A Survey of Undergraduate Research Over the Past ~ e c a d e J a m e s N. S p e n c e r Lebanon Valley College. Annville. PA 17003 and Claude H. Yoder Franklin and Marshall College. Lancaster. PA 17604 Research has heen an important component of faculty activity in most universities for many decades. More recently, however, in institutions without graduate programs, particularly liberal arts colleges, scholarly research has begun to emerge as an important sector of faculty activity. The last comprehensive survey of publications from liberal arts colleges done in 1962concluded that "liberal arts colleges are entering the field of chemical publicationsat a more explosive rate than previous performance would indicate" ( I ) . However, research is expensive, hoth in faculty time and equipment and, consequently, a t some undergraduate institutions research is not encouraged and is often absent ( 2 ) . The Wooster Conference on teaching and research (sponsored by the American Chemical Society and supported by the National Science Foundation) in 1959 divided liberal arts colleges into 4 categories-very productive, productive, borderline, and unproductive-on the basis of the number of chemistry graduates that had earned the PhD in chemistry. The responses to questionnairessent to selected institutions indicated that the more productive colleges believed that "original research by the faculty contributes to the college by keeping the faculty up to date scientifically, by bringing the college favorable publicity and by the acquisition of equipment which would not he purchased hy the college. The less ~roductivecolleees were of the opinion that research takes uo iime needed for ;reparation. . . . A I I classifications agreed that 'research itnoroves teachina.'" All of the particioants in the conference aireed that the p&onal contack between students and faculty stimulate interest in chemistry and a desire to pursue graduate work (3).As pointed out by Mattill. "College and university research is an investment that returns double dividends. While the research results contribute to fundamental knowledge, the project3 in progress strengthen the teaching prngrams and promise more and better qualified scientific manpower for the future" (4). Perhaps the mcst important aspect of research a t the college level is the involvement of undergraduates. Independent study ~ r o v i d e sa review of concept3 learned earlier, an increased repertoire of laboratory techniques, a mastery of some narrow area of a discipline, and an understandina of what it means to hen scientisr. More importantly, it providesan opportunity for creativity and original contr~hutions11, scientific knowledee. A studv hv Chambers (5) . . supeests that research a t the un'dergradua"teSlevel may he a significant stimulant to the potentially creative person. An equally vital aspect of the research experience is the insight into the student's own personality and abilities that it provides: the long hours, the failures and frustrations followed by successful collection of data, the discovery of new patterns, and the final incorporation of these into the thesis may provide the footholds for the climb to higher levels of maturity. The effect of active research on the faculty should not he minimized, however. It provides opportunities for scientific and professional growth, often results in increased vitality and enthusiasm in the classroom and lahoratory, and generally
....
780
Journal of Chemical Education
improves faculty morale. Funding for the research through grants improves the faculty's financial status and brings new facilities into the department. Measures of Research and lnstltutlonal Characteristics While research in the undergraduate environment can take a varietv of forms and levels of soohistication, the participants in a recknt undergraduate reseaich ronferences~onsor~d by Research Corporation concluded "that in order for facultv and students tosustain interest in research they must be involved in real prohlems-problems with the potential for advancing science and able to withstand the scrutinv of the investieator's peers. A good problem would, therefore, appear to be one which has as one of its " eoals a ouhlication which will he of interest to the scientific community" ( 6 ) .Thus, appropriate measures of research activitv a t the undereraduate level for a particular institution woulh seem to be n&ber of publications in chemistw, number of facultv authors, and the number and total dollar amount of grants received for