Mount Holyoke College’s Experience with Optional ...



Mount Holyoke College’s Experience with Optional Submission of SAT Scores: An Analysis of the First Four Years

Cate Rowen, Mount Holyoke College, Director of Special Research Projects, crowen@mtholyoke.edu, (413) 538-2782

Michael D. Robinson, Department of Economics, Mount Holyoke College, mirobins@mtholyoke.edu, (413) 538-3085

Abstract

This paper examines the results of the decision by Mount Holyoke College to make SAT scores optional in the admissions process. Beginning with the class entering in 2001 the college has not required applicants to submit SAT scores. Using data from the first four years of SAT optional admissions we examine a number of implications and consequences of the policy. We investigate both the student’s decision of whether or not to submit her scores and the impact of that decision on the admissions outcome for the student. There is evidence supporting both the claim that SAT scores can be highly useful in making admissions decisions as well as the counter argument that, at least for a college where admissions is driven by time-intensive reading of folders, SATs are not necessary and add little information beyond what is already contained in the qualitative portions of the applicants file.

Narrative

Introduction

Despite the heightened concern with the use of standardized tests in college admissions and the important institutional and public policy implications of the use of SAT scores in admissions, there has been little public empirical analysis of the effects of an optional SAT score submission policy on admission outcomes and the subsequent academic performance of those students who chose not to submit their scores, but were admitted to the institution. This paper attempts to fill this void by examining the results of the recent decision by Mount Holyoke College, a small, prestigious, New England women’s liberal arts college, to make SAT score submission optional in the admissions process.

Using data from Mount Holyoke College, this paper presents a comprehensive and detailed analysis of the impact of the SAT optional policy on admissions and educational outcomes of the College. We employ several different techniques to address various questions about the policy.

The following analyses focus on the effect of the optional SAT score submission policy on:

1. the decision of applicants to either submit or withhold their SAT scores in the admissions process;

2. the treatment of applicants who choose to withhold their SAT scores by the admissions office;

3. the effect on yield of withholding scores;

4. the academic performance of the applicants who blocked their scores during the admissions process; and finally,

5. the perceptions of high school seniors, current students, and high school counselors of the policy.

Data and Methodology

All data used in the evaluation of student score submission and performance was obtained from Mount Holyoke College with the exception of SAT scores. Since the College did not have complete information on the SAT scores of non-submitters, SAT scores for all applicants were obtained from the College Board. We used the College Board-supplied scores throughout the analyses to guarantee that the source of SAT information was consistent between submitters and non-submitters. For each area of the analysis an appropriate statistical model is formulated and estimated.

In order to track potential differences in adjustment to college and non-quantitative student outcomes, we are following a randomly-selected group of SAT-blockers and an equivalent group of non-blockers, interviewing them on an annual basis about their college experience, adjustment to college, time management, social issues, and the development of their academic goals and plans. While this study incorporates a control (SAT submitters) and test (SAT blockers) group, our analysis is thematic and qualitative in nature.

To evaluate the effect of Mount Holyoke’s SAT-optional policy on prospective students during the application process, and to test a number of hypotheses about how students decide whether or not to submit scores (and whether or not to apply to Mount Holyoke in the context of its SAT-optional policy), we conducted a telephone survey of students on Mount Holyoke’s inquiry list who did not apply to Mount Holyoke, during the month of February, after the application deadline. A group of accepted students were interviewed with a similar instrument during June of the same year, after matriculation decisions and deposits were due.

To better understand the opinions of key influencers in the college-admission process and to monitor perceptions in the “marketplace,” we conducted focus groups with college admission counselors at feeder high schools. These groups focused on counselor attitudes about the SAT and the college admission process, as well as exploring the concepts of selectivity and prestige as they are understood by counselors, students, and parents.

Finally, we conducted a folder-reading experiment in which two groups of admission staff read applicant folders with and without SAT scores in order to determine whether or not readers were influenced (in either a positive or negative direction) by the absence of scores.

Results

Analysis of Score Submission Patterns

Table 1 shows the percent of the applicants that were non-submitters[1] by year for race, international status, athletes, artists, legacies, and need. Between 2001 and 2004 there was a 6.9 percentage point increase in non-submission of scores. In 2004 one third of the applicants did not submit scores. ALANA (African-American, Latina, Asian-American, and Native American) and international students are less likely to submit scores than white students, however the growth in non-submission is about the same for each of the groups (slightly higher for whites, slightly lower for ALANA applicants as a whole). It is interesting to note the highest rate of growth of non-submission were among legacy applicants and impact athletes. Identified artists (students who were recruited because of their artistic work) became less likely to be non-submitters over the first three years of the policy.

