HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Kingsley, K. Articles by Galbraith, G. M. Search for Related Content PubMed PubMed Citation Articles by Kingsley, K. Articles by Galbraith, G. M.

Creating an Evidence-Based Admissions Formula for a New Dental School: University of Nevada, Las Vegas, School of Dental Medicine

Key words: Dental Admission Test (DAT), National Board Dental Examination (NBDE) Part I, correlation study, outcomes assessment

Submitted for publication 08/18/06; accepted 11/28/06

	Abstract

Top Abstract Materials and Methods Results Discussion Conclusion References

 
This article reports development of an evidence-based admissions formula that effectively incorporates the admissions criteria most likely to influence dental school performance. This study utilized peer-reviewed literature and analysis of admissions and performance data from the first three classes of students at the University of Nevada, Las Vegas, School of Dental Medicine (UNLV-SDM). We used Pearson’s correlation, linear regression, and ANOVA to determine the strength and direction of association between admissions variables, both singly and in combination, and performance measures. Our initial results revealed no significant relationship between our previous admissions formula, which was adapted from other dental admissions offices, and student performance for our first class and National Board Dental Examination Part I (NBDE-I) (R=.288) or dental school grade point average (DS-GPA) (R=0.193). After using the combined data from the first three classes of students at UNLV-SDM, we confirmed no significant relationship between our previous admissions formula and DS-GPA (R=0.207) and a slight increase in correlation to NBDE-I (R=0.303). More detailed analysis of the admissions variables within the formula revealed that some Dental Admission Test scores, such as Reading Comprehension, Quantitative Analysis, and Biology, were significantly correlated with dental school performance at UNLV-SDM, allowing for revision of the admissions formula to a formula score that is now significantly correlated with student performance for the first class to NBDE-I (R=0.458) and DS-GPA (R=0.368), as well as the combined data from the first three cohorts at UNLV-SDM (R=0.361, 0.218, respectively). In addition, this reformulation did not significantly impact the overall ranking of females or minorities. Although formulaic data can never perfectly predict student performance, this study demonstrated that constant review and revision of relevant admissions criteria are needed for each school to maintain an evidence-based admissions program that provides for fair and effective comparison of student admissions data.

All accredited dental schools in the United States currently require applicants to take the Dental Admission Test (DAT). The DAT serves as the only nationwide, standardized test that is required, and therefore used, by dental admissions officers to compare dental school applicants from the United States and other countries. The American Dental Association (ADA) administers the DAT, which is designed to gauge general academic ability, comprehension of scientific information, and perceptual ability.²

Although the DAT is administered to all prospective dental school applicants, it serves as only one factor, among many others, considered in the admissions and enrollment process.³ Other measures most commonly used are undergraduate science GPA, general or cumulative GPA, letters of evaluation from faculty or members of the profession, relevant work experience in the field of dentistry, community service, and personal on-site interviews.⁴

Early studies concluded that GPA and DAT scores can be used as independent predictors of dental school performance.^5–7 More recent studies have demonstrated that multiple admissions criteria, in combination, provide evidence of stronger, positive correlations between calculated admissions scores and dental school performance outcomes, such as DS-GPA and NBDE-I scores.^8–10 It is also important to note that the dental school admissions process exists not only to select the candidates who are most likely to do well in school, but also to identify the individuals who are most likely to serve the public well in their professional careers.^11–13 Using multiple admissions criteria in combination, including assessment of community service and subjective interviews, may help to identify those applicants most likely to serve their communities.

This study examined the relationships between UNLV’s original admissions criteria and formula with actual student performance in dental school at UNLV. In addition, this study also examined the predictive power of various dental school admissions factors not currently used by UNLV to identify significant trends and relationships.

	Materials and Methods

Top Abstract Materials and Methods Results Discussion Conclusion References

 
Performance and admissions data from 225 students of the first three UNLV-SDM classes (admitted 2002, 2003, 2004) were retrieved, and each record was assigned a numerical, nonduplicated identifier by the Office of Student Affairs. The corresponding dental student performance measures (NBDE-I, DS-GPA) were also matched and assigned the corresponding numerical identifier prior to dissemination of 210 complete records to the study authors, to prevent the disclosure and ensure the confidentiality of personally identifiable private information. Minorities were classified by UNLV to include underrepresented students of black, Native American/American Indian/Alaska Native, Hispanic, Asian, and Pacific Islander descent.