research. The Wooster Conference as well as Mattill's study suggest that "some kind of relation exists between a college&tu&on which produces a relatively large number of scientists and which encourages staff and students to do research" (3). Consequently, a characteristic of the undergraduate institutions used in the oresent studv was the numher of PhDs in chemistry produced by the respective instirutiuns. Undereraduate research mirht also be exoected to imoact on student success in obtaining national awards. ~ h e r e i o r e , the number of NSF Fellows was also incorporated in the present study. Many factors may impinge on undergraduate research. For this study the following factors were examined: abilities of students, as measured by SAT scores; number of chemistry faculty and teaching loads; faculty compensation; college enrollment; numher of majors: number of research students; and number of journals in chemistry. The institutions included in this study are four-year, private colleaes which do not have masters or doctoral proarams in chedistry and are defined by the American ~ ~ s o e i a t i oofn University Professors as IIA or IIB institutions in the AAUP Bulletin, Vol. 63, No. 3, August 1977. The only other available criterion which could be uniformly applied to the selection of the institutions to he included i n thk-survey is the National Academy of Science's "Baccalaureate Origins of the PhD in Chemistrv." Those colleees which had oroduced 8 or more PhDs in chemistry during the period 1967-76 were chosen. This selection procedure eliminated a number of institutions which have significant undergraduate research activity. These institutions include lthaca Colleee (NY) with 44 puhlicati(ms and 10 faculty authors from 197fto 1979, Drury college (MO) with 30 publications and 3 faculty authors, Newberry College (SC) with 20 publications and 2 faculty authors, and Simmons College (MA) with 17 publications and 6 faculty authors. Sources and explanatory notes for the column headings for the data given in Table 1 follow:
044
(U5 066 047
Ole 010
EliSTtRNNUNA EUZABETHTOWIPA EIU*LASTIL EMORY ILHENRYIVA FlilRFiELDlCT FLORIDA S O U M R N f i
019 UWENCEIWI 080 LEBANON V A U Y I P A 081 LEMOYIR1HY 082 LEWISOL 083 LlNFlELDlOR (184 LORASIIA
Volume 58
Number 10
October 1981
781
Table 1. Continued
WB Millll(lNIIL 091 UIIISAPSIMS 098 MlSSlSSiPPl/MS 099 UONMOWHIII I W UONMOWHlKl 101 UO(IEIIWSEIO* 102 UT ST M*R"S/M) 303 UTUNiWlOH 304 UUHLENBERWPA 105 UUNDtLEINIL 108 UUSKINWUIW 107 NEBRASKA WLSLEY*NIK 108 NIAGARAINY
114 116 11(1 111 111)
OHIO NOI1MIUIIW OKLAHOMA CITIIW OUACNITA BIWSTIUI
PhCIFIC LUTlfR&MM)& PARWMO 119 RIILA L SCIRI 120 P W O N A I C A 121 R 1 A T T INSTIMEINI 122 RTUANINDSICA
m m
165 W A I I T B U O l l 158 WASHINOTON&LEEIyA (57 WASHINGTON&.EFFERsoNRA IS8 WASHINGIONIM 168 WELLESLEIIMA 180 WEST YA WESLEIAYIW 161 WESTERN MI\RYUN)IM)
1 13
0
8 1 2 1 1 11 15 3310 1 2 8 711 15 13 12 28 10
8
9 4
I 13 0 7 117
5 I 8 4 0 2 9
24 1
2
2210 YI 18
8
2 0
I7
7 3
4B
3 2 5 8 1 3 > B 2010
(I
0 4
4
1
0 0 0 t 0 1 11
24 0
7 0 0 1 0 0
0 111230 0 10290 0
I 2
90710 0 21130
4
0
2
0 0
0 0 I
2 5 0
0 1 (I
0
2
0
P
1
0
0 1 0 S 0 0 10 0 5
0 32780 0
0 ,704 0 0
0 5 0 0 2 0 0
1) Baccalaureate origins of PhDs in chemistry were obtained from
the survey of earned doctorates conducted by the Commission on Human Resources of the National Research Council (National Academy of Sciences) made available to the Office of Institutional Research of Franklin and Marshall College. The column headed 20 lists the numher of doctorates in chemistry earned hy graduates of each institution from 1920 to 1976. The column headed 67 provides the number earned between 19fi7 and 197fi. 2) Number of publications and number of faculty authors were
found hy a search of the Chemical Abstracts data base for the
782
Journal of Chemical Education
time period January 1970 toOctober 1979 provided by Bibliographic Retrieval Services. Inc.. Scotia, NY. The search was carried out by institution name. Publications resulting from research done a t institutions other than the author's home institution are consequently not included in the Table. Publieations bearing the names of two or mare institutions are included only if the author's home institution is the first listed. Because some institution names are inconsistently nhhreviated in the Chemical Abstracts data base, a variety of plausible shbreviations were searched for each institution. Faculty authors were identified from the American Chemical Society Directory of "College Chemistry Faculties."
