College Admissions Process Explained

Applying to College
1

<p>The following is from this week's Chronicle of Higher Education:</p>

<p>The Power and Peril of Admissions Data</p>

<p>By ELIZABETH F. FARRELL</p>

<p>Inside Baylor University's admissions office, almost anything is quantifiable. When a high-school student calls, an admissions representative ranks the student as soon as he or she hangs up the telephone. Callers who happen to mention that Baylor is their top choice get a 1 (on a five-point scale). If they seem likely to apply, they receive a 2. Those who are not sure about Baylor, but are somewhat interested? They get a 3.</p>

<p>The university records and rates every bit of correspondence with each prospective applicant. This fall even student tour guides have begun to assign scores to high-school students who give any hints of interest in Baylor.</p>

<p>"We want to know if they mentioned something," says James Steen, Baylor's assistant vice president for admission and enrollment services, "like the fact that they're a third-generation legacy and they're not going to apply anywhere else, or if they were not that interested and just happened to stop by Baylor on the drive home after visiting UT."</p>

<p>Mr. Steen's staff enters this information into a database called Bearhaus, which allows admissions officers to keep track of all their communications with students. The database uses statistical models to interpret the numbers, assigning each student an overall score that tells Baylor how likely he or she is to apply, and later, to enroll.</p>

<p>The models also cross-reference demographics and information about a student's academic profile in assigning scores. That way, students who are highly sought after can be identified, sometimes so the college can recruit them more aggressively.</p>

<p>That score helps Mr. Steen and his staff make crucial marketing decisions. Potential applicants with the highest scores receive glossy, full-color brochures. Those who seem less interested may get only an e-mail message or a black-and-white postcard. With limited marketing dollars, the object is to keep the interest of those students who already have Baylor on their radar screens.</p>

<p>As colleges become more like businesses, with strict bottom lines, quantifying a potential applicant's interest in a particular institution has evolved into a science. Almost all colleges are scrambling more than ever before to meet net tuition-revenue goals, increase the geographical and racial diversity of their applicants, and lower their acceptance rates by attracting more applicants.</p>

<p>Until about five years ago, only a handful of colleges used sophisticated statistical formulas. But now there are few four-year institutions that do not.</p>

<p>Many admissions officials now rely on databases like Bearhaus to help them recruit and enroll each new crop of students. The use of statistics is vital in admissions, but it forces officials to walk a fine line between meeting quantifiable objectives and preserving their commitment to judging each student on his or her individual merits, which numbers do not always capture.</p>

<p>"We are the office of revenue and reputation," says Daniel M. Lundquist, vice president for admissions, financial aid, and communications at Union College, in Schenectady, N.Y. "When I started doing this, you could do it with a baseball bat. Now you need forceps. ... The stakes are astronomically higher."</p>

<p>Statistical models like Baylor's work by using the behavior of past applicants to predict how future ones will act. Those data are supposed to help admissions officials meet their enrollment goals, yet many officials say the scramble to collect more and more numbers has placed them on a hamster wheel that never stops spinning.</p>

<p>Furthermore, the same reams of data many colleges collect to simplify the admissions process may also skew their enrollment predictions. After all, if the formulas help colleges market more effectively to prospective applicants, officials will have more applications to wade through. The more applications a college receives, the harder it is to predict which students really want to go there, no matter how many statistics an admissions office has on each applicant.</p>

<p>"When will we say enough is enough and get back to good, old-fashioned admissions practices?" asks Monica Inzer, dean of admission and financial aid at Hamilton College, in Clinton, N.Y. "The transcript and what we value in an applicant should be what matters most, not whether or not a student visited campus."</p>

<p>'Funnels' of Students</p>

<p>As recently as 10 years ago, colleges tracked only a few statistics, including the grades, test scores, and geographic locations of prospective applicants. They would wait until students applied, then plug that information into a simple statistical model to get a reasonably close estimate of how many of their admitted students would enroll. Closely tracking data on high-school sophomores and juniors before they applied was unheard of at all but a few institutions.</p>

<p>During the last five years, however, the way many students apply to colleges has changed drastically, making the models most institutions use much less reliable than before. The number of high-school students applying to college is the largest in history, and those students are applying to more colleges than their predecessors did. And some admissions officers now struggle to predict their enrollment numbers because more students are acting as "secret shoppers," giving no hint of their interest in a college until they submit an application.</p>

