Lies, damned lies, and statistics.

[pritch]

Mark Gardiner is a motorcycling blog writer with whom some of you will be familiar. He recently attended a presentation in Washington DC in which statistics related to motor cycling fatalities featured large.
His take on what it all meant is interesting.

It is quite long, and the graphs work better on a bigger screen than a tablet - but the holidays are coming...

http://backmarker-bikewriter.blogspo...ath-rates.html

[Ocean1]

Mark Gardiner is a motorcycling blog writer with whom some of you will be familiar. He recently attended a presentation in Washington DC in which statistics related to motor cycling fatalities featured large.
His take on what it all meant is interesting.

It is quite long, and the graphs work better on a bigger screen than a tablet - but the holidays are coming...

http://backmarker-bikewriter.blogspo...ath-rates.html

You can tell a lot about the agenda behind the data source from the way it's structured and presented.

And the lack of acknowledgement of the lack of data correcting for distance traveled by a given cohort is suggestive.

Edit: it's virtually the same data profile several authorities present to justify policy here.

[pritch]

Edit: it's virtually the same data profile several authorities present to justify policy here.

It had a definite familiar ring to it which is why I posted it here.

[Viking01]

You can tell a lot about the agenda behind the data source from the way it's structured and presented.

And the lack of acknowledgement of the lack of data correcting for distance traveled by a given cohort is suggestive.

Edit: it's virtually the same data profile several authorities present to justify policy here.

Morning,
I read the referenced article with interest, and the added comments were thought provoking.

And then I read your comments above. The data gathered by government agencies or otherwise
may well be similar between countries, and the same data omissions may be present (whether it
be due say to cost, or to other factors such as politics).

But, asking out of true interest, what are the specific points that you're making or questioning ?

Cheers,
Viking

[Ocean1]

Morning,
I read the referenced article with interest, and the added comments were thought provoking.

And then I read your comments above. The data gathered by government agencies or otherwise
may well be similar between countries, and the same data omissions may be present (whether it
be due say to cost, or to other factors such as politics).

But, asking out of true interest, what are the specific points that you're making or questioning ?

Cheers,
Viking

The lack of the single most significant variable: miles per event.

And the lack of any acknowledgement that it's missing.

The suspicion being that, although it's absence is almost certainly due to it having not fallen into the researchers laps via freely accessible sources, treating the available data as if the confounding variable were irrelevant is almost certainly due to the fact that the data presented fits a narrative required to justify downstream decisions.

ACC used similarly truncated data to conclude that "born again" bikers were more likely to kill themselves. They even went so far as to present the raw data in different sized age bands, EG: fatalities per 20-25, 25-35, 35-45, 45-75.

When roughly corrected for accidents per kilometer and displayed at a similarly sized age demographic the data showed what most consider a more believable picture: Beginners have far more accidents, there's just fewer of them lately, and far more old bastards with more time on their hands reacquiring the riding habit.

[Viking01]

Thanks for the reply. Appreciated.

Understand your point re age bands and miles travelled. And how an invalid
conclusion could well be derived by a researcher (whatever their motivation).

Re your last sentence: When the source data was re-cast and re-presented,
was that analysis performed by another party (other than ACC) ?

I tend to keep links to such types of analysis (or copies of such reports)
whenever I see them published, but I couldn't find anything similar in a
collection of such resources on my PC. Admittedly, there's only about 15
docs and they date back only as far as 2015.

Would you happen to recall the source of the report, or happen to have a
link to it ? Would be good if I could get my hands on a copy.

Cheers,
Viking

[FJRider]

When roughly corrected for accidents per kilometer and displayed at a similarly sized age demographic the data showed what most consider a more believable picture: Beginners have far more accidents, there's just fewer of them lately, and far more old bastards with more time on their hands reacquiring the riding habit.

This may just prove that the same statistic's (put in a different format) can get an entirely different result. It also means ... if the appropriate formatted statistic's are used ... they can "prove" different sorts of results. Depending on what result you want proved.

Was not the learner rider/driver statistic's used ... to get the change to the rider/driver licensing format that we have today ... ???

[Akzle]

The drug companies need to come up with an alternate cure for a MLC to buying a motorbike. That will be the only way to cut MLC rider deaths assuming they are not the fault of anyone else. An interesting statistic would be how many MLC guys are walking into bike shops each week and buying bikes? Having statistics for the reason why guys get them in the first place would be interesting too. Maybe counselling services need to have a MLC division to divert interest away from buying a bike.

you're a fuckwit

[Akzle]

ACC used similarly truncated data to conclude that "born again" bikers were more likely to kill themselves.

ACC (in this instance) must be fuckwits too by your warped logic.

you have no idea what "logic" is. so stop using the word.

you're a fuckwit.

[Ocean1]

Thanks for the reply. Appreciated.

Understand your point re age bands and miles travelled. And how an invalid
conclusion could well be derived by a researcher (whatever their motivation).

Re your last sentence: When the source data was re-cast and re-presented,
was that analysis performed by another party (other than ACC) ?

I tend to keep links to such types of analysis (or copies of such reports)
whenever I see them published, but I couldn't find anything similar in a
collection of such resources on my PC. Admittedly, there's only about 15
docs and they date back only as far as 2015.

