What Makes Research Good

The sample that is not a sample. Whoever agreed to speak, presented as the district. Chapter 1040.

The conclusion that outruns the evidence. Forty interviews in one settlement supporting a claim about the country.

The literature review that is a list. Ten paragraphs summarising ten books without saying what any of them left unanswered. Chapter 1310.

The recommendation that is a wish. That awareness should be improved. Chapter 1440.

And the citation that cannot be checked. A proposition attributed to a source that does not contain it, which is the failure that destroys a reader's confidence in everything else.

What good research does NOT require

It does not require a large subject. Chapter 760's example: the smaller question was the researchable one.

It does not require an original theory. Establishing a fact carefully is a complete contribution.

It does not require agreement with the researcher's expectation. A disproved hypothesis is a finding, chapter 1170.

It does not require perfect data. It requires honest data and stated limits.

And it does not require length. MU prescribes 100 to 150 pages for the dissertation body, chapter 120, but a hundred pages of padding is worse than eighty of work.

A worked example

Two dissertations on the same question, both on legal aid awareness in one taluk.

Dissertation A. Sixty pages on the history and importance of legal aid, then a chapter reporting that the researcher spoke to some people and found awareness to be low, then recommendations that awareness should be increased.

Dissertation B. A stated question: what proportion of a defined population knows that free legal services exist and where to obtain them. A stated method: a schedule of six questions, administered by the researcher because most respondents cannot read, to every fifth household in three named settlements, over two weeks, with refusals recorded. A stated result: the proportions, with the refusal rate. Stated limits: three settlements, one taluk, one season, and the possibility that respondents said what they thought was expected. And a recommendation: that the District Authority hold camps in these three places on the three problems the same schedule found to be commonest, using the clinic register to verify.

A is longer. B is research. Apply the three tests: A's claims cannot be checked because the method is not stated; A's conclusion is not supported because there is no evidence of the sample; and A is not new because its first sixty pages are available in any textbook.

And notice that B is entirely within a student's reach. Two weeks, a printed schedule and a notebook.

Quick revision

Eight characteristics: a clearly defined problem; a design fitting it; a stated method; adequate and honestly collected data; proportionate analysis; conclusions that follow and stop; stated limitations; and a report clear enough to be checked. With ethical conduct throughout.

Three tests a reader applies: is it true, checked by looking up citations; is it supported, by evidence rather than by assertion; and is it new.

Seven common failures: a subject instead of a question; an unstated method; a sample that is not a sample; a conclusion outrunning the evidence; a literature review that is a list; a recommendation that is a wish; and a citation that cannot be checked.

Not required: a large subject, an original theory, agreement with your expectation, perfect data, or length.

Test yourself

1. Give the eight characteristics of good research. A clearly defined problem; a design and method appropriate to it; a method stated openly; adequate and honestly collected data; analysis proportionate to the data; conclusions that follow from the evidence and go no further; candidly stated limitations; and a report clear enough for another researcher to verify or repeat the work.

2. Name the three tests a reader applies, and how each is checked. Is it true, checked by looking up several citations to see whether the sources say what the writer says they say; is it supported, checked by asking whether the evidence reaches the conclusion or whether an assertion of obviousness does the work; and is it new, checked by asking whether anything is established that the sources did not already say.

3. Identify four common failures in student dissertations. Choosing a subject rather than a question, so the work ends without a finding; not stating the method, so claims about respondents cannot be checked; presenting whoever agreed to speak as a sample of the district; and making recommendations that are wishes rather than proposals addressed to a body that can act.

4. Why does stating limitations make a work stronger rather than weaker? Because a finding whose bounds are stated can be relied on within them, so it becomes usable; an unqualified claim invites the reader to distrust the whole, since they cannot tell how far it was meant to reach.