Table 2. Ranklng of Top 30 InstHulions by Instnullon Number No. PhD 1967-76
No. PMI 1920-76
No. Publ 1970-78
No. Aum
WRP
1970-78
1968-77
Table 3. Averages for All InstHulIons and the Top Twenty Ranked by Number ol Publkatlons RIbIIcNions
PhD
- 0 No. Aulh 1920-76 1967-76
All Top
174 20
45 87
18 31
7.8 37
2.4 6.9
(irants SURP 8.690 27.922
$OW 16.751 76.798
The column headed No. gives the total number of publications and the column headed Auth provides the number of faculty authors. The publications tabulated under the column heading ACS are those that appeared in the following journals: Journal of the Amerieon Chemical Society, Journal af Organic Chemistry, Journal physical Chemistry, Inorganic Chemiftry, and Anolytid Chemistry. Thecolumn headed JCE lists publications appearing in the JOURNAL OF CHEMICAL EDUCATION. 3) Information on grants was obtained from the annual "Grants and Awards" listing of the National Science Foundation for 1968-77 (available from Superintendent of Documents, US. Government Printing Office, Washington, D.C. 20402). Research Corporation Quarterly Bulletins for 1968-77, and Petroleum Research Funds grants listings published in Chemical and Engineering News for 1968 through July 1978. The column headed $URPgives the total dollar amount of the URP grants received from 1968-77: Nu. URP gives the number of these grants over the same time period. 5Res designates total dollar amounts from NSF, Research Corporation, and PRF grants: No. Res. is the number of these grants. 4) College board scores (SAT) were taken from the 8th edition of "Comparative Guide to American Colleges" by J. Cass and N. Birnhaum, Harper and Row, NY, 1977. These scores are the average for male and female students for the entire college student body. Full time college enrollments were taken from the 6th edition of thesameguide published in 1973.The year 1973 was chosen because it represents the approximate median year for the time period covered in this study.
NO. URP 1968-77
$Re5
No. Rer
1968-77
1968-77
5 ) Average faculty compensation (all ranks combined, inclusive
of lecturers) for theentire college faculty was obtained from the 1977-78 data of the American Association of University Professors supplied by the Associate Secretary for Research, Maryse Rymonerie, 1 DuPont Circle, Suite 500. Washington. D.C. 7MRR
6) Number of chemistry faculty were taken from the ACS 1977 College Chemistry Faculties Directory or from "Research in Chemistry a t Private Undergraduate Colleges," B. Andreen. (Editor), Council on Undergraduate Research, Suite 275,4570 W. 77th St.. Edina, MN 55436. 7) The average chemistry faculty teaching load, average number of senior chemistry majors ( M j r )for 1973-78, average number of students involved in resenrch (Res) from 1973-78 per academic year and per summer (Smr Res), and the total number of journal subscriptions (No. Jrnl) in 1978 wereobtained from "Research in Chemistry a t Private Undergraduate Colleges." 