We filed, subsequently amended, and received approval for our protocol from the Institutional Research Board (IRB), as an exemption to human subjects research under the Basic HHS Policy for Protection of Human Research Subjects (46.101) Subpart A (b) regarding IRB Exemption for research involving the use of education tests (cognitive, diagnostic, aptitude, achievement) where the subjects cannot be identified or linked, directly or through identifiers, to the individual subjects.

The t-test is a commonly employed statistical procedure used to infer whether differences exist between the means of two population samples. In this study, the means of the admissions and outcome (performance) measures for demographic groups (males, females; minority, nonminority) were analyzed using a t-distribution.¹⁴ We expected to find no significant differences in the means of these groups, based upon either ethnicity or gender. As long as the sample size is even moderate (>20) for each group, quite severe departures from normality make little practical difference in the conclusions reached from these analyses.¹⁵ All samples were measured using two-tailed t tests, as departure from normality can make more of a difference in a one-tailed than in a two-tailed t test. Significance level for these analyses was =0.05. All statistical analyses were completed using SPSS.¹⁶

Simple linear correlation considers the relationship between two variables, but neither is assumed to be functionally dependent upon the other.¹⁷ Based upon this understanding, Pearson’s correlations were performed to analyze the strength and direction of association between individual admissions variables (independent variables) and student performance measures (dependent variables). Results from the multifactor admissions formulas (previous, revised/new) were also analyzed using Pearson’s correlation of formula score to the student performance measures (NBDE-I, DS-GPA) to reveal associations. We expected that most, if not all, admissions variables would correlate with performance measures. Significance level for these analyses was =0.05.

Pearson’s R or correlation coefficients measure the strength of linear relationships and were interpreted using the following:

• –1.0 to –0.7 strong negative association
  
• –0.7 to –0.3 negative association
  
• –0.3 to +0.3 little or no association
  
• +0.3 to +0.7 positive association
  
• +0.7 to +1.0 strong positive association
  

If there is a logical relationship that implies functional dependence of one variable on another, linear regression can help to determine the magnitude of the effect of one variable on another. Based upon this understanding, linear regression was performed using dental student performance (NBDE-I, DS-GPA) as dependent variables to isolate relevant predictors. R and R² were calculated to determine the relative contribution of these factors to the dependent variable assessed. Adjusted R² (AR²) and power (p) were calculated to determine the generalizability of these results to other study populations.^16,17 In many kinds of data involving human subjects, however, the relationship may not be one of function dependence but of correlation. For example, although a student may score well on the DAT Quantitative Analysis (DAT-QA) section, performance on the NBDE-I may not be due to this, but rather to some other quality or characteristic that accounts for both. We expected to find few, or no, significant linear regressions of this data based upon the assumption that the data are correlated via other relevant mechanisms (e.g., hours spent per student for study or preparation).

Because these analyses involve multiple two-sample t-tests and correlations, these data have a higher probability of Type I error (incorrectly rejecting the null hypothesis, H_O). ANOVA was performed to more accurately assess the relationships between the predictor (admissions formula score: previous or revised/new) and the dependent variables (NBDE-I, DS-GPA). Based upon results from the correlations, we expected that DAT Biology (DAT-BIO), Reading Comprehension (DAT-RC), and DAT-QA, but not other DAT variables, would remain significant predictors of student performance. Significance level for these analyses was =0.05.^16,17

	Results

Top Abstract Materials and Methods Results Discussion Conclusion References

 
This study characterized various admissions variables often used by admissions officers for the strength and reliability of these measures in predicting student performance. To determine if there were confounding influences from specific demographic variables, we sorted the admissions variables into the categories of ethnicity and gender to isolate any significant pre-existing differences within these sub-populations (Table 1). The results of two-tailed t tests of population means from the first class at UNLV-SDM revealed no significant differences between females and males or between minorities and nonminorities in most of the admissions variables at a significance level, =0.05. Two differences between males and females were initially identified: DAT Organic Chemistry (DAT-OC) (p<0.05) and Perceptual Ability (DAT-PAT) (p<0.05) scores were higher among males. In assessing the combined data from all three classes, we found males had slightly higher scores on DAT-BIO (p<0.05) and again on DAT-OC (p<0.05), but not on DAT-PAT, as was found in the initial class. No significant differences in DAT scores were found by ethnicity, although incoming science GPA and cumulative GPA were slightly lower for minorities.