<p>The less admissions officers know in advance about their final enrollment numbers, the more challenges they are likely to encounter. If they admit too many students, faculty members and residence-life officials will complain about strained resources. And if fewer students than expected matriculate, trustees and administrators will bemoan the lack of tuition revenue.</p>

<p>Those pressures have led colleges to identify the students they want to pursue long before colleges receive the students' applications.</p>

<p>Various organizations, including the College Board, ACT, and the National Research Center for College and University Admissions, compile vast databases with information on the demographics and credentials of millions of high-school students: their test scores, grade-point averages, intended majors, ZIP codes, extracurricular activities, and colleges in which they have expressed an interest.</p>

<p>Each criterion is known as a "funnel" that helps colleges customize the list of students they wish to recruit.</p>

<p>Don Munce, president of the admissions-research center, says colleges are buying more and more names each year. At large public universities, those lists can have between 50,000 and 150,000 names, and private universities on average purchase lists of about 20,000 names, according to Kevin W. Crockett, president and chief executive of Noel-Levitz, a higher-education consulting firm that advises more than 1,000 colleges a year on enrollment-management practices.</p>

<p>In the past, most admissions officers were more discriminating at the beginning of the recruitment process than they are today. Before the rise of the Internet, when colleges could send promotional materials only by traditional mail, postage costs limited the volume of correspondence. Purchasing 100,000 names in one year was unheard of because very few institutions could afford to send so many glossy brochures.</p>

<p>Even as e-mail and the Internet have lowered the cost of casting a wide net for potential students, more-sophisticated databases and formulas have made marketing efforts more efficient. Having 100,000 names of potential applicants is useless unless admissions staffs know which ones are worth the $3 cost of producing and mailing a viewbook.</p>

<p>The numbers are especially useful at large public universities with shrinking budgets and growing pressure to recruit out-of-state students, according to Rick Burnette, director of student-information management at Florida State University. Mr. Burnette's charge is to figure out which statistics the university should track, and how to interpret them. That Florida State created his position two years ago despite overall cutbacks affirms the high value the university places on evaluating admissions data, he says.</p>

<p>"We're a growing institution with more applicants, and we don't have a growing staff," says Mr. Burnette. "We have to do a lot more with the same resources we've always had, and it's incumbent on us to be as efficient as possible and leverage technology and information in our communications with students."</p>

<p>By tracking data on prospective students, Mr. Burnette has found which towns and high schools in Florida tend to produce the highest proportion of the university's applicants. So he has stepped up marketing to students in those areas. He has also increased the amount of marketing the college does in areas with many prospective first-generation college students, to help Florida State diversify its applicant pool.</p>

<p>At Wilkes University, in Wilkes-Barre, Pa., a similar statistical analysis led Mike Frantz, vice president for enrollment and marketing, to discover that students who listed skiing as one of their hobbies were more likely than others to attend Wilkes, even though the college does not have a ski team.</p>

<p>To attract more of those potential students, he changed the college viewbook five years ago to include a picture of a student skiing. He also added wording that described Wilkes as located "in the foothills of the Poconos," a mountainous region of northeast Pennsylvania known for its ski resorts.</p>

2

<p>ARTICLE CONTINUED:</p>

<p>Signs of Success</p>

<p>Do such tactics help colleges attract more students?</p>

<p>Since Mr. Frantz began using statistical modeling four years ago to identify more prospective applicants, he says he has discovered about 380 students each year who he would not have found otherwise. Each year, about 90 of those students have enrolled at the college.</p>

<p>Wilkes's statistical model identified those students by finding a pattern in their academic and demographic characteristics that matched that of the typical Wilkes applicant. None of the students had previously expressed an interest in the college, and Mr. Frantz would have assumed they were a lost cause if the numbers had not told him differently.</p>

<p>"Some of those students would have obviously inquired about our college on their own at some point," says Mr. Frantz. "But when you look at the data and see those numbers, you don't want to not use it and risk the possibility of losing some of those leads."</p>

<p>Baylor also found success using similar statistical tools. In 2001, Mr. Steen's office replaced its basic database with a more sophisticated one. The same year, he also starting using several predictive models.</p>

<p>Every year since then, his office has added more predictive models to the mix, and it now uses a total of 15. During that time, the annual number of applicants has grown, and the university's admission rate has declined. Five years ago, Baylor had 2,620 applicants and a 79 percent acceptance rate. Last year it had 4,000 applicants and accepted only 42 percent of them.</p>

