Click for next page ( 80


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 79
6 CHANGES IN REGISTERED TIME TO THE DOCTORATE, TIME PRIOR TO GRADUATE ENTRANCE, AND TIME NOT ENROLLED IN THE UNIVERSITY . . This chapter uses the common and unique variables models defined in Chapter 5 to explain changes in registered time to degree (RTD) and the common variables model to explain changes in time prior to graduate entrance CAGE) and time not enrolled at the university (ADIEU). As discussed in Chapter 1, TTD and RTD have a similar time trend, and increases in RTD are largely responsible for increases in l ID. Registered Time to the Doctorate RTD in the Common Variables Model Using Linear and Log Linear Equations Regression coefficients for each field, using both linear and log linear estimating equations, appear in Appendix Tables 6 and 6A. A summary of the findings for each variable in each model is given in Tables 6.1 and 6.2. As was true for TTD, a comparison of the results for the linear and log estimates suggests that the results are different depending on the model used. While the importance of certain variables such as teaching assistantships, foreign baccalaureate, and salary does not change across specifications, the role of others such as age, federal support, and unemployment are affected. In most cases, the signs of the statistically significant variables do not change, and the log linear model explains the variation in the data no better than the linear model does. RTD in the Unique Variables Model Table 6.3 (pp. 82-83) summarizes the results of using a unique model for each of the 11 fields. Age is no longer an important variable in all fields, and no other variable has a significant impact on RTD in every field. 79

OCR for page 79
TABLE 6.1: Summary of Common Linear Model Regression Results for RTD, by Field Variable Fielders) Statistically +/ Significant Female Social Sciences yes + Age Chemistry yes + Mathematics yes + Earth Atmospheric, yes + - & Marine Sciences Social Sciences yes + Federal Support Earn, Atmospheric, yes tic Marine Sciences Biosciences yes Teaching Assistantship Biosciences yes + Research Assistantship no Baccalaureate from Foreign Social Sciences yes + Institution Baccalaureate from Category I Chemistry yes Research School Agricultural Sciences yes + Graduate Degree from Category I no Research School Number of Faculty Earth, Atmospheric, yes + & Marine Sciences Biosciences yes Salary Ratio: New Ph.D.s no to Ph.D.s 10 yrs. after Degree Unemployment Rate of College-Educated Per Capita Doctorates in United States Chemistry Earth, Atmospheric, & Marine Sciences Social Sciences yes yes yes no 80

OCR for page 79
TABLE 6.2: Summary of Common Log-Linear Model Regression Results. for RTD, by Field Vanable Fielders) Statistically +/ Sign~ficant Female . no Age . Earth, Atmosphenc, yes & Manne Sciences Biosciences . yes Federal Support Biosciences yes Teaching Assistantship Research Assistantship Baccalaureate from Foreign Institution Baccalaureate from Category I Research School Graduate Degree from Category I Research School Number of Faculty Salary Ratio: New Ph.D.s to Ph.D.s 10 yrs. after Degree Unemployment Rate of College-Educated Biosciences yes no Social Sciences yes Agncultural Sciences yes Agncultural Sciences yes Earth, Aunosphenc, yes & Marine Sciences Biosciences yes no Earth, Atmospheric, yes dc Manne Sciences Per Capita Doctorates Earth, Atmosphenc, yes In United States & Marine Sciences + + + + Evaluation of the Results A number of observations can be made about Table 6.4 (p. 84), which shows the number of fields in which a particular independent variable was statistically significant. For example, no one variable explains the widely observed increases in RID across fields. Instead, the combinations of variables 81

