During this annual holiday shopping melee, economists project the implications of our spending for the country’s overall economic health in the coming year. That can be like claiming a warm December day proves global warming. What does this have to do with measurement? Quite a bit, actually, because it shows the implications of improper interpretation of outliers in a data set.

In the spirit of the season, when we’re all expected to spend (or charge) large amounts of money for gifts, it can be interesting to see what the newspapers say about that spending. One of the most important days in the retail trade is “Black Friday,” the day after Thanksgiving. That day is “black” because it’s usually so busy that businesses that weren’t making money (in the “red”) finally start making money and go into the “black.” This year, it’s estimated that 5 percent of all holiday shopping will have happened on that day.

Television and newspapers report this information as a three-step process. The first step is reporting the fact itself. The second is a bunch of conjecture, smoke and mirrors, expert analysis, or whatever else about why the numbers came out the way they did. The third step is more smoke, mirrors and conjecture about what the numbers mean about the future. The point here is that these three steps can happen with product measurement data as well.

The three steps of reporting data
Step one: Report the results. About Black Friday 2007, one report says online purchases were up 83 percent over the same day last year. On the brick-and-mortar side, one report says same store sales were up 3.1 percent over the same day in 2006.

Step two: Talk about what that data “signals.” With more people shopping online all the time, an increase in spending there should be expected, and growth is from more transactions, not from increased spending per customer. On the retail-store side, any increase is viewed favorably. With all the doom and gloom about the mortgage mess and the corresponding stock market volatility, any increase in spending could be considered a pleasant surprise. Let’s ignore inflation, which would mean that if there was an increase of 3 percent and the inflation rate is 3 percent, sales were actually flat.

Step 3: What do these Black Friday numbers say about the future? A commonly reported line is that spending is up because U.S. consumers have confidence in the future despite the subprime mortgage mess, declining housing values, and a volatile stock market. Stating that supposition as fact, they then build on it and say that consumer spending may, therefore, be enough to continue overall economic growth into 2008. Of course, on the other side of the coin are the naysayers, who state that this single day of positive data is an exception and doesn’t signal a trend. Keep those thoughts in mind.

Relative to what?
The most common way of presenting Black Friday data is relative to the previous Black Friday. One would expect that there’s some consistency in conditions for that date from year to year. It has always signaled the beginning of the holiday shopping season. It’s always the Friday after the holiday, and stores typically open early and have big sales to lure shoppers.

Reporting Black Friday sales relative to the Wednesday before Thanksgiving would make little sense, nor would comparing it to a random day in the summer. Comparing the same day each year makes sense.

Where the analysis of each year’s data gets convoluted, however, is in interpreting and analyzing the environmental conditions this year that could be affecting the results. For example, gasoline prices are up significantly this year over the same time last year. The previously-mentioned subprime mortgage mess wasn’t a factor last year. I’m sure some pundit somewhere could associate Black Friday sales data to global warming or the devalued U.S. dollar against many foreign currencies.

Then those same pundits can project that one day’s data into a forecast for the future from their “sound” interpretation of the data, producing tantalizing sound bites that might get picked up in the news, especially if they sound ominous or portentous.

Black Friday or the whole holiday?
The shopping data more easily communicate my point. Data from the entire holiday shopping season of 30-plus days are more telling of the state of consumer confidence than one day’s data. To most that seems obvious.

As I write this, media are already reporting that holiday spending is down compared to 2006. You might argue from this that Black Friday dat aren’t a leading indicator of how the entire holiday shopping season will turn out. You can hear or read what the analysts are saying about that in the other two steps of the three-step process. A strong day might simply be a big reaction to one-time sales and promotions. Or a weak Black Friday might be attributed to poor weather across the country (or an increase in unemployment, or higher gas prices, or the uncertainty in the upcoming election, or higher taxes, etc.).

Analysis of economic data also follows the analyst’s outlook. The “glass half-full” people will see negatives, even in positive data. The “glass half-empty” people will see positives, even in negative data. “Seasonal shopping levels are lower than 2006 but better than many had predicted despite the mortgage mess, declining housing values, etc.” Or, “Sales levels are up, but this is probably just a temporary situation as the housing situation seems unlikely to improve in coming months.”

My point
Inferring too much about the past or the future from one data point can be costly. It’s better to look for trends based on an understanding of the data. In the case of data analysis, one person’s pessimistic view might be offset by another’s optimistic view of the very same data. Interpret the numbers yourself. A relevant old adage goes, “It’s not the newspapers’ job to tell the news. It’s their job to sell newspapers.”

Until next year, yes, measurement matters.

Happy shopping.