This simple examination, however, does not control for a variety of factors that might influence the decision of whether of not to submit SAT scores. It would be logical to assume that applicants submit scores when they believe it will help their chance of admission. Following this logic, we would anticipate that the higher the applicant’s high school GPA, the less likely the applicant would be to submit her SAT scores. In order to see which factors were important in the decision to submit SAT scores, we estimated a probit model where the dependent variable was a dummy indicating whether the SATs were submitted. Table 2 reports the results of this model for all four years. Model 1 uses the SAT scores the College obtained from the applicants and Model 2 the SAT scores obtained from the College Board. The model controls for ability to pay, race, region, early decision status, legacy status, and high school GPA. Applicants with higher ability to pay, lower SAT scores, and higher high school GPAs are less likely to submit their SAT scores. This is consistent with a prediction that scores will be submitted only if the applicant thinks they will help her chance of admission. Black applicants were significantly more likely to submit SAT scores than white students in both models and Hispanic applicants were more likely to submit in Model 1. This result casts some doubt on the possibility that dropping the SAT requirement is useful in minority recruiting. While the dummy variables for year of application reveal differences in score submission patterns over time, they do not suggest a trend of either increasing or decreasing submission of scores.

Analysis of Submitters and Non-Submitters in Admission Process

In order to determine the impact of applicants’ score submission decisions, we estimated models to determine the factors related to admission and admission rating. Table 3 reports the results of two acceptance models estimated. Model (1) controls for SAT scores, while model (2) does not. In model (1) we see that the coefficient on non-submission is positive, which means that, conditional on their SAT scores, non-submitters are more likely to be admitted (about 9 percentage points more likely). However in model (2), which does not control for SAT scores and measures quality only based on high school GPA, we see that non-submitters are less likely to be admitted by about five percentage points.

Table 4 reports the results of a linear regression designed to test for the impact of the submission of SAT scores on the admission rating received by the applicant. Again two models are estimated, one that controls for SAT scores and one that does not. In the model controlling for SAT scores, non-submitters are rated more highly by about one third of an admission rating. However, in the model that uses only high school GPA as the quality measure, the non-submitters receive worse ratings than the submitters and by about one third. It is interesting that in both the analysis on acceptance and admission rating the results indicate that the non-submitters seem to be rated more positively than would be expected based on their SAT scores, though this result is not fully explained by higher GPAs. There does not seem to be a clear advantage for submitting or not submitting SAT scores; rather, the benefits depend on other aspects of a prospective student’s application.

Of course, the preceding empirical analysis is problematic because of the endogenous nature of SAT score submission: students are making judgments as to whether it would be advantageous to submit SAT scores. To attempt an alternate approach to estimating the impact of submission status on admission rating we designed an experiment. We randomly selected a stratified sample of one hundred domestic first-year, first-time college applicants to the class of 2006, including SAT submitters and non-submitters. Students were included in the study based on having complete score information available from files provided by the College Board. We stratified the sample based on students’ initial admission rating in 2002, including an equal number of students in each rating, with equal numbers of students who chose to block their scores as applicants and those who did not. This stratification allowed us to analyze any variation within individual rating groups and between those who chose to block or not.

Once the students were randomly selected, their admission folders were reconstructed using archived information and two copies were made for each student. All traces of scores, including SAT and ACT, were blocked from view. Two sets of summary sheets were produced and attached to the folders, one with SAT scores shown and one with all scores blocked. ACT scores were not available for either group.

Two teams of experienced admission readers were formed, and a full set of 100 folders were given to each team, with a random distribution of blocked and unblocked folders going to each team. Therefore each team read the same 100 folders, but with the opposite status. Following a procedure similar to Mount Holyoke’s standard reading format, readers read folders individually, then met in a committee format to review reader ratings. Table 5 reports the results of the experiment in the form of the difference in ratings generated by reading with SAT scores and without SAT scores.

When comparing applicant ratings when SAT scores were blocked to when they were not blocked, 65% of the time Mount Holyoke’s admission committees gave the same rating. A third of applicants had ratings that did differ: one quarter (24%) were rated higher when their scores were blocked, and one tenth (11%) were rated lower. The average difference in ratings was about one fifth of a point (0.17) on a scale of 1 through 8. When ratings did differ, they differed no more than two points, with the majority of differing ratings being one point away from the rating given when scores were not blocked.

