Citation
The Impact of Standardized Testing on High-Achieving Students Postsecondary Acceptance Outcomes

Material Information

Title:
The Impact of Standardized Testing on High-Achieving Students Postsecondary Acceptance Outcomes
Creator:
Hasbini, Leena
Place of Publication:
[Gainesville, Fla.]
Florida
Publisher:
University of Florida
Publication Date:
Language:
english
Physical Description:
1 online resource (109 p.)

Thesis/Dissertation Information

Degree:
Doctorate ( Ed.D.)
Degree Grantor:
University of Florida
Degree Disciplines:
Educational Leadership
Human Development and Organizational Studies in Education
Committee Chair:
ELDRIDGE,LINDA BURNEY
Committee Co-Chair:
REDDING,CHRISTOPHER HYDE
Committee Members:
MILLER,DAVID
DANA,THOMAS M

Subjects

Subjects / Keywords:
act -- admissions -- college -- sat -- testing
Human Development and Organizational Studies in Education -- Dissertations, Academic -- UF
Genre:
bibliography ( marcgt )
theses ( marcgt )
government publication (state, provincial, terriorial, dependent) ( marcgt )
born-digital ( sobekcm )
Electronic Thesis or Dissertation
Educational Leadership thesis, Ed.D.

Notes

Abstract:
Standardized testing, specifically geared toward postsecondary admission, such as the SAT and ACT, were designed to measure readiness for higher education. The purpose of this study was to examine the degree to which ACT or SAT scores impact college acceptance outcomes for high-achieving students with similar high school grade point averages. This research study also compared individual and familial variables and the parallel impact they have on postsecondary acceptance. Significant differences in acceptance outcomes were observed among the high-achieving student sample. The sample, largely homogenous with respect to high school GPA, experienced notable differences in acceptance outcomes based on test scores (SAT and ACT): higher scores were associated with acceptance to more selective institutions. Additionally, significant differences were observed in acceptance outcomes due to the demographic characteristics of gender, race, and parental demographics (parental educational attainment and household income). ( en )
General Note:
In the series University of Florida Digital Collections.
General Note:
Includes vita.
Bibliography:
Includes bibliographical references.
Source of Description:
Description based on online resource; title from PDF title page.
Source of Description:
This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Thesis:
Thesis (Ed.D.)--University of Florida, 2018.
Local:
Adviser: ELDRIDGE,LINDA BURNEY.
Local:
Co-adviser: REDDING,CHRISTOPHER HYDE.
Statement of Responsibility:
by Leena Hasbini.

Record Information

Source Institution:
UFRGP
Rights Management:
Applicable rights reserved.
Classification:
LD1780 2018 ( lcc )

Downloads

This item has the following downloads:


Full Text

PAGE 1

THE IMPACT OF STANDARDIZED TESTING ON HIGH ACHIEVING STUDENTS' POSTSECONDARY ACCEPTANCE OUTCOMES By LEENA HASBINI A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF EDUCATION UNIVERSITY OF FLORIDA 2018

PAGE 2

2018 Leena Hasbini

PAGE 3

To my family: Baba, Mama, Zeina, & Jad: you have supported me through every jou rn ey with wisdom, patience, and understanding this is for you

PAGE 4

4 ACKNOWLEDGMENTS My acceptance to the UF LEAD program commenced an incredible journey filled with challenging, rewarding, and life changing experiences. Our cohort was tremendously unique, and every individual offered significant contribution s that only brought us closer together. I would especially like to thank my chair, Dr. Linda Eldridge, for being so encouraging and understanding throughout this program I am truly appreciative of your patience as I transitioned through life stages and career changes. I would also like to take the time to thank my committee members, Dr. Christopher Redding, Dr. David Miller, and Dr. Tom Dana, for offering guidance and feedback. I would also like to thank Ryan Thurman at the University of South Florida for his support and feedback I am incredibly grateful to my parents, Mohamad Ali and Sawsan, for always pu tting our education first and t o my brother, Jad, numerous extended family members, friends, and co workers for supporting, motivating and influencing my decis i on to pursue my doctorate. Finally, I would be remiss not to mention Joanne and Pat, who in my very early years, propelled my desire to make a difference for those most in need. I wouldn't have come this far if not for you. To whom much has been given, much is expected

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ ...... 4 LIST OF TABLES ................................ ................................ ................................ ................ 7 LIST OF FIGURES ................................ ................................ ................................ .............. 8 LIST OF ABBREVIATIONS ................................ ................................ ................................ 9 DEFINITI ON OF KEY TERMS ................................ ................................ .......................... 11 ABSTRACT ................................ ................................ ................................ ........................ 12 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ ........ 13 Overview of the Topic ................................ ................................ ................................ 13 Statement of the Problem ................................ ................................ ........................... 13 Research Questions ................................ ................................ ................................ ... 14 Limitations and Assumptions ................................ ................................ ...................... 14 Limitations ................................ ................................ ................................ ............ 14 Assumptions ................................ ................................ ................................ ......... 15 Significance of the Study ................................ ................................ ............................ 15 Summary ................................ ................................ ................................ ..................... 16 2 REVIEW OF THE LITERATURE ................................ ................................ ................ 17 Or igins of Postsecondary Admissions Criteria ................................ ........................... 17 Current Admissions Practices ................................ ................................ .................... 19 High School Grades in the Admissions Process ................................ ....................... 20 Test Scores in the Admissions Process ................................ ................................ ..... 20 Standardized College Admissions Tests ................................ ................................ .... 21 SAT ................................ ................................ ................................ .............................. 22 ACT ................................ ................................ ................................ ............................. 22 Concordance between SAT and ACT ................................ ................................ ........ 23 Predictive Validity ................................ ................................ ................................ ........ 23 Test Optional Admissions Policies ................................ ................................ ............. 24 Discrepant Performance ................................ ................................ ............................. 25 Model of College Choice ................................ ................................ ............................. 27 College Choices Of High Achieving Students ................................ ........................... 28 Socioeconomic Status ................................ ................................ ................................ 29 Parental Educational Attainment ................................ ................................ ................ 30 College Selectivity ................................ ................................ ................................ ....... 30

PAGE 6

6 CIRP Freshman Survey ................................ ................................ .............................. 31 3 METHOD OLOGY ................................ ................................ ................................ ........ 32 Research Design ................................ ................................ ................................ ........ 32 Data Source ................................ ................................ ................................ ................ 33 Data Collection ................................ ................................ ................................ ............ 33 Data Analysi s ................................ ................................ ................................ .............. 33 4 RESULTS ................................ ................................ ................................ .................... 36 Summary ................................ ................................ ................................ ..................... 36 Findings ................................ ................................ ................................ ....................... 36 Descriptive Statistics ................................ ................................ ................................ ... 36 One Wa y Frequencies (Categorical and Ordinal Variables) .............................. 36 Two Way Frequencies (Categorical and Ordinal Variables) .............................. 41 Descriptive Analysis (Interval and Ratio Variables) ................................ ............ 41 Inferential Statistical Analyses ................................ ................................ ............. 42 Research Question 1 ................................ ................................ ........................... 45 RQ1 SAT ................................ ................................ ................................ .............. 45 RQ1 ACT ................................ ................................ ................................ .............. 46 Research Question 2 ................................ ................................ ........................... 47 RQ2 SAT ................................ ................................ ................................ .............. 48 RQ2 ACT ................................ ................................ ................................ .............. 52 Research Question 3 ................................ ................................ ........................... 55 RQ3 SAT ................................ ................................ ................................ .............. 56 RQ3 ACT ................................ ................................ ................................ .............. 59 Limitations ................................ ................................ ................................ ................... 62 5 DISCUSSION AND RECOMMENDATIONS ................................ .............................. 64 Implication of the Findings ................................ ................................ .......................... 64 Recommendations for Future Research ................................ ................................ .... 66 APPENDIX A 2006 CRP FRESHMAN SURVEY ................................ ................................ .............. 67 B CIRP FRESHMAN SURVEY DATA FILE ................................ ................................ 71 C CIRP FRE S HMAN SURVEY (CODEBOOK) ................................ ............................. 84 D ACT SAT CONCORDANCE: A TOOL FOR COMPARING SCORES .............. 98 LIST OF REFERENCES ................................ ................................ ................................ 100 BIOGRAPHICAL SKETCH ................................ ................................ .............................. 109

PAGE 7

7 LIST OF TABLES Table page 4 1 Odds ratio for race by gender controlling for SAT ................................ .................... 51 4 2 Odds ratio for race by gender controlling for ACT ................................ ................... 54 4 3 Odds ratios for highest parental attainment by household income, controlled for SAT ................................ ................................ ................................ ............................. 58 4 4 Odds ratios for highest parental attainment by household income, controlled for ACT ................................ ................................ ................................ ............................. 61

PAGE 8

8 LIST OF FIGURES Figure page 1 1 Model of College Choice (Hossler & Gallagher, 1987) ................................ ......... 27 4 1 Distribution of College Acceptance Outcomes ................................ ...................... 37 4 2 Distribution of Race ................................ ................................ ................................ 38 4 3 Distribution of Gender ................................ ................................ ............................ 38 4 4 Distribution of Household Income ................................ ................................ .......... 39 4 5 Distribution of Highest Parental Educational Attainment ................................ ...... 40 4 6 Distribution of Age (In Years) ................................ ................................ ................. 40 4 7 Distribution of SAT Composite Scores ................................ ................................ .. 43 4 8 Distribution of ACT Composite Scores ................................ ................................ .. 43 4 9 Distribution of SAT composite scores by acceptance outcome ............................ 44 4 10 Distribution of ACT composite scores by acceptance outcome ............................ 44 4 11 Odds Ratios for SAT by Acceptance Outcome ................................ ..................... 47 4 12 Odds Ratios for ACT by Acceptance Outco me ................................ ..................... 48

PAGE 9

9 LIST OF ABBREVIATIONS Catholic 4yr Col. L Catholic 4 year College Low Selectivity Catholic 4yr Col. M Catholic 4 year College Medium Selectivity Catholic 4yr Col. H Catholic 4 year College High Selectivity CIRP Cooperative Institutional Research Program ETS Educational Testing Service FATHEDUC FTIC First Time In College HERI Higher Education Research Institute at UCLA HSGPA High School Grade Point Average INCOME Household Income MOTHEDUC Other Relig. 4yr Col. VL Other Religious 4 year College Very Low Selectivity Other Relig 4 yr Col. L Other Religious 4 year College Low Selectivity Other Relig 4 yr Col. M Other Religious 4 year College Medium Selectivity Other Relig. 4yr Col. H Other Religious 4 year College High Selectivity Private Uni. L Private University Low Selectivity Private Uni. M Private University Medium Selectivity Private Uni. H Private University High Selectivity Private Uni. VH Private University Very High Selectivity Public 4yr Col. L Public 4 year College Low Selectivity Public 4yr Col. M Public 4 year Colleg e Medium Selectivity Public 4yr Col. H Public 4 year College High Selectivity Public Uni. L Public University Low Selectivity Public Uni. M Public University Medium Selectivity

PAGE 10

10 Public Uni. H Public University High Selectivity RACEGROUP R ace TFS The Freshman Survey TOP Test Optional Admissions Policy UCLA University of California, Los Angeles

PAGE 11

11 DEFINITION OF KEY TERMS ACT Formerly known as the American College Testing Program; one of two main standardized examinations utilized in th e postsecondary admissions process College The level of preparation a student needs to enroll and succeed Readiness at an institution of higher learning (Geiser, 2009) High Achieving Respondents on the HERI CIRP Freshman Survey who Students self (CIRP) SAT Formerly known as the Scholastic Assessment Test; one of two main standardized examinations utilized in the postsecondary admissions process Socioeconomic The social standing or class of an individual or group as Status (SES) measured by a combination of education, income and occupation (American Psychological Association) Standardized Standardized testing used solely to make admissions College Admissions determinations for undergraduate admissions applications Testing (Zwick, 2007) Test Optional An institutional admissions policy in which applicants can elect not to provide their standardized college admissions test score

PAGE 12

12 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Education THE IMPACT OF STANDARDIZED TESTING ON HIGH ACHIEVING STUDENTS' POSTSECONDARY ACCEPTANCE OUTCOMES By Leena Hasbini August 2018 Chair: Linda B. Eldridge Major: Educational Leadership Standardized testing, specifically geared toward postsecondary admission, such as the SAT and ACT, was designed to measure readiness for higher education. The purpose o f this study was to examine the degree to which ACT or SAT scores impact college acceptance outcomes for high achieving students with similar high school GPA's. This research study also compared individual and familial variables and the parallel impact they have on postsec ondary acceptance. Significant differences in acceptance outcomes were observed among the high achieving student sample. The sample, largely homogenous with respect to high school grade point average (HSGPA) experienced notable differences in acceptance ou tcomes based on test scores (SAT and ACT) H igher scores were associated with acceptance to more selective institutions. Additionally, significant differences were observed in acceptance outcomes due to the demographic characteristics of gender race, and parental demographics (parental educational attainment and household income).