Would you happen to recall the source of the report, or happen to have a
link to it ? Would be good if I could get my hands on a copy.

Cheers,
Viking

I don't have a link to the original source material, it was ACC though. The raw data wasn't made available at the time, but I later stumbled upon it here somewhere and rearranged the age grouping myself. I wasn't interested in the results outside of confirming that the original format was heavily biased in favour of the report's recommendations, so I didn't bother keeping it.

As for motivation, Occam's razor suggests it was an attempt to justify the massive increases to ACC levies on licencing, (presented by Nick Smith and co at the time as a way to redress what was claimed was an unfair extra burden motorcyclists put on the system). The fact that the target demographic, (older, high earning big bike riders) represent both lower hanging fruit in terms of a levy increase and a higher cost in terms of income insurance adds significant weight to Mr Occam's diagnosis.

He might also have been interested in the fact that this was the first historic instance of any attempt to levy ACC directly in terms of claimed associated activity costs.

[Ocean1]

This may just prove that the same statistic's (put in a different format) can get an entirely different result. It also means ... if the appropriate formatted statistic's are used ... they can "prove" different sorts of results. Depending on what result you want proved.

Was not the learner rider/driver statistic's used ... to get the change to the rider/driver licensing format that we have today ... ???

Yeah. Nah, not only are statistics no part of the group "lies, damned lies etc" but it's literally correct that statistics don't lie. Which says nothing about their suitability in supporting any given conclusion, in the case of the original ACC biker beat-up the data was presented in such an amateurishly distorted way that if I'd been minister anyone associated with the report and it's recommendations would have been sacked.

[pritch]

The drug companies need to come up with an alternate cure for a MLC to buying a motorbike. That will be the only way to cut MLC rider deaths assuming they are not the fault of anyone else. An interesting statistic would be how many MLC guys are walking into bike shops each week and buying bikes? Having statistics for the reason why guys get them in the first place would be interesting too. Maybe counselling services need to have a MLC division to divert interest away from buying a bike.

Using the ancient standard of three score years and ten your MLC happens about age 35 when people are riding bikes less because wives and kids tend to be the priority. Even if you extend that life expectancy by ten years, which may not be actually justified, mid life would be forty.

Once again you seem to miss the mark completely.

[pritch]

You have missed my point completely too in that I was not wanting to enter into a debate over at what age a MLC happens but I was just highlighting the point made by ACC that "Born Again" (MLC) riders are at a high risk of killing themselves. There needs to be an alternative cure to buying a bike for guys who suffer a MLC so there are fewer MLC motorcycle deaths I feel. Possibly the best way of avoiding becoming a MLC motorcycle accident statistic is to never give up riding in the first place.

No you are completely confused. SNAFU. Born agains and MLC are different age groups completely and the argument Mark Gardiner presents is that the statistics are misleading. If you had actually read and comprehended what was written you would understand that Gardiner's argument indicates that it is possible the ACC may be misreading the statistics too. As have you.

[GazzaH]

The article is a mess of opinion, conjecture and assertions by a blogger who appears distinctly biased or prejudiced. It lacks hard data and statistical analysis of the data presented in Washington. Is the original data source even cited? Has it been duly peer-reviewed and published? If so, I missed it.

As described by the blogger, the original presentation smacks (to me, a cynic) of death-by-data: 'We have a bunch data points, so let's graph them and blabber as if they mean something and we know what we're talking about'. The data could be used to develop theories and testable hypotheses for further assessment but, without additional work, pure conjecture such as the blogger's commentary is almost worthless. We have no reasonable, rational basis for deciding between the original presenter's findings and conclusions (which don't appear to have been properly cited and referenced by the blogger), the blogger's alternative opinions, and other possibilities (of which there are many, mostly unstated).

But as a device to drive up blog readership, or for political purposes, it's OK I guess. Who cares about science, eh? Who needs statistics? Does it matter if the data concern the US, not NZ? And what have the scientists and statisticians ever done for us?

[pritch]

The article is a mess of opinion, conjecture and assertions by a blogger who appears distinctly biased or prejudiced. It lacks hard data and statistical analysis of the data presented in Washington. Is the original data source even cited? Has it been duly peer-reviewed and published? If so, I missed it.

As described by the blogger, the original presentation smacks (to me, a cynic) of death-by-data: 'We have a bunch data points, so let's graph them and blabber as if they mean something and we know what we're talking about'. The data could be used to develop theories and testable hypotheses for further assessment but, without additional work, pure conjecture such as the blogger's commentary is almost worthless. We have no reasonable, rational basis for deciding between the original presenter's findings and conclusions (which don't appear to have been properly cited and referenced by the blogger), the blogger's alternative opinions, and other possibilities (of which there are many, mostly unstated).

But as a device to drive up blog readership, or for political purposes, it's OK I guess. Who cares about science, eh? Who needs statistics? Does it matter if the data concern the US, not NZ? And what have the scientists and statisticians ever done for us?

Holy shit! Your comprehension skills are a match for Cassina's. Why don't you two get a room?