8) Numbers of NSF Fellows for the period 1974 to 1979 were compiled by Brian Andreen. Research Corporation Regional Director, from NSF announcements of these awards. Table 2 gives t h e rankings of t h e t o p institutions in each of 9 characteristics. T h e institutions are designated by t h e number associated with each in T a b l e 1. W h e n two o r more institutions have t h e s a m e rank this is designated by a bracket. T a b l e 3 shows averaged data for all 174 institutions listed in T a b l e 1, a n d also for t h e t o p 20 institutions ranked according t o t h e number of publirations. T h e average number of puhlications for t h e period 1970 t o 1979 is 7.8 for all institutions. whereas for t h e to^ 20 it is considerably higher-37. T h e s a k e is t r u e for t h e n h b e r of P h D s f r o m i 9 2 6 t o 1976; t h e averaee number for all 174 institutions is 45, whereas for t h e &p 20the number is 87. It is also clear from these data t h a t for those too 20 ranked accordina t o publications t h e amount of grant support is also larger. ~ * b l 4e provides a summary of t h e averages for t h e CUR institutions, i.e.. t h e institutions listed in t h e CUR directory (these institutions tend to be more Volume 58
Number 10
October 1981
783
Table 6. Top Ranking Colleges In Number of Publications
Table 4. Averages for CUR Colleges' PhO. 1920-76 1967-76 Publications. No. Auth ACS JCE $Grant. URP OTH NO. Grants. URP OTH VSAT QSAT NO. Faculty Facully Compensation Teaching Load Enrollment NO. Maiors NO. Research Students NO.Summer Research Students No. Journals NO. NSF Fellows FW me 46
P~biiMtions
69 25 17 4.5 3.8 3.9 19.106 39.407
Pomna Hope Amhsrst ~ N B YMudd Williams Lsfayene Wooster Pran Institute Lebanon Valley Franklin & Marshall
PhO
No.
Aum
ACS
JCE
1967-76
$OM
117 86 57 52 48 4 36 34 31 30
9 12 14 10 9
11 40 20 25 19
7 12
15 60 21 39 29 43 57 12 35 45
84.415 370.585 197.579 216.330 52.650 0 51.722 0 38.500 35.070
1
6 9 2 5 5
1
5 2 24 11 8
8 13 2 11 2 10 5
Table 7. Conmarlsons wlth Universilles PubiIFaclYr Collegesd Universitiesb
instihlfions tor which all intormation war available.
---
0.5 3.2
% ACS Pubi $OlantslFaclYrc SGrantslPubl 27 34
1.440 16.200
2.690 5.100
Top 20 in pullcatlono. 11 relected Innlop RwseAndersongwplng. Not including NIH.
Table 5. TODRanklna Colleaes In Analnment 01 PhD No. PhO
HOP^ Franklin 8 Marshall St. Olaf Wwster Kslamazm Reed Manhallan Juniala Carleton Lafayene
P~bllcatl~n~
GIBnts
1920-76
1967-76
No.
Auth
$URP
$OTH
166 161
60 45
86 30
12 5
127.510 0
370.585 35.070
156 147 123 123 113 109 108 101
47 57 39 29 62 41 38 43
8 36 4 9 5 14 14 41
5 9 4 6 3 7 6 6
13.960 79.189 8.760 42.770 23.170 67.780 61.340 25.010
16.045 51.722 28.000 53.200 20.000 25.000 50.820 0
active in research than the average of all 174 institutions). Here, the numher of publications have been broken down into the two categories mentioned previously-ACS and JCE. These CUR institutions produced an average of 17 publications, about 4 of which appeared in the set of selected ACS OF journals and about the same numher in the JOURNAL CHEMICAL EDUCATION. The averaee CUR institution also has a total of 6 faculty members in chemistry. with an average faculty compensation of 921,048, a teaching load of 13, an enrollment of 1579.15 chemistry majors, 8.5 research students, and 41 journals in the library. Hence, Table 4 gives a representative profile of a college with a moderately active research program. Table 5 lists the top ranking colleges in attainment of the PhD from 1920 to 