Colleges For Coeducation

get lesser-quality students, or it can switch to coeducational status, have reduced giving among prior alumni, but gain better (and possibly more) students.37 Intelligent administrators will discount the two streams and choose the optimal switching date when long-term gains from switching just begin to outweigh short-term losses. One of the model’s predictions in the face of secularly rising student demand for coeducation is that current student quality will decline (particularly relative to competing coeducational institutions) before a single-sex school switches, and possibly long before. In addition, after the switch, alumni donations will initially decline relative to what they would have been in the absence of the switch to coeducation. Donations will eventually improve as the composition of the alumni shifts toward newer cohorts who attended in the coeducation regime. Nevertheless, under the optimal policy, an institution’s president will choose to “bite the bullet” and make the switch before the net financial gains from coeducation become apparent. The trade-offs suggested by the model are borne out in the histories of single-sex schools. According to Karabel (2005, chap. 14), in 1956, Yale’s admissions officer observed that many of the school’s best admits had chosen coeducational institutions and that student quality at Yale was declining. But Yale alumni were strongly resistant. By the mid-1960s, current students at Yale were demanding a switch to coeducation. At Princeton, 55 percent of alumni polled in 1969 were opposed to co- education, but the trustees supported the switch to stem declining qual- ity.38 Similar changes occurred at the other all-male institutions in the 1960s and 1970s. It should be emphasized that antidiscrimination legislation did not play a quantitatively important role in the switch to coeducation. Title
37 The switch to coeducation is assumed in the model to be irrevocable, and, in fact, almost all were. 38 A dissent, filed by director of development Arthur J. Horton ’42, noted: “I fear that there will be alumni who, liking the University as an all-male institution, could lose much of their present ardor” and reduce their alumni giving. He asked: “Can we really argue that we are not getting the best applicants when over 46% of our senior class graduated last June with Honors?” The Patterson Committee advocated the switch over member Horton’s lone dissent (Horton Dissent, August 28, 1968, available in the Arthur J. Horton Collection on Coeducation, Seeley G. Mudd Manuscript Library, Princeton University, Princeton, NJ, box no. 4, folder no. 5).
400 Journal of Human Capital
IX of the Education Amendments of 1972 was passed and its imple- mentation written long after most male-only institutions had decided to become coeducational.39 There were, to be sure, other reasons for switching, such as institution- specific fiscal shocks, depletion of potential students due to wartime draft and fatalities, and general economic downturns, the effects of which could be smoothed by a sufficiently large endowment. Public institutions often face different constraints than private colleges because pressures exist for public dollars to serve all people.40 State legislatures often forced coeducation on an institution to save expenses on building a separate women’s college. We first examine the institutional characteristics associated with faster or slower transitions of single-sex 4-year colleges to coeducational in- stitutions. We estimate formal hazard models of the duration spent as a single-sex school for all colleges starting as a single-sex school in our 1934 and 1980 Coeducation College Databases (including all originally single-sex schools present in 1897, 1924, 1934, and/or 1980). The es- timation sample consists of 511 schools, of which 281 started as men’s colleges and 230 started as women’s colleges. Cox proportional hazard models are estimated for the duration of a spell as a single-sex school using a nonparametric (fully flexible) base- line hazard.41 The time at risk for becoming a coeducational institution is assumed to begin in 1835, the year in which Marietta College opened and a year after Oberlin College began coeducational classes. In 1837, Oberlin was the first to accept female students into a BA-granting pro- gram and to switch from a single-sex to a coeducational institution.42 Thus, the time at risk begins in 1835 for schools founded before 1835 and at the actual opening date for schools founded in 1835 or after. A “failure event” is a transition to being a coeducational school. Schools remaining single-sex institutions today (three male-only and 35 female- only in our data set) are treated as censored spells, with 2010 as the
39 The switch of most institutions to coeducation also preceded the interpretation of Title IX in Mississippi University for Women v. Hogan, 102 Supreme Court Reporter 3331 (1982), that private single-sex undergraduate institutions can be exempt from the ad- missions requirements of Title IX but must comply with constitutional equal protection requirements in admissions. 40 As M. Carey Thomas, then president of Bryn Mawr College, noted around 1900, “public opinion in the United States almost universallydemandsthatuniversitiessupported by public taxation should provide for the college education of women” (1900, 358b). 41 The findings are similar to those from standard parametric models such as with a Weibull hazard. 42 Fletcher (1943) claims that, in 1834, when the first college classes began at Oberlin, “[male] college students shared their classrooms and class instructions with women” and also that “in 1837 four ladies were admitted to the Collegiate Course [at Oberlin] with the men and in 1841 three of them received the A.B. degree, the first bona fide college degrees ever granted to women” (379–80). Marietta College began in 1835 as a coedu- cational institution but may not have granted a BA degree until later.
College Coeducation from 1835 to the Present 401

Note.—The sample includes all 4-year institutions starting as single-sex institutions in our 1934 and 1980 Coeducation College Databases. In other words, the sample consists of all originally single-sex schools present in our 1897, 1924, 1934, or 1980 institutional samples. The estimation uses Cox proportional hazard models for the duration of a spell as a single- sex school with nonparametric baseline hazards estimated via maximum likelihood using the stcox command in STATA. The time period at risk in the duration models begins in the year of opening for institutions founded after 1835 and in 1835 for institutions that opened before 1835. The failure event is the transition to a coeducational institution. Schools that continue today as single-sex institutions are treated as censored spells, with 2010 as the date of censoring. Schools that closed as single-sex institutions are treated as censored at the date of closing. The reported coefficients are hazard ratios. The standard errors for the hazard ratios are in parentheses. The base region is the Northeast (New England plus the Middle Atlantic states). There are no land grant and no technical institutions among the women’s colleges.
censoring date; schools that closed as single-sex schools are treated as spells censored at the date of closing.43 The basic hazard models for all single-sex schools and men’s and women’s colleges separately are presented in table 5. The models in- clude as the covariates time-invariant institutional characteristics, in- cluding a continuous measure of the year of opening and indicator variables for private control (vs. public), religious affiliation, and region. The year of opening indicates the strength of alumni resistance to a switch. The religious affiliation shows the particular ideology and, in the case of Catholic institutions, the degree to which their decisions are dictated by a higher authority and are therefore coordinated. The re- ported coefficients are hazard ratios. (A coefficient greater than one indicates that a variable increases the hazard rate of being coeduca- tional; a coefficient less than one implies that it shrinks the hazard rate.)
43 We know of only one school, Wesleyan College, that began male only, switched to coeducational status, returned to male only, and later became coeducational (Potts 1992). Colby College began coeducational, created a coordinate women’s institution, and later returned to being coeducational