PAGE 13

13 CHAPTER 1 INTRODUCTION Overview of the Topic Attending an institution o f higher education provides the most substantial conduit for econ omic and social mobility in the United States (Paulson & St. John, 2002; Thompson, 2017). The correlation between the attainment of a college degree, and earnings realized over a lifespan is indisputable (Baum, Ma, & Payea, 2013; Thompson, 2017). The type and selectivity of an institution also critically impacts intergenerational mobility (Thompson, 2017). Research consistently demonstrates that attendance in a higher quality postsecondary institution is positively associated with an increased four year co llege graduation rate (Melguizo, 2008). Where first time in college (FTIC) freshmen matriculate can innumerably impact the future trajectory o f their career and life decisions (Melguizo, 2008). To that end, the postsecondary institution to which a student applies and i s admitted, and the factors that encompass the admissions decision making process are o f tremendous significance. Statement of the Problem This study aims to determine the extent to which standardized college admissions testing, namely the SAT and ACT, impact college acceptance outcomes for students that are deemed high achieving; furthermore, this study examines the impact that race, gender and parental demographics have on postsecondary acceptance outcomes when controlling for standardized college admissions testing. There is a significant benefit linked to a better educated populace, and as such, issues o f unequal access to higher educati on are o f great consequence (Baum, Ma, & Payea, 2013). Research has consistently demonstrated that the achievement of higher

PAGE 14

14 education levels leads to decreased rates of joblessness and limits the impact of poverty (Paulson & St. John, 2002; Baum, Ma & P ay ea, 2013; Thompson, 2017). College graduates are more likely to be gainfully employed, and to pay taxes They are less likely to be unemployed, incarcerated, or dependent on social safety net programs. Higher education is also correlated with increased lev els of civic engagement (Baum, Ma & Payea, 2013). Recognizing the implications, an examination is made within the current study o f determining factors related to who will or will not be provided an opportunity to attain and share in the advantages of post secondary education has serious and enduring repercussions. Research Questions In reviewing current research literature, the impact o f standardized test results on students postsecondary acceptance outcomes provided the foundational background for the r esearch questions formulated. The practitioner in this study focused uniquely on the following research questions: 1. T o what degree do standardized college admissions test scores impact college acceptance outcomes for high achieving students? 2. Controlling f or test scores, do race and gender impact college acceptance outcomes for high achieving students? 3. Controlling for test scores, do parental demographics (household income and highest parental educational attainment) impact college acceptance outcomes for high achieving students? Limitations and Assumptions Limitations This study examined the relationship between high achieving students' postsecondary outcomes and their postsecondary admissions standardized test scores A s such, the following limitations should be taken into consideration when reviewing the

PAGE 15

15 results o f this study. The postsecondary admissions outcome process is inherently complex, and consists of a significant number of longitudinal, and interactional aspects that can be difficult to measu re (Hossler & Gallagher, 1987). This study does not examine these factors and their impact on the postsecondary admissions outcome. Furthermore, although the normative sample was large, and the calculated standard error was small, there are limitations in regard to the validity and reliability o f the CIRP Freshman Survey. Additionally, a s the data utilized was obtained from the CIRP study, the use o f secondary data precludes the ability to determine whether respondents' postsecondary outcomes were singular ly based on standardized test scores, or o ther factors not explored in this study, such as athletics, romantic relationships, peer influence, family connect i ons or other non cogn i tive factors. Assumptions The most significant assumption that underlies th is study is the utilization of secondary data. Practitioners and researchers should be cognizant of the intended utilization o f instruments, specifically, the method in which they were designed to be used. For the purposes of this study, the practitioner m ade the assumption that the respondents' understanding of the questions utilized in the survey instrument mirrored that o f the practitioner. Finally, the responses on the survey instrument were self reported T herefore the practitioner makes assumption s that the respondents were honest and truthful in their answers. Significance of the Study While there is abundant research literature regarding the SAT and ACT as variables in the college admissions process, to date no comprehensive study has been publi shed contrasting high achiev i ng students' postsecondary acceptance outcomes

PAGE 16

16 based on standardized admissions test scores. A thorough examination o f high achieving students' test scores that demonstrates a correlation between test scores and postsecondary acceptance outcomes would yield implications of great significance to educators, policy makers, enrollment managers and college administrators seeking to understand the specific postsecondary acceptance outcomes of high performing students. This correlatio n is meaningful as more postsecondary institutions move in the direction of employing test optional admissions policies. Additionally, educators and leaders are able to gain additional insight into the necessary preparation and planning needed to help stud ents who have discrepancies between their HSGPA and test scores P articularly as admission to more selective institutions become more competitive, a greater awareness will enable education leaders to help students better meet their postsecondary goals and attain the outcomes they aspire to. Summary The purpose o f this study is to determine the overarching impact o f standardized college admissions testing on the postsecondary acceptances of high achieving students T his research study also examines the eff ect of race, gender and parental demographics (parental educational level and parental income level) on the postsecondary acceptance outcomes of the same population.

PAGE 17

17 CHAPTER 2 REVIEW OF THE LITERATURE Origins o f Postsecondary Admissions Criteria Cab rera and Burkum (2001) examined the evolution o f postsecondary admissions in the United States, delineating the time periods into four distinct eras: subjectivity (1600's 1800's), uniformity (late 1800's to early 1900's), objectivity (1900's to 1960's), an d holistic approach (1970's to present). Era o f Subjectivity. From the founding o f Harvard College in 1636 until the early 19th century, admissions criteria were primarily determined by social background, moral character, religious denomination, and profi ciency in Greek and Latin. College presidents conducted admissions interviews, with applicants o ften traveling for weeks at a time to take institution specific entrance exams (Cabrera & Burkum, 2001; Crouse & Trushe i m, 1988). Ultimately, the ability to pay tuition was a precluding criterion for most applicants (Cabrera & Burkum, 2001). As the number of colleges in the United States grew in scope, candidate interviews were delegated to faculty members, leading to increasing discrepancy in admissions standard s (Cabrera & Burkum, 2001 ; Furuta, 2017). Era o f Uniformity. In 1893, a number o f private universities voted to enact common admissions requirements and recommended a core high school curriculum to better prepare applicants for admission (Cabrera & Burkum, 2001; Atkinson & Geiser, 2009). In 1900, the College Entrance Examination Board (known as The College Board) was formed and the first standardized test, in the form o f subject specific essay questions was administered in 1901 (Cabrera & Burkum, 20 01; Atkinson & Geiser, 2009; Gambino, 2013).

PAGE 18

18 Era o f Objectivity. As criteria for admissions became progressively uniform and objective, the interpretation of essays remained largely subjective. To that end, the College Board sought an alternative that was initially based off the works of Binet and Simon (Cabrera & Burk um, 2001; Crouse & Trusheim, 1988; Jencks & Crouse, 1982). When initially administered in 1926, the Scholastic Aptitude Test (SAT) was touted by supporters as a gauge of intelligence, an apparatus of social and educational opportunity and a measure o f fut ure potential (Cabrera & Burkum, 2001; Lemann, 1995). During a period of time when admission to institutions of higher education was preeminently defined by status, the SAT was designed to differentiate between scholastic capability rather than fortune and birth, in order to classify gifted students who would have been otherwise overlooked (Lemann, 1995). A new system of meritocracy materialized with the advent of the SAT, which allowed for the classification of students on the "basis of academic potential rather than social status" (Jencks & Crouse, 1982; Alon & Tienda, 2007). In 1937, IBM developed technology that could grade SAT score cards by using electric currents to detect pencil marks, enabling the efficient, mass administration of the test (Cabrera & Burkum, 2001). In the 1940's, the University of Calif system elected to require the SAT, solidifying its role as a critical admissions criterion (Cabrera & Burkum, 2001). Over the course o f the following 40 years, the SAT became a mainstay at high er education institutions throughout the United States; by the late 1960s, the annual administration of the exam was given to "more than half a million high school students" (Jencks & Crouse, 1982). The American College Testing Program, now known as the AC T, was created in 1959 W hile similar to the SAT in offering an objective measurement o f verbal

PAGE 19

19 and mathematical skills, its emphasis focuses on college related knowledge and skills (Cabrera & Burkum, 2001). Era o f Holistic Approach. A s a cost effective m echanism through which aptitude could be assessed, standardized tests were viewed as the panacea to college admissions, particularly after World War Il, when college enrollment soared (Alan & Tienda, 2007; Cabrera & Burkum, 2001 ; Crouse & Trusheim, 1988; Crouse & Jencks, 1982; Lemann, 1995). However, standardized testing has fallen under strident criticism in recent years, as detractors have questioned their validity as a benchmark of college success (Cabrera & Burkum, 2001; Geiser & Santelices, 2007; Men drinos, 2014). Critics argue that standardized postsecondary admissions tests have an inconsistent impact on racial minorities and women (Cabrera & Burkum, 2001; Geiser & Santelices, 2007; Mendrinos, 2014; Sanchez & Lin, 2017). Current literature suggests that HSGPA is a better predictor o f college graduation rates much more so than ACT or SAT test scores (Atkinson & Geiser, 2009; Belasco, Rosinger, & Hearn, 2015; Geiser & Santelices, 2007; HSS & Franks 2014). Edmunds (2010) found that students with greate r high school GPA's but lower standardized test scores were as likely to persist to sophomore year as students with consistent achievement. Geiser and Santelices (2007) found that a significant amount of the SAT's extrapolative power originates from a dire ct relationship with students' high school demographics and socioeconomic background. Current Admissions Practices Currently within the college admissions process, higher education institutions employ a number of academic factors, which include high schoo l GPA, standardized college admissions test scores, and class rank, in combination with other non academic