1976. It is clear that there is no one-to-one corresoondence between the number of PhDs and the other characteristics; for example, there are some colleges with a large number of PhDs and alsoa large number of puhlications, hut there are also colleges with large numbers of PhDs and a small numher of puhlications. However, when these colleges are compared with others that rank lower in attainment of the eeneral relationshiw between number of PhD (see Tahle 3).. a .. PhDs and research activity becomes more apparent (vide infro). Table 6 lists the top ranking colleges in numher of puhlications. This number varies from 117 to 30. The published articles alsovary in composition. Somecolleges have published quite extensively in "ACS" journals while others publish elsewhere. All of these institutions contribute to the JOURNAL OF CHEMICAL EDUCATION. Eighteen universities selected from the Roose-Anderson 784
Journal of Chemical Education
(RA) study (7) were alsoexamined. Table7gives a comparison between the 11 universities selected from the top RA grouping (Harvard, Princeton, Purdue, Stanford, Cornell, Ohio State, Michigan, Florida State, Penn State, Carnegie-Mellon. and Rochester), and the top 20 colleges ranked according to the number of puhlications. It is no surprise that the number of puhlications per faculty member per year is quite a hit higher for the universities. The comparison of the composition of the nuhlications is. however. somewhat surwrisine-the oercent ;d'prlhlications'in the 1\65 category is nEar~ythesamf for the two rrouos .. . ( 2 7 9 for the colleees. 34% for the r~nivers.ties).The dollar amount of grants per faculty member per year is given in the next column-$1440 for the colleges, $16,200 for the universities. This dollar amount does not include NIH money which for the universities is very sianificant. One of the university departments in this surv&, for example, received about a million dollars over a one-year period, and if NIH money were included that figure of $16,200 would roughly douhle according to our best estimates. Inclusion of NIH monev for the colleees would increase the fieure onlv.sliehtlv. . The next column tabulates the amount of research grant support per publication and indicates that it "costs" about $2700 for every publication a t the undergraduate institutions as comwared to S5100 for universities. If NIH monev is included'the figure is probably closer to $10,000 for tile universities.
-
Regression Analyses The qualitative relationship (demonstrated in Tahle 3) between the numher of PhDs originating from an institution and the research oroductivitv of that institution was investigated more quantitatively by performing regression analyses between the variahles listed in Tahle 1. Table 8 lists the correlation coefficients for the bivariate (one to one) relationships between the variahles PhD20 throueh NoRes for all institutions. An examination of the first coiumn (which lists all the correlation coefficients for the relationships between the numher of PhDs from 1920 to 1976 with the other variables) reveals that the number of faculty authors correlates more strongly than the number of publications with the number of PhDs from 1920-1976. Indeed, this correlation is the strongest one for the variahle PhD20 (omitting "trivial" correlations such as PhD20 versus PhD20 and PhD20 versus PhDGi) and the second stroneest with the variable PhD67. The third andfourth columns in Tahle 8 reveal that the highest correlations with the numher of publications and
