Professional Societies Workshop Series
Qualitative Approaches to Studying Graduate Student Attrition
Qualitative Approaches to Studying Graduate Student Attrition
I am Carla Howery from the American Sociological Association. Our afternoon session is going to look at some research that three scholars are undertaking, that use more qualitative approaches to studying attrition. Our first presenter is Beatriz Chu Clewell, who is a principal research associate at the Urban Institute here in Washington, and has migrated from an English literature background to educational policy, and is now undertaking an important study for NSF on graduate attrition.
Current NSF Project on Graduate Attrition
Contextual Factors in Graduate Student Attrition
Research on Graduate Student Attrition
Current NSF Project on Graduate Attrition
Good afternoon. Usually the speaker right after lunch is the one who has to be very lively, otherwise people will fall asleep. Although I do not really promise to be lively, some of the information that I will be sharing with you is information you will be interested in. This morning we talked about databases and attrition and the role of large-scale databases in computing attrition and understanding attrition. I see that as answering the "what" question. How many students are dropping out? I see the qualitative pieces as answering the "why" and "how" questions. If we were to characterize this today, I think it would be going from the large picture to focusing a lens on the actual graduate experience. I would like to talk first about the reasons why qualitative research methods are important to the study of graduate attrition. Then I would like to illustrate this by talking about the study that Vincent Tinto and I are doing on the effect of financial aid on graduate persistence.
Why are qualitative research methods important to the study of graduate attrition? First of all, graduate education is decentralized; a lot of the decisions are made at the departmental level. The graduate institution makes overall policies, but the way they are interpreted, often at the departmental level, is what counts. Selection – admission decisions – are an example of this, as are decisions on the awarding of research assistantships and teaching assistantships. Graduate education is field- and institution-specific. We talked a lot about that before, so I would not go into that. In the final analysis, decisionmaking takes place within the context of the department. I cannot tell you how many students I have interviewed and talked with about the institutional climate, and then they go right back to their department context when they answer me. Graduate experience is shaped by departmental policy. Charlotte Kuh said this morning that there is a diverse set of institutional and departmental practices. Departments are very different, even departments in the same field. Institutions are very different, so when we are interpreting data we really have to take that into account. The graduate experience is also shaped by the field of study, and there is a lot of variance across fields. We talked about that this morning too.
Third, graduate experience is heavily influenced by relationships with faculty members and graduate students' peers in the department. I think Vincent Tinto said this morning that some of those experiences might involve two or three faculty members, but they are very intense relationships. So quantitative data need to be interpreted within the context in which they are collected. That context is usually the department, and it takes qualitative research to provide the context and to tell what was referred to this morning as the "complex story."
I am going to describe briefly our study and the qualitative research design for the NSF study. The qualitative design focused on minority students or underrepresented minority students, more specifically African-American and Latino students. We were seeking to find out how different types of financial aid affect time to degree and completion of doctoral degrees among graduate students, how the timing of different types of financial aid affect time to degree and completion, and how these effects vary across different fields and types of programs. I am just going to give a brief description of the study, because it is not the main focus of my talk.
We selected two institutions and, since the departmental context is so important, looked at a natural science department, a physical science department, a social science department, and an engineering department. And as you can see from the departments selected, in the first institution we had to add the English department because of one constraint that we come across a lot when we are doing research on under-represented minority students: you do not have the numbers. I am sure everybody in this room can agree with that. So those are the departments that we selected from within the larger sample to focus on. At the departmental level we interviewed the graduate chair; three or four faculty members; administrative staff who worked directly with students, especially around financial aid and financial support; and African-American and Latino students. We ended up interviewing and having focus groups with about 30 students at each institution.
At the graduate school level, we talked with the graduate dean or the associate dean, and the administrator of fellowships – that person could be an associate dean or somebody in the financial aid office. As I mentioned in choosing the departments and the students, we had to consider a couple of issues. One was that we had to choose the departments that had a high enough enrollment of minority students. That was very difficult because many of these especially the science departments had one or two minority doctoral students in a department. So that really drove our choice of departments.
We collected data via semi-structured interviews, as we had an instrument, and focus groups. The interviews with the faculty members and administrators focused on the financial aid policies both at the institutional level and at the departmental level – the availability and distribution of graduate assistantships (teaching assistantships and research assistantships) and how they were distributed by racial ethnic group and gender. We also looked with great care at the procedures for awarding these teaching and research assistantships at the departmental level and at the institutional level. We looked at how fellowships were awarded, because those decisions are often made at the graduate institution level. In our interviews and focus groups with the graduate students, we looked at their expectations of financial support (how financial support affected their enrollment decisions, for example), what they expected when they got there, whether or not their expectations were met, the adequacy of financial support, and what they thought the advantages and disadvantages of differing financial aid were. We found this to be a very rich source of information. The participants had some very strong ideas about the disadvantages and advantages. I will share those in a little while. They had lots of suggestions for change and the conditions and terms of the different types of financial aid.
