Refining a dissertation topic

Ross Woods, 2020

Having a dissertation idea is one thing; refining it is another. The idea will normally change several times, whether it is expressed as a problem, a question, an hypothesis, or a title.

Changes are a normal part of the process. Some students feel threatened by any suggestion that their original idea is not good enough, but refining is a good thing; it means that you're learning more about it as you go and closing in on a useful topic. For example, you start to consider other factors, assumptions, definitions, variables, what if scenarios, and the kinds of data that would be needed. Beware, however, that it might be good to have multiple ideas for a topic, becuase your favorite might not work out.

Discussion greatly helps in refining a topic and feedback will help you crystallize your ideas. Take up the other students' offers to discuss it. Your supervisory committee will almost certainly want to discuss it with you.

Put another way, you are looking at the topics that come up in research methodology textbooks:

The research problem. What is the actual problem you want to research? Does it need to be narrowed down?
Defining populations. Who are you actually researching? How can you define them to neutralize extraneous factors?
Defining terminology. Did you use the right word? If so, do you need to give a definition?
Identifying assumptions. You'll list them in your introduction if they are reasonable.
Considering factors over which you have no control. You might use them to qualify your conclusions or put in the list For further research.

The example of Sam

Sam suggested the title: Online education might improve academic and socio-emotional outcomes of frequently relocated K-12 students compared to classroom-based schooling. The following comments came out of Sam's discussions:

It is an hypothesis, not a title.
A good hypothesis is a good place to start, better than a title. A question or a problem statement also works. A title doesn't necessarily imply the problem you want to solve or the question you want to answer. (Expressed as a title, Sam's title would be: The academic and socio-emotional outcomes of online education for frequently relocated K-12 students compared to classroom-based schooling.)

In essence, Sam wants to compare online education with classroom-based education, but only academic and socio-emotional outcomes. This seems like a good start.
It is a broad generalization and Sam should significantly narrow it.
Hmmm. Let's keep going. It needs some specificity.
It presumes that academic and socio-emotional outomes will correlate positively.
What if they don't? It might be better to choose only either academic or socio-emotional.

An alternative might be to measure both and compare them, but then Sam would be comparing more variables. This might be more practical in a doctoral dissertation, which is big enough to include the extra level of complexity.

Is Sam simply trying to prove that online education gets better outcomes? (Some students have such a strong a preconceived idea of what they want to find that they are at risk of circular logic. I believe X and want to prove X ...)
Might is vague.
For example, if Sam found only one case of improvement, then the hypothesis would be proved correct, even if that were not so in all other cases. He could replace it with tend to. He'd then only need to demonstrate a tendency, but he would probably need to quantify data in some way.
Would researching it be ethicalIy feasible?
We might be dealing with rather large polulations of minors. Would getting permission be difficult?
What if Sam finds that online education does not improve outcomes?
This would be a valid conclusion, as long as he identified other relevant factors. In other words, he'd need to qualify his conclusion.

It would also be a valid conclusion to find that online and classroom-based education produce the same outcomes.
How frequent is frequent?
This is not a problem as long as Same defines it. For example, it could be twice in the last three years.
K-12 is very broad.
Could different grade levels respond differently to online education? Should Sam define the polulation more narrowly?
It assumes that all parents are the same.
What if they respond to online education in widely different ways? Some might be supportive of online education, while others see it as a second choice. Some might encourage their children, while others might be more ambivalent. Should Sam define the sample of parents clearly and broadly enough to represent a defined category of parents? (For example, they could be military parents, because military families move relatively frequently.) Should Sam also identify their attitudes, both their actions and their perceptions?
It assumes that all students are the same.
What if they respond in very different ways? Should Sam define a sample of students broadly enough to represent a defined category of students? Would this cancel out individual differences? Should Sam explore their attitudes in both (or either of) their actions and their perceptions?
It assumes that all demographics are the same.
What if different demographics respond in very different ways? Should Sam either select a defined demographic, or include multiple demographics but identify respondents according to their demographic? If he chose the latter, he could then compare results to see whther or not they varied greatly by demographic.
It seems to assume that frequent relocation negatively affects students.
What if it doesn't? What if it actually creates benefits, such as flexibility, resiliance. and adaptability? Could different demographic groups respond in very different ways? What if families adjust, perhaps by having different kinds of family structures?

However, the current hypothesis does not test this. It would need to compare frequently-relocated people with non-relocated (and perhaps seldom-relocated) populations of students. It also does not include the role of family structures.

Then Sam started to think more about academic outcomes.

It assumes that all classroom-based schools are the same.
Should Sam define a a category of classroom-based schools so he can choose a sample? Does this mean he will need to work with several schools?
It assumes that all online education is the same.
Should Sam define a particular category of online education? It could be a highly automated instruction package, or it could be something more like face-to-face classroom instruction done over the Internet.
It assumes that all curricula are the same.
Should Sam define a category of curricula? Wouldn't it be even better if both populations followed the same curriculum?
It assumes that the implementation of curricula are the same.
Would Sam be comparing the best of classroom-based schooling with the worst of online education, or vice-versa? Even if different populations followed the same curriculum, the implementations of curriculum could still be very different. For example, the two implementations might vary; what if one version had been used and improved for several years while the other was new and untested? What if the online version had a novelty factor of graphics, games, and creative interactions? Perhaps they differ in other ways:
1. Theoretical assumptions of curriculum delivery or teaching.
2. Assumptions about students.
3. Sequencing of topics.
4. Time spent on each topic.
5. Choice of examples.
6. Serious flaws in program design.
It assumes that curricula will be implemented equally effectively across curriculum areas (mathematics, language, science, etc.).
Should Sam choose only one curriculum area (mathematics, language, science, etc.), or several and compare them?
It assumes that students will perform equally well across the currculum.
Should Sam choose only one curriculum area (mathematics, language, science, etc.), or several areas and compare them?
It assumes that all teachers have the same level of ability.
Could variations in the abilities of classroom teachers skew the results? They might be very gifted and creative, or exactly the opposite. Should Sam define a big enough population of teachers to get a representative sample? Could Sam make the same assumptions about on-line instructors?
It assumes that teachers will teach equally effectively across curriculum areas.
Could it be that they tend to teach some curriculum areas better than others?
Could the population of online students and classroom-based students be intrinsically different?
Could their choice of education represent a pervasive difference between populations? For example, what if online students (or their parents) have a different psychological profile from those of classroom-based students (or parents)? Online students or their parents might be early adopters, that is, they like to try something new and less conventional. Should Sam note this possibility in his research?

If this could be established, it would show that comparisons are, at least to some extent, unrealistic and unfair. It would be the classic problem of comparing apples and oranges.

Next version of the topic

Sam then asks: Does online education tend to improve socio-emotional outcomes of frequently relocated middle school students compared to classroom-based schooling?

The population is children of U.S. military families who have been relocated twice in the last three years and currently in middle school.
All students will be studying the same curriculum.
The implementations of curriculum are assumed to be radically different, and any conclusions must be qualified for this factor.
Etc.