Not too long ago, I read that suicides by gun are more common where guns are easy to obtain. As far as I'm concerned, that's a worthless statistic. Of course, suicidal individuals with guns are more likely to use that method than people without guns.
What was the point of this statement? If guns lead to more suicides, then perhaps we can persuade people to support gun control. The problem is that this is not what the stats showed. No evidence was presented related to suicide rates in general.
I have seen similar tactics from businesses. Leading to the launch of one of Sony's game consoles, they had a contest that allowed a handful of gamers to get a console early. Reports emerged that a number of these consoles had reliability issues. The day prior to launch, Sony stated that the number of consoles with reported problems made up a small percentage of consoles that had been shipped to date.
The number of reported issues was likely low. Not everybody contacts support the moment a problem arises. There were likely some consoles won that had not been used. Some might have been saved as a gift.
The other side of this statement was more problematic. How many consoles had been shipped prior to launch that were not yet available for purchase? This number was likely heavily inflated.
How did people respond? They looked at the numbers provided as proof that there wasn't much of a problem. Some members of the gaming media even misquoted Sony and insisted that the rate provided was what they expected to see for the failure rate. For those of us who paid attention, that's not what Sony said.
In another instance, there were reports of a model of iPhone that could be warped if kept in pockets. Apple pounced on this and mentioned the small percentage reported compared to shipped products. Keep in mind that this type of problem typically takes time to emerge and even more time to be reported. Apple fans enthusiastically bragged about the small percentage of devices that had this problem. Like Sony, this was not what Apple actually stated.
In the three cases that I just mentioned, I suspect the numbers provided were accurate. Most people don't intentionally create such underwhelming arguments for those who pay attention to what the numbers actually mean.
Of course, not all numbers are accurate. There are a lot of people who misrepresent numbers, cherry pick numbers that make their views look better, or even flat out lie. If you want proof, just look at politics.
You could view polls as an example of this problem. Polls typically try to portray public views in a statistical manner. Political figures commission polls with intentionally skewed wording. They downplay the wording in the polls to make the numbers seem more credible. When results don't match what they want, they simply omit the data when discussing results.
The AFT has provided a good example of this. They tried to push a poll to show that voters are sick of Republicans harming the schools by politicizing them. The even introduced poll results by insisting that it proved that Republicans were out of touch with voters because of the dislike of the politicization. The poll backed concerns with politicization, but there was a big contradiction in their own poll that they had to downplay. The poll in question showed that voters trusted the Republican party more than Democrats on the issue of education.
As a general guideline, I don't trust any data that reflects the bias of those funding the collection of data, actually collecting the data, or reporting the data. There are far too many ways to skew the portrayal. Most people, however, will not misrepresent things to make themselves look bad. The numbers that I'm most likely to trust are those that contradict the desired outcome.
Statistics are frequently skewed and misleading, but that doesn't mean that we should pretend that they can't be of any use. Numbers can have value, but you have to be careful about how you interpret them. We need to recognize bias and pay attention to what the numbers actually mean.
No comments:
Post a Comment