We wanted to know about their perceptions of how to change the packaging of financial aid to increase persistence. There was a bit of discussion at the end of the last session about that, and I will share our findings about that. We talked to them about not only the packaging of financial aid, but also how they felt having adequate financial support increased persistence and time to degree. We collected background questionnaires on all of the students that included demographic information, work information, whether they were working, how many hours they were working, their status in the graduate program, whether they had completed their course work, whether they had been advanced to candidacy, and the types of financial support that they had received throughout their time in the program.
I will now mention some of the issues to be addressed in collecting qualitative data. In obtaining access to student information – I think Vincent Tinto talked about that a bit – confidentiality is an issue. We asked the graduate deans for a letter introducing us to the departments and to the faculty, and that worked pretty well. We also worked with the departments directly, and we had an onsite person who worked directly with the departmental staff in choosing students. Scheduling the interviews and focus groups with the students and faculty worked out pretty well. We had an onsite person doing that also. Confidentiality, of course, is always an issue, and we had a statement that we sent with our letter of introduction plus we read a statement before we started our interviews and focus groups.
Doing this kind of work is expensive. There are high costs associated with qualitative research. It is very labor intensive if it is to be done well; for example, for a focus group, you really need two people and it costs a lot.
Now I would like to share, in light of the discussion that emerged this morning, some of our findings from the focus groups and interviews. Also, in doing this I want to show how qualitative research can really provide more depth of analysis. Students seemed to be very dissatisfied with the amount of the TA and RA support they received. However, by being on campus and talking with people, we found that in the past six years the cost of living had almost doubled in that area and the RA-ships and TA-ships had not increased to cover that. This lends another explanation as to why students would be so vehement about their dissatisfaction, and it also will lead us to re-analyze the data and to look at whether the cohorts in the last six years are much more dissatisfied than the previous charts. That is an example of how you can use qualitative data to enrich your survey findings.
We also found that health benefits were considered to be a real disadvantage of the fellowship. Students on fellowships did not receive health benefits, and the fellowships were taxable. That is something that did not emerge in the survey. Conversely, we found that students who were on fellowships did not feel alienated. I know we talked about that this morning. From their point of view, getting a fellowship put them in line to get really good research assistantships because the professor in the lab did not have to support them, and so they were very happy to get them. That was their take on fellowships. Students on fellowships did not report feeling alienated; I think there was only one student out of the 30 who said that he felt alienated.
In talking about the disadvantages and advantages of the TA-ship, we thought that they would be much more negative about TA-ships than they actually were. They felt that having a TA-ship helped them to learn the material, to find out whether they wanted to teach rather than do research in an academic position. Their big complaint about TA-ships was that it took more time than they actually had. Their suggestion was that the system be structured so that students with expected TA-ships are only to give a certain amount of time. Apparently there is a wide variation in the amount of time that is expected from a TA, and they felt that it should be more closely monitored and structured so that they were not expected to give that much time. There were two complaints about RA-ships. One was that they forced them to focus too closely on a specific area and did not allow them to get a broad range of experience in their research. The other, which was really much more important for them, was the tyranny of the faculty. They felt that faculty were very, very influential in how the TA experience turned out, and they had total control. There was no recourse.
The participants were pretty positive about fellowships in general. I mentioned the health benefits and having to be taxed. They did not mind having to be taxed as much as not being told ahead of time that this chunk was coming out of their check. In terms of the packaging, and we talked about that this morning, they felt that getting a TA-ship at the beginning of the graduate experience was probably the best arrangement, the best packaging and then at the end, they were very adamant about the need for a fellowship. They thought the RA-ship was probably the ideal in between.
What is the contribution of the qualitative component to the overall study? The qualitative component allows us to explore the quality of student experiences, specifically those of minority students. It tells us how the events and outcomes identified in the survey results are linked in time; it adds depth to the analysis; it helps us to cross-validate research findings I think almost everything we found in our qualitative component confirmed our survey findings – and it opens up the analysis to unexpected findings. The net result is not only more reliable research, but also research that is better geared to the development of future policy.
