Thread by @Mark_Goldberg

Normally, I would keep clear of the controversy about @AmiraElghawaby's appointment as Canada’s Special Representative on Combatting Islamophobia, but I think there is an unfortunate overlap with the #LaithMarouf debacle, which gives me a modicum of authority to weigh in

1/13

Apparently, @CdnHeritage does not have access to internet search engines, which should concern all those watching the various pieces of Digital legislation passing through Parliament. But I digress.

I'll focus this thread on the charge of @AmiraElghawaby being anti-Quebec

2/13

The offending column, co-authored by @BernieFarber appeared in the @OttawaCitizen 3½ years ago: ottawacitizen.com/opinion/column…

The key offending paragraph cites data sourced from a @leger360 poll

3/13

Note that the 88% figure is a hyperlink, that in turn links to a @mtlgazette article by @JasonMagder from two months earlier: montrealgazette.com/news/local-new…

It describes a poll of 1,212 Quebecers commissioned by @CanadianStudies, offered exclusively to @mtlgazette and @LeDevoir

4/13

The @JasonMagder article is indeed the source of the 88% figure: "Among those who have negative feelings about Islam, 88 per cent support a ban on religious symbols for public school teachers."

But here is the rub: It doesn't say what % has negative feelings about Islam

5/13

I think that there is a measure of innumeracy that is driving this matter - at least as far as @AmiraElghawaby's views of Quebec.

The @mtlgazette article tells us 28% of those polled had a positive view of Islam and 37% had a positive view of Muslims, but ...

6/13

But it is silent on how many gradients were in the scale, how many were neutral, and most importantly, how many were counted as having negative feelings about Islam.

Without that, the 88% figure cited in the @BernieFarber - @AmiraElghawaby article is useless.

An example:

7/13

It is mathematically possible that only 9 of the 1,212 people polled answered in a way that got categorized as "negative feelings about Islam". If 8 of those 9 supported the ban on religious symbols for school teachers, that would yield the 88% figure.

8/13

Unless @BernieFarber and @AmiraElghawaby had the full study, there is no way they could conclude (as they did) that "the majority of Quebecers appear to be swayed not by the rule of law, but by anti-Muslim sentiment."

That was the premise for the entire column.

9/13

Speaking as a statistician, I am concerned about the possibility that innumeracy led to an offensive generalization about Quebec, and indeed that may have inappropriately shaped the views of the authors.

It would be fascinating to see the entire poll, questions, results.

10/13

I can't find links to the poll itself

As noted by @CanadaCRIC (@leger360 is a member): “CRIC’s standards provide an important tool for journalists to ask the right questions about polls and to do a thorough job of vetting facts before they publish them with their byline.”

11/13

Bottom line: if @AmiraElghawaby, @BernieFarber misread the 88% figure and concluded it was the percentage of Quebecers supporting a ban on religious symbols on teachers, then they should retract the @OttawaCitizen article in its entirety, since it's based on flawed premise

12/13

Finally, the innumeracy among so many (and I see it a lot in my day job, as frequent followers know) concerns me.

Too many people don't have an ability to look at numbers in the paper and sense that something doesn't smell right.

But I'll save that for another day

13/13

About @leger360 web-based survey cited by @AmiraElghawaby and @BernieFarber: In 2019 (when the survey was done), Quebec had Canada's lowest rate of internet use. @Statcan shows ~1 in 6 Quebecers (age 15+) didn't go online, so it couldn't really be a representative sample

PS1/13

The survey has a very misleading statement about accuracy: "For comparison purposes, a probability sample of 1000 respondents would have a margin of error of ±3.09%, 19 times out of 20"

But this wasn't a random sample, so no margin of error could actually be given

PS2/13

Lessons:
• Journalists need to learn how to question surveys, per @CanadaCRIC: canadianresearchinsightscouncil.ca/wp-content/upl…
• Before you write: "the majority of Quebecers appear to be swayed ... by anti-Muslim sentiment", be sure you have reliable data, and someone to check the data twice.

PS3/13