PAGE 20

20 factors, such as extracurricular involvement, first generation status, recommendation letters, and personal essays. As the relationship between previ ous and future academic performance is essentially linked, most higher education institutions use high school GPA as a critical determinant within the admissions process. In combination with standardized test scores, which are presently utilized these sco res act as a critical decision making factor in admitting a student to a particular college or university (Geiser & Santelices, 2007; Soares, 2012). High School Grades in the Admissions Process Numerous research studies indicate that the most robust and meaningful predictor of college grades are high school grades (Geiser, 2009). Specifically, high school GPA is significantly related to four year graduation rates, freshman GPA, and overall GPA (G eiser & Santelices, 2007; Geiser, 2008). As high school and college grades measure similar outcomes (Geiser & Santelices, 2007), it is of natural consequence that this becomes a criterion in the postsecondary admissions process. Furthermore, utilizing high school grades to make admissions decisions yields "less adverse impact than standardized tests on disadvantaged and underrepresented minority students" (Geiser & Santelices, 2007, p. 1). Edmunds (2010) found high school GPA to be a clearer predictor of bo th persistence during the sophomore year and four year college graduation rates than standardized test scores. Test Scores in the Admissions Process Employing standardized test scores is widely prevalent in college admissions decisions. "One reason why i nstitutions consider using standardized test scores in making admissions decisions is to increase the proportion of their enrolled students who are academically successful" (Sawyer, 2007, p. 260). The utilization o f standardized

PAGE 21

21 testing provides a common m etric that can be applied across numerous high school backgrounds and learning environments. The uptick in the average number of college applications submitted per student demographics the highlighted emphasis on standardized testing as a criterion in the admissions decision making process (Clinedinst & Koranteng, 2017; Sawyer, 2007; Zwick, 2007). Increased application volume creates an additional filter on w hich college admissions officers can rely, particularly at more selective institutions (Clinedinst & Koranteng, 2017). Soares (2012) found that employing admissions test scores to determine admissions selection stratified postsecondary education admissions into a tiered system, creating unintended social disparity. Colleges with an increased test score t hreshold were more likely to attract and retain students (and families) o f a higher socioeconomic status (Soares, 2012). Thus, the more selective a college became, the greater the concentration o f high Socio E conomic Status ( SES ) students within that colle ge's student body (Soares, 2012). Standardized College Admissions Tests The utility of standardized testing for postsecondary admissions rests on the assumption that the tests are dependent upon what is actually measured, and subsequently interpreting an d applying those measurements (Kane, 2013). Most significantly, while the SAT and ACT both offer standard measurements of verbal and quantitative reasoning abilities aimed at determining college readiness, the content and structure differ considerably (Dic kenson & Adelson, 2016). As of 2015, SAT and ACT takers surpassed 1 .9 million (Belasco, Rosinger, & Hearn, 2015). A significant number of students attempted both exams, often taking them repeatedly to improve test scores

PAGE 22

22 (Lewin, 2013). Both tests are norm referenced, meaning they compare test takers to each o ther, rather than measuring for knowledge mastery (Geiser, 2009) SAT When initially introduced the SAT was an abbreviation of "Scholastic Aptitude Test F or a significant period, the SAT's emphasis on aptitude enabled college admissions officers to adopt the mindset that results were unalterable predictors o f ability. This became problematic, as some test takers who performed exceptionally well in high school began to receive sub standard scores on t he SAT (Atkinson & Geiser, 2009). In 1991, "aptitude" was replaced by "assessment with ETS emphatically stating that the SAT was actually a measure of abilities acquired over an extended period of time (Crouse & Trusheim, 1988). With the retooling toward s achievement, the utility of the SAT could be fashioned to determine the accumulated mastery of knowledge, identifying prodigies in lower performing high schools or those w hose full potential was not demonstrated in a traditional environment (Jencks & Phi llips, 1998). In 1995, ETS relinquished the idea that the SAT measured either achievement or aptitude, and renamed the test SAT Reasoning They then claimed that it "developed reasoning" (Jencks & Phillips, 1998). Students taking the SAT today receive scor es in reading and math. ACT The ACT, first administered in 1959, measures general educational achievement (ACT Inc., 2017). It is comprised of four tests: English, mathematics, reading and science. The composite score o f the ACT is a mathematical average o f all four subsections. The ACT also established benchmark scoring that is designed to establish specific sub scores that indicate college readiness in each subject tested. Lichtenberger

PAGE 23

23 and Dietrich (2012) examined the linkage between ACT benchmarks and postsecondary outcomes. The results indicated that the more ACT benchmarks that were met, the higher the rate of acceptance at selective colleges They also found an increased rate of college graduati on. Additional studies demonstrated that ACT scores were highly correlated to college freshman GPA, and related to notable creative accomplishments a number o f years after the examination (Marsh, Vandehey, & Diekhoff, 2008; Dollinger, 2011). Concordance b etween SAT and ACT While each admissions test is on a different end of the aptitude achievement spectrum, both admissions tests share overlapping constructs Concordance tables can be utilized to examine the relationship between both tests, and also to ev aluate concurrent or convergent validity. T est scores are considered concordant at a similar percentile rank for a test taker who took both tests. ACT provides a concordance table (see Appendix D) to conduct comparisons between scores on both tests. As an example, a student obtaining a score of 32 on the ACT, would utilize the concordance table to obtain a predicted range of SAT scores. In this case, the SAT score range for an ACT score of 32 is 1400 1430 (ACT, 2014). Predictive Validity Research studies on predictive validity, or the extent to which measures of current performance impact future performance (Kane, 2013), have identified a correlation between students' admission test performance and their subsequent performance and achievement in the postse condary environment (Bett i nger, Evans & Pope, 2013; Buckley, J, Letukas, & Wildavsky, 2018; Mattern et al., 2011; Patterson & Mattern, 2012). The predictive validity o f HSGPA and admissions test scores are

PAGE 24

24 positively correlated with a number of college out comes including first year college GPA, sophomore year persistence, and graduation from college (Allen, 2013; Kobrin et al., 2008; Mattern et al., 2011; Patterson & Mattern, 2012). Furthermore, existing research literature found that the relationship betw een increased test scores and increased HSGPA with successful postsecondary outcomes is consistent across race, gender, ethnicity, and socioeconomic status (Patterson & Mattern, 2011; Shaw, Marini & Mattern, 2013; Zwick & Sklar, 2005). Rubin (2014) found that course rigor in high school was more significant in determining academic merit and ability rather than standardized test scores or HSGPA. Atkinson & Geiser (2012) stated that : irrespective of the quality or type o f school attended, cumulative grade point average (GPA) in academic subjects in high school has proved to be the best overall predictor of student performance in college. This finding has been confirmed in the great majority of 'predictive validity' studies conducted over the years, includin g studies conducted by the testing agencies themselves (p. 24). Test Optional Admissions Policies The majority of postsecondary institutions have historically utilized standardized test scores to make admissions decisions But an increasing number o f sel ective postsecondary institutions have shifted towards allowing applicants to elect not to submit standardized admissions test scores as part of the application process This is commonly known as test optional policies (TOP's) (Espanshade & Chung, 2012). The test optional movement took root at private, liberal arts colleges that found it an appealing method to grow and diversify their applicant pool (Mendrinos, 2015). As of 2018, there were over 1 ,000 postsecondary institutions that deemphasized the utiliz ation o f standardized admissions test scores in admissions practices, with over 100 joining this group in the past four years alone (Syverson, Hiss & Franks, 2018; Fair Test

PAGE 25

25 List, 2018). Syverson, Hiss, & Franks (2018) conducted an extensive, longitudinal study that examined data from 28 colleges and universities and 955,774 applicants. The researchers compared four year colleges that utilized TOP's with their mandatory testing counterparts. The results demonstrated that standardized admissions tests failed to fully identify high achieving applicants. Furthermore the study yielded notable results, specifically that : a) more than a quarter of all applicants applying to test optional colleges elected not to submit their test scores ; b) institutions with TOP's had gains in the number o f underrepresented and m i nority applicants ultimately enrolling ; c) underrepresented, minority students did not submit scores in greater numbers ; d) While non submitters (the term for those who did not submit scores) were accepted at marginally lower rates, those that were admitted were much more likely to enroll ; e) non submitters graduated at equivalent or slightly higher rates than those who submitted test scores. The momentum shift towards TOP's evolved as a result of overlapping changes, particularly, research and interpretation of academic outcomes and widening scope of testing as a more limited determinant of academic potential (Syverson, Hiss, & Franks, 2018). Across the postsecondary spectrum, TOPs have proven to be beneficial and constructive tools that support enrollment practices (Syverson, Hiss & Franks, 2018) Discrepant Performance A review o f the literature indicated that there are scenarios in which HSGPA and test score performance combinations a re inconsistent. This phenomena is generally defined as "HSGPA discrepant where a student's HSGPA is significantly higher than their performance on standardized admissions testing or "test score discrepant" in which their test scores are sign i ficantly h igher than their HSGPA performance (Buckley

PAGE 26

26 et al., 2018; Mattern et al., 2011, Patterson & Mattern, 2012; Shaw, Marini & Mattern, 2013). Research on discrepant performance generally categorizes the discrepancy by standard deviation. Students with standar dized admissions test scores and HSGPA that fall within one standard deviation are labeled as having consistent academic achievement students with standardized admissions test scores greater than one standard deviation of their HSGPA are labeled as "test score discrepant and students with HSGPA greater than one standard deviation of their standardized admissions test score are labeled as "HSGPA discrepant" (Buckley et al., 2018; Edmunds, 2010; Geiser & Santlices, 2007; Patterson & Mattern, 2012). HSGPA d iscrepant students were more likely to be female, minority, and have low socioeconomic status (Buckley et al., 2018, Holland, 2014; Kobrin & Patterson, 2011; Patterson & Mattern, 2012; Sanchez 2010). Sanchez and Lin (2017) found a direct correlation betwe en these variables, identifying 66% o f students with lower ACT composite scores and higher high school GPA's to be female, and greater than 50% to be minority and have low socioeconomic status. The researchers also found that less than 9% of minority stude nts tended to have consistently high HSGPA and high test score correlation (Sanchez and Lin, 2017). Hiss and Franks (2014) reviewed 123,000 student records from 33 public and private postsecondary institutions that chose to go test optional They found th at students' college GPA s varied less than .05, and that there was only a variance o f 6% in college graduation rate s between students who initially submitted admissions test scores and those who did not. They also found that students who scored lower on a dmissions tests but had a stronger HSGPA, were more likely to achieve in college as

PAGE 27

27 opposed to those who had higher standardized admissions test scores and a mediocre HSGPA (Hiss & Franks, 2014). The researchers also noted that elevating HSGPA over standar dized testing was o f greater benefit to a significant number o f minority, first generation, and low SES students (Hiss & Franks, 2014) The researchers concluded that students with solid HSGPA's, even without standardized admissions test scores, were more likely to succeed in college as compared to students with lower HSGPA's and higher admissions test scores F urthermore students in the latter group were more likely to have a lower college GPA and less likely to ultimately graduate (Hiss & Franks, 2014). Syverson, Hiss and Franks (2018) conducted a meta analysis o f discrepant performance related research that revealed that across the studies examined, between 11% and 18% of the sample population demonstrated discrepant performance. Model of College Choice Significant research has examined college decision making models over the past three decades (Hossler et al., 1998; Kinzie et al., 2004; Paulsen, 1990). Several frameworks have been used to describe the process of college choice. Hossler and Gallag her (1987) developed a three stage model which focuses on three conceptual phases: predisposition, search and choice (Hossler and Gallagher, 1987). Figure 1 1. Model of College Choice (Hossler & Gallagher, 1987) The predisposition stage, generally occu r s between 7th and 9th grade, and is characterized by the decision to pursue college attendance rather than alternative educational attainment paths such as technical or trade school, military service, or direct