• [(overall GPA/4) + (science GPA/4)] x [(DAT-AA + DAT-PA)/60] x 300=Objective score
  
• (subjective interview score₁ + subjective interview score₂)/2=Subjective score
  
• Note: 1=faculty interview, 2=student interview
  
• Objective score (range: 0–600) + Subjective score (range: 0–400)=1000, maximum score
  

To determine if these admissions variables used by the UNLV-SDM, and by other dental school admissions officers, were independent predictors of dental school performance at UNLV-SDM, we analyzed the strength and direction of correlation between these admissions variables and both NBDE-I scores and DS-GPA (Table 4). The results of the Pearson’s correlation analysis for the first class initially revealed several significant relationships among admissions variables. First, DAT-BIO was found to be a statistically significant predictor of both DS-GPA (R=0.310) and NBDE-I scores (R=0.383), supporting the observations made by De Ball et al.¹¹ In addition, DAT-RC scores were also significantly correlated to DS-GPA (R=0.332) and NBDE-I scores (R=0.367), but were slightly less robust than DAT-BIO in their predictive capabilities, an observation supported by Bergman et al.¹² Although DAT-QA was initially predictive for NBDE-I for the first class (R=0.318), when the data from all three classes were reviewed, only DAT-BIO remained a significant predictor of NBDE-I score (R=0.304), but not DS-GPA (R=0.148). Finally, we found that the subjective evaluation, which included the faculty interview, was not helpful in predicting student academic performance (NBDE-I, DS-GPA), as previously reported by Stacey et al.¹³

• {[(science GPA + cumulative GPA)/8] x [DAT-RC/30] x [(DAT-QA + DAT-BIO + DAT-OC)/90]} x 75=Objective score
  
• [(subjective interview score₁ + subjective interview score₂)/2] x 25=Subjective score
  
• Note: 1=faculty interview, 2=student interview
  
• Objective score (range: 0–75) + Subjective score (range: 0–25)=100, maximum score
  

To determine if the new admissions formula scores significantly lowered the rankings of individuals along categorical demographics, we analyzed our results from the previous and new calculated formula results after sorting by ethnicity and by gender to test for any pre- and post-revision changes in the ranking of these subpopulations. The results of two-tailed t-tests demonstrated significant differences between females and males with both the previous and new formula scores (Table 7). The combined data from all three classes revealed that males had higher pOBJ (p<0.05) and pOVL scores (p<0.05) than females. This difference between males and females, although remaining significant in both nOBJ (p<0.05) and nOVL score (p<0.05), was reduced by the new/revised formula. The comparison of minority and nonminority students using the combined, three-year data set revealed that minority students had slightly lower pOBJ (p<0.05) and pOVL (p<0.05) scores than nonminority students, differences that remain significant under the new formula score, nOBJ (p<0.05) and nOVL (p<0.05). These differences in minority student score are neither reduced nor amplified by the new formula.

	Discussion

Top Abstract Materials and Methods Results Discussion Conclusion References

The results of this study found that some of the admissions variables cited in the literature did correlate well with dental student performance, while other variables lacked significant correlation with student performance. Our study confirmed that DAT-BIO, DAT-RC, and DAT-QA were the variables most strongly associated with NBDE-I scores and DS-GPA, among all groups at UNLV-SDM.^12,18 One seemingly contradictory finding was that DAT-OC was not associated with either NBDE-I scores or DS-GPA, although this factor had been identified as a predictor of NBDE-I performance in other studies.¹² This anomaly may be attributed to random data variation, a temporal statistical anomaly, or perhaps a true difference in the population in this study. We are confident that, as additional future classes are admitted, we can assess their academic progress and, over time, increase the relative power (p=0.38) and applicability of the findings in this study.