Even though the difference in ratings was small when we compared whether the scores were considered or not, we estimated several regressions to test whether the difference in ratings was related to a variety of demographic and applicant qualification variables. For example, we hypothesized that applicants with high SAT scores might fare better when their scores were considered or that applicants with high school GPAs might fare better when scores were blocked. It might also have been the case that the difference in ratings was produced by the process itself unrelated to any characteristic of the applicant. The only significant relationship we discovered was between SAT scores and the rating difference. Not surprisingly, applicants whose combined SAT scores were low (lower than 1327) were rated more highly when their scores were blocked and those whose combined SATs were high (over 1327) were rated more highly when their scores were not blocked. The overall magnitude of the effect, however, was small. For example an applicant with combined SATs of 1600 would only be rated 1/3 of a rating higher when her scores were considered compared to when they were not considered. For no observation in our sample was this SAT score effect large enough to change a rating from one level to the next. These results confirm those found using the actual admissions data, though perhaps suggest that the magnitudes suggested by the non-experimental approach overestimate the impact of non-submission on ratings.

Analysis of Performance of Submitters and Non-Submitters

Of primary importance to the College was the academic performance of non-submitters. One fear expressed in discussions about the policy change was that the College would admit non-submitting applicants unable to perform at the same academic level as SAT score submitters. Table 3 shows the average Mount Holyoke College grade point averages for the submitters and non-submitters overall and by admission rating and SAT score band. Overall the non-submitters perform at a slightly lower level having a GPA that is 0.09 lower than that of the submitters. This difference while significant is quite small amounting to one letter grade lower out of ten courses taken. Because the non-submitters had lower admission ratings than the submitters it was anticipated they would have lower GPAs and so this is not a surprising result. Though the difference is seems very small, especially given the fairly large difference in average SATs between the groups. In the section of Table 3 that reports GPA by admission rating, in ratings 1 to 5 the non-submitters have lower GPAs which are not statistically significant (in part to do the smaller numbers of observations). While for those rated 6 and 7 the non-submitters GPAs are higher. On the other hand the non-submitters had higher GPAs than the submitters for each of the SAT score band groups, though again these differences are not significant in all but the 1000-1100 SAT score band. It appears as though non-submitters do slightly worse than would be predicted by admission rating and slightly better than would be predicted by SAT scores, though these effects are very small and often not significant.

To further explore this Table 4 reports the results of regressions of GPA on admission rating, SAT scores, year of enrollment and a dummy variable for the grades from the sophomore year. Model 1 controls for admission rating and model 2 for SAT scores. In the admission rating model the non-submitters perform more poorly by 0.07 grade points than the submitters, while in the SAT equation the non-submitters have GPA’s higher by 0.06 points (though the latter effect is not significant). Both effects remain small and it can be concluded that both admission rating and SAT scores predict the GPAs of the submitters and non-submitters equally well. The size of the non-submitter effects -0.07 and 0.06 are both smaller than the typical GPA difference between admission ratings (which ranges from .37 between ratings 5 and 6, to 0.08 between ratings 1 and 2).

Perhaps a more important question is how well SAT scores and admission rating predict GPAs for submitters and non-submitters. Table 5 reports the results of separate regressions for submitters and non-submitters of GPA on admission rating and SAT scores. Since admissions did not see the non-submitters SAT scores when assigning their rating the admission rating for non-submitters does not contain information about SAT scores. This means that the non-submitters regression provides a test of whether having SAT scores could have improved the admission rating at least as a predictor of college performance. Since SAT scores are not significant in this equation we conclude that in fact there would have been no additional predictive value of having the SAT scores for the non-submitters. However, SAT scores are significant in the submitters equation which suggests that the SATs are a better predictor of performance for submitters than for non-submitters.

To further explore this question of the usefulness of SAT scores for predicting the GPAs of both groups these regressions were estimated again without controls for admission rating. These are reported in Table 6. SAT scores are significantly related to both the GPA of the submitters and non-submitters however they explain more of the variation for submitters (15.5 percent compared to 3.8 percent) and the estimated impact of SAT on GPA is twice as large for submitters and non-submitters. Thus it would appear that both those who believe that SAT scores would be predictive for non-submitters and those who think they would be less useful are correct.