PAGE 28

28 entry into the workforce (Hossler & Gallaghe r, 1987; Hossler & Stage, 1992). An intensified focus on academic preparation and coursework necessary for postsecondary attendance becomes important at this stage, in addition to the development of college aspirations (Hossler & Gallagher, 1987). The sea rch stage occurs in 10th 12th grade of high school enrollment (Hossler & Gallagher, 1987). This phase is characterized by an increased concentration on finding information and literature related to colleges, delineating criteria of interest, and narrowing choices. This time period of gathering information plays a critical role in the college choice process as students build a list o f colleges based on specific benchmarks (Hossler & Gallagher, 1987). The choice stage culminates in the college selection proce ss, and occurs in the 12th grade; at this phase, the student has applied to a number of institutions and is preparing to make a decision regarding which postsecondary institution to attend (Hossler & Gallagher, 1987). College Choices Of High Achieving Stu dents After identifying the appropriate framework for understanding the college choice process that students experience as they make postsecondary enrollment decisions, factors that influence this pivotal life decis i on become paramount. DesJardins et al. (2006) provided appropriate terminology for a wide range of smaller influential factors identified throughout the college choice literature that falls within the larger categories In addition, McDonough (1997) identified an independently influential conce pt, bounded rationality, that explained the relationship between types of influential characteristics (DesJardins et al. 2006, p. 383; Hossler et al., 1998; Kinzie et al., 2004; McDonough, 1994; Paulsen & St. John, 2002). Student characteristics directly impact a student's perspective and experience. The literature has shown that gender, race ethnicity,

PAGE 29

29 parental education, and socioeconomic status notably influence postsecondary acceptance outcomes of enrolling freshmen (Geiser & Santelices, 2007; Zwick & Himelfarb, 2011). Socioeconomic Status Within the study o f student characteristics in college choice literature, socioeconomic status is the most commonly researched factor. The literature recognizes the sign i ficant, and often complex, ways that socioec onomic status affects the college choice process This research recognizes that students from higher socioeconomic backgrounds tend to apply to and choose from more selective universities (Goenner & Snaith, 2004; Rubin 2014) Additionally, socioeconomic stat us has a greater impact on the college choice process o f students by coloring their lens o f opportunity with a financial shade, meaning that students will consider options within the framework o f affordability and cost of attendance (Paulsen & St. John, 20 02; Rub e n, 2014). Soares (2012) found significant correlation between family income and average standardized test score s Students from the lowest SES quartile scored 100 points lower than students whose annual income was in the median range of income They also found that students w hose family income was in the highest 5% tended to score more than 200 points higher than those in the lowest SES sub group. In another study, Geiser and Santelices refuted claims that HSGPA was more likely to be correlated with SES status. The researchers found that : T he SAT V correlated at the .32 level with family income, and at the .39 level with parents' education; similarly, SAT M scores correlated respectively at .24 and .32, but HSGPA correlated with family income a t the .04 level, and with parents' education at the .06 level (p. 2).

PAGE 30

30 Hoxby and Avery (2012) examined a sample of 35,000 high achieving students in the lowest SES (defined by students with standardized admissions test scores in the highest 10% and pare ntal SES in the lowest quartile), and found that these students are less likely to apply to the most selective institutions. This phenomena, commonly referred to as "undermatc hin g occurred primarily because applicants did not have an awareness of the adm issions process, did not know how to utilize application fee waivers, applied to a limited number of colleges, or did not have awareness of the financial aid process (Hoxby & Avery, 2012; Rubin, 2014). Parental Educational Attainment Similar to socioecon omic status, parental education plays a complex role in the admissions decision making process of students (Cabrera & LaNasa, 2000). The postsecondary parental educational experiences of parents impact ed students directly in their understanding o f costs an d financing an education as well as indirectly in their ability and support to navigate steps to college admission (McDonough, 1994). Race/ethnicity, parental education, and household income correlate strongly with standardized admissions testing scores. As a result, test scores accentuate socioeconomic and racial disparities among applicants more so than o ther selection criteria. The hard and soft cost of testing, particularly for minority and low income applicants, is disproportionate with any marginal b enefit it provides as an indicator of how students will perform. College Selectivity Students with higher academic ability are more likely to enroll in more selective colleges and universities (Brand & Halaby, 2006; Rubin, 2014). Furthermore, Hispanic and African American students who attend extremely selective colleges are more likely

PAGE 31

31 to attain a degree (Bills, 2003). The perception and reality that a superior quality education enables graduates of more selective institutions to obtain more economical ly desirable skills and abilities (Rubin, 2014). Second, notwithstanding of ability or aptitude, an institution's "brand name" may attract potential employers seeking highly capable graduates (Rubin, 2014). Institutions ranking higher in selectivity can p rovide access to alumni networks, adding significant benefit in social capital through employment opportunities, mentoring, and inside information. Rubin (2014) also found that more selective institutions have a greater inclination to give preference in co llege admissions towards high SES students. Students not in need of financial assistance were grouped and categorized into separate "pools" and had a greater likelihood of being admitted (Rubin, 2014, p. 2). DesJardins, Ahlburg, and McCall (2006) examined institutional admissions practices across the U.S. and concluded that selective colleges base their admissions decision making on institutional goals, diversity level and academic congruency. They found that test score thresholds and individual reviews va ried across college type and student type (DesJardins et al., 2006). CIRP Freshman Survey The Cooperative Institute Research Program's Freshman Survey, governed by the Higher Education Research Institute at UCLA, is administered annually to over 400,000 entering freshmen at more than 700 two and four year universities and colleges (Keup, 2004). First published in 1966, the CIRP Freshmen Survey is one o f the oldest surveys still continually administered today to college students. The representative and com prehensive nature of the battery has led to extensive use by researchers and practitioners wanting to gauge various issues confronted by incoming college freshmen (Keup, 2004).

PAGE 32

32 CHAPTER 3 METHODOLOGY A summary o f the methodology utilized is presented in this chapter. The research design, data source, sample, data collection, and data analysis methods are all described in detail. Research Design Through the utilization of a correlational research design, this pa rticular research design will apply multinomial logistic regression to determine whether substantial discrepancies exist in postsecondary acceptance outcomes between high achieving students with lower test scores on the ACT and SAT and high achieving stude nts with higher test scores on the ACT and SAT. The independent variables are based on (1) students' cumulative high school GPA (HSGPA) (2) students' individual characteristics, including race/ethnicity, gender, and (3) parental demographics including par ents' educational attainment, and family income. Utilizing this particular design enables the investigation of potential roots of the phenomenon being researched, evaluating individuals for whom a trait is ex i stent compared to similar individuals for whom the trait is existent to a lesser extent or absent entirely (Creswell, 2009). The research data used originates from the Cooperative Institutional Research Program (CIRP) 2006 Freshman Survey, managed by the Higher Education Research Institute (HERI) at t he University of California Los Angeles (Eagan, Lozano, Hurtado, & Case, 2016). Annually administered, the CIRP is given at approximately 800 institutions of higher learning to roughly 400,000 freshmen. The Freshman Survey consists of items that relate to college choices, future goals, and demographic information (Including but not limited to parental demographics race, and religious preferences). The CIRP also

PAGE 33

33 collects information regarding established behaviors in high school, educational preparedness, c ollege expectations, factors in admissions, faculty and peer interaction, student goals and values, demographic characteristics of students, and financial concerns associated with postsecondary education (Eagan, Lozano, Hurtado, & C ase 2016). Data Source Data from The Freshman Survey (TFS) were collected from the Higher Education Research Institute at UCLA (HERI) data archives. The practitioner requested and was granted access to the database for the purpose o f conducting this study. The TFS data files a vailable for download are relatively massive (SPSS files), consisting of all survey data from 1966 to 2006. As such, a great deal of computational memory and power was devoted to the creation of subsets. The most recently available data (2006) was used for the current study. Data Collection As the study focused on h i gh ach i ev i ng students, only those with a self reported high school GPA o f A or were retained. Further, all survey responses collected from 2 year colleges (public or private) were omitted du e to the relatively low representation of such institutions in the data. Thus, the final sample consisted of 103,267 unique survey responses. Due to item non response, however, not all statistical processes utilized all cases Data Analysis All data filtering was conducted using R3.3.1. The SPSS files obtained from the HERI data archives were read into R using the functions available as part of the foreign package. Once the appropriate cases and variables were determined, a final CSV file

PAGE 34

34 was written. The file, which consisted of 103,267 cases and 10 variables (SUBJID, COMPGROUPI, AGEI, RACEGROUP, GENDER ACTCOMP, SATCOMP, INCOME, MOTHEDUC, and FATHEDUC), was then imported into SAS 9.4, for further statistical analysis. Acceptance outcomes were measur ed using COMPGROUP 1 which is a classification of institution type and institution selectivity Institution type consisted o f public and private universities, as well as public, private/nonsectarian, Catholic, and other religious 4 year colleges. Selectivi ty consisted of very low, low, medium, high, and very high. Some institution/selectivity pairs (such as very low selectivity public universities and very high selectivity C atholic 4 year colleges) were not represented in the data. Test scores were measure d using SAT and ACT composite scores. At the time of the 2006 data collection, SAT scores ranged from 400 to 1600 in increments of 10, while ACT scores ranged from 1 to 36 in increments of unity. Any test scores that did not adhere to these scoring constra ints were omitted (at the procedure level). The survey item corresponding to test scores was were your scores on the SAT and/or ACT?" RACEGROUP and GENDER were used to measure the demographic constructs o f race and gender The race categories were Wh ite Black, Hispanic Asian, and "Other." Due to their low representation among high achieving students, American Indians and those reporting "Two or more race/ ethnicity were included in the "Other" category. The gender categories were Male and Female.

PAGE 35

35 Parental demographics were measured using INCOME, MOTHEDUC, and FATHEDUC, which are measures o f household income and parental educational attainment (for the mother and father, respectively). Due to the high degree o f (ordinal) correlation (Y = 0.61 r = 0.59) between the respective parental educational attainment measures, a variable representing highest parental educational attainment was created by taking the maximum of both parents' educational attainment This created variable appears as PARENT.EDU in the tables and figures throughout this study. The survey items corresponding to household income and parental educational attainment were W hat is your best estimate of your parents' total income last year? Consider income from all sources before taxes and "What is the highest level o f formal education obtained by your parents?" Age was measured using a recode of the AGEI categorical variable. The total number o f categories was reduced (due to low cell counts) from ten to four: "less than or equal to 17," "18," "19," and "greater than or equal to 20." The survey item corresponding to age read, How old will you be on December 31 of this year?"

