, the NCQ was found to be more predictive for whites than blacks in both sample years. Further, the same non-cognitive variables were related to first semester grades for each racial group. The variables that were found to be predictive for both races were: positive self-confidence (items 3 and 8, as listed in Table 2) and realistic self-appraisal (items 9, 12, and 13). For the white subsample only, community involvement (item 14c) leadership (item 4) and preference for long range goals (item 7) were also predictive of first semester grades. Thus, the non-cognitive variables (particularly self-confidence and self-appraisal) were predictive of first semester grades for both races but this relationship was stronger for whites than blacks.

A similar picture appears from the analyses performed using three semester cum as the criterion. The non-cognitive variables of positive self-concept (items 8, 11, and 3 alone for whites) and realistic self-appraisal (item 2 for blacks and items 2, 6, and 13 for whites) were highly related to cum for both races. And like the previous analyses, more of the non-cognitive variables were predictive for whites. For whites only, the variables of preference for long range goals (item 1) and recognizing racism (item 10) were related to cum. It is noteworthy that the multiple correlation coefficients in these analyses were higher than those coefficients of the analyses done on first semester grades. The NCQ items, and implicit variables, were more predictive with increasing time.

The final analyses related the NCQ items to enrollment status after three semesters. It was here that a strong relationship was found between the non-cognitive variables and college success for the black subsample but not the white subsample. Only one of the non-cognitive variables (realistic self-appraisal) was predictive of enrollment for whites, while four of the variables were predictive for blacks. Realistic self-appraisal (item 2), positive self-confidence (items 3 and 8), support (item 5) and community involvement were significantly related to continued enrollment. Of all the analyses done, this was the only set where the predictive power of the resulting equation for the blacks was higher then the prediction of the equation for the whites. So, for blacks, the non-cognitive variables are most predictive of continued enrollment and moderately predictive of grades; while these variables are predictive of grades for whites but not particularly predictive with regard to enrollment status.

Discussion

The results of this study support the increase in predictive power gained by using non-cognitive variables, as measured by the NCQ, in addition to the usual academic predictors, i.e., SAT scores. In every one of the analyses performed, the addition of the non-cognitive items to the SAT scores significantly increased the prediction of grades and enrollment status. Further, the use of the NCQ items alone (without SAT scores) yielded a significantly higher relationship to college success (GPA and enrollment status) than did the SAT scores alone.

The predictive power of the NCQ was evident in each of the racial subgroups studied. In fact, using the NCQ added slightly more to the prediction of college grades for whites than it did for blacks. But this result was probably more due to the far greater number of whites in the sample than blacks. With this much higher number, any relationship evidenced in the regression would more likely attain significance even though the level of the relationship (r) in the two samples was equal.

Generally, it was slightly easier to predict grades after one and three semesters for white students than it was for black students, even with the inclusion of the NCQ which was designed to increase prediction with blacks. But when a different criterion of collegiate success was examined, that of enrollment status after three semesters, the opposite relationship was evidenced. The enrollment status of blacks was much better predicted from the NCQ whereas using this questionnaire yielded little predictive power for whites. So it appears that different processes are operative for each race with regard to collegiate success.

For whites, the non-cognitive dimensions (Sedlacek & Brooks, 1976) of self-confidence, preference for long range goals over short-term or immediate needs, and realistic self-appraisal were most strongly related to GPA. In addition, some items relating to leadership, community service and not understanding racism entered some of the predictive equations but the relationship was not as strong as those above. But with regard to enrollment status, the only items that were significant, albeit marginally, were those reflecting positive self-concept. So the non-cognitive dimensions of positive self-concept, ability to delay gratification, and realistic self-appraisal were highly related to doing well academically in college for whites. The dimensions related to continued perseverance in school for whites were not generally related to the variables measured in the NCQ and proposed by Sedlacek and Brooks (1976).

