Statistics Tricks

Monday, August 11, 2014

Gary Smith, economics professor at Pomona College in Claremont, CA, looks at all the ways data, big and small, can be manipulated and offers a guide to gleaning the truth behind the trickery. He's the author of Standard Deviations: Flawed Assumptions, Tortured Data, and Other Ways to Lie with Statistics.


Gary Smith

Comments [15]

Gary Smith from California

Here are my thoughts on a recent statistical debate:

Aug. 12 2014 07:39 PM
gene from NYC

This item purports to give us "real world" interpretations of certain statistics.

IMO, Mr. Smith should have issued a disclaimer, or at least put a bit more emphasis on REAL "real world" experiences, ie, human interventions that influence the stats.

For example, if a coin turns up heads 4x in a row, it behooves us to consider if the coin has been "loaded."

If a basketball player misses shots 4x in a row, it behooves us to consider if he/she has been paid to thrown the game.

ESPECIALLY if we're betting on the next event.

Aug. 11 2014 11:45 AM
Tim from Upper Saddle River

You are presenting such a cynical view of research and researchers as if there are no honest scientists doing research.

Aug. 11 2014 11:39 AM
Jaime from elenville, ny

Do you know anyone who answers political polls? Do you know anyone who is part of nelson ratings, etc. ?

Aug. 11 2014 11:36 AM
Chriss from Montclair, NJ

The 1 in 5 women on college campuses have been sexually assaulted. It's been debunked, because it's based on a nonscientific internet poll, yet continues to be used.

But using the numbers, 1 in 5 is 20%.

NYU has about 22,000 students. Let's assume half are women (probably more, but that's another story).

20% of 11,000 is 2200 sexual assaults. Detroit had in 2012, a combined rate (murder, rape, robbery, and aggravated assault) was 2 percent. Yet folks trip over themselves to send their daughters to NYU.

Sexual assault is such a serious issue, it doesn't need inflated data. It requires the truth.

** Perhaps a show on what this number really is at area Colleges/Universities?**

Aug. 11 2014 11:35 AM

Hahaha - I think this applies to all of Economics.

Who paid for the research and then the methods section is what I read first.

Aug. 11 2014 11:31 AM
Alyson from Manhattan

How about the statistic of the amount of women vs the amount of men in NYC? Do they ever account for LGBT people?

Aug. 11 2014 11:31 AM
juan ito from Mount Vernon

When I took a stats course in business school, I remember taking a final exam where we were presented with problems where we were asked to determine whether certain conclusions based on the data presented were valid. We had yes or no, because of this reason or that reason. It seems to me that there are definite specific impartial criteria by which a statistician can say that this study or a conclusion from statistics is valid, and it occurred to me there should be some kind of nationwide or international board of impartial PhD statisticians that would say that this study is valid, and this one is not. Is it feasible?

Aug. 11 2014 11:30 AM

Personally, I dislike the statistics that get shared on facebook through images/posts. Many of them don't site their sources. How do you know they're backed up at all without this info?

Aug. 11 2014 11:29 AM
Steve from Rockville Centre, NY

I remember a headline: "Deaths in U.S. at an all time high". Well, duh.

Aug. 11 2014 11:29 AM
jgarbuz from Queens

"Lies, Damned Lies, and Statistics" - made by Benjamin Disraeli, but often quoted by Mark Twain.

Aug. 11 2014 11:26 AM
Stan from Harlem

Favorite New Yorker cartoon: "Dad, if practice makes perfect and nobody's perfect, does that mean that nobody practices?"

Aug. 11 2014 11:25 AM
deborah in newark

FYI: For all news Big Data, consult the "Big Data" app, a kind of rss-type feed of news stories, videos, and other media sources concerning Big Data. Available for download FREE on Google Play Store.

Aug. 11 2014 11:23 AM
Roshen from Midtown

You might find this appropriate:
"Spurious Correlations"

Aug. 11 2014 11:21 AM

Is there any good reason why the CPI - consumer price index - an artificial basket of goods the contents of which are somewhat arbitrary - has become the most cited measure of loss (or growth) in the average person's buying power?

Why not just measure it against GDP and be done with all the guesstimates?

Most Americans would be shocked to know that CPI has trailed GDP growth by 2% over the last 50 years. As a result, lots of things that were affordable on a middle class wage are now out of reach.

Aug. 11 2014 11:11 AM

Leave a Comment

Email addresses are required but never displayed.

Get the WNYC Morning Brief in your inbox.
We'll send you our top 5 stories every day, plus breaking news and weather.