PAGE 36

36 CHAPTER 4 RESUL TS Summary The purpose of this study was to determine to what degree standardized college admissions test scores impact ed college acceptance outcomes for high achieving students, and the impact that race, gender and parental demographics have on college acceptance outcomes. A review of college admissions practices, standardized college admissions testing, college choice model theory, and college selectivity literature provided context for this study. Data from the UCLA Higher Education Research In stitute's CIRP Freshman Survey were analyzed and the methodology and design was addressed in Chapter 3. The specific questions examined included: 1. T o what degree do standardized college admissions test scores relate to college acceptance outcomes for high achieving students? 2. Controlling for test scores, do race and gender relate to college acceptance outcomes for high achieving students? 3. Controlling for test scores, do parental demographics (household income and highest parental educational attainment) re late to college acceptance outcomes for high achieving students ? Findings A main focus of this study was to determine the quantitative connection between standardized college admissions test scores and postsecondary admissions outcomes for high achieving students. The findings in this study indicate a significant relationship Descriptive Statistics One Way Frequencies (Categorical and Ordinal Variables) From Figure 4 1, we see that a large number o f high achieving students chose to attend a highly sel ective public university (22.37%). The least common choice was "Other Religious 4 yr. Colleges very low," which accounted for only 0.52% of total

PAGE 37

37 choices. Overall, we see that within each institution type, acceptance outcomes are skewed left with respect to selectivity, indicating that the high achieving students preferred to attend more selective institutions. Figure 4 1. Distribution o f College Acceptance Outcomes

PAGE 38

38 From Figure 4 2, we see that most of the high achieving students were white (74%), with Asians, Blacks, and Hispanics accounting for only 9.76%, 3.52%, and 4.03% respectively. A total of 4,537 students did not provide a response to the race question. Additionally, in Figure 4 3 we see that 62.76% of the high achieving students were female, and that 131 students failed to provide a response to the gender question. Figure 4 2. Distribution o f Race Figure 4 3. Distribution of Gender In terms of household income as shown in Figure 4 4, a majority of students had a combined household income of $75k or more per year, with $100 $ 149.9k accounting for roughly 20% o f total responses. A significant number o f students (10,181) chose to leave the income question blank, e ither because they did not know their parents'

PAGE 39

39 combined annual income or chose not to provide the information. In regard to highest parental educational attainment as shown in Figure 4 5, over 70% of students had a parent who had earned at least a college degree, and 38.75% had a parent with a graduate degree. Those families in which neither parent graduated from high school accounted for only 2.4% o f total responses. Figure 4 4. Distribution o f Household Income

PAGE 40

40 Figure 4 5. Distribution of Highest Parental Educational Attainment From Figure 4 6, we see that most o f the high achieving students were either 18 or 19 years old, and that less than 5% of the sample fell outside of this range. Figure 4 6. Distribution o f Age (In Years)

PAGE 41

41 Two Way Frequen cies (Categorical and Ordinal Variables) In order to examine the pairwise relationships between the predictor variables and the response, a series o f two way tables were produced (along with their corresponding statistical tests of dependence and associat ion). Race and gender were significantly related to acceptance outcome (X 2 (76, n = 98730) = 8958.39, p 0.001 and X 2 (19, n = 103136) = 2061.16, p 0.001; respectively). Further, household income and highest parental educational attainment were significa ntly related to acceptance outcome ( X 2 (247, n = 93086) = 8045.05, p 0.001 and X 2 (133, n = 100839) 9762.94, p 0.001; respectively). Household income was significantly related to highest parental educational attainment ( X 2 (91, n = 91831) = 21269.91, p 0.001). The correlation between household income and highest parental education was moderate (Y = O.43, r = 0.42). Descriptive Analysis (Interval and Ratio Variables) As previously mentioned, test scores were measured using SAT and ACT compos i te scores. Almost sixty thousand students (n = 59,450) reported a SAT score, wh ile somewhat less (n = 46,753) reported an ACT score. Almost twenty thousand students (n = 19,921) reported scores for both tests. Due to issues inherent in making comparisons u sing concordance scoring between the two tests, each test was analyzed separately. The analysis indicated that SAT and ACT were highly correlated (r(n = 19921) = 0.78, 0.001). The average SAT score among the high achieving students (with scores that adher ed to the scoring constraints) was 1310.45, and the minimum and maximum scores were 450 and 1600, respectively. The average ACT score was 27.36, and the minimum and maximum scores were 1 and 36, respectively. While all reported ACT

PAGE 42

42 scores made sense in the context of the ACT scoring, 283 students provided SAT scores that did not adhere to the scoring constraints; the potentially inaccurate SAT scores were not completely divisible by 10. It is conceivable that such values were obtained by averaging the score s for multiple attempts. However, to mitigate potential issues in interpretation, those 283 SAT scores were omitted. The SAT and ACT distributions are presented in Figures 4 7 and 4 8, respectively. Both distributions were mostly normally distributed. Last ly, the distributions for SAT and ACT by acceptance outcome are presented in Figures 4 9 and 4 10, respectively. Both figures indicate that an increase in selectivity corresponds to an increase in average test score for a given institution type Inferentia l Statistical Analyses In cases involving the need to analyze the effect o f one or more predictors on a polychotomous response variable, multinomial logistic regression is a natural choice. W hile each student chose to attend one and only one institution, it cannot be determined whether students would remain in the same institution category if alternatives were added or subtracted from their choices of institution type/institution selectivity. Thus, the Independence of Irrelevant Alternatives assumption may not be entirely valid. For the sake of model interpretation, SAT scores were divided by 10. Such a division facilitates understanding, as SAT scores are in increments o f 10 points (i.e., the score immediately following a 1310 is a 1320, not a 1301). As previously mentioned, SAT and ACT scores were analyzed separately, to avoid any issues related to concordance score interpretation, high correlation between the tests, and the exclusion of stud ents who did not report scores for both tests. For all of the statistical tests below,

PAGE 43

43 Figure 4 7. Distribution of SAT Composite Scores Figure 4 8. Distribution of ACT Composite Scores

PAGE 44

44 Figure 4 9. Distribution of SAT composite scores by acceptan ce outcome Figure 4 10. Distribution of ACT composite scores by acceptance outcome

PAGE 45

45 a type I error rate of 0.05 was assumed (i.e., a 0.05). This value was chosen based solely on current scientific convention. The phrase "odds ratio" is used in lieu of the more general (yet more appropriate) phrase "relative risk ratio" in all o f the resulting interpretations below. The less general phrase has been chosen due to its overw helming presence (relative to the more general phrase) in the education literature (Creswell, 2009). Research Question 1 To what degree do test scores impact college acceptance outcomes for high achieving students? Two multinomial logistic regressions wer e conducted with acceptance outcome as the response variable and SAT and ACT as the predictors. The first MLR included SAT only, while the second included ACT only. Low selectivity public universities were used as the reference category for the response va riable. As such, all other response levels (19 in total) were compared in this reference category. RQ1 SAT The global test for non contribution of the predictor was significant (X 2 (19, n = 59167) = 13776.57, p 0.001), indicating that SAT score does a sig nificantly better job o f predicting acceptance outcomes than an intercept only model. Almost all relative log odds estimates were significantly different from zero, with the exception of medium selectivity C atholic 4 year colleges (p = 0.115) and medium se lectivity o ther religious 4 year colleges (p = 0.517). For these institutions, there was no significant increase or decrease in the likelihood of attendance (relative to low selectivity public universities) for each 10 point increase in SAT score. The odds ratios provided in Figure 4 11, indicate s that higher SAT scores were associated with a higher likelihood of attending a more selective college and a lower likelihood of attending a less selective college (relative to

PAGE 46

46 the reference d institution). The largest positive effect was observed for very high selectivity private universities, where for each additional 10 points on the SA T, a student was 16.2% more likely to attend a highly selective private university than a low selectivity private university. The smallest negative effect was observed for low selectivity public 4 year colleges (OR 0.958). RQ1 ACT The global test for non contribution of the predictor was significant (X 2 (19, n = 46753) = 8254.17, p 0.001), indicating that the ACT score does a significantly better job o f predicting acceptance outcomes than an intercept only model. All relative log odds estimates, with the exception of medium selectivity other religious 4 year colleges (p = 0.530), were significantly different from zero. The o dds ratios provided in Figure 4 11 indicate that higher ACT scores were associated with a higher likelihood of attending a more select ive college and a lower likelihood of attend ing a less selective college (relative to the referenced institution). As was the case for SAT scores, the largest positive effect was observed for very high selectivity private universities, indicating that for each additional point on the ACT, a student was 61% more likely to attend a highly selective private university than a low selectivity public university. The results indicate that test scores play a significant role in the acceptance outcomes o f high achi eving students. The ACT model displayed a better model fit than the SAT model (AIC = 237204.13 vs. AIC = 261364.08, respectively) as shown in Figure 4 12 Thus, the ACT did a better job of predicting acceptance outcomes among high achieving students than the S AT.

PAGE 47

47 Figure 4 11. Odds Ratios for SAT by Acceptance Outcome Research Question 2 Controlling for test scores, do race and gender impact college acceptance outcomes for high achieving students? Two multinomial logistic regressions were conducted with acceptance outcome as the response variable and race and gender as the predictors. The first MLR controlled for SAT, while the second controlled for ACT. Low selectivity public universities were use d as the reference category for the response variable, while white males were used as reference category for the predictors. Reference coding was used to create dummy variables for the predictors. This type of

PAGE 48

48 coding was chosen to allow for the comparison of w hites to each o ther race category, and females to males. Figure 4 12. Odds Ratios for ACT by Acceptance Outcome RQ2 SAT The global test for non contribution of the predictors was significant ( X 2 (190, n = 56800) = 16729.76, p 0.001), indicating that SAT score, race, and gender do a significantly better job of predicting acceptance outcomes compared to an intercept only model. A significant interaction effect between race and gender was observed and, therefore, included in the m odel (p 0.001). N o predictor by covariate interactions were included (i.e., the SAT by gender and SAT by race interactions were excluded from the

PAGE 49

49 model). All ma i n effects were significant (p 0.001). However, not all category effects were significant Due to the inclusion o f an interaction term, one must practice caution when interpreting the results of this section. To simplify the matter, the odds ratios for race are provided by gender in Table 4 1. From Table 4 1, we see that (compared to white fem ales) black females were over nine times more likely to attend a very high selectivity private university than a low selectivity public university, when controlling for SAT scores. O n the other hand, Black males (compared to white males), were over twenty one times more likely to attend a very high selectivity private university than a low selectivity public university, when controlling for SAT scores. The magnitude and sign (negative vs. positive) of the differences in o dds ratios between males and females throughout Table 4 1 highlights the interaction between race and gender in the current model. The largest difference in odds between males and females was observed for low selectivity private/non sectarian 4 year colleges (black versus white), where the o dds rat io for males (OR = 16.71) was over four times larger than that for females (OR = 3.88). The smallest negative effect observed indicates that Asian males (compared to white males) were 97.1% less likely to attend a low selectivity other religious 4 year college than a low selectivity public university, controlling for SAT scores. In terms of non significance, there were 22 (of the 76 possible) race category effects that were not significant for either gender Overall, race and gender significantly a ffect a high achieving student's acceptance outcomes when SAT scores are held constant (controlled for). The model including race, gender and SAT displayed a significantly better fit than the SAT only model (X 2 (171) 14991.28, p 0.001). Viewed alternativ ely to

PAGE 50

50 the current research question, the increase in model fit indicates that SAT does a significantly better job o f predicting acceptance outcomes when race and gender are controlled for.

PAGE 51

51 Table 4 1. Odds ratio for race by gender controlling for SAT

PAGE 52

52 Table 4 1. Continued RQ2 ACT The global test for non contribution of the predictors was significant ( X 2 (190, n = 44909) = 10968.46, p 0.001), indicating that the ACT score, race, and gender do a significantly better job of predicting acceptance outcome s compared to an intercept only model. A significant interaction effect between race and gender was observed and, therefore, was included in the model (p 0.001). N o predictor by covariate int eractions were included (i.e., the ACT by gender and ACT by race interactions were excluded from the model). All main effects were significant (p 0.001). However, not all category effects were significant. The odds ratios for race are provided by gender in Table 4 1

PAGE 53

53 From Table 4 2, we see that (compared to white females) Hispanic females were almost thirteen times more likely to attend a very high selectivity private university than a low selectivity public university, when controlling for ACT scores. O n the other hand, Hispanic males (compared to white males), were over twenty one times more likely to attend a very high selectivity private university than a low selectivity public university, when controlling for ACT scores. As was the case for the SAT, the magnitude and sign of the differences in odds ratios between males and females highlights the interaction between race and gender in the current model. While ten of the category effects were significantly positive for one or both of the gender es when controlling for SAT, twenty nine were significantly positive when controlling for ACT. In terms o f non significance, however, there were 31 (of the 76 possible) race category effects that were not significantly different from 0 for either gender A few no teworthy differences can be inferred from Table 4 2. Black females (compared to white females) were roughly eight times more likely to attend a low selectivity Catholic 4 year college than a low selectivity public university, while black males were not sig nificantly more or less likely than white males to make the same acceptance decision. Black males were roughly 88% less likely than white males to attend a medium selectivity private non sectarian 4 year college (relative to the reference institution), whi le no significant difference was observed for females (for the same institutional comparison).