For blacks, the opposite pattern emerged. The only non-cognitive variables that were related to academic achievement, i.e., GPA, were positive self-concept and realistic self-appraisal. The strength of the relationship of these dimensions to GPA was not as high as it was for the white subsamples. But while there was little, if any, ability to predict enrollment status using the NCQ for whites, there was a strong relationship for the black subsample. For blacks, the dimensions that were related to continued enrollment were positive self-concept, support and community service. Having a person(s) available to support the black student when needed and having had experience in community service were strongly related to staying in school. This support person does not have to be a member of the family (as these items did not load into the analyses). What seems to be measured by these dimensions is an ability to ask for help when it is needed. This ability was more crucial for continued existence in college for blacks than whites.

It is interesting to note that the variables of support and community service are related to continued enrollment for blacks and not to GPA. This seems to indicate that those blacks who do get good grades and stay in school have similar levels of self-confidence and ability to realistically appraise themselves as those blacks who do not persevere. The key difference between these two groups of blacks is that those who continue have more support in the family and community to continue. This continuance relationship does not appear at all for whites. This result indicates that the process of succeeding in college varies between the races. Success for whites should be examined in terms of grades; while success for blacks should be examined first with regard to enrollment status and then with regard to grades as different processes appear involved.

The results of this study demonstrate that the Non-Cognitive Questionnaire is both reliable and valid as an aid in predicting collegiate success for both blacks and whites. The exact relationship of the NCQ to collegiate success varies between blacks and whites. For whites, the NCQ significantly adds to the prediction of grades while for blacks it is related to both grades and enrollment status. Lockett (1980), using a modification of the NCQ presented here, found that for blacks as the University of Missouri positive self-concept, community participation, leadership, and understanding racism correlated with grade point averages. Lockett further found that long range goals, lower self-concept and realistic self-appraisal correlated with satisfaction with the college environment for black students. Given this reliability and validity, the NCQ items can be used as a beneficial addition to those collected in initially selecting students. Also, the NCQ could be of value post-admission. Students could be given the NCQ during orientation, as was done here, and those students lacking in the dimensions related to collegiate success could be identified. Programs aimed specifically at these students could then be developed and implemented. Thus, efforts could be directed where they are most needed such as aiding black students that do not have the self-confidence, support and community service experience to keep them in school.

References

Astin, A.W. Preventing students from dropping out. San Francisco: Jossey-Bass, 1975.

Baggaley. A.R. Academic prediction of an Ivy League college, moderated by

demographic variables. Measurement and Evaluation in Guidance, 1974, 6, 232-235.

Borgen, F.H. Differential expectations? Predicting grades for black students in five types

of colleges. Measurement and Evaluation in Guidance, 1972, 4, 206-212.

Farver, A.S., Sedlacek, W.E. & Brooks, G.C., Jr. Longitudinal predictions of university

Grades for blacks and whites. Measurement and Evaluation in Guidance, 1975, 7,

243-250.

Goodrich, A. A data-driven minority student retention model for faculty and

administrators in predominantly white institutions. Paper presented at the

annual meeting of the American College Personnel Association, Detroit,

March, 1978.

Lockett, G.C. A study of traditional measures and non-traditional measures used to

predict the success of black college students. Unpublished doctoral dissertation,

University of Missouri, Columbia, 1980.

Ott, L.S. Admissions management with the focus on retention. In L. Noel (Ed.)

Reducing the drop-out rate. San Francisco: Jossey-Bass, 1978.

Pfeiffer, C.M., Jr. & Sedlacek, W.E. Non-intellectual correlates of black and white

Student grades at the University of Maryland, College Park. Cultural Study Center

Research Report #3-70, University of Maryland, College Park, 1970.

Pfeiffer, C.M., Jr. & Sedlacek, W.E. The validity of academic predictors for black and

white students at a predominantly white university. Journal of Educational

Measurement, 1971, 8, 253-261.

Pfeiffer, C.M., Jr. & Sedlacek, W.E. Predicting black student grades with non-

Intellectual measures. Journal of Negro Education, 1974, 43, 67-76.

Sedlacek, W.E. & Brooks, G.C., Jr. Racism in American education: A model for

change. Chicage: Nelson-Hall, 1976.

Sedlacek, W.E. & Pelham, J.C. Minority admissions to large universities: A national

survey. Journal of Non-White Concerns in Personnel and Guidance, 1976, 4, 53-63.

Sedlacek, W.E. & Webster, D.W. Admission and retention of minority students in large

universities. Journal of College Student Personnel, 1973, 19, 242 –248.