Chris Golde, University of Wisconsin, Madison: I have a question about context, going back to what we talked about this morning in terms of disciplinary differences. I was speculating (and I have done some qualitative research in the area as well) that the definition of things, such as what it means to have a research assistantship and what it means to pick a dissertation topic, varies among the disciplines. To say that a fellowship allows one to have freedom of choice in terms of dissertation topic or gives one freedom seems to me a much more relevant point for a scientist, who is intensely dependent on the principal investigator whose research is funding your graduate work typically. A fellowship gives you some freedom and flexibility yet it allows you to work on a project because you are free to the researcher and it allows you to pick the researcher with whom you work, as opposed to having to pick a researcher who has a slot for a student. That has a really different meaning than for a historian or someone in English where a fellowship means that you get to sit at home and work on your dissertation or sit in the library. I just wanted to hear a little bit more about context. If we do not contextualize our research discipline by discipline, we start to speak in such vague generalities that it almost loses meaning.
Toni Clewell: You are right. We did get some disciplinary and field differences in the way people looked at their TA-ships and RA-ships. Actually, there were not as much as I expected. The biggest difference was around the RA-ship, as you mentioned. The people in the sciences felt that the RA-ship was much more helpful to them, whereas the people in the social sciences felt that the TA-ship was much more of a positive thing. People in the sciences felt that being a TA was a waste of their time and they wanted to be in the lab all the time. Like someone said, "Gee, they are paying me to do what I love to do." So that did emerge as a difference.
Carla Howery: We are going to continue the conversation about departments with Barbara's [Lovitts] presentation. Barbara is going to talk more about the departmental climate and what it is about departments that make it more less likely for attrition or retention to occur. She has made the case that looking at factors that affect the individual is either misplaced or overstated, and that we need to look at the contextual effects of departments using this qualitative approach. Barbara is a sociologist who has worked at the AAS and now works in the graduate school at the University of Maryland.
Contextual Factors in Graduate Student Attrition
I am probably one of the least familiar faces on this panel, and my former colleagues here at NSF, where I worked for five years until about a year ago, are probably wondering what I am doing on this panel. I had a double life while I was here. Perhaps the best autobiographical introduction of myself is the title of the press release that went out on my research from the University of Maryland last spring. The title was " Two-time Ph.D. Dropout Succeeds Third Time with Dissertation on Ph.D. Dropouts." Unlike Charlotte's phantom attrition, my attrition was real. There were two of them and they were profound experiences. I first came to the idea of pursuing the issue of the causes and consequences of departure from doctoral study while I was leaving my second doctoral program in November 1985. I was finally able to complete a dissertation on that subject in December 1996.
In my talk today, I am going to focus on contextual factors in graduate student attrition. If you look at most studies of graduate student attrition, they begin with the assumption that there are differences in the background ability and motivation between those who complete the degree and those who do not. These studies contain an implicit deficit model of the student who leaves. They ask the implicit question, "What is wrong with the student?" Rather than focusing on differences among students, I am going to focus how contextual factors shape student attrition independently of any characteristics students bring with them to the university. I am going to focus on department environments for integration, and I am also going to look at some of the differences between high and low productive Ph.D. faculty. My research is highly exploratory and most of the analyses I did on the data I am going to present today were unplanned and serendipitous.
This slide shows the departments in the study. I was interested in labor market consequences of attrition, which is irrelevant to this talk. The study included three departments from the sciences, three from the humanities, and three from the social sciences.
I want to give you a brief overview of the sources of data I used in my entire study. I conducted a survey of the 1982 to 1984 entering cohort in 9 department in two universities, for a total sample of 816 students?that includes both completers and non-completers. The universities in the study were among the top 40 Ph.D.-producing institutions in the United States. Their identities are confidential. However, one was in an urban location and one was in a rural location. The reason for that selection was for studying labor market issues as well.
I conducted hour-long telephone interviews with 30 of the non-completers, roughly two from each department, in order to get at issues that could not be explored adequately by the survey instrument. I also conducted half-hour telephone interviews with the directors of graduate study at each of the 18 departments in order to get background information on the departments' formal and informal structures and processes for educating graduate students. I made site visits to both campuses and walked around all the departments, every floor; poked my head into student TA offices; sat in graduate student lounges; when I could, eavesdropped and took notes. You get a lot of information when you eavesdrop.
While I was at the campuses, I also conducted half-hour, face-to-face interviews with 30 faculty members, again roughly two from each department, who would have been on the faculty when my cohort was going through. One faculty member from each department was chosen because he or she was high-productive of Ph.D.s and the other was low-productive of Ph.D.s. I had Institutional Review Board permission to withhold that information from the faculty members that I was interviewing. The results would have been different if they knew that was why they were selected. These interviews were conducted in order to explore systematic differences in faculty members' attitudes and beliefs and behaviors towards graduate students and graduate education.
