What would readers conclude after looking at the graph above?
- Best Buy sales have gone up at least 10 fold in the last 2 years. (Recession, what recession?)
- Best Buy’s monthly sales at the end of 2009 were roughly $22 million, at the peak of holiday season.
- Newegg is the #2 electronics retailer, after Best Buy. Amazon, buy.com, Walmart and Target aren’t players in electronics.
That’s what the graph says, but all of those conclusions are wrong.
In reality, Best Buy did $8.5 billion in revenue in December. In fiscal year 2009, Best Buy did $45 billion in revenue. $22 million in revenue is a rounding error.
What the graph really shows is how useless these Mint.com data are for analyzing consumer spending by category.
The reporter apparently didn’t understand the methodology behind Mint’s data. Mint allows consumers to track their credit and debit transactions. Consumers enter their account information and mint pulls transaction data from their banks and presents it online. Here’s how that methodology leads to the above erroneous conclusions:
- The enormous growth in the graph likely represents the growth in mint.com’s user base. Mint launched in September 2007 and has grown to 1.8 million users. (Mint was recently acquired by personal-finance software maker Intuit for $170 million.) Best Buy’s comparable store sales were down for fiscal year 2009 and up marginally for fiscal 2008.
- Only transactions by Mint’s users are included in this number, accounting for the multibillion-dollar difference. Even then, those numbers may not be complete as not all users import all accounts.
- Newegg is the #2 dedicated electronics player as classified by Mint. Because Mint collects transaction level data, not item level data, it doesn’t know what to do with purchases from diversified retailers like Amazon, buy.com and Walmart. (Amazon wouldn’t even show up as a leading bookseller; on my Mint account, it shows up as “Shopping”.) By Best Buy’s own definitions, consumer electronics only account for 36% of its revenue.
There’s also zero justification in the Mint data to support the chart’s caption that Best Buy generated these huge returns by “offering discounts on laptops and flat-panel televisions.”
These kind of errors are rampant in our business news because many reporters don’t understand numbers or methodology. When I was in school at the Medill School of Journalism, most classmates took the minimal math and economics classes they could and still get a journalism degree.
One of the most frequent numerical mistakes by journalists is confusing percent increases with percentage point increases. If your credit card’s interest rate goes from 10% to 15%, it’s a 5 percentage point increase, but a 50% increase.
This journalistic innumeracy hurts us all. We can’t make informed decisions about what government is doing if people don’t understand the numbers. Even basics get confused: journalists frequently confuse millions with billions (via Paul Kedrosky).
Journalists are at least partly to blame for the dot com bubble and the housing crash. Not understanding the economics, they repeated the lines of “experts,” such as investment bank analysts and real estate agents — most of whom had clear incentives to keep pumping air into the bubbles. I read many stories about how exotic mortgages were making housing more affordable. What was actually happening was that the easy availability of credit and flood of otherwise unqualified buyers in the marketplace drove up price. Basic supply and demand.
Unfortunately this problem is only going to get worse. Many of the sharpest minds I know from the journalism business have left to go on to other careers for many reasons — the difference in pay (often 2x or more), instability and constant layoffs and backward thinking management.
Additional analysis of the data also takes time, which is becoming harder to come by as newsrooms across the nation shrink. The pressure to do more “quick hits” like charts of the day will lead to more sloppiness and misinformation.