Analysis of Yield

Applicants with low SAT scores (who did not submit those scores) might be more likely to be accepted at Mount Holyoke than at other institutions that require the SAT because of the SAT optional policy. If this happened it would seem likely that Mount Holyoke would see higher yields on non-submitters than on submitters. Table 10 reports these yields. We see that the evidence here is mixed. In admission ratings 1, 4 and 5 there is not a significance difference between the yields of submitters and non-submitters. In ratings 2 and 3 there is a higher yield for non-submitters but in rating 6 the yield is higher for submitters. Since we would expect this effect to be most pronounced in the lower rating we interpret these results as suggesting that there are not many additional options to submitters whose ratings are low. In addition a probit model of enrollment to the college, that controls for a number of other demographic and financial characteristics found no impact of score submission on yield.

In-Depth Interviews

In our longitudinal qualitative interviews with submitting and non-submitting students, we have found few differences between the two groups, based on preliminary analyses. When we asked students to self-report their submission status, we found that a surprisingly large number reported their status inaccurately, with most of these being non-submitters who believed they had submitted their scores. In many cases this came from confusion about the process—students had indicated that they did not want their scores considered, and so they were not, but their scores were sent to Mount Holyoke anyway. Not realizing that their scores were actually removed from their records, these students assumed that their scores had been included in their admission materials when they were received. This inference echoed questions from guidance counselors about how the process works when scores are submitted inadvertently, as they often are.

Although adjustment to college, reports of stress levels and social and academic involvement themes were consistent among SAT submitters and non-submitters, in the third year we are starting to observe themes related to study abroad participation, involvement in academic activities related to majors, and graduate/professional school aspirations and achievement that may diverge between the two groups. Because of this we have conserved some grant funds to continue these interviews for a fourth and final year in order to track these students’ experience throughout the full four years of college. These interviews have also provided the College with valuable information about student experiences in the first years here and we hope to continue to elicit insights both related to score submission and to the College as a whole.

Inquirer and Accepted Student Survey

We conducted interviews with 389 non-applying inquirers (students who had requested information about Mount Holyoke either through SEARCH direct mail or another route, but who did not apply to the College), and with 300 applicants who were accepted to the college.

We queried inquirers and applicants about their college choices and plans, then focused on their test-preparation activities and perceptions of the SAT and their own scores. Our questions were designed to help us understand how these students perceived and prepared for the SAT, and how they made decisions about whether or not to submit their scores when given the option. In part, we hoped to explain the differential rates of submission by race and income when scores were held constant.

Not surprisingly, student use of test preparation and other college preparatory services was higher among higher-income students. The proportion of students who took the SAT was relatively consistent across income groups, however, lower income students were much more likely to take the ACT, with twice the proportion of inquirers with incomes under $75,000 taking the ACT (40.3%) compared to the wealthiest inquirers, those with incomes over $150,000, taking this test (20.0%).

In contrast, SAT II tests were more likely to be taken by wealthier inquirers, with over ninety percent (83.6%) of inquirers with incomes over $150,000 taking one or more of these subject tests, while only two-thirds (67.4%) of the sampled population of inquirers took the test.

Nearly all of the inquirers surveyed (99.6%) took the PSAT, the preliminary SAT test that is normally taken in the sophomore or junior year of high school. This is most likely a result of the sampling methodology, which used Mount Holyoke’s inquirer list to sample students, a list largely drawn from mailings to PSAT-takers. However, this pattern holds true for accepted Mount Holyoke applicants as well, with 94.2% of these students taking the test.

Family income appears to be highly related to test preparation activity, at least among Mount Holyoke’s student inquirers. While a little under half (44.4%) of sampled inquirers took a course to prepare for the SAT, nearly two-thirds (63.0%) of inquirers with family income over $150,000 took one, though differences were not observed among admitted students. While one-sixth (16.7%) of sampled inquirers had a private tutor, one-third (34.5%) of the wealthiest inquirers used this service. Similarly, among admitted students, 17.0% of all admits had a tutor or coach, while 32.6% of the wealthiest group had one.

In fact, higher income inquirers were more likely to say they participated in all of the test preparation activities we queried them about, with the wealthiest inquirers being more likely than all inquirers to study on their own (92.6% of those with incomes of $150,000 and up vs. 83.3% of all inquirers) through taking practice tests or working from study books.

Looking at all forms of test preparation involving a course or tutor, lower-income inquirers were again less likely to participate, with more than half (59.5%) of those with family incomes below $50,000 saying they did not participate in test preparation through a course or tutor, and less than one-third (29.1%) of those with incomes over $150,000 doing without these services.

While students’ perceptions of the involvement of their college counselors did not vary by income, with roughly one-third (33.5%) of all inquirers and two-fifths (39.3%) of admitted students saying their counselor was very involved in their college search, there were significant differences in terms of race. Half of African-American inquirers (50.0%) and admitted students (53.1%) and nearly as many Latina inquirers (44.2%) and even more Latina admits (56.5%) report having a college counselor who was very involved.