Turning now to the department environments for integration. The first column on the slide shows the basic issues that I asked the directors of graduate studies about. I have broken them down into academic and social substructures or subscales. I asked the directors of graduate studies, "Do you have these different activities and events in your department?" or "How frequently do you have them?" I also used my field notes come up with some additional information about the departments. Based on the information provided, I computed an integration score for each department. Some of the data were weighted. For instance, if a department had colloquia once a week, it got more points than a department that had colloquia once a month, and more points than a department that had them occasionally.
The student survey asked a parallel set of questions, but the questions were not "Does your department have these structures and activities?" Rather, the questions asked, "Did you participate in these activities?" or "How frequently did you participate?" I was able to extract a department mean for student participation based on the student responses, so we are not going to be looking at differences in participation of completers and non-completers, rather we are going to be looking at department averages.
The next slide shows what happens when I correlated each department's integration score with its attrition rate. For all of the factors that you saw, we get a significant negative .41 correlation. When you correlate the departments' attrition rates with their academic integration sub-scores the correlation rises to .54 and becomes more significant. For those people from the hard sciences in the audience, I know you like to see .9s and higher, but when you deal with social phenomena, there is a lot of noise. For people who work with social data, .4s and .5s are actually pretty good. What these data are saying is that the more resources a department has for integration, academic integration in particular, the lower the department's student attrition rate is going to be.
The next thing I looked at was the student-integration-in-the-department scores (the department averages for student participation, against the department's attrition rate). Although none of the correlations was significant, they were all negative, they were all going in the predicted direction, supporting the hypothesis that the more integrated students are in the department, the lower the students' attrition rate is going to be. So there is some evidential support that students' participation in departmental activities affects their attrition rates.
Finally, I looked at the relationship between the department environments for integration and the student-integration-in-the-department scores. I got a fairly strong and positive correlation between the department's academic environments and the student-academic-integration-in-the-department scores. What this is saying is that the more resources a department has for academic integration, the more academically integrated its students are going to become.
The bottom line is that graduate student attrition appears to be, in part, a function of the social organization of the department. It can be explained, in part, in terms of departmental factors and without reference to individual student characteristics.
I should also note at this point that I was able to calculate a faculty turnover rate for a 10-year period for each department, and when I correlated the faculty turnover with the student attrition rate I came up with a positive and significant .6 correlation. That finding indicates that factors in the department environment affect the attrition of both its faculty and its students.
I am now going turn to the role of advisor. If you look at the literature on advisors and mentors, it presents faculty as monolithic entities. It assumes that they are all the same and that they all interact with students in the same way. Similarly, if you look at studies that ask graduate students to rate their advisors on a variety of dimensions, you find that students who completed the Ph.D. rate their advisors quite differently from those who did not. These studies try to explain the differences in terms of students or in terms of the differences in judgments faculty make about students.
I am going to present data that suggest that it is not so much the differences among students, but differences among the faculty with whom they are working. I am going to explore the differences I found between high and low Ph.D. productive faculty, and then I am going relate these differences to students' satisfaction with their advisors and use student satisfaction as a proxy for advisor productivity; I will make that connection as we go along.
No matter how I asked questions about productivity, the high producers worked with more students. They produced more students and they were currently working with more students than were the low producers. In fact, they had been on the faculty fewer years than the low producers, so you cannot explain the difference in their interaction with students and their productivity in terms of their tenure in the departments.
Turning now to their academic interactions, I asked the faculty members how they established relationships with their students. High producers tended to cite two to three ways in which they came to work with their students. Low producers typically only told me one way. There was an overlap between the two groups in the methods by which they acquired students, and I am going to highlight a few of the things that really distinguish them.
High producers were more likely to tell me that students took classes with them and asked them to be their advisor, that they had interests in common with their students. High producers, unlike low producers, expected their RAs and TAs to ask them to be their advisors. Four of the low producers which was about 25 percent of the low producers sampled acquired their students by default. They told me that the student's advisor had left or died or the student did not get along with his or her first advisor so they ended up with the student. Basically, the connection between students and low producers was not made in the same way that it was between students and high producers.
I asked the faculty about the amount of time they spent interacting with their students typically. The difference adds up to roughly 20 minutes a week, which at one level is not a lot, but if you consider 20 minutes over the course of a several-year relationship, it is actually a lot of time. What is more important here – and this is one of the advantages or benefits of using qualitative data – was the gloss that surrounded the answers. High producers were much more likely to say "It depends," and to pay attention to variation in student needs. They were much more sensitive to the nature of the individual student and modulated their degree of interaction with the student based on their perceptions of how much the student needed to interact with them.
