Why the world going gaga over statistics is misplaced?

One oft-cited statistic shows that messages delivered as stories can be up to 22 times more memorable than straight facts.

Well, I am a statistician with more than a decade of experience in the said field. I find numbers without a story or a context half-baked. Have you ever imagined how statistics can go wrong in so many ways? Do statistics add realism or tell a half-truth?

Well, let’s look at some of the statistics paradox that I have come across in my 14 years of professional life. What we are going to talk about may not be exciting or fun but very much needed. Especially in times when the whole world around us is shouting numbers, big data, machine learning, artificial intelligence & whatnot.

You must be wondering why I am talking about “what can go wrong” instead of going gaga over my statistic experience and the hottest topic of the century- data analytics? I love statistics and data analytics. I believe data analytics has accelerated the human development cycle. But being a little cautious while performing this artistic data science can do wonders.

Using data to back our argument is such a cliché now!!

Has this ever happened to you that you’ve discovered a fascinating statistic that clinches your persuasive argument? Numbers can be irrational and transcendental at a given point in time.

There are Lies, damned lies, and statistics

Mark Twain

Above is a popular expression to describe the influential power of numbers, particularly the use of statistics to augment weak arguments. In our science-backed real world, typically mathematical approach to data is employed to exaggerate, inflate & complicate.

Nobody uses statistics to controvert themselves.

Statistics are often used to serve an agenda that someone believes in. This is a dangerous world to live in. Instead of getting an unbiased look, we’re getting a biased look toward crucial matters. These views are what other people want us to see by backing a statement with a data set (relevant or not, that’s a different question).

For example, a very common thought-provoking statistically significant statement used by so-called statisticians is the “On an average” statement. Data says that On an average a person will sleep 229,961 hours, eats & drinks 32,098 hours, drive 37,935 hours, work 90,360 hours, internet surfing 28,300 hours, watch TV 80,486 hours, cleaning 12,896 hours.

Well, what kind of average did the research includes? The arithmetic mean, weighted mean, geometric mean, median, or mode as these can be radically different. This misleading average is a classic example of Paradox.

Similarly other interesting statistics are correlations.

Does the correlation imply causation? Certainly not. A mere connection doesn’t mean that a relationship exists.

There is an interesting & real story that is related to the Pontiac, car division of General Motors. The team received a complaint, saying, “My Pontiac car doesn’t like vanilla ice-cream.”

This is the second time I have written you, and I don’t blame you for not answering me, because I kind of sounded crazy, but it is a fact that we have a tradition in our family of ice cream for dessert after dinner each night. It’s also a fact that I recently purchased one of your new automobiles and since then my trips to the store have created a problem. You see, every time I buy vanilla ice cream when I start back from the store my car won’t start. With other flavors it’s okay. What is there about this car that makes it not start when I get vanilla ice cream, and easy to start whenever I get any other kind?

The complaint letter

The president of the automotive giant sent an engineer to check anyway. The engineer went with the customer for a few days for ice-cream. He recorded all sorts of data, time of day, type of gas used, time to drive back and forth, etc. He refused to believe that a car was allergic to vanilla flavor.

But the engineer got his answers. The customer was spending relatively less time in buying the vanilla flavor due to easy access for the shopkeeper. But the time was not enough for the engine to cool down and start again efficiently. The point here is that vanilla flavor had nothing to do with the car not starting. Though the events were correlated there was no causation. It was a pure logical fallacy.

However, it doesn’t mean that correlation is useless.

Correlations are important in helping us determine the degree of relationship between variables. But next time when you see those fancy doted charts moving along each other remember not to fall in the trap & look for solid evidence for it. Some of the established spurious correlations are like “No. of people who drowned by falling into a swimming pool correlates with No. of films Nicolas cage appeared in”.  Find more such interesting examples at tylervigen.com.

Oh no! We are just good friends (trends), it doesn’t mean relationship exists

However, this logical fallacy is not only limited to amateurs statisticians. A good percentage of data practitioners admitted to questionable practices. Practices include modifying results to improve outcomes, subjective data interpretation, withholding analytical details, and dropping observations because of a gut feeling.

Some of the big brands have used misleading statistics to prove they are the BEST.

Colgate ran a classic ad in the early 2000s that received a lot of attention: 80 percent of dentists recommended Colgate toothpaste. Sounds like a stellar endorsement, right? But this statistical scandal came to light and put Colgate under fire.

The company used statistics to work in its favor by manipulating the survey. It turns out that when dentists were surveyed, they could choose several brands — not just one. So other brands could be just as popular as Colgate.

Statistics are important but without a story they are worthless

Stats are finicky, they are important but without the perspective, they are pretty hollow. So one needs to be well conversant and know the problem they are solving, the data they are handling before making any conclusion i.e. one needs domain knowledge to infer/interpret the stats.

Some of the fabricated statistics prove that even the most reliable statistical gatekeepers are not immune to biases that can arise out of statistical miss-interpretations.

If you are not attuned to its subtleties and trickeries, you may be none the wiser. But the more aware you are of statistical interpretation, the lower you risk implementing faulty stats.

It ain’t so much the things that we don’t know that get us in trouble. It is the things that we know that ain’t so?

– Artemus Ward

Interested in reading our Advanced Strategy Stories. Check out our collection.

Also check out our most loved stories below