A Screwed-Up Life

"Data is life" is a motto I've heard, and it's a good one. In other words, without data, we find ourselves just mucking about from one opinion to another, one intuition to another. Data can keep us rational and reasoned.

Better is information and good analysis of the data leading to a better life.

However, a bit of data can be dangerous. It's also said that a little of knowledge can be a dangerous thing. Today on a public radio broadcast, hosted by Kerri Miller, they were talking about news and the objectivity and fairness of it. One of the problems with the news business is that they have to "create" news by stretching the data. They need to take a 3-minute story and fill 30 minutes (or worse, 3 hours) by adding analysis and commentary. The news agency will bring in lots of experts (or worse, opinion makers) to provide the analysis. They only have a bit of data. It's not complete. They often put in juxtaposition to other bits of data and assume causality. "The Dow Jones Index went down 10 points today because it rained in Brazil this morning..." I mean, "Come on! What are you trying to infer?" Statistically, there's no cause for a 10 point drop. It's normal random variation. Or take another story: "Education System Has Failed; Half of the Kids are Below Average" and then the analysts extrapolate all kinds of causes that make the education system disappointing...when, in fact, this is also a insignificant bit of data. (Obviously, half of anything will be below average given a normal distribution of data.)

One datum is hard to make significant without understanding the context. "Sales are up this month" is a datum. However, we cannot claim success because we haven't put it into the context of how much sales varies from month-to-month. "Profits are down" may mean our strategy is not working, only if there's a statistically significant trend. Comparing one period's (month's, year's) results with another is meaningless. Any result will always be in one of three comparable states: up, down, or the same. Most of the time, it's like a coin-flip whether sales or profits or some other metric is going to be higher or lower than the previous datum; there's a 50/50 chance that it will be better.

Two correlated facts or consequent events do not mean causality. "We re-organized the company" and "Our profits are up", though occurring within the same fiscal year, does not mean that there's a causal relationship. Increased profits have many influences. Stating one event at the same time as another does not mean they're related. "It's warm today" and "My gas mileage was better than normal" is similar faulty reasoning if I try to link the two events. One company's profit improvement or marketing failure will not cause the Dow to change but a microscopic bit. And yet, most news reports will link the two claiming that one company's results will influence ALL of the buying/selling decisions on the other companies in the Index.

Third, we often fail to find other data to corroborate our conclusion, or ignore data that contradicts our conclusion. Recently, Jim Paulsen, an economist for Wells Fargo Capital Management, has pointed out (on his website and in the Financial Times) that the US recovery from the latest recession has been the strongest in the past 25 years. Most of the news still points to a slow recovery. True. However, most analysts fail to compare it to other recent recoveries. There is a significant difference in recoveries pre-1985 and post-1985. There are also significant differences in the global market place pre-1985 and post-1985. One is the the number of people entering the workforce. All of the Baby Boomers were in the labor force by 1985. There are other considerations also that may mean we'll never see pre-1985 rates of recovery. Likewise, in most companies, we'll look at one bit of data that hoped to be better. We'll compare the results to marketing efforts, operation efforts, etc. We'll only look internally for the reasons, and not look at external factors--changing competitive landscape, supply chain factors, demand curves, etc.

Unless you're the CEO...then you'll take credit for a good or great year, and blame the external factors for a bad year. Now that's data stretch. "Data is life" but misused it paints a picture of a screwed-up life.