Yet Another Dangerous Graph and Poll
Written by Gloria Metrick on . Published as written. It may not reflect how these systems work today.
Just yesterday and today, I ended up in a probably too-long discussion about an graph of industry products and whether it was credible or not. The problem is that the results graphed out to be a distribution that didn't seem possible.
Personally, getting into a discussion about the graph was interesting and helped give me a great reason to avoid the work I should have been doing. ;-) However, it also gives me yet one more opportunity to give an example of a graph that could be misused.
It turns out that the errant graph was part of a real poll of a specific sector of people regarding products within our industry. When I was able to read the selection of people included in the poll and all the supporting text, the graph did seem to make sense.
The Problems
What's the problem, then? The problems with this graph were these:
1. The graph ended up taking on a life of its own being taken from the poll and supporting data that produced it. Hopefully, it wasn't used to prove or predict anything, but once these graphs and charts go out on the internet, all bets are off - people will now find them, think they're standalone pieces of information and use them for important purposes.
2. Even when the graph remains with the supporting information, most people won't bother reading the supporting information but will look just at the graph they're interested in, take it as a standalone item, and use it for an important purpose.
One More Problem
Besides the fact that too many of us find these standalone bits of statistics that we use to prove our point to our bosses, our customers and ourselves, the other issue is that most people don't actually understand the statistics as well as they think they do. When information is boiled-down into a simple chart, it's easy to read and seems straightforward. Most people don't take statistics in school nor learn much of it later in life, except for a few bits such as the mean average, standard deviation and possibly a few other bits, which is not actually enough to get us thinking about what went into the graph or chart.
What you have to do when you see these is ask these questions, as a minimum:
1. Where did this come from?
2. What group of people was polled?
3. How many people were included in the poll and what percentage of all eligible respondents is that?
Gloria Metrick
GeoMetrick Enterprises
http://www.GeoMetrick.com/