PAGE 54

54 Table 4 2. Odds ratio for race by gender controlling for ACT

PAGE 55

55 Table 4 2. Continued Overall, race and gender significantly affect a high achieving student's acceptance outcomes when ACT scores are held constant (controlled for). The model including race, gender and ACT displayed a significantly better fit than the ACT only model (X 2 (171) = 12886, p 0.001). As was the case for SAT, the increase in model fit indicates that ACT does a significantly better job of predicting acceptance outcomes when race and gender are controlled for. The results indicate that demographic characteristics (race and gender ) play a significant role in the acceptance outcomes of high achieving students, when controlling for test scores. The ACT model displayed a better model fit than the SAT model (AIC = 224660.13 vs. AIC = 246714.80, respectively). Thus, the model that contr olled for ACT scores did a better job of predicting acceptance outcomes among high achieving students than the model that controlled for SAT scores Research Question 3 Controlling for test scores, do parental demographics (household income and highest pa rental educational attainment) impact college acceptance

PAGE 56

56 outcomes for high achieving students? Two multinomial logistic regressions were conducted with acceptance outcome as the response variable and household income and highest parental educational attain ment as the predictors. The first MLR controlled for SAT, while the second controlled for ACT. Low selectivity public universities were used as the reference category for the response variable. As mentioned above, household income and highest parental educ ational attainment were treated as continuous in the statistical procedures o f this section. RQ3 SAT The global test for non contribution of the predictors was significant (X 2 (76, n = 53863) = 13392.65, p 0.001), indicating that SAT score, household inc ome, and highest parental educational attainment do a significantly better job of predicting acceptance outcome compared to an intercept only model. A significant interaction effect between household income and highest parental educational attainment was o bserved and, therefore, included in the model (p 0.001). N o predictor by covariate interactions were included (i.e., the SAT by household income and SAT by highest parental educational attainment interactions were excluded from the model). All main effects were significant (p 0.001). However, not all comparison effects (the coefficients for each acceptance outcome relative to low selectivity public university for a given predictor) were significant. The odds ratios for highest parental educational attainm ent are provided by household income (at the lowest, most frequent, and highest income class values) in the following section From Table 4 3, we see that while several of the marginal odds ratios were significant, only one of them was more than 10% diffe rent from 1: students whose household income was less than ten thousand dollars per year were 10.3% more likely

PAGE 57

57 to attend a medium selectivity o ther religious 4 year college (relative to a low selectivity public university) for each one unit increase in highest parental educational attainment. For those in the lowest household income class, all of the odds ratios for o ther religious 4 year colleges were significantly greater than I indicating that students in the lowest household income class were significantly more likely to attend an "other religious" 4 year college (of any selectivity) for each unit increase in highest parental educational attainment, when controlling for SAT scores. This increased likelihood appears to decrease for a ll levels o f selectivity as household income increases (i.e., the odds ratios decrease across the rows of Table 4 3 corresponding to other religi ous 4 year colleges). While the odds ratios for high and very high selectivity private universities were no t significantly different from 1 for those in the lowest household income class, a significant increase was observed for those in the highest household income class. Those students in the highest household income class were 2.7% and 4.6% more likely to att end a high and very high selectivity private university (respectively and relative to the reference institution) for each one unit increase in highest parental educational attainment. Thus, for a seven unit increase (the difference between "Grammar School or Less" and "Graduate Degree"), students in the highest household income class were 20.5% and 37% more likely to attend a high and very high selectivity private university (respectively and relative to the reference institution), controlling for SAT score s. Overall, parental demographics (household income and highest parental educational attainment) significantly affected a high achieving student's acceptance outcomes when SAT scores are controlled for. The model including parental demographics and SAT

PAGE 58

58 Ta ble 4 3. Odds ratios for highest parental attainment by household income, controlled for SAT s cores displayed a significantly better fit than the SAT only model (X 2 (57) = 23800.28, p 0.001). This increase in model fit indicates that the SAT does a si gnificantly better job of predicting acceptance outcomes when parental demographics are controlled for. The

PAGE 59

59 interactions between SAT and parental demographics were not included in the model above, as their inclusion would make interpretation more difficult Additionally, all of the comparison effects for the SAT by household income and SAT by highest parental educational attainment interaction effects were virtually zero, indicating that the inclusion of these interaction terms would not have added much val ue to the current discussion. RQ3 ACT The global test for non contribution of the predictors was significant (X 2 (76, n = 42751) = 8890.76, p 0.001), indicating that ACT score, household income, and highest parental educational attainment do a sign i ficantly better job of predicting acceptance outcome compared to an intercept only model. A significant interaction effect between household income and highest parental educational was observed and, therefore, included in the model (p 0.001). No predict or by covariate interactions were included (i.e., the ACT by household income and ACT by highest parental educational attainment interactions were excluded from the model). All main effects were significant (p 0.001). However, not all comparison effects were significant The odds ratios for highest parental educational attainment are provided by household income (at the lowest, most frequent, and highest income class values) on the following page. From Table 4 4, we see that more than half of the margina l odds ratios were non significant (for the income classes included in the table). Further, none of the marginal odds ratios were more than 10% different from 1. The largest positive effect indicates that students in the highest household income class wer e 10% more likely to attend a very low selectivity other religious 4 year college (relative to a low selectivity public university) for each one unit increase in h i ghest parental educational attainment.

PAGE 60

60 Conversely to the relationship observed for very low selectivity Other religious 4 year colleges in the SAT controlled model, the odds ratios appear to increase as income values increase, when ACT scores are controlled for. This reversal of the direction o f the highest parental educational attainment o dds r atios across the different household income values may be due to the stronger predictor by covariate interaction effects in the ACT model (relative to the SAT model), which were excluded (after much consideration) from the analysis due to the potential for interpretation issues and overall understanding. More than half of the response levels (labeled "Effect" in Table 4 4) saw no significant difference at one or more of the household income values provided (the lowest, most frequent, and highest). In o ther words, the overall strength o f the interaction between household income and highest parental educational attainment was lower when controlling for ACT score s than when the SAT score was controlled for. The magnitude o f the overall main and interactio n effects was greater for SAT than ACT: household income ( X 2 = 690.21 vs. X 2 = 411.47), highest parental educational attainment ( X 2 = 496.78 vs. X 2 = 312.53), and their interaction ( X 2 = 774.36 vs. X 2 = 612.06). However, the overall model fit was better fo r ACT (AIC = 215215.63) than SAT (AIC = 237677.80). Overall, parental demographics (household income and highest parental educational attainment) significantly affected a high achieving student's acceptance outcome when ACT scores were controlled for. The model including parental demographics and ACT displayed a significantly better f it than the ACT only model ( X 2

PAGE 61

61 (57) = 22102.5, p 0.001). This increase in model fit indicates that ACT does a significantly better job of Table 4 4. Odds ratios for highest parental attainment by household income, controlled for ACT

PAGE 62

62 predicting accept ance outcomes when parental demographics are controlled for. The interactions between ACT and parental demographics while stronger than those between SAT and parental demographics were not included in the model above, as their inclusion would make interp retation more difficult. While some of the comparison effects were significantly different from 0 for the ACT by highest parental educational attainment interaction (especially among the high and very high selectivity institutions), all interactions betwee n parental demographics and ACT were excluded from the current analyses for the sake of clarity. Limitations The responses to The Freshman Survey were self reported. Therefore, any analyses conducted on such information make s the inherent assumption that the students responded honestly and that the responses are true. The current study defined a high achieving student as one with a self reported high school GPA of "A or A+" Differences in GPA within this range may account for some of the differences obser ved in this study. Two o f the predictors (household income and highest parental educational attainment) were ordinal in nature, and as such, their treatment within certain statistical modeling procedures is not as clearly defined as for nominal or interval variables. Many educational researchers claim that if the number of categories is larger than, say, 5, then one is justified in using such a variable as though it were an interval (or ratio). Others claim that ordinal variables should be treated as categ orical, to avoid any additional assumptions. Recoding household income into a variable with fewer categories was not possible as information on the size of a student's household was not available. Further, the use o f percentile based recoding was avoided due to the rather large number of ties in household income. To determine the appropriateness

PAGE 63

63 (with respect to information gain /loss) of treating household income and highest parental education as continuous predictors, models treating these predictors as c ategorical were compared to the same models treating these predictors as continuous. No significant differences were observed between the continuous and categorical runs of the models. Therefore, household income and highest parental education were treated as continuous. Lastly, the mechanisms governing the relations between the variables of interest may have changed drastically in the past twelve years. As such, any sequential generalizations should be approached with caution. Concerning the sample, a few important limitations are explained. First, due to the fact that the part i cipating institutions (and their participating students) were not chosen at random, the sampling method of the HERI Freshman Survey essentially falls into the category o f non probability sampling (more specifically, voluntary response sampling). W hile i t may be logical to assume that the students at non participating institutions did not differ from those at participating institutions in any meaningful way, it is not quit e as easy to assume that the students who chose to respond to the survey were the same as those who chose not to respond. Secondly, the high achieving student sample utilized in the current study was large (over 100k). As such, more importance should be pl aced on effect sizes, rather than p values when interpreting the results. The interpretation of p values in light o f large sample sizes is an issue that is well documented, but not generally well understood.

PAGE 64

64 CHAPTER 5 DISCUSSION AND RECOMMENDATIONS Implication of the Findings Significant differences in acceptance outcomes were observed among the high achieving student sample. While the students in this sample are identical with respect to high school GPA, the differences they experienced in accept ance outcomes based on test scores (SAT and ACT) were significant: higher scores were associated with an increased likelihood of acceptance to more selective institutions. In other words, a student's score on a standardized examination related to their cha nces o f attending a more selective institution. There were also identifiable differences in acceptance outcomes due to the demographic characteristics o f gender and race. Students who were wealthy, or White and/or Asian, were significantly more likely to b e accepted at a selective four year college. Controlling for test scores and (by design) high school GPA, certain students were more or less likely to attend certain institutions based on whether they were male vs. female, Black vs. White Hispanic vs. Bla ck, etc. Some of these differences could have been due to the low representation of certain race groups at certain institutions (i.e., an institution may more readily admit a member o f an underprivileged race group). However, such policies do not account f or the differences between males and females observed in Tables 4 1 and 4 2. The independent variables utilized in this study were selected on the basis of their relevance to the underlying research questions. The inclusion or exclusion of other variables or responses may or may not have skewed the result s, and as such further studies using the HERI Freshman Survey variables may yield more specific outcomes.

