98 8 Outcomes and Results
each of the main chapters was built around a theorem and a proof. Another instance
is a thesis in which the student used three case studies (her own patients) to explore
the relationship between childhood neglect, high social status, and psychopathy; a
chapter was dedicated to each of the case studies, which led straight into the final
discussion. A third instance was a historical study weighing up the evidence for
and against water contamination as the real cause of a ‘plague’ in the 1600s, where
much of the challenge of the thesis was the fact that the data was sparse and un-
reliable. However, the ‘how to present results’ thinking discussed below can be as
relevant to such work as it is to a thesis with a more quantitative organization.
Quantitative or Qualitative Data?
A common categorization of research is that it is either quantitative or qualitative,
or perhaps more accurately, whether it is closer to one or another of these extremes.
The experiences of two students I worked with are good examples of the challenges
in quantitative work. One, Jorge, had had the luxury of being able to test his idea
(a way of reducing the time required to compute some kinds of simulations) over a
great many data sets. He had small sets he had used for preliminary measurements
and much larger sets for the final evaluation. However, the data was not always
consistent, and statistical evaluation of the data, and visualization to confirm his
understanding of the statistics, had played a part in forming conclusions. As a con-
sequence Jorge had some hundreds or more of graphs and tables to draw on, reflect-
ing tens of thousands of automated experiments (and he could easily have run many
times that number), and now he needed to use this material to construct a narrative.
Despite the strong quantitative basis, however, the data fell into cases that needed
qualitative analysis.
The other student, Dai, had painstakingly gathered a small set of data through
work with genetics researchers, tracking the extent to which a software tool for
comparing chemical structures was reliable enough for practical use; this data was
counts of effects such as the number of incorrect comparisons, and human estimates
of factors such as the severity of the error or the extent to which it caused the re-
searchers to waste their effort. Here the challenge was that the results were to some
extent preliminary, but nonetheless strongly indicative; they went beyond a simple
case study, but didn’t easily fit into a tabular or graphical presentation, and was
insufficient for thorough quantitative analysis.
The task faced by both Jorge and Dai was to take a chaotic mass of information
and turn it into a results chapter. In both cases it was reasonably clear that the results
supported their initial hypothesis, and the challenge was coherent presentation. For
another student, Jackie, there was a deeper challenge because the data was contra-
dictory. She had come to me through a mutual friend because, as she assembled
her results, she increasingly wondered if her interpretation of the results was strong
enough to stand scrutiny.