OCR for page 79
TABLE 6.3: Summary of Unique Variables Model Regression Results for RTD, Field Variable~s) Correlation Comment Chemistry Age Baccalaureate from + Foreign Institution Salary Ratio: New Ph.D.s to Ph.D.s 10 yrs. after Degree These three variables accounted for 91 percent of the variation in RTD. A . . One-year Increase In age boosted RTD by 1.5 years. A 1 percent increase in doctorates with degrees from r Iorelgn Institutions increases RTD by about a week. Physics and Marital Status - These three variables Astronomy Graduate Degree from - accounted for 91 percent of Category I Research variation in RTD. A 1 School percent increase in married Teaching Asst. + students lowered RTD by nearly two weeks. A similar increase in percentage of students from Category I school decreased RTD by a little over two weeks. ~ Earth, Marital Status - These four variables Atmospheric, Baccalaureate from - explained 89 percent of the & Marine Category I Research variation in RTD. Sciences School Temp. U.S. Residents + Receiving Ph.D.s Baccalaureate from + Top-20 School Mathematics Female - + The two variables explained & Computer Salary Ratio: New - 97 percent of the variation Sciences Ph.D.s to Ph.D.s in RTD. 10 yrs. after Degree EngineeringBaccalaureate from + These three variables Foreign Institution explained 93 percent of the Undergraduate Degree - variation in RTD. in Same Field Definite Employment 82

OCR for page 79
by Field Field Variable~s) Correlation Comment ( I) Agricultural Teaching Asst. + These four variables Sciences Baccalaureate from + accounted for 82 percent of Foreign Institution the variation in RTD. Definite Employment Salary Ratio: New Ph.D.s to Ph.D.s 10 yrs. after Degree Biological Research Asst. + These Tree variables Sciences Percent Cohort + explained 95 percent of the Seeking Emp variation in RTD. The Salary Ratio: New - Durbin-Watson statistic for Ph.D.s to Ph.D.s this regression is in the 10 yrs. after Degree indeterminate range. Health Federal Support - These three variables Sciences Salary Ratio: - explained 85 percent of the Doctorates to variation in RTD. A 1 Baccalaureates percent rise in federal Temp. U.S. Residents + support decreased RTD by Receiving Ph.D.s about two weeks. Psychology Federal Support Salary Ratio: New Ph.D.s to Ph.D.s 10 yrs. after Degree Temp. U.S. Residents + Receiving Ph.D.s Economics Private Support Baccalaureate from + Foreign Institution Temp. U.S. Residents + Receiving Ph.D.s These three variables accounted for 96 percent of the variation in RTD. These three variables explained 95 percent of the variation in RTD. A 1 percent increase in those with baccalaureate from foreign institution lowered RTD by nearly a month. Social Private Support - These three variables Sciences Salary Ratio: New - explained 99 percent of the Ph.D.s to Ph.D.s variation in RTD. A 1 10 yrs. after Degree percent jump in private Temp. U.S. Residents + support increased RTD by Receiving Ph.D.s about a month. 83

OCR for page 79
TABLE 6.4: Number of Fields in Which Variable Has Statistically Significant Effect on AD . . . . MODEL C O MM ON UNIQUE Linear Log Linear . Variable POS NEG POS NAG POS BEG WOMEN 1 0 0 , 0 1 0 AGE 4 0 2 0 1 0 SUPPED O SUPTA 2 o 0 1 0 2 1 0 1 0 2 0 SUPRA O O O 0 1 O' FORBACC 1 O 1 0 4 0 BCARN1ST 1 1 1 0 0 PCARN1ST 0 0 0 1 0 FACULTY 1 1 1 1 0 0 UNEMP4YR 0 3 0 1 O O PERPOP 0 0 0 1 0 0 MARRED TEMP SAMEFLD - SUPPRIV BTOP20 SDRSAL10 SALRAT1 SAIRATIO SEEK DEFIN 0 2 5 0 o O 2 I O o O o o 1 I O NOTES: (1) "Pos" indicates a positive regression coefficient. "Neg" indicates a negative regression coefficient. (2) Variables below the dotted line were not entered in the common variables models. (3) Acronyms are defined in Appendix B. pp. 175-177. 84