PAGE 65

65 Lastly, there are the differences in acceptance outcomes based on parental demographics o f household income and highest parental educational attainment. While statistically significant, these differences were not as clearly defined as those observed for other characteristics. While increases in highest parental educational attainment were associated with an increased likelihood of attending the most selective institution s of the study (very high selectivity private universities) for students whose household income was greater than 250k per year, such increases were not observed for all household income va lues. Additionally, the study did not examine the implications of those students who were living with one rather than both parents, and the impact this had on postsecondary acceptance. Practitioners should keep in mind several things when interpreting the results of the current study: some students may have been accepted to a more selective institution, but chose not to attend due to the cost or the distance from their parents' home; some institutions are male or female only; and some responses may not hav e been entirely truthful. In spite of these potential issues, the current study indicates that a relationship exists between standardized test scores and the race, gender, and parental demographics of a high achieving student. The differences identified in this study warrant additional examination to provide more definitive insight on how standardized testing differences ultimately impact postsecondary acceptance outcomes, and is intended to be the beginning of a dialogue about how standardized testing and demographic variables are interrelated and how these metrics are utilized by high school and college admissions leaders, professional organizations, and research institutions This research effort may be critical to understanding the relationship between race, gender, parental demographics and the

PAGE 66

66 stratification of higher education, paving the road toward a more consistent set of admissions standards and practices that promote equitable access to all institutions for an increasingly diverse student populat ion. Recommendations for Future Research The manner in which postsecondary institutions evaluate potential applicants has shifted very little in the past half century. The results of this study indicate that the way selective colleges and universities ma ke determinations about who to admit is still primarily based on high school grades and standardized test scores. While these cognitive demographics are significant they cannot fully predict a student's career success, academic achievement, college persis tence, campus contributions, and ultimate success down the road. This study clearly merits further examination relating to the relationship between race, ethnicity, gender and parental demographics and college acceptance outcomes. Additionally,f urther rese arch exploration into the utilization and development o f assessments that examine non cognitive qualities, such as creativity, determination, problem solving skills, interpersonal ability emotional intelligence and grit, would yield promising results to i nnovate the state of college admissions. In an ever evolving world that isn't primarily centered around percentages, criteria, and data points, further research on the creation of structured admissions processes that identify leadership teamwork, and gene ral success traits may provide more equitable and diverse populations matriculating to college campuses.

PAGE 67

67 APPENDIX A 2006 CRP FRESHMAN SURVEY

PAGE 68

68

PAGE 69

69

PAGE 70

70

PAGE 71

71 APPENDIX B CIRP FRESHMAN SURVEY DATA FILE

PAGE 72

72

PAGE 73

73

PAGE 74

74

PAGE 75

75

PAGE 76

76

PAGE 77

77

PAGE 78

78

PAGE 79

79

PAGE 80

80

PAGE 81

81

PAGE 82

82

PAGE 83

83

PAGE 84

84 APPENDIX C CIRP FRE S HMAN SURVEY (CODEBOOK)

PAGE 85

85

PAGE 86

86

PAGE 87

87

PAGE 88

88

PAGE 89

89

PAGE 90

90

PAGE 91

91

PAGE 92

92

PAGE 93

93

PAGE 94

94

PAGE 95

95

PAGE 96

96

PAGE 97

97

PAGE 98

98 APPENDIX D ACT SAT CONCORDANCE : A TOOL FOR COMPARING SCORES

PAGE 99

99

PAGE 100

100 LIST OF REFERENCES ACT, Inc. (2017). The ACT technical manual. Retrieved from https://www.act.org/content/dam/act/unsecured/documents/ACT_Technical_Man ual.pdf ACT SAT Concordance: A Tool for Comparing Scores. (2013) Retrieved from https//www.act.org/content/dam/acb'unsecured/documents/reference.pdf Alon, S., & Tienda, M. (2007). Diversity, opportunity and the shifting meritocracy in higher education, 72(4), 487 511 Atkinson, R. H. (2001). Standardized tests and access to American universities: UC and the SAT. Speech delivered at the 83rd Annual Meeting of the American Council on Education Washington, D.C., February 18, 2001. Retrieved from http://ucop.edu/news/sat/speech.html Atkinson R., & Geiser, S. (2009). Reflections on a century of college admissions tests. Educational Researcher 38(9), 665 676. Atkinson, R., & Geiser, S. (201 2). Reflections on a century of college admissions tests. In J. A. Soares (Ed.), SAT wars: The case for test optional college admissions (pp. 23 49). New York, NY: Teachers College press. Baum, S. Ma, J., & Payea, K. (2013). Education pays, 2013: The be nefits of higher education for individuals and society. Trends in Higher Education Series. College Board. Belasco, A. S. Rosinger, K. 0., & Hearn, J. C. (2015). The test optional movement at America's selective liberal arts colleges: A boon for equity o r something else? Educational Evaluation and Policy Analysis 37(2), 206 223. Bettinger, E. p. Evans, B. J. & pope, D. G. (2013). Improving college performance and retention the easy way: Unpacking the ACT exam. American Economic Journal: Economic Policy 5(2), 26 52. Bills, David B. 2003. "Credentials, signals, and screens: explaining the relationship between schooling and job assignment." Review of Educational Research 73(4): 441 49. Booker, K. C. (2004). Exploring school belonging and academic achievement in African American adolescents. Curriculum & Teaching Dialogue 6(2), 131 143. Bradshaw, G., Espinoza, S., & Hausman, C. (2001). The college decision making of high achieving students. College and University 77(2), 15. Brody, L. E. & Benbow, C. p. (1990). Effects Of high school coursework and time on SAT scores. Journal of Educational Psychology 82(4) 866 75.

PAGE 101

101 Bulman, G. (2015). The effect Of access to college assessments on enrollment and attainment. American Economi c Journal: Applied Economics 7(4): 1 36. Buckley, J. Letukas, L. & W i Idavsky, B. (Eds.). (2018). Measuring success: testing, grades, and the future of college admissions. JHU Press. Cabrera, A. F. & Burkum, K. R. (2001). College admission criteria in the United States: An overview. Madison, WI: University of Wisconsin at Madison. Cabrera, A. F. & La Nasa, S. M. (2000). Understanding the college choice process. New Directions for Institutional Research 107, 5 22. Camara, W. J. (2009). College ad mission testing: Myths and realities in an age of admissions hype. In R. P. Phelps (Ed.), Correcting fallacies about educational and psychological testing (pp. 147 180). Washington, DC: American Psychological Association. Chenoweth, E. & Galliher, R.V. (2004). Factors influencing college aspirations Of rural West Virginia high school students. Jou rn al of Research in Rural Education 19(2 ). 66 75. Clinedinst, M. E. & Koranteng A.M. (2017). State Of College Admission. Alexandria, VA: National Association for College Admission Counseling. Cole, N. S. (1973). Bias in selection. Journal of Educational Measurement 10, 237 255. Creswell, J W. (2009). Research design. Qualitative, quantitative and mixed methods approaches. Thousand Oaks, CA: Sage Crouse, J., & Trushe i m, D. (1988). The case against the SAT. Chicago, IL: The University of Chicago Press. Dale, S.B., & Krueger A. (2002). "Estimating the payoff to attending a more se lective college: An application of selection on observables and unobservables." Quarterly Journal of Economics 117(4): 1491 1527. DesJardins, S. L. Ahlburg, D. A., & McCall, B. p. (2006). An integrated model o f application, admission, enrollment, and fi nancial aid. Journal of Higher Education 77(3), 381 429. Dickinson, E. R., & Adelson, J. L. (2016). Choosing among multiple achievement measures: Applying multitrait multimethod confirmatory factor analysis to measures o f state assessment, ACT, and student GPA data. Journal of Advanced Academics 27, 4 22. Dollinger, S. J. (2011). Standardized minds or individuality? Admissions tests and creativity revisited. Psychology of Aesthetics, Creativity and the Arts, 5, 329 341.

PAGE 102

102 Dorans, N. J, Lyu, C. F., Pommerich, M., & Houston, W. M. (1997). Concordance between ACT assessment and recentered SAT I sum scores. College and University 73(2), 23 34. Eagan, K. Lozano, J. B. Hurtado, S. & Case, M. H. (2016). The American freshm an: National norms fall 2015. Los Angeles: Higher Education Research Institute, UCLA. Edmunds, A. O. (2010). An examination of the likelihood of persistence of students with discrepant high school grades and standardized test scores (Order N o 3423016). A vailable from ProQuest Dissertations & Theses Global; Social Science Premium Collection. (758391786). Elliott, R., and Strenta, A. C. (1988). Effects of improving the reliability of the GPA on predict' on generally and on comparative predictions for gende r and race particularly. Journal of Educational Measurement 25(4), 333 347. Espenshade, T., & Chung, C. Y. (2012). Diversity outcomes of test optional policies. In J. Soares (Ed.), SAT wars: The case for test optional college admissions (pp. 177 200). Ne w York, NY: Teachers College Press Evans, B. (2015). College Admission Testing in America. In Stead, V. (Ed.), International Perspectives in Higher Education Admission Policy: A reader. New York: Peter Lang Publishing. Everson, H. T. & Millsap, R. E. (2004), Beyond individual differences: exploring school effects on SAT scores. Educational Psychologist 3 9(3 ), 157 172. Fa i rTest List, (2016 2018) Retrieved from https://www.FairTestorg/university/optional Fleming, J. (2000). Affirmative action and standardized test scores. The Journal of Negro Education 6 9( 1 /2), 27 37. Freedle, R. O. (2003). Correcting the SATs ethnic and social class bias: A method for re esti mating SAT scores. Harvard Educational Review 73, 1 43 Furuta, J. (2017 ) Rationalization and student/school personhood in LIS college admissions: The rise o f test optional policies, 1987 to 2015. Sociology Of Education 90(3), 236 254. Gambino, M. (2013, April 12). Document deep dive: What was on the first SAT? Smithsonian Magazine Retrieved from http://v.ww.smithsonianmag.com/history/document deep dive vvhat was on the first sat 21720496pno ist the first sat 21720496/ Geiser, S., & Studley, R. (2002). UC and the SAT: Predictive validity and differential impact of the SAT I and SAT Il at the U niversity of California. Educational Assessment, 8, 1 16

PAGE 103

103 Geiser, S. & Santelices, M.V. (2007). Validity of high school grades in predicting student success beyond the freshman year: High school record vs. standardized tests as indicators of four year college outcomes. (Research & Occasional Paper Services CSHE.6.07). Berkeley, CA: University of California, Center for Studies in Higher Education. Goenner, C., & Snaith, S. (2004). Assessing the effects of increased admission standards. College and University, 80(1) 29 34. Goodman, S. (2016). Learning from the test Raising selective college enrollment by prov iding information. Review of Economics and Statistics 98(4), 671 634. Griffore, R. J. (2007). Speaking o f fairness in testing. American Psychologist, 62, 1081 1082. Hbert, T. P ., & Reis, S. M. (1999). Culturally diverse high achieving students in an ur ban high school. Urban Education, 34(4) 4 28 457. Hill, C., Winston, G., & Boyd, S. (2005). Affordability: Family incomes and net prices at highly selective private colleges and universities. Journal of Human Resources XL : 4 Hiss, W.C., & Franks, V.W. (2014). Defining Promise: Optional standardized policies in American college and university admissions. Retrieved from https://offices.depaul.edu/enrollment management marketing/testoptional/Documents/HISSDefiningPromise.pdf Holland, M. M. (2014). Navigating the road to college: Race and class variation in the college application process. Sociology Compass 8(10), 1191 1205. Hoover, E. (2015 May 29). College Admissions, Frozen in Time. Chronicle of Higher Education p. 24. Hossler, D, & Gallagher, K. S. (1987). Studying student college choice: A three phase model and the implications for policymakers. Colle ge and University 62(3), 207 21. Hossler, D. & Stage, F. K. (1992). Family and high school experience influences on the postsecondary educational plans Of ninth grade students American Educational Research Journal 29( 2), 425 451. Hossler, D., Schmit J., & Vesper, N. (1998). Going to college: How social, economic, and educational factors influence the decisions students make. Baltimore: The Johns Hopkins University Press. Hoxby, C. M., & Avery, C. (2012). The missing "one offs": The hidden sup ply of high achieving low income students. NBER Working Paper No. 18586.