Tracey, T.J. & Sedlacek, W.E. A description and illustration of a model for conducting

student retention research. National Association of Student Personnel Administrators

Journal Field Report, 1981, 5, 2, 5-6.

Table 1

Summary of the Factors and Largest Loading

Items[1] Identified in the Black Sample[2]

Factor I Leadership (32.8% of common variance)

Item Loading

I am sometimes looked up to by others. .73

If I ran into problems concerning school,

I have someone who would listen to me

and help me. .41

In groups where I am comfortable, I am

often looked to as leader. .55

Factor II Fair academic opportunity (13.9%)

Item Loading

I want a chance to improve myself

academically. .62

If course tutoring is made available

on campus at no cost, I would

attend regularly. .54

I expect I will encounter racism at UMCP. .33

Factor III Preferring long-range goals (13.5%)

Item Loading

Once I start something, I finish it. .90

When I believe strongly in something,

I act on it. .54

Factor IV Academic Self-Appraisal (10.9%)

Item Loading

I am as skilled academically as the

Average applicant to UMCP. - .58

I expect to have a harder time than

most students at UMCP. .45

Table I (Continued)

Factor V Family Support (9.1%)

Item Loading

My family has always wanted me to

go to college. - .62

My friends and relatives don’t feel

I should go to college. .62

Factor VI Lack of Perseverance (7.7%)

Item Loading

I get easily discouraged when I try to do

something and it doesn’t work. .51

People can pretty easily change me even

though I thought my mind was already

made up on the subject. .44

Factor VII Self-Confidence (6.5%)

Item Loading

When I believe strongly in something, I act

on it. .39

My high school grades don’t reflect what I

can do. .38

Rated difficulty of three best accomplishments. .31

Factor VIII Academic Familiarity (5.6%)

Item Loading

It should not be very hard to get a B

(3.0) average at UMCP. .43

Rated degree of academic relatedness of

three most primary goals. .41

Table 2

Summary of the NCQ Items that were Significant1 and Corresponding Beta Weights for Each of the Analyses Performed2

			Regressions on first			Regressions on three			Discriminant analyses
			semester GPA			semester GPA			on enrollment status
									after three semesters
			1980 Sample		1979 Sample		1979 Sample		1979 Sample
		Whites Blacks	Whites	Blacks	Whites	Blacks	Whites	Blacks	Whites	Blacks
Items			N=355	N=89	N=1339	N=190	N=1027	N=158	N=1046	N=154
1) Three goals that
you have for your-
right now. Rated
for:
a) Time to
complete							-.09
2) Uncertainty of graduation	nty of grad
given that 50% do not.							-.07	-.22	.36	.53
3) List three things
that you are proud
of having done.
Rated for degree			.19	.28			.12	.19		-.43
of difficulty.
4) I am sometimes looked
up to by others.					-.09
5) If I run into problems
concerning school, I
have someone who would
listen to me and help						.07				-.52
me.
6) I expect to have a
harder time than most
students.					.12		.06
7) Once I start something
I finish it.					-.07
8) When I believe strongly
in something, I act on
it.					-.08	.10		.22		-.38
9) I am as skilled aca-
demically as the
average applicant.						-.10
10) I expect I will
encounter racism
at UMCP.							-.08
1) People can pretty
easily change me even
though I thought my						-.17		-.16
mind was made up on
the subject.
12) I want a chance to prove
myself academically.					-.06
13) My high school grades
don't really reflect
what I can do.			.34	.21	.31		.32		-.68
14) List of offices held
and activities.
Rated for:
a) leadership
b) academic related-
ness			.16				.61
c) community involve-
ment			.09							-.35
Multiple R for analyses			.48	.38	.39	.29	.44	.40	.18	.49
Multiple R for analyses
with SAT included			.59	.51	.51	.37	.54	.41	.24	.52

---

[1] Only those items with loadings above .30 are reported.

[2] Complete factor and intercorrelation matrices for this sample and the white sample are available upon request from William Sedlacek, Counseling Center, UMCP, College Park, MD, 20742.

1 (p 105)

2 Copies of complete instrument are available from William E. Sedlacek, University of Maryland, College Park 20742