<p>Although statistical models can increase a college's applications, the data those models provide can be biased. Mr. Munce, of the admissions-research center, says colleges that rely strictly on statistics to devise their recruiting strategies can inadvertently exclude underrepresented groups of applicants. Mr. Munce discovered that problem last month when he completed a study of the criteria 300 private colleges use to select names from his organization's database.</p>

<p>Although those colleges sought to increase their enrollment of minority students, the factors they used to find prospective applicants — including ZIP codes and students' stated career interests — had the opposite effect. Many of the institutions he studied were categorically excluding students from inner cities, even if they had 4.0 grade-point averages. The criteria the institutions were using generated recruitment lists on which minority students were included with 30 percent less frequency than white students.</p>

<p>Mr. Munce has been advertising those findings, hoping that his clients will more carefully select the types of data they use to weed out prospective students. He cannot, however, regulate how colleges use the lists of names they purchase.</p>

<p>"We give them the data they ask for because they are ultimately the customer," says Mr. Munce. "But they are often using criteria that are too simplistic. They try to find kids who meet certain criteria without considering it from a race standpoint, and they don't use common sense."</p>

<p>'Mass Customization'</p>

<p>After drawing up their initial list of thousands of names, admissions officials begin to track who responds to their materials, how they respond, and when they respond. Eventually, colleges separate students into smaller, more stratified groups. Officials assign high scores to prospective applicants who respond early and enthusiastically, signifying that such students are likely to apply and enroll. Just because certain students do not respond to a mailing does not mean that an admissions office will stop pursuing them, however.</p>

<p>In many cases, colleges aggressively pursue students with high test scores or good grades. Those students may receive telephone calls from professors in academic departments in which they have indicated an interest. Or a college might send them a T-shirt.</p>

<p>Such gestures are the result of strict calculations, but are supposed to give students the impression that the college cares about them personally. This practice is known in the industry as "mass customization."</p>

<p>As data crunching becomes more widespread, many admissions officials wonder if the practice will lose its effectiveness. If every college sends more personalized mailings and makes more personal telephone calls, will they cancel out each other's efforts?</p>

<p>According to Mr. Munce's figures, any high-school student with an average of A-minus or better usually receives 200 to 300 mailings from colleges after taking the SAT or the ACT.</p>

<p>"As we all raise the bar on how we market our colleges and use more of the data that's available, how are we actually differentiating ourselves?" asks Ms. Inzer, at Hamilton. "We reach the point of diminishing returns because even if these approaches work, the end result is that students apply to more colleges, and it gets harder to predict what they're ultimately going to decide."</p>

<p>Colleges that refrain from data mining are generally small private institutions that have had consistent numbers of applicants and students who accept offers of admission, such as Pomona College, in Claremont, Calif.</p>

<p>"The only calculations I do are in my head," says Bruce J. Poch, vice president and dean of admissions at Pomona. "People spend thousands to have companies tell them how to interpret their data, and they break the numbers down so finely that it doesn't mean anything."</p>

<p>Yet many of Mr. Poch's counterparts do not seem willing to stop collecting and using data on students. In fact, the more data colleges track, the more they feel they need.</p>

<p>Many colleges are reaching students even earlier in their high-school careers, according to Mr. Crockett, at Noel-Levitz. Admissions offices used to send detailed information about academic programs and campus life only to seniors. Now high-school juniors are getting brochures on course offerings in their intended major. Sophomores, who used to receive little more than a generic postcard, now get viewbooks and personalized letters.</p>

<p>Even if a college were to determine that its mailings to sophomores were ineffective, it probably would not stop the practice. After all, its competitors are doing it, too.</p>

<p>"In that type of situation, you can't say, We don't want to bombard them because everyone else is doing it," says Mr. Steen, at Baylor. "It's quite the opposite. We have to be out there in front of these students as much if not more because they are so highly recruited."</p>

<p>The Final Equation</p>

<p>After all the prospective students have been funneled into their various categories and have submitted their applications, colleges have the closest possible approximation of the number of students who are definitely interested in their institution. At this point, the data is at its most refined, and the statistical models used to interpret it are most sophisticated.</p>

<p>These final numbers are the ones colleges are most likely to use unscrupulously.</p>