PAGE 104

104 Hoxby, C. M., & Tu rn er, S. (2015). What high achieving low income students know about college. American Economic Review 105(5), 514 17. Jencks, C., & Crouse, J. (1982). Aptitudes vs. achievement' Should we replace the SAT? The P ublic Interest (67), 21 Jencks, C. & Phillips, M. (1998). The Black White test score gap: An introduction. In C. Jencks, & M. Phillips (Eds.), The Black White test score gap (pp. 1 54). Washington, DC: The Brookings Institution. Kane, M. T. (2013) Validating the interpretations and uses of test scores. Journal of Educational Measurement 50. 1 73. Keup, J. R. (2004). The cooperative institutional research program Freshman Survey and Your First College Ye ar: Using longitudinal data to assess to first year of college. Assessment Update 16(2), 8 10. Kim, M. (2010) Preferences of high achieving high school students in their career development. Gifted and Talented International 25(2), 65 74 Klasik D. (2012). The college application gauntlet: A systematic analysis of the steps to four year college enrollment. Research in Higher Education 53(5), 506 549 Kobrin, J., Patterson, B. Shaw, E., Mattern, K, & Barbuti, S. (2008). Validity of the SAT fo r predicting first year college grade point average. (College Board Research Rep. No. 2008 5). New York: College Board. Kobrin, J. L., & Patterson, B. F. (2011). Contextual factors associated with the validity of SAT scores and high school GPA for predict ing first year college grades. Educational Assessment 16(4), 207 226. Lemann, N. (1995). The great sorting. The Atlantic Monthly 276(3), 84 100. Lichtenberger, E. J. & Dietrich, C. (2012). College readiness and the postsecondary outcomes Of Illinois high school students. policy Research: IER& 2012 1. Edwardsville, IL: Illinois Education Research Council. Linn, R. L. (1992b). Admission testing on trial. American Psychologist 37, 279 291. Long, M. (2008)_ College quality and early adult outcomes. Eco nomics of Education Review 27(5), 588 602 MacAllum, K, Glover, D. M. Queen, B. & Riggs, A. (2007). Deciding on postsecondary education: Final report; NPEC 2008 850. Washington DC: National Postsecondary Education Cooperative.

PAGE 105

105 Mattern K. D., Shaw, E. J., & Kobrin, J. L. (2011). An alternative presentation of incremental validity: Discrepant SAT and HSGPA performance. Educational and Psychological Measurement 71(4), 638. Marsh, C. M., Vandehey, M. A, & Diekoff, G. M. (2008)_ A compar ison of an introductory course to SAT/ACT scores in predicting student performance. JGE: The Jou rn al of Genera l Education 57(4), 244 255. McDonough, P.M. (1994). Buying and selling higher education: The social construction Of the college applicant. Journ al Of Higher Education 65, 427 446. Melguizo, T. (2008). Quality matters: Assessing the impact of attending more selective institutions on college completion rates of minorities Research in Higher Education 49, 214 236. Mendrinos N. (2014). Beyond the SAT/ACT: An examination of non cognitive factors that contribute to students' college success (Order No. 3671934). Available from ProQuest Dissertations & Theses Global. (1651614968). Miller, S. Walker, M., & Letukas, L. (2017). Redesigning the SAT Using Principles Of Fairness and Equity. In Jiao, H. & In Lissitz, R. W. (2017). Test fairness in the new generation o f large scale assessment., Charlotte, NC : Information Age Publishing, Inc Muhammad, C. G. (2008). African American students and college choice: A consideration of the role of school counselors. National Association of Secondary School Principals. NASSP Bulletin, 92(2), 81 94. National Association for College Admission Counseling. (2008). Report of the commission on th e use of standardized tests in college admission. Alexandria, VA: Author. Niu, S. X., & Tienda, M. (2012). Test scores, class rank and college performance: Lessons for broadening access and promoting success. Rassegna italiana di sociologia, 53(2), 199 22 6. Noble, J. p. & Sawyer, R. L. (2004). Is high school GPA better than admission test scores for predicting academic success in college? College 8 University, 7R4), 17 22. Noftle, E. E., & Robbins R. W. (2007). Personality predictors of academic outco mes: Big five correlates Of GPA and SAT scores. Journal of Personality and Social Psychology 93, 116 130 Paulsen, M. B. & St. John, E. p. (2002). Social class and college costs: Examining the financial nexus between college choice and persistence. Jou r n al of Higher Education 73(2), 189 236.

PAGE 106

106 Perna, L. (2000). Differences in the decision to attend college among African Americans, Hispanics, and Whites. Journal of Higher Education 71(2), 117 141 Perrone, KN., Tschopp, MM., Snyder, E R Boo, J.N., & Hyatt, C. (2010). A longitudinal examination of career expectations and outcomes of academically talented students 10 and 20 years post high school graduation. Journal of Career Development 36(4), 291 309. Posselt, J.R., Jaquette, 0., Biel by, R., & Bastedo, M. N. (2012). Access without equity: Longitudinal analyses of institutional stratification by race and ethnicity 1972 2004. American Educational Research Journal, 49(6), 1074 1111 Radford, A. W. (2013). Top student, top school?: How s ocial class shapes where valedictorians go to college. University Of Chicago P ress Robbins, S. Allen, J. Casillas, A, Peterson, C. H. & Le, H. (2006). Unraveling the differential effects o f motivational and skills, social, and self management measu res from traditional predictors o f college outcomes. Journal of Educational Psychology 98, 598 616. Robinson, M. Monks, J. (2005). Making SAT scores optional in selective college admissions: A case study. Economics of Education Review 24, 393 405. Ro driguez, A. & Martell, C. (2016). Average students and college match: Looking beyond the elite. In A. P Kelly, J. S. Howell, & C. Sattin Bajaj (Eds.), Matching students to opportunity: Expanding college choice, access, and quality (pp. 53 77). Cambridge, MA: Harvard Education Press. Rosner, J. (2012). "The SAT: Quantifying the Unfai rness Behind the Bubbles," in SAT Wars: The Case for Test Optional College Admissions, edited by J. A. Soares, Teachers College, Columbia University, New York and London. Rothstein, J. (2004). College performance predictions and the SAT. J ournal of Econometrics, 121, 297 317. Rubin, R. (2014). Who gets in and why? An examination of admissions to America's most selective colleges and universities. Inte rn ational Education Re search 2(2), 1 18. Sackett, P R, Kuncel, N. R, Beatty, A. S, Rigdon, J. L., Shen, W., & Kiger, T. B. (2012). The role of socioeconomic status in SAT grade relationships and in college admissions decisions. Psychological Science 23(9), 1000 1007. Sacke tt p. Kuncel N. Arneson J. Cooper S. Waters S. (2009 ) Does socioeconomic status explain the relationship between admissions tests and post secondary academic performance? Psychological Bullet i n 135, 1 22

PAGE 107

107 Sanchez, E.I., and Lin, S. (2017). Effect s of discrepant ACT and high school GPA on enrollment and persistence. Unpublished manuscript, ACT, Iowa City. Santelices, M. & Wilson M. (2010). Unfair Treatment? The Case Of Freedle the SAT, and the standardization approach to differential item functioning. Harvard Educational Review, 80( 1), 106 133. Sawyer, R. L. (2007). Indicators of usefulness of test scores. Applied Measurement in Education 20(3), 255 271. Sawyer, R. (2010) Usefulness Of High School Average and ACT Scores in Making College Admission Decisions. ACT Research Report Series 2010 2. Shanley, B. J. (2007). Test optional admission at a liberal arts college: A founding mission affirmed. Harvard Edu cational Review 77(4), 429435 Shaw, E. J., Marini, J. P., & Mattern, K. D. (2013). Exploring the utility of Advanced Placement participation and performance in college admission decisions. Educational and Psychological Measurement 73(2), 229 253. Smit h, J., Pender, M., & Howell, J. (2013). The full extent of student college academic undermatch. Economics of Education Review 32, 247 261. Soares, J. A. (2012). The future of college admissions: Discussion Educational Psychologist 47(1), 66 70. Syvers on, S. (2007). The role of standardized tests in college admissions: Test optional admissions, New Directions for Student Services 118(2), 55 70 Syverson, S. Hiss, W.C., & Franks, (2018). Defining access: How test optional works. National Association fo r College Admissions Counseling. Retrieved from https://wuvw.nacacnet.org/globalassets/documents/publicatlons/research/definln g acces s report 2018.pdf Thompson, J. (2017). Access, outcomes, and social mobility in a stratified system of postsecondary education (Order No. 10192286). Available from ProQuest Dissertations & Theses Global. (1937951166). Westrick, P. A., Let H., Robbins, S. B., Radunzel, J. M., & Schmidt, F. L. (2015). College performance and retention: A meta analysis of the predictive validities of ACTO scores, high school grades, and SES. Educational Assessment, 20 ( 1), 23 45. Williams J. M. & Bryan, J. (2013). Overcoming adversity: High achieving African American youth's perspectives on educational resilience. Journal of Counseling 8 Development, 91(3), 291 300

PAGE 108

108 Young, J. W. & Fisler, J. L. (2000). Gender differences on the SAT: An analysis Of demographic and educational variables. Research in Higher Education 41(3), 401 16. Yuan IN., & Soares J.. (2017). Excellence, equity, and diversity: An exploration of the admission goals at selective colleges and universities in the : United States Journal of China University of Geosciences (Social Science Edition), 17, (4): 148 156. Zwick, R. (1999). Eliminating standardized tests in college admissions: The new affirmative act i on? Phi Delta Kappan 81 Zwick, R. & Sklar, J. C. (2005). Pred icting college grades and degree completion using high school grades and SAT scores: The role o f student ethnicity and first language. American Educational Research Journal 42(3), 439 464. Zwick, R. & Green, J. (2007). New perspectives on the correlati on o f SAT scores, high school grades, and soc i oeconomic factors. Journal of Educational Measurement, 44(1), 23 45. Zwick, R. & Himelfarb, l. (2011). The effect of high school socioeconomic status on the predictive validity of SAT scores and high school grade point average. Journal of Educational Measurement 48(2), 101 121. Zwick, R. (2013). Disentangling the role o f high school grades, SAT scores, and SES in predicting college achievement. ETS Research Report Series, 2013(1), 1 20. Zuckerman, M. (1979). Attributions o f success and failure revisited, or: The motivational bias is alive and well in attribution theory. Journal Of Personality 47, 245 287.

PAGE 109

109 BIOGRAPHICAL SKETCH Leena Hasbini was born and raised in Tampa, Florida. She rec eived her bachelor's degree in business administration at nineteen from the University of South Florida, and subsequently went on to earn her master's degree in counselor education also from the University of South Florida at 21. She has been privileged to serve in various capacities within the education field, instilling a sense o f confidence and competence within students as they prepared for their transition to higher education. Leena currently serves as Director of Marketing for Sunrise Homes.