Despite the active preparation done by the wealthiest group of inquirers, this group was more critical of the test’s value. When asked the open-ended question “In your own words, how do you feel about the SAT overall?,” more high-income inquirers talked about feeling that the test was not a good representation of their knowledge or intelligence, compared to lower income inquirers, with more than forty percent (42.4%) of inquirers with incomes over $150,000 mentioning this concern, and only one-fifth (21.6%) of those with incomes under $75,000 raising the issue. In terms of race, Asian and African-American inquirers were more likely to mention concern about how well the test represents their ability (38.2% and 36.0%) compared to White and Latina inquirers, who mentioned these concerns 29.7% and 23.5% of the time. Conversely, almost one-fifth (19.2%) of lower-income inquirers (under $75,000) volunteered that they thought it was a good assessment, while only 5.1% of the wealthiest group made this statement. Most races were similarly likely to praise the test as an assessment, but African-American inquirers were much less likely than the group as a whole to say this, with only 2.3% making this statement.

Inquirers have a relatively stable perception of their own scores. When controlling for actual scores, few differences were seen by income in terms of student perceptions of their own scores as compared with others at their high school. Asian and Black inquirers, however, were slightly more likely to downgrade their scores in comparison to peers, though even this downgrade was slight.

Next, we looked at inquirers’ perceptions of their own scores. Latina inquirers placed less personal importance on their own SAT scores than did their peers, and Asian accepted students indicated that their SAT scores were more important to them than did other students. Asian and Black inquirers and accepted students reported that their scores were more important to their parents when compared to their peers. Higher income inquirers said they were less concerned about their own scores, while high-middle-income inquirers ($75 to $150K annual income) reported less concern by parents about SAT scores, though the highest income category, those earning over $150K, did not differ from the population as a whole in terms of parent concern about SAT scores. This same high-middle income group ($75 to $150K) reported less concern on the part of their school about scores as well.

Given a hypothetical scenario where their first-choice school was SAT-optional, Latinas and higher-income inquirers were less likely to submit scores when compared to other inquirers with similar scores, while inquirers who took a course, studied on their own for the exam or took the exam multiple times were more likely to say they would submit.

Guidance Counselor Focus Groups

In 2003 and 2004, we conducted focus groups at the National Association of College Admission Counselors conference. In 2003 we conducted two groups, one with private school counselors and one with publics, and in 2004, we conducted one group with private school counselors. Respondents were recruited by Mount Holyoke staff from a list of targeted schools gleaned from NACAC’s member directory.

The discussion guide focused on eliciting counselor opinions about Mount Holyoke’s SAT-optional policy, as well as perspectives on the SAT and college admission in general. The counselors interviewed were very supportive of Mount Holyoke’s decision, did not have any significant reservations about the policy, and were generally highly critical of the SAT. Counselors expressed enthusiasm about SAT-optional policies, but also stressed that one school’s initiation of such a policy would not change the overemphasis counselors felt students place on the test. There was some suspicion about the extent to which the decision might be used to inflate Mount Holyoke’s average SAT scores for ranking purposes, but counselors felt that Mount Holyoke was already holistic in its approach to admission. Some counselors, observed that colleges face pressures not unlike those faced by applicants who must participate in the SAT “game” from another vantage.

The following broad themes emerged:

o Private school counselors spoke candidly about the work they do to prep students for the SAT. At some schools, students are tested regularly so that they become comfortable with the “bubble test” environment, most schools offer and/or pay for test-prep programs, students are coached and advised on testing, retesting, and maximizing scores. One counselor remarked, "Beyond the misuse and over-reliance on the SAT, my concern is not with what the SAT measures but what it doesn't measure. It misses all kinds of traits, habits of mind, dispositions, habits of heart, spirit, excitement. My other concern is what the SAT does to kids in an era when college admissions has been hijacked by the media and there is this feeding frenzy among kids and parents--the messages it gives them about their worth and the limits of their worth and the ways that their value and their potential and their identity is measured by the broader market."

o While reactions to the College Board’s development of a new writing section in the SAT were mixed, counselors mentioned that the writing SAT II was very easy to prepare for and that some students, who were “far beyond the five-paragraph essay” actually needed to “dumb down” their writing in order to get a good score. Some counselors thought that the writing section would simply be a valuable source of a proctored writing sample, and preferred to have the actual writing sample forwarded to colleges unscored, rather than seeing it as an important enhancement to the test.