Table 8.
Phi320 PhD67 Pub Aulh ACS JCE SURP NoURP $Res NaRes
Correlatlons for All Instllutlonsa Conelalion Coenicienls (R) ACS JCE
RD20
PhD67
Pub
Aulh
1.00000 0.82697 0.46336 0.56362 0.39378 0.42731 0.44300 0.41603 0.43666 0.45602
0.62697 1.00000 0.40266 0.47342 0.39857 0.36011 0.46147 0.42567 0.46772 0.49203
0.46336 0.40266 1.00000 0.77145 0.76764 0.62566 0.64364 0.57196 0.68263 0.69226
0.56362 0.47342 0.77145 1.00000 0.65184 0.63624 0.58644 0.57609 0.89915 0.70064
0.39376 0.39857 0.76764 0.65164 1,00000 0.47359 0.51139 0.39005 0.60467 0.79664
DURP
No URP
$Re$
No Res
0.44300 0.46147 0.64364 0.58644 0.51139 0.48431 1,00000 0.95269 0.69310 0.71199
0.41603 0.42567 0.57196 0.57609 0.39005 0.44155 0.95269 1.00000 0.56443 0.61012
0.43886 0.46772 0.68263 0.69915 0.80467 0.45131 0.69310 0.58443 1.00000 0.96656
0.45602 0.49203 0.69226 0.70064 0.79664 0.49502 0.71 199 0.61012 0.96656 1.00000
SURP
No URP
$Re*
NORBS
0.44242 0.49394 0.61522 0.64574 0.47920 0.42029 1.00000 0.94316 0.71931 0.71947 0.06004 0.10653 0.17947 0.17569 -0.24615 0.00435 0.39176 0.61251 0.57961 0.74762 0.39012
0.40222 0.44827 0.51503 0.60030 0.29227 0.33657 0.94316 1.00000 0.55319 0.56246 0.14430 0.16049 0.25574 0.13079 -0.20261 -0.03570 0.33279 0.46426 0.53678 0.64142 0.31367
0.43734 0.46657 0.56230 0.62944 0.66461 0.41596 0.71931 0.55319 1.00000 0.98044 0.05917 0.05646 0.08216 0.20804 -0.23466 0.01579 0.36119 0.71619 0.35004 0.63691 0.44434
0.44347 0.46952 0.57694 0.63595 0.83969 0.42541 0.71947 0.56248 0.98044 1.00000 0.05265 0.05193 0.07466 0.19361 -0.26277 0.02251 0.39697 0.66796 0.34866 0.83547 0.44342
0.42731 0.36011 0.62566 0.63624 0.47359 1,00000 0.48431 0.44155 0.45131 0.49502
'174 InslIMlonr Table 9. Correlatlons tor CUR lnstltullons'
PhD2O PhD67 Pub Aulh ACS JCE $URP NoURP $Re$ NoRes VSAT QSAT C ~ P NoFac
Load Enrl
Mjr Res Jrnl Smr FeI
Conelalion Coefficienls (R) ACS JCE
WDZO
PhD67
Pub
Aum
1.00000 0.64442 0.44105 0.56770 0.47101 0.31062 0.44242 0.40222 0.43734 0.44347 0.31695 0.43036 0.31325 0.55576 -0.33662 0.18522 0.65756 0.54871 0.27007 0.49663 0.43566
0.64442 1.00000 0.32754 0.50454 0.44223 0.31048 0.49394 0.44827 0.46657 0.46952 0.03172 0.15276 0.13030 0.62697 -0.24040 0.33116 0.66662 0.60237 0.16806 0.46773 0.32053
0.44105 0.32754 1.00000 0.72796 0.69671 0.57107 0.61522 0.51503 0.58230 0.57694 0.22439 0.32324 0.39206 0.36119 -0.23926 0.02674 0.33602 0.42266 0.78231 0.51311 0.41625
0.56770 0.50454 0.72796 1.00000 0.69463 0.63558 0.64574 0.60030 0.62944 0.63595 0.29292 0.34464 0.35341 0.38150 -0.36861 0.00615 0.44666 0.56561 0.47614 0.55315 0,42796
0.47101 0.44223 0.69671 0.69463 1.00000 0.50066 0.47920 0.29227 0.66461 0.83989 0.12930 0.17033 0.12975 0.32629 -0.26643 -0.00265 0.40769 0.63413 0.33266 0.65455 0.43145
0.31062 0.31046 0.57107 0.63558 0.50066 1.00000 0.42029 0.33657 0.41596 0.42541 0.01193 0.01709 0.03036 0.26492 -0.13260 0.06467 0.30691 0.50914 0.26232 0.32645 0.25542
fatha 46 lnotnutions tm whichall i n l m l l m was avalbble.
