About Us
The Research Process
Research Techniques
Seminars
- > About the University
- > Research
- > Faculties
- > Library
- > IT Services
- > News
- > Events Calendar
Research Questions
Research involves a process of finding out. You would not set out on a research project if you already knew all the answers.
So the very beginning of the research enterprise is jotting down all the things you need to find out. These should arise out of your topic and the reasons you want to do it. You may need to read quite a lot before you even know what you don't know. And you may find that many of your questions may be answered by further reading. Don't worry about that at this stage. You need to think what you want to know before you can work out how to find out.
The critical point to remember at this stage is that your questions must be-in principle-answerable. This means they must
- be specific and
- they must refer to something you can look at or find out about.
In The Hitch Hikers' Guide to the Galaxy, comedian Douglas Adams tells of a computer which took 7.5 millions years to calculate that the answer to 'the ultimate question of life, the universe and everything' was 42. When its operators objected that the answer made no sense, the computer pointed out that 'once you know what the question actually is, you'll know what the answer means'.
Adams, Douglas. The Hitch Hiker's Guide to the Galaxy. London: Pan Books, 1979. chapters 27-28
Questions or hypothesis?
People often ask whether research must test a hypothesis. This all depends on the style of research you are engaged in. A hypothesis is simply a way of generating research questions within the argument structure of a particular type of inquiry. Other types of inquiry - including those common in legal research - generate their questions out of other sorts of arguments. We consider two different types of argument/question structures below.
Why do people test hypotheses? This is usually related to the 'hypothetico-deductive method', typical of some of the 'pure' sciences. The idea behind this method is that we are trying to improve a theory of the general laws of nature. From the theory we may deduce that 'x' will happen under certain circumstances, so that becomes a hypothesis. If 'x' does happen as expected, then we have disproved the 'null hypothesis' and the theory is strengthened.
However, in legal and related research we are very rarely interested in testing a grand theory. As discussed in regard to the Purposes of Research, we are more often interested in some improvement of legal or social arrangements, or exploring ways of interpreting and understanding particular laws or cultural phenomena.
In this type of research you still need research questions, but you develop them within a different argument framework. Two of the most difficult sorts of questions are hypotheticals and questions of quality. We need to discuss these because of their difficulty. They also serve very well to illustrate two of the ways in which questions typical of legal research may fit into the structure of an argument. Of these, hypotheticals are the most difficult, so I deal with them first.
A hypothetical question asks what ... if...? For instance: If we increase penalties for breaches of workplace safety, would that reduce work related injury? Now, unless we actually do it as an experiment, this question is technically unanswerable. In general, hypotheticals are like that, and should be avoided as research questions.
However, in dealing with areas of policy, we are often faced with choices of this sort. We can deal with this problem by clarifying an argument which may include hypothetical arguments, but reduces the research questions to ones which can be answered through research.
If one purpose of our research was to reduce work related injuries, then we would need information which could suggest what the causes of injury in a particular site or industry were (answerable, depending on existing records), what laws and policies were currently in place (answerable), how they were applied in practice (answerable) and how managers and workers responded to this application of the law (answerable). We may be able to use some of this information, then, to mount a plausible argument working from the question whether penalties are, at present, a significant factor in decisions relating to workplace accidents, to some informed guess as to whether a change in penalties would affect a change in accident rates. A sensible approach to this problem, then, is to break down an unanswerable (or hypothetical) question into a number of answerable questions, which may be embedded in an argument.
The worst thing to do with hypothetical or unaswerable questions is to ask someone else. In the workplace accidents example, we might be tempted to ask managers and workers whether they would act more safely if there were stiffer penalties. The problem here is that we have stopped asking ourselves an unanswerable (because hypothetical) question, and asked them "what... if...?" As we will see again in discussing interviews, never ask people a question they do not know the answer to (specially if they may think they do!).
Questions of quality ask 'how good?' or 'which is better?' These are not so difficult as long as we know how we are defining 'good' or 'better'. Such questions are typical of evaluation research, where we need to find out whether a particular program is achieving its aims (or perhaps some others). As long as we are asking questions about quality, we need to be aware of the values we are bringing to our inquiry. Questions of quality are absolutely central to many policy fields, and must be very clear about the criteria they use to determine what is 'better' or 'worse'.
Here again, we are framing our questions within a set of arguments about what is a good program, or a better one. By clarifying what constitutes 'good' or better' within a reasoned argument, we can then develop questions which research can answer. For instance, if we were evaluating a case management program in a court, we may have a number of criteria, including the speed of disposal of cases, and whether all parties receive a fair hearing. In this case, we could quantify speed, but would still need to do more work to find out how the 'fairness' of a hearing could be measured. Once we have worked out a valid measure for these variables, then research questions could be framed in those terms.
(Note: 'validity' means measuring what we set out to measure; 'reliability' means that if we measure the same thing several times, we should get consistent results.)
There is a useful literature on evaluation research, and since it is particularly helpful in framing questions of quality, it is referred to here.
Michael Q. Patton has written a comprehensive, sometimes amusing, but not very thought provoking book on evaluation, Practical Evaluation (Newbury Park, CA: Sage, 1982).
Yoland Wadsworth has written one of her well illustrated and readable handbooks on evaluation (Everyday Evaluation on the Run. Melbourne: Action Research Issues Centre, 1991), though hers is a specific form of self-evaluation, which may not be suited to all research purposes. William Shadish gives a good overview of the field, telling us of the historical policy context in which it arose, and relating evaluation research to such everyday and common sense decisions as what sort of car to buy. Michael Scriven asks what criteria we should use in an evaluation, and goes so far as to suggest we may be able to look beyond the stated goals of a program to see what other benefits or disadvantages may come with it ("goal free evaluation").
Scriven, Michael. “Evaluation ideologies.” Evaluation Studies Review Annual 9 (1984): 49-80.
Shadish, William R. “Sources of Evaluation Practice: Needs, Purposes, Questions, and Technology.” Evaluation Studies Review Annual 12 (1987): 520-554.
Your questions.
This discussion of hypothetical and evaluative questions is intended to give examples only. You do not have to use these sorts of questions any more than you have to test hypotheses. But you do need questions which are specific and answerable, and which will provide data or backing for your arguments (discussed further in that section).
Organise your questions in hierarchies if necessary. For instance, as we saw above, one unanswerable question may be broken down into a number of answerable questions, linked by a chain of argument. This exercise may take several pages, but it is advisable to be inclusive, even if many of the questions may have self evident answers. Once you can see all your questions, lined up according to their contribution to your topic and their place in your argument, then you are in a better position to work out what methods you can use to answer them.