o Counselors gave Mount Holyoke admissions high marks for consistency and quality in its admission process. Some counselors commented that Mount Holyoke was already known for taking an holistic approach to admission, saying that the SAT-optional policy was consistent with their experience and expectations of the College’s admission process. While counselors praised Mount Holyoke for this approach, they did not feel it was a radical departure from past practice.

o Counselors perceive that both students and colleges are playing a game. There was an interesting response among the private schools when asked whether they were confident that a student admitted under the SAT-optional policy would be successful at Mount Holyoke. One counselor said that she felt that almost any student in her school could do the work at any college in the country; and said she told her students that that was not what the process was about. Most other counselors in the private group agreed with her.

o Counselors wanted reassurance that blocking SAT scores would not disadvantage a student. Some private school counselors critiqued the fact that so many colleges listed as being SAT “optional” in fact had convoluted policies involving substitute requirements like SAT II.

o Some counselors felt that SAT-optional policies were sometimes used to allow a school to accept less-qualified students, but none felt that this was Mount Holyoke’s objective, seeing it as a more selective institution than the ones from which they might suspect such motives. They were more likely to direct their skepticism about Mount Holyoke’s motives to the “rankings game,” but counselors were relatively sanguine about this potential motivation, feeling that the good outweighed the possible harm.

o All counselors agreed that the SAT-optional policy would not affect their decision to recommend Mount Holyoke to students, though some indicated that they might move it from the “reach” category to the “core” category for some potential applicants.

Discussion

The use of SAT scores in college admissions has come under heightened scrutiny of late. This analysis attempted to examine whether selective college admissions could be successfully performed without the requirement of standardized test scores from all applicants. The primary conclusion from this analysis is that selective college admissions can indeed be carried out under an optional SAT score submission policy. It appears from the case study of Mount Holyoke College that some of the potential benefits of the optional SAT policy may have been achieved. We find that the students who withheld their SAT scores and ultimately were admitted and enrolled had a lower average GPA than their peers with comparable admissions ratings but who submitted their SAT scores. The magnitude of this effect is quite small and additionally, the statistical evidence suggests that the college could not have made better predictions of the college GPAs of the non-submitters had their SAT scores been available.

It remains to be seen if these preliminary results remain consistent over time. It may be the case that as applicants become better informed about the optional SAT policy at Mount Holyoke, the percentage and profile of applicants choosing to not submit their scores will change rendering the admissions office task more difficult. The experience at Bates College, however, suggests that this will not be the case. Additionally, the long run success of non-submitters in terms of cumulative GPA, graduation rates, and satisfaction of these students with their educational experience is yet to be determined and warrants future examination. The long run impact of this policy on the composition of the student body and the overall academic quality of the institution is also fertile ground for future analysis.

Evidence from inquirer/applicant surveys and focus groups show positive responses from key constituencies. Inquirers and applicants generally perceive SAT-optional policies as neutral or positive, and college counselors see the move as holistic and supportive of academic preparation over test preparation. These results provide a clear answer to concerns that an SAT-optional policy might cause the College to lose prestige: in fact, the response has been positive both in our research and anecdotally.

While this analysis suggests that there are both benefits and costs to an institution of pursuing an optional SAT admissions policy these results may not be universally applicable. Mount Holyoke is a small liberal arts college with fewer than 3000 applications. The time and resources it is able to devote to each individual application may not be available at larger institutions with many more applications. Additionally, if more institutions follow this admissions approach the impact on the admissions environment would appear to be minimal. The yield on non-submitters, conditional on admissions rating, is not significantly different from the yield on submitters suggesting that Mount Holyoke College does not appear to be treating these individuals substantially differently than most other comparable institutions. Should more institutions pursue this policy it is not apparent that the admissions and enrollment decisions would be dramatically different.

Bibliography

Bradley, Drake R., Committee on Admissions and Financial Aid. 1990. “The Optional SAT Policy at Bates: A Final Report.” February. Bates College, Lewiston, Maine.

Brownstein, Andrew. “Colleges Debate Whether Dropping the SAT Makes Them More Competitive.” 2001. Chronicle of Higher Education. October 26.

Clotfelter, Charles; Vigdor, Jacob. 2001. “Retaking the SAT.” Sanford Institute of Public Policy Working Paper SAN01-20. July 2001.

Ehrenberg, Ronald G. 2001. “Reaching for the Brass Ring: How the U.S. News and World Report Rankings Shape the Competitive Environment in U.S. Higher Education.” May. Cornell Higher Education Working Paper 17.

