So, I'm now reading the article he cites to support this.

It's an old article, from 2013. It states that:

"Police statistics show 25 children of the 46,718 reported missing in 2011 listed as "abducted by stranger."

And then goes on to show that in the majority of cases that the abductor was known to the parents.

So, the useful dataset would be.

* How many of the children were abducted with unsupervised?

* In the Canadian population how many children in total are left unsupervised regularly.

From that we could calculate:

Percentage of children left unsupervised regularly who are abducted.

Percentage of children not left unsupervised regularly who are abducted.

Of course, correlation != causation. But if there is no difference between the two populations then that would tell us something. However, there isn't enough data in the article to make this assertion.

Fair enough. The original statistic could well be inaccurate. I was simply responding to the comment that it's irrelevant without knowing the number of underage bus riders.

What is the relevance of the statistic present? I can't see is? (honestly).

If no (or very few) children are left unsupervised in Canada, then the statistic isn't relevant to calculating the probability of unsupervised (or lone traveling) children being abducted is it?

I'm using personal experience. I live in Canada, and I'm a parent, so I tend to frequent places that kids congregate. From my experience there are plenty of opportunities out there for someone to abduct children if they were halfway intelligent and motivated to do so. If indeed there was only one incidence over a two year span in the entire country (although your reference is apparently 25, not one?) it therefore suggests a lack of potential abductors.

> it therefore suggests a lack of potential abductors

Or a lack of unsupervised children?

I don't know. I think you're probably correct. But really the assertion would be might stronger if there was more than anecdotal evidence to back it up.

This is what I'm trying to say the data presented in the article, is on the level of anecdotal evidence. Which, tends to weaken the argument rather than strengthen it.

Isn't it possible that very few children ride the bus alone in Canada? Particularly given the fact that there seems to be governmental/social pressure against this?

So the population of unsupervised children riding buses might be 10? or 100? or 1000000? we don't know.

1 in 10 would be high, though given the limited size of the dataset, possibly not very statistically relevant.

It seems doubtful that this is a very useful dataset.

The entire population of children do not all ride alone on a bus though. It's % of those who do that that are abducted is important.

Typically not possible to estimate for either murders or abductions which ate already extremely rare nowadays in general.

Are you suggesting chasing accuracy in a fraction of a fraction?

The post I replied to said:

> One incident in two years, out of a population of tens of millions

I'm pointing out that overall population is not relevant.

The "fraction of a fraction" is the ratio of children who ride the bus alone, and get abducted. It seems that is relevant information when trying to contextualize the danger in allowing a child to ride a bus alone.