<p>Mr. Crockett says it is unethical for a college to use the data on phone calls, e-mail messages, and campus visits to make admissions decisions. Some students may not know that such information is collected, and others may not have the wherewithal to make campus visits, for example. Instead, he says, admissions officials should read each application carefully and judge the candidates on their individual merits. Only after officials have decided whether to admit or deny each student should they turn to the models to see what adjustments they need to make to the number and type of students they have accepted.</p>

<p>At Union College, Mr. Lundquist, the director of admissions, says he does just that. After a human being has read each application and debated its relative merit with the admissions committee, the predictive model's final calculations of such factors as who will accept an offer, or how much an applicant will contribute to tuition revenue, may determine whether the college accepts or rejects a particular student.</p>

<p>"It's like balls running through a pinball machine," says Mr. Lundquist. "We put in our picks and run the model to see what the head count looks like, and if it spits out a number that's higher than our financial-aid target, or has too many engineers, or whatever, we have to go back and take some people out and put others in."</p>

<p>In recent years, Mr. Lundquist has had to make some difficult decisions. He has, for instance, told his application readers to pull out names of financially needy students that they had planned to accept. Why? The statistical model had predicted that Union's net tuition revenue would be lower than expected given the students he had chosen to admit.</p>

<p>Although no admissions official interviewed for this article would admit to using statistical models to make admissions decisions, one who asked not to be identified said he was fired twice for refusing to employ such criteria. The official, who has worked for two highly selective colleges, is now employed at a less competitive one.</p>

<p>"The pressure from the rankings guides and the need to meet tuition-revenue targets drives a lot of manipulation of data in ways that are very ethically questionable," he says. "If colleges have information that tells them how to tweak their admissions numbers to accept more students who enroll and can pay the full price, it's very hard to stop them from using it."</p>

<p>Other officials say that while they cannot name any institutions that use such tactics, they believe the practice is ubiquitous. That many admissions officials assume the worst of their competitors reveals the cynicism in the field. And as that cynicism grows, so does the appetite for numbers.</p>

3

<p>ARTICLE CONTINUED - 2:</p>

<p>THE SCIENCE OF RECRUITING</p>

<p>With 1.8 million high-school students applying to college every year, colleges must decide which are the most sensible targets for marketing efforts. Institutions use various funnels to narrow down the number of students they encourage to apply. Here is one scenario: </p>

<p>About 10 million names of high-school students, along with data about their test scores, demographics, and preferred college type, are available for purchase from various companies. </p>

<p>Initial list of possible applicants: 100,000 to 250,000 </p>

<p>Admissions officers pick criteria of desired students, and the companies make a list of possible applicants. Typically, a large private university will purchase anywhere from 100,000 to 250,000 student names. </p>

<p>The admissions office will divide those names into two groups--"high priority" students (25% of the names) with the best test scores and grades, and "bread and butter" students, (the remaining 75%), with average test scores and grades--and send mailings to them. The final list of possible applicants may be as many as 100,000 names. </p>

<p>After sending mailings, the college sees who responds. Students who were not on the mailing lists will also contact the institution. Together, those names comprise the list of possible applicants. </p>

<p>The college tries to figure out how many applications it will have. Potential applicants can be broken up into dozens of different groups, but those groups fall into one of three categories: </p>

<p>Likely to apply
May apply
Unlikely to apply
Applicant pool: About 10,000 to 15,000 students (about 10 to 15 percent of the final list of possible applicants) will end up applying to the average large private institution. Of those, about half will be admitted.</p>

4

<p>interesting article! I think the phenomenom may be self-limiting, though. How many times have we seen an action by government or business that treats the decisions made by people as if they were hardwired and that people don't take into account new information? Or, as they say in econ, people respond to incentives.</p>

<p>So once it became known that having a few ECs enhanced the ability to get into a top college, everyone had a few and then the competition heated up to make them more and more prestigious. If you have a few gray hairs you can remember when just belonging to a club or two was enough for a top college; these days you need awards and leadership, on a city level as a minimum.</p>

<p>The same thing is going to happen to these admission models. As people realize what's going on (exactly what your posting leads to) then people start to respond. Who wouldn't read this article and then gush over Baylor when they take the tour? Who won't respond to mailings if they suspect colleges care about it? And so indicators that once carried significance will become nothing more than markers separating the sophisticated from those less so.</p>

5

<p>I would respond to each of ur comments if you guys hadn't spent hours upon hours deciding to write a 2,000 word essay for a post. I could've walked to Mexico and gotten back before you guys finished typing your posts...ok, hyperbole, but sorry for wasting your times making you read my stupid post.</p>