When I asked them where they interacted with their students, both groups of faculty mentioned roughly the same number of settings but the settings were different. Naturally, everybody mentioned their offices, but high producers were much more likely to talk about informal settings and the low producers were more likely to talk about formal settings. High producers mentioned coffee shops, lunch places, the hallway, their homes; the low producers were much more likely to talk about the classroom, a seminar room, a conference room, or telephone contact.
I also asked about collaboration, which, of course, varied by discipline because people work much more collaboratively in the sciences. Regardless of discipline, the high producers were much more likely than the low producers to work collaboratively with their students. I also found some other interesting things. The high producers seemed more likely to scaffold their students' development; that is, they were more likely to work closely with them in the early stages of the relationship and then pull back and let the students assume a greater and greater degree of control over the research as time when on. By contrast, it seemed to me that low producers were more likely to engage in direct instruction and then leave their students to their own devices. This type of finding, which is derived from looking at people's language, is not something that you can get at in survey research.
Another thing that I was struck by when I looked at the language the faculty used was that high producers were much more likely to talk in terms of "we," when they talked about their graduate students "We do this," "We sit down together," "We planned this together." Low producers, by contrast, were much more likely to talk in terms of "I" and "them." Language gives you insight into the way faculty think about graduate students, their world view.
I also asked about issues like co-authorship. The results again varied across disciplines. Not surprisingly, high producers were more likely than low producers to collaborate with their students. But the high producers were also more likely to let their students be first or sole author, whereas the low producers tended to have rules and policies about authorship.
I asked faculty whether they encouraged their students to engage in the professional tasks of the discipline like subscribing to journals, joining professional associations, and attending meetings. Again, high producers were more likely to actively encourage their students than low producers, whereas the low producers were more likely to defer to the students' subculture, saying, "Oh, they pick it up from other graduate students," but that is not always true. High producers were also more likely to pay for or go through departmental channels and to come up with funds for their students to attend professional meetings, whereas the low producers tended to comment, "Oh, when the funding allows."
Finally, another important academic difference relates to helping students find jobs. High producers used more means to help their students find jobs and they worked a lot harder at it. In fact, several of them prefaced their remarks by saying, "That is the most important thing I do." They were also more likely to actively network, whereas low producers were more likely to simply pass job notices on to their students and leave the rest up to the students.
Looking at social interaction, these are the same activities I asked the directors of graduate studies if they existed in their department. This time, I asked the faculty if they themselves participated and how frequently. Again, across the different activities, the high producers engaged in these activities more frequently and more enthusiastically than did the low producers. A few things stand out about the high producers. They were more likely to take the initiative to interact with their students, to engage in social interactions with them. In fact, several of the high producers but none of the low producers referred to their graduate students as "friends." I did not hear a single low producer refer to his or her graduate students as friends.
In terms of having students to their homes, high producers had students to their homes at least once a year and sometimes up to five times a year. Several had holiday parties. Some were concerned about students being orphaned over the holidays. High producers were also more likely than low producers to have their students over individually as opposed to in groups earlier in their graduate careers. When low producers had students over individually it was usually to celebrate the completion of the Ph.D., which obviates any integration benefits that are going to come from that type of relationship.
Overall, the data begin to suggest that there are some large differences in the way high producers and low producers interact with their graduate students and the way they help them through their programs and into their careers. Although I did not match students with their advisors, I do want to discuss some evidence that suggests completers and non-completers are working with different types of advisors. Completers were more likely than non-completers to have advisors. Completers were twice as likely as non-completers to select advisors whose intellectual interests were closest to theirs. Non-completers were six times more likely than completers to have been assigned to their last advisor, and changing advisors does not account for that difference.
Some of the students I interviewed mentioned their advisors by name and in other instances, I could figure out who some students' advisors were. I also had a check of sorts on advisor productivity. I hired an undergraduate to go through about seven years worth of Dissertation Abstracts International and tally each time he saw the faculty in the departments listed as chair of a students' dissertation committee. So I had a sense of who was producing Ph.D.s and who was not. When I looked at the productivity of the advisors of students in the interview sample who had negative experiences with their advisors, their advisors were definitely low productive. Those who had better experiences with their advisors had advisors who tended to be in the top half in their departments in terms of productivity. Students who had positive experiences with their advisors were also more likely to have left the university for external reasons like marriage or injury.
The survey also asked, "How satisfied were you with your last advisor?" That question was followed up with an open-ended question: "Why was that?" Using students' case IDs, I matched their rankings with their open-ended responses and analyzed them. Completers were clearly much more satisfied with their advisors than non-completers, and the non-completers were much more dissatisfied with their advisors than the completers. When you look at the open-ended responses, which fell into seven basic categories, regardless of completion status, students who were satisfied with their advisors wrote about them in terms that were consonant with the manner in which high producers talked about the way they interacted with their students. Students who were not satisfied with their advisors tended to write about them in ways that were consonant with the attitudes and behaviors of lower productive faculty.
