martine wrote:Rolyan wrote:We could start by getting the full data.
Who says there's a problem? This increase may be part of the normal variation expected in a downward trend.
I may have missed the graphs and/or analysis, but this appears rather ill informed and sensationalist.
Yes the full analysis will be interesting but it does say in the article that only 3.5% of the total 7.7% increase can be explained by the US economic recovery (and hence increased mileage). I'm not a statistician but I'm pretty sure 4.2% rise in such a large number (32,675) is beyond normal variance.
https://crashstats.nhtsa.dot.gov/Api/Public/ViewPublication/812269
I hear where you're coming from, but the problem is that we simply don't know what the normal variance is. So we cant judge (believe/think/guess) that this increase is above it (or below it).
In one of the training courses, I show three graphs reflecting the share price of genuine companies, measured in the last 12 months. I ask attendees to think about their own approach to risk, and decide which, if any, of the three companies they would invest in. Graph one shows a consistent large increase but with an occasional large drop, two is fairly flat but a very small increase and three is a huge continuous drop. After discussions, its revealed that they are all the same organisation and more importantly,
all the same graph.
There are lots of other examples, but I've seen so many poor uses of data in regard to road safety that I don't trust anything reported without seeing the data and the analysis that goes with it. The time period, context, collection method and analysis all affect the results.
Please note: I'm not suggestion that this data is 'normal' or otherwise, I'm simply saying that we can't know from the evidence provided.