Based upon these findings, we constructed a revised, evidence-based formula for admission to UNLV-SDM that incorporated the admissions criteria most likely to predict dental school performance outcomes. Using this revised formula, we performed a retrospective evaluation of the dental students’ completed NBDE-I scores and DS-GPA as dependent outcomes. Our results found a significant correlation between the revised formula admissions score and NBDE-I scores (R=.361), representing a significant increase in correlation over the previous admissions formula results (R=0.303). Correlation of new formula score, nOVL to DS-GPA, increased (R=0.218) compared with the previous formula score pOVL (R=0.208) although the correlation is not statistically significant. Our results from the ANOVA and linear regression support our use of DAT-RC, DAT-QA, and DAT-BIO to predict dental school performance. Two-tailed t-tests revealed that scores and overall ranking of females and minorities were not significantly altered using the new formula, although the new formula reduced the difference in scores between females and males. These data further suggest that the revised formula neither favors nor discriminates against either subgroup.

It is clear that no admissions formula models can uniformly and unerringly predict clinical, didactic, or academic performance. Furthermore, the temporal nature of these results indicate that these correlations are not uniformly generalizable, but rather that some of the NBDE-I score and DS-GPA may be attributed to the variables used in this revised formula—a useful indicator to other admissions officers regarding the most important variables to measure and analyze within their applicant pools. The potential for, and necessity of, constant and continual revision of admissions formula models is evident. It is essential to incorporate the latest peer-reviewed evidence, as well as integrative self-study and analysis, to improve current standards for admission. While this study successfully integrated the peer-reviewed evidence into a model that improves the predictive capability of the admissions formula at UNLV-SDM, continued evaluation and recalibration should ensure the reliability and validity of these results.

This study has limitations that must be considered to accurately evaluate if its results can be generalized to other institutions. First, this study is based primarily on data from the first three classes of dental students at UNLV-SDM, and therefore does not provide a long-term longitudinal representation of student admissions data or performance outcomes. Second, this study is also limited to the students who have enrolled at UNLV-SDM and does not include larger populations from other institutions, which may reveal regional and local differences from this applicant pool.

In addition, Bergman et al. reported that DAT-QA was the least accurate predictor of dental school performance, while at UNLV-SDM the predictive value of DAT-QA was significant in the initial evaluation from the first class of students at UNLV-SDM.¹² Although many schools may not weight this score heavily in the admissions process, these data suggest that further research may be necessary to elucidate the relationship that may exist between this score and student performance. Although analysis may fluctuate from year to year and may also be region- or school-specific, we anticipate that further review over time will help to provide more definitive answers regarding these initial findings.

	Conclusion

Top Abstract Materials and Methods Results Discussion Conclusion References

 
These results suggest that the relationship between admissions criteria and academic success in dental school may be more complex and dynamic than previous studies have concluded. Although this study is limited by the effects of a small clustered sample and retrospective data analysis, it nonetheless demonstrates that the relationships that exist between admissions criteria and dental student performance measures at UNLV-SDM can be utilized for their predictive potential. This study further demonstrates that constant review and revision of the many factors that predict academic performance in dental school are necessary to maintain a relevant, evidence-based admissions program that fairly and effectively compares student admissions data. Based upon the results of this study, UNLV-SDM has adopted the revised formula, and the first students admitted using the revised formula are currently engaged in their first semester of study.

Although the DAT was designed for predictive validity, the changing nature of dental medicine and the challenges of discovering new and ever more complex interactions between systemic health and oral health necessitate review and perhaps revision of admissions criteria to more successfully identify those students with the capability and potential to succeed in the dental school environment and ultimately as dental professionals in the health care environment of the twenty-first century. These results can provide guidance for admissions officers seeking to select the candidates most likely to excel in dental school and to serve the community, but these findings may also benefit dental school applicants by allowing them to identify both their strengths and deficiencies that may affect their performance in dental school.

	Footnotes

 
Dr. Kingsley is Assistant Professor, Department of Biomedical Sciences; Mr. Sewell is a second-year dental student; Dr. Ditmyer is Assistant Professor in Residence, Department of Professional Studies; Ms. O’Malley is Staff Research Associate, Department of Biomedical Sciences; and Dr. Galbraith is Chair and Professor, Department of Biomedical Sciences—all at the University of Nevada, Las Vegas, School of Dental Medicine. Direct correspondence and requests for reprints to Dr. Karl Kingsley, Department of Biomedical Sciences, University of Nevada, Las Vegas, School of Dental Medicine, 1001 Shadow Lane B315, Las Vegas, NV 89106-4124; 702-774-2623 phone; 702-774-2721 fax; karl.kingsley{at}unlv.edu.