Hiss, William. 2001. “Optional Sat’s and Bates: Seventeen Years and Counting.” Chronicle of Higher Education. October 26.

McCarty, Barry. 2001. “Lafayette’s Comfort Level is Higher With SAT’s.” Chronicle of Higher Education. October 26.

Meng, Chun-Lo; Schmidt, Peter. 1985. “On the Cost of Partial Observability in the Bivariate Probit Model.” International Economic Review. vol. 26 (1). p 71-85. February.

Monks, James; Robinson, Michael D. ”Making the SAT Optional at a Selective College. “ Economics of Education Review, Forthcoming.

Yablon, Marcia. 2001. “Test Flight: The Scam Behind SAT Bashing.” New Republic. October 30. pp. 24-25.

Tables

Table 1

Percent Non-Submitters 2001-2004

| | | | | |

|Variable |Coefficient |T-Statitstic |Coefficientnt |T-Statitstic |

|Intercept |3.652 |13.63 |4.085 |13.16 |

|2002 |-0.289 |-3.48 |0.145 |2.23 |

|2003 |0.101 |1.69 |0.089 |1.41 |

|2004 |0.318 |5.29 | | |

|Math SAT |-0.005 |-18.00 |-0.006 |-13.75 |

|Verbal SAT |-0.004 |-14.00 |-0.004 |-10.75 |

|High School GPA |0.236 |4.55 |0.133 |2.22 |

|Family Contribution |0.004 |2.64 |0.004 |1.80 |

|Legacy |0.195 |1.85 |0.202 |1.64 |

|Early Decision |0.204 |2.62 |0.385 |4.67 |

|Black |-0.251 |-2.71 |-0.239 |-2.39 |

|Hispanic |-0.244 |-2.44 |-0.033 |-0.30 |

|Asian |-0.024 |-0.32 |-0.093 |-1.11 |

|Native American |-0.052 |-0.22 |-0.236 |-0.92 |

|Race Unknown |0.133 |1.58 |0.167 |1.80 |

|International |-0.082 |-0.76 |0.252 |1.78 |

|N |5841 | |3469 | |

|Log -l |-1861.43 | |-1515.51 | |

Model also includes variables for region not reported.

Math and Verbal SAT scores are high score. In Model 1 the scores are from what Mount Holyoke was able to obtain from applicant records. In Model 2 the scores are from the College Board. Omitted race category is white.

Table 3

Probit Model for Acceptance to Mount Holyoke College.

| |Model 1 | |Model 2 2 | |

| |Coefficient |T-Statitstic |Coefficientnt |T-Statitstic |

|Intercept |-9.01 |-21.05 |-4.89 |-25.59 |

|Year 2002 |1.59 |16.37 |1.51 |19.91 |

|Year 2003 |0.27 |4.12 |0.16 |3.48 |

|Year 2004 |na |na |0.76 |14.74 |

|Non-Submitter |0.23 |3.07 |-0.12 |-2.79 |

|High School GPA |1.42 |17.85 |1.22 |27.43 |

|Math SAT |0.00 |6.40 | | |

|Verbal SAT |0.00 |7.00 | | |

Model also includes regional dummy variables, family contribution, race dummy variables and legacy status.

Bold represents significance at the 95 percent level.

Table 4

Regression Model for Admission Rating

| |Model 1 | |Model 2 2 | |

| |Coefficient |T-Statitstic |Coefficientnt |T-Statitstic |

|Intercept |10.63 |87.16 |14.74 |76.23 |

|Year 2002 |-0.76 |-19.54 |-0.68 |-17.32 |

|Year 2003 |-0.43 |-12.29 |-0.38 |-9.89 |

|Year 2004 |-0.36 |-9.94 |na |na |

|Non-Submitter |0.29 |9.54 |-0.26 |-6.31 |

|High school GPA |-1.89 |-67.85 |-1.60 |-43.35 |

|Math SAT | | |-0.01 |-19.10 |

|Verbal SAT | | |-0.01 |-16.78 |

Model also includes regional dummy variables, family contribution, race dummy variables and legacy status.

Bold represents significance at the 95 percent level.