Using satisfaction with advisor as a proxy for faculty productivity, this slide shows that the more satisfied students were with their advisors, the more likely they were to engage in sports activities, the more likely they were to engage in other social activities with faculty and graduate students, and the more likely they were to attend colloquia. The more satisfied the students were with their advisors – non-completers in particular – the more likely they were to do things like subscribe to a journal, attend professional meetings, and belong to professional associations. So I have some evidence, although it is only preliminary evidence, that completers and non-completers are working with different types of faculty and that these interactions affect the nature of their experiences in graduate school and quite probably their persistence outcomes as well.
To conclude, I would say that the social structural context of graduate education affects persistence outcomes independent of student characteristics, and that there is a lot more research to be done on the social forces affecting graduate students as a whole.
Participant: We have tried using semantic differentials, where you ask people to pose themselves on the extremes of two oppositional descriptions of a context, for instance for faculty, a warm welcoming inclusive faculty, a cold exclusive. You can construct the semantic differential scales as a way of capturing that with one or two questions.
Barbara Lovitts: I have quantified some of these data, and I certainly think that it is much easier to quantify the department environments for integration that I just presented. In fact, I developed a preliminary survey for departments. Addressing the faculty issue is a whole different can of worms. There are a lot of methodological caveats that I have not mentioned that go along with it. There are a lot of different reasons why people are high and low productive. You have to take differences among departments into consideration as well. A high producer in mathematics might produce 6 Ph.D.s over a 25-year career, whereas a high producer in a chemistry department producing 66 Ph.D.s over a 25-year career is not unheard of. Also, you have to look at the size of the department. In large departments with a lot of graduate students, faculty have more opportunity to work with graduate students than do faculty in departments with few students.
If you were also to do a survey, you would get different non-response rates from high and low productive faculty, and I think there would be differences in reasons for those non-responses. The high productive faculty tended to have very well-funded research programs. They tended to be more eminent, and they were also out of town more because as researchers they were so productive. Whereas, low productive faculty, based on personality differences, might simply not be disposed to answering surveys. Ultimately, a lot of the differences between high and low producers boil down to differences in personality. Of course, the next question is, "How do you deal with that issue?" It is true that you cannot change faculty's personalities, but I do believe that departments have corporate responsibility for their students. The department has to find a way to compensate for students who are working with low productive faculty.
Participant: You keep saying a lot of this was independent of student characteristics, but I did not see any point at which you controlled for student characteristics.
Barbara Lovitts: I deal with student characteristics in other parts of my research that I have not been able to discuss here. It is clear that students are having different experiences with different faculty. As someone indicated there might be a selection effect with bad students choosing bad advisors, but I don't really believe that.When you look at the assignment to advisor issue, I think that boils down to students not being told that they can change advisors and feeling that they have a choice between staying on with a person that they do not like and get along with or leaving – and many end up leaving. As I said earlier, all of this is highly exploratory. I think the point that you are raising is valid, but the point that I am making is that advisors do not interact with their students in the same way and that these different types of interactions shape students experiences in graduate school. I would encourage a lot more research on this issue.
Carla Howery: Let us turn to our third presenter who is Willie Pearson, professor of sociology at Wake Forest University. Around our office we call him Mr. Science because he studies careers of scientists. In this presentation, he is going to talk about some of the factors that affect attrition or retention of African-American and Hispanic students in graduate school.
Research on Graduate Student Attrition
Thank you. In the interest of time, I will summarize my report. This study was co-authored in 1990-91 with Gail Thomas of Texas A&M University and Toni Clewell, a previous speaker. The methodology was essentially the same as that described earlier by Toni and Vince Tinto. We conducted case studies of the graduate education experiences of African-American and Hispanic doctoral students at six highly rated public institutions. All student respondents were U.S. citizens. I point out citizenship status because in the initial stages of identifying a pool of institutions, we discovered that one institution reported large numbers of Hispanics in its biology department but that none were U.S. citizens.
The study focused on three departments. At each institution, we visited the departments of psychology and education. However, the third department varied from one institution to another because of the small numbers of African-American and Hispanic doctoral students. The third department included the following: history, criminology, sociology, biology, and life sciences. The site visits were conducted over two to three days. All of the interviews were conducted by the principal investigator and an assistant who had considerable experience conducting interviews and focus groups.
