Bad data leads to bad decisions

Russell Wangersky
Published on May 6, 2014

It may seem trite to say so, but your decisions are only as good as your baseline data. Start with flawed data and it’s pretty hard to make the right decision by anything except pure chance.

This week, the Globe and Mail revealed that one of the main planks that’s been used to justify the Temporary Foreign Worker (TFW) program may have been critically flawed. The federal government had been arguing, in part, that the program was needed because there were so many unfilled jobs in Canada.

The government had based that “knowledge” on a web program called Wanted Analytics that scans online job boards and counts openings. The problem is, the program apparently didn’t take into account jobs that were posted multiple times, particularly on Kijiji, where posters can place the same job under several different headings. Each posting was counted as a separate empty job; knock out those numbers and that dropped the national job vacancy rate from the four per cent used to prove the need for the massive growth in TFW positions, to a much-different 1.5 per cent.

In other words, the skills shortage that meant workers had to be imported was not nearly as serious as was thought.

No wonder, then, that a growing number of Canadians have come forward to say that they’ve been turned away from jobs that were subsequently filled by a burgeoning number of foreign workers. (Lots has been written on the ethics of the TFW program — much of it that we’re using temporary workers to fill jobs we don’t want, and holding out a thin, thin chance of citizenship as a fake reward. If we really wanted to be honest, we should be allowing people to immigrate fully if they’re willing to take on work. That seems like a much more ethical trade, and much less likely to be a case of taking advantage of the disadvantaged.)

But back to data.

In a lot of ways, the huge amount of data at our fingertips is making us sloppy. Just in this last couple of weeks, for example, there have been diametrically opposing views on the dangers of closed-pen aquaculture in this paper. The aquaculture industry has tooted its own horn and said that science backs up that it doesn’t do damage; opponents have shown data that proves the industry is either being wilfully blind or remarkably selective in the data it uses.

Is there data on both sides? Of course there is.

Is the data of even close to equal value? No. Far more data suggests that the escape of farmed salmon is doing damage to wild stocks. You can choose not to believe that, but that’s the same kind of choice as choosing not to believe that global warming is occurring, even as businesses as staid as the insurance industry calculate both the empirical damage of strengthening storms and jack up your premiums based on actuarial forecasts of what’s coming next.

The thing is that no single source is simple proof. Being convinced by a single study suggests you’re picking a study that confirms what you believe, or want to believe, already.

You need a preponderance of data, and good data at that.

You can, of course, choose to put your head in the sand instead, by merely refusing to look at information that contradicts your beliefs.

The move from a full long-form census to a voluntary effort is just one example of how the federal government has muddied the data waters. The strangling of federal science is another.

And it may work for a while.

In the end, though, it may well be a case of cutting off its nose to spite its face. Governments that choose to fly blind can’t help but fly into  the occasional wall they didn’t see coming.

Good numbers can at least give you a little early warning, even if you don’t like what they’re showing.

Russell Wangersky is The Telegram’s news editor. He can be reached by email at