Demonstrating Editorial Focus and Learning About Your Data - Data Visualization: A Successful Design Process - page 60

At this point, we are starting to consider the relationship between our editorial focus
and the potential visualization design options.
As Amanda Cox describes earlier, the way that you choose to represent your
data—the form you give it through your selection of chart type—should be
influenced by the questions you are trying to answer.
For instance, if you are asking a chart to facilitate a comparison between the values
of different categories, you might deploy a bar chart. You wouldn't use a line chart
to achieve this, but you would if you wanted to show how a value or values change
over time. The scatter plot we just saw was the perfect method of comparing two
quantitative values for all those different countries. It was the right form to answer
the specific data questions identified.
So we need to know what questions we're trying to answer.
Unless you've already had them specifically outlined to you, an effective approach
to tackling this can be drawn from the practice of logical reasoning, specifically
induction and deduction. These techniques are common to academic and scientific
research.
Deductive reasoning involves confirming or finding evidence to support specific
ideas. It is a targeted and quite narrow approach concerned with validating certain
hypotheses. A deductive approach to defining your data questions will involve
a certain predetermined sense of what stories might be interesting, relevant, and
potentially available within your data. You are pursuing a curiosity by interrogating
your dataset in order to substantiate your ideas of what may be the key story
dimensions.
Inductive reasoning works the opposite way. It is much more open-ended and
exploratory. We're not sure precisely what the interesting stories might be. We use
analytical and visualization techniques to try and unearth potentially interesting
discoveries, forming different and evolving combinations of data questions. We
may end up with nothing, we may find plenty—the insights we observe may be
serendipitous as we follow our nose for the scent of evidence. Fundamentally,
this is about using visual analysis to find stories.
For most visualization projects, if we have the time, ideally we would seek to use
both deduction and induction in conjunction in order to learn as much as possible
about what stories the dataset can reveal about the given subject matter.