OCR for page 79
with statistically significant effects on RTD vary by field. In both the linear common variables model and the unique variables model, female gender was significant and positive in just one field. In the log-linear model, gender was not significant in any field. In those equations where age is statistically significant, it tends to have a large impact on RTD, suggesting that as more older students enroll in doctoral programs, RTD will increase. However, as noted earlier, age may act as a proxy for cohort differences rather than for physiological or other effects of aging. This possibility deserves more study before conclusive statements can be made. The role of financial support in affecting RTD is mixed. In a number of fields, financial variables did not enter the equation at all and, in a few, they had a positive partial correlation, contrary to intuitive expectations. This finding suggests that the effects of financial aid are field- specific and the type of aid provided influences whether students complete the doctorate more or less rapidly. The data do not allow firm conclusions about the effects of increasing financial aid as the primary source of support. The analysis suggests that in some fields increases in the number of foreign students or in the percentage of students with foreign baccalaureates have led to increased RTD. Finally, analysis supports the belief that changes in market variables unemployment rate, salaries, and salary ratios affect RTD. The results of this inquiry are best viewed as suggestive rather than conclusive. Problems of multicollinearity, aggregation, and limited data suggest the need for study of these issues in a cross-seciion and/or pooled time-series cross-section framework. Further research is needed to affirm the role of age, to elaborate on the role of financial aid, and to provide greater insight into the role of student ability (see Chapter 7~. Time Spent Prior to Graduate School Entrance (TPGE) The results summarized in Table 6.5 were obtained using the linear common variables model to explain changes in TPGE (see Appendix Table 7~. The implicit assumption in the use of these variables is that students have prior knowledge of how their cohort is likely to fare in terms of receiving financial aid and entering the labor market. The R2 for the individual field equations are lower for TPGE than for 1-1 D or RTD and, for three fields, the equations themselves are not statistically significant. In part, this results because decisions made at the time of undergraduate graduation are more likely to be based on family background and undergraduate performance factors not contained in the model (see Chapter 2~. It may also be that new variables are needed to adequately capture conditions at the / 85

OCR for page 79
TABLE 6.5: Summary of Common Linear Model Regression Results for TPGE, by Variable Variable Fielded Statistically Correlation Significant Female no Age Chemistry yes Mathematics yes Engineering yes Biosciences yes Health Sciences yes Social Sciences yes Federal Support no Teaching Assistantship Social Sciences yes Research Assistantship Chemistry Baccalaureate from Foreign Mathematics Institution Baccalaureate from Category I Research School Graduate Degree from Category I Research School Number of Faculty Salary Ratio: New Ph.D.s to Ph.D.s 10 yrs. after Degree Unemployment Rate of College-Educated Percent Population with Doctorates yes yes no no no Mathematics yes Mathematics yes Mathematics yes + + + + + ; + + NOTE: No variables were significant for the following fields: earth, atmospheric and marine sciences; agricultural sciences; and economics. 86

OCR for page 79
time the decision to enter graduate school was made. For example, the relevant financial variable may be the percentage of the prior year's entering class with financial aid and the relevant market variable may be the percentage of doctorates who found jobs in the year in which the person decided to enter graduate school. Analysis of these issues may explain why fewer variables are statistically significant in the TPGE equations than in the RID equations. It's interesting to note that in math, biosciences, psychology, and social sciences, the equations explained better than 90 percent of the variation in the data. As was true for the linear analysis, in the log-linear analysis (Table 6.6), the equations for earth, atmospheric, and marine sciences; agricultural sciences; and economics were not statistically significant. Also, the R2s were generally lower on these equations than for TTD and RID. Several points can be made about the determinants of TPGE based on the findings in this section. First, in most of the fields, the variables that explained most of the change in TPGE were demographic and economic in nature. With rare exceptions, institutional factors did not affect the TPGE. However, in the log equations the unemployment rate and salary variables were statistically significant determinants of TPGE. Second, the financial aid variables did affect TPGE in some fields, although not always in the expected direction. TPGE in chemistry and physics and astronomy was consistently affected by financial aid. Finally, in most fields neither the percentage of women nor the percentage of students with foreign baccalaureates had a statistically significant effect on IPGE. Time Not Enrolled in the University (TNEU) TNEU, time the student spends away from his or her studies after registering for graduate school, is affected by such factors as illness or financial exigency, frustration with the doctoral program, and the need to take a break from dissertation work (see Appendix Table 8~. Since the variables in the common variables model do not specifically address these concerns, this model is not expected to explain as much of the variation in TNEU as it did for other dependent variables. Tables 6.7 and 6.8 summarize the results from the linear and non-linear regression equations. The analysis shows no one variable consistently explained changes in TNEU in all fields. Compared to TPGE, unemployment and salary variables do not appear to have a strong effect on TNEU. This is surprising. One would expect student decisions to leave graduate school to be more affected by market conditions. And, as with TPGE, factors such as gender and percent with foreign baccalaureates do not appear to exert a strong influence on TNEU. 87