Table 5: Difference between Admission Rating when Score was Blocked vs. Unblocked

(Blocked Rating minus unblocked Rating)

Rating Scale of 1-8, 1 is best possible score, 8 Includes all unadmissable candidates

|Difference: Blocked |Frequency |Percent |

|vs. Unblocked | | |

| |-2.00 |4 |4.0% |

| |-1.00 |20 |20.0% |

| |.00 |65 |65.0% |

| |1.00 |9 |9.0% |

| |2.00 |2 |2.0% |

| |Total |100 |100.0% |

Table 6

Mount Holyoke First and Second Year GPA

For Submitters and Non-Submitters

| |Submitters |N |Non-Submitters N | |Difference |

|Total |3.36 |1137 |3.27 |443 |* |0.09 |

| | | | | | | |

|Admission Rating| | | | | | |

|1 |3.84 |45 |3.73 |9 | |0.11 |

|2 |3.70 |128 |3.64 |39 | |0.06 |

|3 |3.51 |319 |3.44 |110 | |0.07 |

|4 |3.33 |359 |3.24 |142 | |0.09 |

|5 |3.10 |206 |3.07 |116 | |0.03 |

|6 |2.87 |76 |2.90 |25 | |-0.03 |

|7 |2.00 |3 |2.78 |2 | |-0.78 |

| | | | | | | |

|By SAT Scores | | | | | |

|1300-1600 |3.54 | |3.70 | | |-0.16 |

| | |135 | |10 | | |

|1200-1300 |3.37 | |3.46 | | |-0.09 |

| | |119 | |30 | | |

|1100-1200 |3.24 | |3.38 | | |-0.16 |

| | |62 | |47 | | |

|1000-1100 |2.89 | |3.13 | |* |-0.24 |

| | |27 | |55 | | |

|900-1000 |2.88 |2 |3.06 |32 | |-0.18 |

|Less than 900 |2.84 | |3.08 | | |-0.24 |

| | |4 | |9 | | |

Table 7

Mount Holyoke GPA Regressions

| |Model 1 | | |Model 2 | | |

| |Coef. | |T-Stat. |Coef. | |T-Stat |

|Intercept |13.39 | |0.21 |-170.00 |* |-2.52 |

|Non-Submitter |-0.07 |* |-2.19 |0.06 | |1.68 |

|Rating 1 |1.01 |* |10.84 | | | |

|Rating 2 |0.93 |* |14.04 | | | |

|Rating 3 |0.71 |* |12.71 | | | |

|Rating 4 |0.52 |* |9.41 | | | |

|Rating 5 |0.37 |* |6.34 | | | |

|Math SAT/100 | | |0.14 |* |6.52 |

|Verbal SAT/100 | | |0.13 |* |6.37 |

|Sophomore |0.13 |* |3.74 |0.13 |* |3.37 |

|Year |-0.01 | |-0.16 |0.09 |* |2.55 |

| | | | | | | |

|N |1041 | | |1041 | | |

Table 8

GPA Regressions for Submitters and Non-submitters.

| |Non-Submitters | | |Submitters | | |

| |Model 1 | | |Model 2 | | |

| |Coef. | |T-Stat. |Coef. | |T-Stat |

|Intercept |-210.12 |* |8.87 |52.00 |* |52.00 |

|Rating 1 |0.59 |* |2.21 |0.89 |* |0.89 |

|Rating 2 |0.64 |* |3.61 |0.85 |* |0.85 |

|Rating 3 |0.52 |* |3.70 |0.66 |* |0.66 |

|Rating 4 |0.30 |* |2.22 |0.52 |* |0.52 |

|Rating 5 |0.24 | |1.66 |0.34 |* |0.34 |

|Total SAT/100 |0.04 | |1.74 |0.07 |* |4.42 |

|Year Enrolled |0.11 | |0.18 |-0.02 | |-0.72 |

|Sophomore |0.20 | |0.03 |0.10 |* |2.73 |

|N |275 | | |776 | | |

|R-Squared |13.6 | | |30.9 | | |

Omitted Rating 6.

Table 9

GPA Regressions for Submitters and Non-submitters.

SAT Scores only

| |Non-Submitters | | |Submitters | | |

| |Coef. | |T-Stat. |Coef. | |T-Stat. |

|Intercept |2.38 |* |9.03 |1.50 |* |9.54 |

|Combined SAT/100 |0.08 |* |3.28 |0.15 |* |11.94 |

| | | | | | | |

|R-Squared |3.8 | | |15.5 | | |

|N |275 | | |776 | | |

Table 10

Yield by Submission Status

| |Non-Submitters |Submitters |

|1 |16.0 |15.5 |

|2 |36.2 |19.2* |

|3 |30.6 |23.7* |

|4 |30.3 |29.3 |

|5 |40.3 |34.1 |

|6 |53.2 |68.9* |

* Significant Difference at the 95 percent level.

-----------------------

[1] Non-submitters are defined as applicants that did not submit either an ACT or SAT score for consideration by the college.

................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download