Histograms – the findings aren’t always so obvious

In each of my last two posts, I showed how histograms help identify distinct subgroups in our data, focus our analysis, and produce more powerful findings. I used real-world examples, but made-up data.

Why?

Because real-world data is messy, often full of random noise, and with odd proportions, making it less-than-ideal for illustration purposes. This is not a limitation of the histogram, but rather a challenge for us as we use them.

Consider the following example exploring the frequency of a product’s weekly sales.

As you can see, our data breaks cleanly into two distinct subgroups – weeks when we advertised the product, and weeks when we didn’t. As you’d expect, we sell more when we advertise – in fact, a lot more, making it easy to see these two subgroups.

But what if our advertising was less effective? And what if our product had a higher base sales level? The two subgroups still exist, but they’re less distinct, as in the following histogram.

Now let’s assume there was greater variance in weekly sales, due to any number of factors: volatile consumer demand, product stock-outs, uncertain economic conditions, etc. Our weekly sales would be less consistent and distributed more broadly as below (note: this would be true both when we advertise and when we don’t).

What happens when we look at the data not as two subgroups, but as one population? Our two subgroups add together to create the following overall sales distribution (the red line).

Hey, wait a minute! This looks like a homogenous, normally distributed set of data!!

It’s important to note, in the real world, THIS is the view we’d start with. We only get to the more revealing views of sales with and without advertising, when we imagine the differences, sort the data, and run a histogram to compare the two subgroups. Looking at the above combined view, it’s easy to see how this can be missed.

The histogram is a great analytical tool. But like most tools, batteries aren’t included – tools don’t work on their own. Thinking on our part gives it its power.

This entry was posted in Uncategorized. Bookmark the permalink.

Leave a Reply

Your email address will not be published. Required fields are marked *

*

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>