We examined documents on the history of the university and the school's mission statement. We also observed the physical surroundings, especially departmental bulletin boards. In some instances, publications were available on minority graduate student initiatives. This allowed us to construct more detailed case studies. From these case write-ups, we produced a shorter version for the final report.
Now I will summarize some of our findings from the case studies and cross-case study analyses. I should mention that the larger study also included an analysis of survey results. I will ask Toni to fill in any omissions.
To place the findings in context, the larger study examined the following issues: admissions, financial aid, campus and departmental racial climate, and retention. I will focus primarily on the retention and racial climate issues. We interviewed students, faculty, department chairs, graduate school administrators, and minority affairs administrators with regards to minority graduate student recruitment and retention efforts. Not surprisingly, there was considerable divergence in interviewee responses.
Focus group responses pointed out several factors related to retention. For example, most students pointed out the importance of minority faculty and financial aid. The interviewees revealed that the financial aid issue dealt with both amount and duration. Many students reported that most of the aid was for 9 months, which caused considerable financial strain because their apartment leases were for 12 months.
Students reported that most classmates who failed to complete their programs left because of finances, not academics. However, response patterns varied across disciplines. For example, at five of six institutions, the retention rates were high for psychology students. We found policies and practices regarding minority student retention to be more similar for psychology departments than other departments. In sharp contrast, we found only two of six education programs with high retention rates. In psychology, the typical financial aid package was a mix of funds. For example, during the first and dissertation years, most students held fellowships, and in the middle years they held assistantships.
We found that most departments that had access to university-wide minority fellowships usually funded minority graduate students from those funds rather than from departmental funds. Sometimes this resulted in differential professional experiences. For instance, some African-American students in one psychology department were surprised that most of their white peers at the dissertation stage had co-authored articles with professors with whom they served as research assistants.
It is noteworthy that students who had mentors seemed to have higher retention rates. In a few instances, the mentor was in a department other than that of the student. For example, the retention rates were very high in the biology department at one institution. The success was attributed to a strong mentoring program. Early in their graduate studies, students were engaged in collaborative research with faculty members. Students were also very supportive of each other. Students reported that the faculty members were very supportive.
Students in education programs seem to be a special case. Compared to other minority students in other disciplines, education students tended to be older and more likely to work while pursuing their doctoral studies. In particular, they were more likely to complain about inadequate financial aid. They were less well-integrated into their departments. Also, they complained that white faculty held low expectations of them. Students from other disciplines also reported this attitude.
Education students complained about the narrowness and the Eurocentric focus of their curricula. They noted that the multicultural perspective was denigrated and students interested in research focused on minorities required extraordinary justification. Many education students reported that when their professors covered minority issues in class, they were asked to express an opinion only because of their minority status. Most reported that they were uncomfortable with this situation because they lacked knowledge of the particular subject matter and felt professors were insensitive.
Returning to psychology, most of the faculty across institutions believed that all of their students were capable of doing well and that student/faculty collaborations were encouraged. These departments were characterized by mutual respect among students and faculty. Faculty members were usually viewed as supportive. Faculty also encouraged student study groups. A mechanism was in place for students to resolve grievances. Further, students played active roles on departmental admissions committees. We discovered that even in psychology there were some program differences. For example, the clinical programs tended to have more supportive environments, more positive racial climates, and more minority faculty.
There were interesting findings from some of the third departments. In the life sciences department at one institution, the retention rates were very high. However, minority students were not well-integrated into the student culture. This appeared to be related to age differences because the minority students were older than the white students. Students tended to socialize along age lines. In sociology, the retention rate was very high. This was due more to student rather than faculty efforts. The minority students organized study groups for their statistics and methods courses. We found similar activities in other disciplines as well. However, a faculty member did offer an orientation program on departmental academic expectations. Additionally, most minority students worked very closely with the minority faculty in the department. Although the retention rate was high in the criminology department at one institution, many students were leaving early to accept job offers. The retention rate in the history department at one institution was very low. According to student interviewees, faculty members were alienated from each other. Students complained of inadequate funding and a poor environment in which to learn.
In sum, we found not only institutional differences in terms of policies and practices regarding retention-related issues but also program/departmental differences within the same institution. Within departments, there were some program differences as in the case of the aforementioned clinical psychology program. In departments where the racial climate was positive or respectful, minority students had high graduation rates. In short, in departments where all students are supported and valued, minority students fare equally as well as their non-minority peers. For example, this was the case in the criminology department, where there were no minorities on the faculty. However, minority students pointed out that the faculty was supportive and that mentoring was greatly valued. Thank you.
Charlotte Kuh, National Research Council: Do we already know what we need to know about attrition? Should we just go into these departments and do something about it, or is there something NSF should be doing?
