"I've heard some very scary stories about professors demolishing candidates' dissertations during the final defense. How do I avoid that happening to me?"

The usual ultimate step in the pursuit of a doctorate involves the candidate meeting with a panel of professors to defend the dissertation. Whether a master's degree thesis must also be defended orally depends on the policies of the institution the student attends.

It is not uncommon for the oral defense to be only the penultimate event--the next-to-last step--if the examining committee decides that improvements are needed in the student's product. In that case, the session with the candidate includes committee members specifying desired changes. Then the candidate's chief advisor or another member of the committee accepts responsibility for ensuring that the revisions are completed satisfactorily before the committee members affirm their approval by placing their signatures on the work.

Difficulties that may arise during the oral defense can often be foreseen, so that candidates who are aware of potential problems can be prepared ahead of time to wend their way safely through the minefield of professors' questions and suggestions. The purpose of this chapter is to identify some of the more common problems and to propose ways of solving them. The style of presentation is the same as that of Chapter 9: Things That Go Wrong. In each of the following cases, a student describes worrisome incidents that may occur during the oral defense, and a faculty advisor suggests ways to cope with such incidents.

The seven cases concern (a) the validity of research findings, (b) a study's significance, (c) the candidate's proper role, (d) the advisor's proper role, (e) professors objecting to the student's research method, (f) committee member debates, and (g) inadequate proofreading.

THE QUESTION OF VALIDITY

Student: "Dr. Johnson looked over my research plan and said my study lacked validity. He says there's no way to validate people's expressed feelings about being Latin American immigrants. When I asked what kind of validity he wanted, he said I needed to do a study in which I could use predictive validity or concurrent validity or something of the kind. How can I defend my study as valid?"

Advisor: "You can start by adopting a general definition of validity, then show how your study qualifies under that definition. Perhaps Dr. Johnson has a different definition, so what you are doing in your project may not meet his criteria. In that case, when you defend your work before the thesis committee, you can hope that the majority of the committee members--and maybe Dr. Johnson as well--will be persuaded by the way you argue your case. I'll suggest one approach you might try.

"You start with some such definition as this: 'Research is valid to the extent that its outcomes convincingly answer the questions on which the study has focused--including both the original questions and ones that may have emerged during the study.' This definition should encompass the kinds of validity that Dr. Johnson seems to have in mind as well as a kind you can offer for your study. You should realize that Dr. Johnson is a psychologist who works with tests. Psychologists typically say that a test is valid if it measures what it is supposed to measure. So, a test has predictive validity if it accurately prophesies some future outcome. For example, a college aptitude test is valid if it accurately predicts the grade point average students will have compiled by their senior year in college. Thus, the question answered by predictive validity is: How well does an evaluation of a phenomenon at an earlier time foretell the status of that phenomenon at a later time?

"Concurrent validity means that an assessment--as with a test or interview-correlates highly with a different sort of appraisal made during the same period of time. Therefore, a type of interview that leads to the same assessment of workers' abilities as that found in their on-the-job productivity can be considered to have high concurrent validity. In the case of both predictive and concurrent validity, statistical evidence--such as a correlation coefficient--is typically used to reflect the level of validity. The closer a coefficient approaches 1.00, the greater confidence one can place in predictive or concurrent validity. So a correlation of .85 is more convincing than one of .36 or .19.

"But many studies don't lend themselves to quantitative appraisals, so you need to depend on other means of convincing people that your results satisfactorily answer your research question. For instance, in writing history, you are obliged to furnish persuasive evidence and a line of logic to support your contention that the events you depict did, indeed, happen as you declare. This can involve your drawing from different descriptions of the same episodes to show consistencies among different witnesses' accounts, or it may include demonstrating how the varied motives of different narrators explain their conflicting interpretations of events."In your study of Latin American immigrants' attitudes, your problem is to present a compelling argument that their expressed attitudes are your respondents' true feelings. There are several ways to build your case. First, you can demonstrate that your methods of gathering information--your interview technique--did not cause the interviewees' to give false answers, either inadvertently or intentionally. Inadvertent distortion could result from a respondent's not understanding the language adequately, so you need to show that the interviews were conducted in language that respondents could comprehend. And your immigrants could have intentionally falsified their answers if they thought you wanted them to report a particular kind of feeling, so they expressed such an attitude in order to please you. Or they may have feared that admitting to certain feelings would get them into trouble, so they feigned opinions that they thought would be more acceptable. Therefore, you need to convince readers of your dissertation that your data-gathering procedures were ones that would evoke respondents' authentic beliefs."Next, in your presentation of your interview results you can buttress your claim of validity in several ways. First, you can identify common themes that run through several respondents' tales, thereby suggesting that Latin American immigrants share certain attitudes in common because of their similar origins and their current status as immigrants. Second, when you find differences among respondents' expressed attitudes, you can hunt for reasonable explanations of those exceptions. Frequently the explanation will be reflected in the respondents' own phrasings, which you can quote in support of your interpretation. Third, you can compare your findings with ones in the professional literature to show how your results are in concert with other researchers' views."Some professors avoid debates about traditional experimental concepts of validity by adopting other terms--verisimilitude, authenticity, trustworthiness, cogency, quality, or conviction--to represent the truthfulness of research interpretations. For instance, consider this passage from Lee Shulman's discussion of such matters:

Narrative modes [of research] are specific, local personal, and conceptualized. We do not speak of the validity of a narrative, but of its verisimilitude. Does it ring true? Is it a compelling and persuasive story? A good piece of physics [research] demonstrates its validity through meeting standards of prediction and control. A good work of tragedy demonstrates its verisimilitude by evoking in its audience feelings of pity and fear. ( Shulman, 1992, p. 23)