	REFERENCES

Top Abstract Materials and Methods Results Discussion Conclusion References

 

1. Weaver RG, Ramanna S, Haden NK, Valachovic RW. U.S. dental school applicants and enrollees: 2003 and 2004. J Dent Educ 2005; 69:1064–72.[Abstract/Free Full Text]
2. American Dental Association. At: www.ada.org/prof/ed/testing/dat/index.asp. Accessed: August 18, 2006.
3. Kramer GA. Predictive validity of the Dental Admission Test. J Dent Educ 1986; 50:526–31.[Abstract]
4. Sandow PL, Jones AC, Peek CW, Courts FJ, Watson RE. Correlation of admission criteria with dental school performance and attrition. J Dent Educ 2002; 66:385–92.[Abstract]
5. Kreit LH, McDonald RE. Preprofessional grades and the dental aptitude test as predictors of student performance in dental school. J Dent Educ 1968; 32:452–8.[Medline]
6. Thompson GW, Ahlawat K, Buie R. Evaluation of the dental aptitude test components as predictors of dental school performance. J Can Dent Assoc 1979; 45:407–9.[Medline]
7. Boozer CH, Lee MM, Rayson J, Weinberg R. Prediction of academic success: a study with dental students using noncognitive and cognitive variables. J Am Coll Dent 1984; 51:14–21.[Medline]
8. Kress GC Jr, Dogon IL. A correlation study of preadmission predictor variables and dental school performance. J Dent Educ 1981; 45:207–10.[Abstract]
9. Potter RH, McDonald RE, Sagraves GD. A derived basic ability criterion for predicting dental students’ performance. J Dent Educ 1982; 46:634–8.[Abstract]
10. Staat RH, Yancey JM. The admission index in the dental school admission process. J Dent Educ 1982; 46:500–3.[Abstract]
11. De Ball S, Sullivan K, Horine J, Duncan WK, Replogle W. The relationship of performance on the Dental Admission Test and performance on Part I of the National Board Dental Examinations. J Dent Educ 2002; 66:478–84.[Abstract]
12. Bergman AV, Susarla SM, Howell TH, Karimbux NY. Dental Admission Test scores and performance on NBDE Part I, revisited. J Dent Educ 2006; 70:258–62.[Abstract/Free Full Text]
13. Stacey DG, Whittaker JM. Predicting academic performance and clinical competency for international dental students: seeking the most efficient and effective measures. J Dent Educ 2005; 69:270–80.[Abstract/Free Full Text]
14. Hinkle DE, Wiersma W. Applied statistics for the behavioral sciences. 5^th ed. New York: Houghton Mifflin Company, 2003.
15. Hays WL. Inferences about population means. In: Statistics. 5^th ed. New York: International Thomson Publishing, 1994:311–42.
16. SPSS for Windows, Rel. 14.0.2. Chicago: SPSS Inc., 2006.
17. Heiman GW. Basic statistics for the behavioral sciences. 5^th ed. New York: Houghton Mifflin Company, 2006.
18. Park SE, Susarla SM, Massey W. Do admissions data and NBDE Part I scores predict clinical performance among dental students? J Dent Educ 2006; 70:518–24.[Abstract/Free Full Text]

This article has been cited by other articles:

				D. C. Holmes, J. V. Doering, and M. Spector Associations Among Predental Credentials and Measures of Dental School Achievement J Dent Educ., February 1, 2008; 72(2): 142 - 152. [Abstract] [Full Text] [PDF]

				K. Kingsley, S. O'Malley, T. Stewart, and G. M. Galbraith The Integration Seminar: A First-Year Dental Course Integrating Concepts from the Biomedical, Professional, and Clinical Sciences J Dent Educ., October 1, 2007; 71(10): 1322 - 1332. [Abstract] [Full Text] [PDF]

This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Kingsley, K. Articles by Galbraith, G. M. Search for Related Content PubMed PubMed Citation Articles by Kingsley, K. Articles by Galbraith, G. M.