Once you've found something in the data,

you probably want to start telling your co-workers.

But before you go telling your results to everyone you know,

there's one very important thing you need to know.

Correlation does not imply causation.

Here is a graph of ice cream sales versus crime rates.

It's clear that higher ice cream sales correlates with higher crime rates.

But that doesn't mean that higher ice cream sales causes more crime.

That'd be ridiculous.

When you're presenting your findings, you need to remember that even though two

things might be related, that doesn't mean they have anything to do with each other.

The most you can say is that they're correlated.

And on that note, I wonder what a graph of our age data would look like.

Let's hop back to Google Sheets to find out.

To make a graph of ages versus counts, let's select both the age and

count columns, including the headers.

And then let's click on the Insert Chart button.

And there we go.

Thanks to all these spikes, it's clear clear that some ages are overrepresented.

Let's drag this chart to the top.

And then since we don't really need this legend over here, let's remove it by

going into the chart editor, clicking the customize tab, selecting the legend and

for position, set it equal to none to remove the legend from the chart.

Perfect.

One thing you want to be careful of though,

is the charts can be unintentionally misleading.

Having all that data represented graphically

is a lot noisier than having it in text.

If you're trying to report something,

don't rely on others to interpret your graphs.

Make it easy for them by spelling out exactly what you found.

A good model to follow when explaining your findings is the outline of

a scientific article.

Start with an introduction where you introduce the problem and

how it came to be.

Then, formally state the hypothesis and

describe the procedure used to test that hypothesis.

Finally, you want to report the results of your testing and

then share any conclusions.

Here's what this might look like for the problem we just tackled.

Introduction, we've received complaints that some ages

have an easier time qualifying than others.

We aim to assess the truthiness of those claims.

Hypothesis, some ages have an easier time qualifying than others.

Procedure, to find this out, we looked at the number of qualifying runners for

adjacent ages.

If no ages have an advantage,

the difference between adjacent ages should be relatively small.

We picked a figure for what would be an unacceptable difference, and

then tested adjacent ages.

Results, we found four differences that exceeded our maximum.

Conclusions, we conclude that some ages have an easier time qualifying

than others.

Breaking it down this way makes it easy to understand what's going on.

There are many things you can do to present your findings, and

this is just one way.

But it's nice to have a format in mind, at least to get you started.

Another thing to mention is that unlike a scientific article, at the end,

you might want to include a recommended solution.

If you think you've found a solution to the problem, make sure to share it.

It really helps bring things full circle.

Also, before we go,

you should know that we haven't found anything particularly alarming here.

The Boston Marathon uses different qualifying times for different age groups.

So it's sort of an open secret that you'll have an easier time qualifying

as a 35 year old instead of a 34 year old.

So maybe our recommendation would be to get rid of age groups and,

instead, have a different qualifying time for each age.

There's so much you can do with data analysis,

from figuring out which peanut butter to buy to finding a good deal on a house,

data analysis informs our every decision.

But this is just the beginning, there's lots more to learn about data analysis.

Until next time.