Warning: this post is not to be taken seriously. At least not entirely.

Previously this week, I named the most influential business book of the past decade. As I pointed out in that post, few business books have a lasting influence. One book that has had a lasting influence is How to Lie with Statistics by the late Darrell Huff.

While Huff intended his book to be an expose, it's actually ended up being something of an handbook for how to present statistics graphically so that they support your narrative, regardless of whether the actual data support that narrative.

I have seen Huff techniques used in business presentations, in media reports, in political press releases, even in "scientific" writing... basically by anybody whose desire to tell a good story exceeds their sense of obligation to be truthful.

Huff died in 2001, which is unfortunate because How to Lie with Statistics is badly in need of an update. Since his death, and particularly since November of 2016, some entirely new and effective methods to spin statistics (i.e. lie) have emerged.

Here are three of them:

1. Multiple Rounding Methodologies

Let's suppose you're a national sales manager and your sales growth dropped from 50.9 percent to 50.1 percent year to year. While that's not much of a decline, if you graph it in Excel, the result gets scaled like so:

 inline image

Who would want to stand up in front of investors or top management with a slide like that? Is there a way to present that data in a more positive manner?

There is indeed: creative rounding. Now, I'm going to get a bit technical here, but I'll keep it as simple as possible, so bear with me, OK?

Now, you're undoubtedly familiar with grade school rounding, where less than "5" rounds down while more than "4" rounds up. This is represented in Excel as the "ROUND()" function.

For example, in the final count of the popular vote in the 2016 presidential election, Trump received 62,984,828 votes while Clinton received 65,853,514 votes. Rounded to the nearest million, Trump got 63 million while Clinton got 66 million, for a net difference of 3 million.

However, it turns out there are two other ways to round numbers:

  • FLOOR(), which rounds everything down. For example, 10.9 rounds down to 10.
  • CEILING(). This rounds everything up. For example, 10.1 rounds up to 11.

In the vote count example above, some news sources have been using CEILING() for Trump and FLOOR() for Clinton. Under that method, Trump got 63 million while Clinton got 65 million for a net difference of only 2 million. Even today, you sometimes see the popular vote difference stated as "2 million" or "more than 2 million" rather than the more accurate "3 million."

The point of that brief discussion is that, since it's now apparently OK for political parties and news sources to use this method, sales managers now have a new tool to make their numbers look more positivel.

For example, if you need to "fix" the graph shown above in the post, simply use FLOOR() for the 2019 sales numbers and CEILING() for 2020 sales numbers. Excel then gives you the following graph, which is clearly more palatable to a business audience:

 inline image

2. Geographical Scaling

Once again, suppose you've lost your jog as national sales manager but made responsible for sales in New York City. Unfortunately, your team made 2,000 sales in the city in 2019, but under your leadership the number plummetted to only 1,000 sales. That's going to look might bad when you use Excel's defaults. You get this depressing graph:

 inline image

Not to worry. You can make these numbers look somewhat more positive by using the same technique that's employed by conservative politicians to display voting patterns in the United States.

Perhaps you may have seen it: a map of the US with counties that voted for Trump in red and counties that voted for Clinton in blue. The map presents a vast sea of red surrounded tiny splatters of blue on the coasts. The message: most of the country voted for Trump.

Not to quibble, that widely-distributed map actually illustrates voters per square mile rather than raw number of voters. Since most underpopulated areas voted for Trump, the red areas look huge.

Since it's clear that much of the country can be impressed (i.e. fooled) by adding a geographical component to a graphic, you should now be able to use a similar method to "fix" the negative graph above.

All you need do is restate the number of customers as the number of customers per square mile.  Then (and this is the important part):

For 2019 you use the square mileage of the STATE of New York. And for 2020, you use the square mileage of the CITY of New York. SInce the city takes up much less square mileage than the state, Excel provides you a graph that tells a far more positive story.

 inline image


3. Creative Improvisation

So, you're no longer working in sales because your CEO says you've "got a way with numbers."  But now you face a new challenge: presenting year-to-year profits. Alas, under your direction, the sales team has been giving massive discounts to close sales and as a result margins are sharply down. You plug your numbers into excell and get this:

 inline image

Obviously, you don't want to present such a downbeat and depressing graphy to investors, customers, regulators, etc.

Not to worry. Remember when President Trump on national news used a sharpie to extend path of a hurricane on a weather map? Well, what's good enough for a POTUS should be good enough for a CFO. You whip out your pen and (voila!) Problem solved:

 inline image

OK, you've probably figured out by now that I'm kidding by poking fun at the way extremely stable geniuses creatively lie with statistics. But the truth is that businesses and executives use similar techniques all the time to spin their corporate stories.

So, if haven't read How to Lie with Statistics, you're doing yourself a disservice because unless you know these techniques, you'll be constantly fooled by those who use How to Lie with Statistics as a handbook rather an expose.

Have a fabulous and well-informed New Year!