Table 10.
faculty authors appear with the dollar amount of research grants and the numher of these grants. The correlations of the number of articles in those selected ACS journals with $Res and NoRes are even more striking and represent some of the strongest "non-trivial" correlations in the table. A greater number of variahles are available for the CUR institutions and these correlations are presented in Table 9. The same basic trends are visible in these correlations except that (a) the number of majors correlates more strongly than the number of authors with PhD20, (b) the number of majors, number of faculty, and numher of research students correlate more strongly than the number of authors with PhD67. (c) the number of journals correlates very strongly with number of publications. (d) the dollar amount of URP grants correlates more strongly than $Res with both Pub and Au, and (e) the correlation between the numher of "ACS" articles and $Res and NoRes is even higher (R = 0.865,0.840) for these institutions. The results of multiple regression analyses of these variables are given in Table 10. In the left-hand portion of this table are listed the three independent variahles that account for the greatest variability in the dependent variable when the independent variahles are added to the regression equation one a t a time. The first variahle to enter the equation is always the one with the highest hivariate correlation: the second variable is the one that accounts for more of the remaining variability (after accounting for that due to the first variable) in the dependent variable than any of the other independent variables. and so on. In the right-hand portion of the table are listed the
b.
Variable
Variablgs Emered Stepwise" 1 2 3
PM20 RD67 Pub Auth ACS JCE $URP $Res
A m NaRes NaRes NoRes $Res NoRes pub ACS
PhD20 PhD67 Pub Auih ACS
Mjr
JCE $URP $Res NoFaC
Mjr Res Smr Fel
Multlple Regression * Variables Entered Simullaneou41yc 1 2 3
Variables PhD2GNoRes (N = 174) $URP JCE AWh $URP NoRes Aufh SURP Aufh NoRes SURP SURP PhD2O R D 2 0 PhD67 WRP RD2O NoURP PhD20 NoURP IURP NoURP NoRes NouRP $Res NoRes RD2O $URP R D 2 0 NoRes $Res P h ~ 6 7 Auth Pub PhD67 Auth Auth PhD67 ACS AuM PhD67
Rd
F*
0.568 0.548 0.753 0.794 0.823 0.560 0.690 0.845
10.9 6.85 36.5 47.5 58.4 14.1 25.3 69.7
Jrnl OURP $Res
Variables PhD20--F~I OSAT Smr Mlr Res NoFac NoFac Jrnl $Res Load PhD2O NoRes PhD20 NoURP OSAT $Re$
(N = 46) 0.662 3.9 NoFac Cmp Mjr Fmp 0.862 4.0 PhD20 VSAT 0.895 6.6 NoURP NoFac 0.827 3.6 Enrl NoFac 0.937 11.9
Res Smr ACS End PhD67 SRes $Res QSAT
SURP Jrni Smr RD2O Enrl RD67 NoURP Smr
OSAT RD20 Smr QSAT ACS NoFae
Smr PhD67 NoFac VSAT Res
Mjr
Fel JCE
Mjr
Smr Auth PhD2O OSAT PhD20 JCE RDZO JCE
NoFac Smr ACS Enrl Enrl End RD20 OSAT
NoFac
Mir
0.636 1.1 0.906 7.5 0.954 16.7 0.941 9.7 0.657 3.5 0.895 5.0 0.916 6.5 0.732 1.4
TrlvIaI" ca%sminBd. O T I M flrn Wee vatlabler entered lo rlwuiu, regesalon.
' b m t e e variable9 wM7 h w s l Fvalver WMl a I l a r e a & d s l ~ l t a ~dMuUlpls ~. -lation don.
mnicient
IW
simu~nemsregrersim.
F-value tor sirnuitanears regre*
Volume 58 Number 10 October 1981
785
three most significant variables (as determined by their Fvalues) in the regression equation that results when all the variables are entered simultaneouslv. In this case the most significant variahle is the one that accounts for most of the variahilitv in the dependent variahle after the variabilitv due to all theither variahles has been accounted for. That thk two procedures sometimes produce different sets of "most significant" variahles IS o h v i o ~ from ~ s an inspection of the table. The multiple cnrrelation coeffirientaand the F-values for the simu~tanebusregressions using all variables (except those giving "trivial" correlations) are also included in Table 10. Using the stepwise regression as a better procedure to measure significance, the data in Table 10 show that the QSAT scores and numher of academicvear research students are the sewnd most significant (the number of majors is first) of all the v a r i a l h affi>ctlnathe number of PhDs ~roducedfor the time period 192&197