Participant: I am always in favor of more study, just on principle, but I think we have some really good information about what kinds of things tend to work to ensure that students at least have a better shot at being successful. This goes back to what we were saying: It has a lot to do with how you fund students and the extent to which we can require faculty to behave responsibly by structuring into support programs certain kinds of behaviors that will sustain students over a period of time – whether we do it through traditional grants and simply require more of PIs. in terms of accountability for graduate students, or whether we do it through moving in a more aggressive way toward traineeship programs. While I think we should continue to study, we need to begin to apply what we know to the way we provide funding for graduate students in the U.S.
Harvey Waterman, Rutgers University: I think we have a political problem. I have gone to I do not know how many meetings at which somebody has stood up and told me that the attrition rate for graduate students is 50 percent. That sounds horrible, but it is not as bad as it sounds. Create a picture that is credible and is backed up by careful study of the actual attrition situation, so that we can say in more than an anecdotal way that yes, there is some real attrition, and yes, there are some things that we could do better in some places. But 50 percent of people do not drop out because we have screwed up and the programs are no good. Obviously, it is an empirical question.
We have heard a lot of reasons today for why attrition might not necessarily be bad, and we have heard some reasons why attrition can be explained in other ways. We just heard about criminology losing students because of the job market. A week does not go by when I hear from somebody in computer science or mechanical engineering to the same effect. We need to be able to display this information in such a way that we can make a case for the reality.
Vincent Tinto, Syracuse University: It seems that we are moving from posing the question "Why do students leave?" to "What can institutions do to enhance the likelihood of students who want to stay will stay and finish?" That is the appropriate question that we can learn from.
Catherine Gaddy, Commission on Professionals in Science and Technology: It is interesting that psychology programs, at least in clinical counseling schools, are accredited, so somebody is watching their data and they are accountable. Just thinking about racial and ethnic retention rates, medicine and law have done a better job than any of the other sciences. If you ask for some of the data and you make people accountable for it and there are some incentives, there are better retention and completion rates. These other fields do not have those same kinds of structures, though.
Harvey Waterman, Rutgers University: They seem to have had a longer history of minority funding and were pretty well-organized on that across the institutions also.
Participant: I see one of the major advantages of qualitative research as helping us to formulate what we are going to study quantitatively. Even those of us who do quantitative research use tools like focus groups to help us with the questions. That helps to bring us along on some of these last comments from the qualitative side to understanding how people's insights, which come from talking to people, play out in the bigger picture. We could be asking questions such as "Do you have a mentor?," preferably in a longitudinal survey. If we go back after the student has dropped out, then that whole experience may be affecting their perspective, so I do see a lot of interplay here.
Participant: Let me respond to the question of do we need more research or should we be implementing programs. Action research actually does both. Action research is familiar to people in education and many of the social sciences but it probably not a concept familiar to people in the humanities and the hard sciences. It is a type of research in which you are implementing change and conducting research on the change as its happening.
Claudia Mitchell-Kernan, University of California, Los Angeles: I just want to make a couple of observations because I think that the presentations that have been made this morning point us in certain directions. One of the things we have known for a long time, or thought we knew, was that environments could be more or less supportive. We also thought we knew that getting students involved in research early on is very critical. It is a complex ecology, and it is very surprising to me that Barbara [Lovitts] did not find any disciplinary effect, because getting students involved in research early on and building collaborative undertakings that involve several graduate students and faculty very much are related to research grants and funding. Thus, in my own examinations on my campus, those patterns were most common in the life sciences and natural sciences, and least common in the areas were there is very little funding such as the humanities and social sciences.
The other thing that I want to mention is that the faculty time spent with students also seemed to vary a great deal. The kinds of questions we asked on our exit survey are not really as comprehensive and did not give as many angles of vision as your particular research. We did ask, "How may times a week do you see your advisor?" Students in the sciences typically said three to five times. Students in the social sciences departments that we looked at most closely said once a quarter; that is once every 10 weeks. When we asked the faculty what was going on in the natural and life sciences, they said, "We really work hard over here in the life sciences. We don't know what you people are doing over in the social sciences." But I know better because I have lots of very fine colleagues who are in both areas of the campus. The social science departments have tended to be much, much larger. When you get an entering cohort of 80 students in biology versus 200 in history, and you have a faculty size that is roughly the same, you have a very, very different kind of environment. I think in the Bowen and Rudenstine study, retention was indeed related to size of departments.
The last thing I want to say with regard to the supportive environment issue is that you have to remember that there are cohort effects within departments, too. As a department goes through a life cycle, you can find that the quality of environment for graduate students can differ very, very radically. If you end up with too many people on either end of the age range, you may find something less than a supportive environment.
