How to Interpret Data: 2 Habits That Stop You Getting Lost in the Numbers

prashant — Fri, 28 May 2021 08:40:13 +0000

Most people who struggle to interpret data do not have a maths problem. They have a starting-point problem: they open the file before they know what they are looking for. Two habits fix most of it, and both take minutes to apply.

Key takeaways

Form a view before you open the data, so the numbers can confirm or challenge it.
A surprise against your hypothesis is the most useful result, because it tells you where to dig.
Qualitative data generates hypotheses. Quantitative data sizes and tests them.
A small gap between two numbers is not a finding until it is statistically significant.

Habit 1: Start with a hypothesis

When I started my career in market research in 2011, my biggest mistake was looking at data with no point of view. I would get so lost in the tables that I forgot what I was looking for. I still see the same thing in people new to the industry.

So before you open any dataset, ask one question: what do I expect this to tell me?

A simple example

Say you are about to review monthly car sales. Before looking, form a view. Perhaps you expect sales to be falling because the economy is weak. Now one of two things happens:

The data agrees. Sales are falling. You have your answer quickly and can move on to the size of the decline.
The data disagrees. Sales are rising. Now you are curious, and that curiosity sends you looking for the reason: a new launch, a price cut, a segment behaving differently.

Either way you are reading with purpose. The hypothesis does not need to be right. It needs to exist.

Habit 2: Know what type of data you are reading

The second common mistake is reading qualitative data as if it were quantitative, or the reverse. They answer different questions.

Data type	What it tells you	What it cannot tell you
Qualitative	What exists and why: the range of segments, motivations and language people use	How big or how common anything is
Quantitative	How much and how many: size, share, growth	The reasons behind the numbers

Qualitative research can tell you exhaustively which car segments exist in the market. Only quantitative research can tell you the size and growth of each one. If one segment seems bigger in a set of interviews, that is a hypothesis to test, not a conclusion.

Check significance before you call it a difference

With quantitative data, significance testing matters. Suppose the hatchback segment measures 25% in one sample and 26% in another. It looks like a one-point gap. Depending on the sample sizes, that gap may well sit inside the margin of error, in which case you cannot say the two are different.

Putting the two habits together

Before your next analysis, write one line stating what you expect to see, and one line stating whether the data in front of you can size things or only describe them. Those two lines will save you most of the wandering.

These were the two biggest mistakes I made early in my career, which is why I am sharing them.

FAQ

What is the first step in interpreting data?

Decide what you expect the data to show before you look at it. A hypothesis gives you something to confirm or challenge.

What is the difference between qualitative and quantitative data?

Qualitative data describes what exists and why, using words and observation. Quantitative data measures how much or how many, using numbers.

What does statistically significant mean?

It means a difference between two numbers is unlikely to be explained by sampling chance alone, given the sample sizes involved.

Data Interpretation Archives - PRASHANT AGGARWAL