OCR for page 79
TABLE 6.6: Summary of Common Log-Linear Model Regression Results for TPGE, by Variable Variable Fielders) Statistically Correlation Significant Female Biosciences yes + Age Chemistry yes + Physics & yes + Astronomy E . ngmeenng Biosciences Health Sciences Psychology Social Sciences Federal Support Teaching Assistantship Physics & Astronomy Research Assistantship Chemistry Baccalaureate from Foreign Mathematics Institution Baccalaureate from Category I Research School Graduate Degree from Category I Research School Number of Faculty no yes yes yes no no no yes yes + yes yes yes + + Salary Ratio: New Ph.D.s Physics & yes to Ph.D.s 10 yrs. after Degree Astronomy Unemployment Rate Physics dc yes + of College-Educated Astronomy Psychology yes + Mathematics yes Percent Population no with Doctorates NOTE: No variables were significant for the following fields: and marine sciences; agricultural sciences; and economics. 88 earth, atmospheric

OCR for page 79
TABLE 6.7: Summary of Common Linear Model Regression Results for TNEU, by Vanable Vanable Fielders) Statistically Correlation significant (+/-) Female Age Federal Support Teaching Assistantship Biosciences Research Assistantship Biosciences Baccalaureate from Foreign Biosciences Institution Baccalaureate from Category I Research School Graduate Degree from Biosciences Category I Research School Psychology Number of Faculty Salary Ratio: New Ph.D.s to Ph.D.s 10 yrs. after Degree Unemployment Rate of College-Educated Percent Population with Doctorates no no yes no yes yes Psychology yes yes yes no no no Biosciences yes Psychology yes Summary of the Findings The common variables model appears to be more effective for understanding changes in RTD than for interpreting changes in TPGE and TNEU. No one variable is responsible for the increase in RTD over time, although in fields in which it is statistically significant, age has a relatively large effect. Moreover, the mix of variables that affect RTD is different among fields, although all five vectors described in Chapter 3 come into play. 89

OCR for page 79
TABLE 6.8: Summary of Common Log-Linear Model Regression Results for TNEU, by Field Variable Fielders) Statistically Correlation Female no Age Health Sciences yes + Federal Support Chemistry yes + Physics & yes + Astronomy Biosciences yes + Teaching Assistantship no Research Assistantship no Baccalaureate from Foreign Biosciences Institution Baccalaureate from Category I Research School Graduate Degree from Category I Research School Number of Faculty Salary Ratio: New Ph.D.s to Ph.D.s 10 yrs. after Degree Unemployment Rate of College-Educated Percent Population with Doctorates yes no no Chemistry yes no no Biosciences yes NOTE: No variables were significant for the following fields: mathematics; engineering; and agricultural sciences. Only biosciences and economics had R2s greater than 90 percent. 90

OCR for page 79
The linear model suggests that age has the largest impact on RTD; the percentage of students with foreign baccalaureates and who are female also consistently increases RTD. These results are field-specific and are not generalizable to all 11 fields, however. The role of financial aid is ambiguous, and different types of aid affect RTD differently. The models explain less of the variance in TPGE and TNEU than in TTD and RTD. In some fields, the models do not produce statistically significant results. While generalizing across fields is difficult, the equations for IMAGE and INEU have fewer statistically significant variables than those for RTD and TTD. Interestingly, market variables explain time spent prior to entering graduate school while, for the most part, they are not statistically significant in He TNEU equations. Additional work is needed to understand the factors that cause changes in TPGE and TNEU.16 It is likely that institutional and psychological factors beyond those captured in this common variables model affect the decision to postpone entry to graduate school and/or to delay completion of the doctorate. 16 Knowledge of the determinants of TPGE would be useful, since it tells us how long students take to move from undergraduate to graduate school. TNEU is important because substantial differences exist across fields and we have little understanding of Be underlying reasons: it may be that market opportunities for ABDs are substantially different among fields or that some field work is useful before obtaining the doctorate. 91

OCR for page 79