OnPoint: Don't put words in our mouths, Rob
142 Responses
First ←Older Page 1 2 3 4 5 6 Newer→ Last
-
Even Statistics NZ says it has difficulty collecting all the data, even when its 'counting everybody'. eg
" males are more likely to be missed than females; young adults are more likely to be missed than older adults; Maori, Asian and Pacific people are more likely to be missed than other ethnic groups."
Asian and pacific peoples ? Oh well let's throw it all out then..For every fault that labours data has , there seem to be false flags planted about 'seasonal variation, or specifically 'targeting chinese buyers', or even the most ludicrous 'hurting asian people'
Even Keith Ng, who seems to be an amatuer statistician and even a volunteer with St John Ambulance is pushing beyond his particular speciality- an angry journalist.
It would be good to hear more people who work in everyday statistics -
Sacha, in reply to
Keith Ng, who seems to be an amatuer statistician
Professional data journalist, thanks. Unlike you or me. Show some respect.
-
Craig Ranapia, in reply to
Even Keith Ng, who seems to be an amatuer statistician and even a volunteer with St John Ambulance is pushing beyond his particular speciality- an angry journalist.
-
Joe Wylie, in reply to
It would be good to hear more people who work in everyday statistics
You mean the kind of folks who dreamed on untroubled while this turkey was flapping loose?
-
steve black, in reply to
It would be good to hear more people who work in everyday statistics
You have heard from me. I am a PhD qualified working statistician. What you are saying isn’t helping to create a quality discussion of the issues. I spent some time constructing my criticism avoiding all mention of the Chinese Ethnicity issue because I feel the Chinese Ethnicity is but one of the methodological issues.
That quote from the Stat Dept is about recognizing that all data has issues and limitations. But no we don’t throw it away. We write them up in the methodology section and use the limitations to inform our analysis so we don’t overgeneralize. Or we take action to adjust for the defects in sampling, and we test our assumptions about how these adjustments affect estimates. These are the sorts of things which weren’t done adequately in the report.
You have also heard from Thomas Lumley on Statschat.
If it was a medical study they would have measured residency, not just imputed it from ethnicity.
link: statschat blog
and he’s a statistican too. Failure to measure what they needed to measure (residency) is still the biggest flaw for me. But I’ve discussed others. Mostly I say "there are things I'd need to know more about to judge the weight of evidence". And I'm all in favor of collecting the right data.
Thank you Craig, for the double facepalm. It’s probably a more eloquent response than mine.
-
Tze Ming Mok, in reply to
I thought your contribution earlier was really useful.
Also while we're on methods, although I didn't mention it in my original post last week, I thought Rob's triangulation exercise was quite weak.
-
Russell Brown, in reply to
Rob’s latest column is just barefaced partisan hackery. I know Russell reposted it to encourage generate debate, but I’m embarrassed that it might be seen as an endorsement of Rob’s independence. Russell, I think that perhaps for Speaker posts it’s a good idea to include a line about the author’s political party affiliations and employment.
I’ve had it since Saturday and I gave myself yesterday off doing anything with it in part because I didn’t think it fairly characterised yours and Keith’s positions (the other part was that I had a very late and jolly Saturday night out and felt I was due a day off grown-up editorial decisions). I decided to go with it this morning partly on the reasoning that it gave you a place to respond.
Rob’s desire to have it discussed was genuine, I think. It’s not like it being here was going to get it in front of many eyeballs compared to being front-of-section in the Star Times.
Rob’s position as a sometime contractor to the party has been evident in some of his previous posts, and, indeed, is implicit in both the posts he’s made in this issue, so I’m not so concerned about that.
I’ve never been in a position quite like this before, with bloggers on my site on opposite sides of a live issue. I think letting the discussion happen has mostly worked. But I hear you.
-
Also, we’ve just recorded a pretty good Media Take discussion with David Soh, editor of Mandarin Pages, and Raybon Kan – in the third part of tomorrow night’s show.
I really liked meeting David and his friends. His English isn’t great, but he got through and his concluding comments were thoughtful and positive. I'm going to see if he wants to work with someone on a post for us expanding on those comments.
-
So we have established Keith Ng is a computer programmer who works in the journalism area ( among others)
I loved his graphical representation of the government budget in the Herald a while backBut I presume he wasn't responsible for this foray into statistical data journalism
http://www.statschat.org.nz/2015/07/20/pie-chart-of-the-day/ -
Emma Hart, in reply to
So we have established Keith Ng is a computer programmer who works in the journalism area ( among others)
No, that's okay, Steve. What's really important is the way you acknowledged your fuck-up and owned it.
-
TracyMac, in reply to
I had no idea what his relationship to the party was till he implicitly, as you say, mentioned it in the article, and it was discussed in the ensuing comments.
Not only in relation to this, as someone non-resident in NZ, I really often don't know the personalities floating about in the media space. A byline outlining someone's affiliations and why they're commenting on Issue X would be great as a standard practice.
I was under the impression that Rob Salmond was an independent statistician engaged by the Labour party, not a member.
-
TracyMac, in reply to
could you elaborate on what you see as the seasonal effects on name frequency?
End of financial year in China? Exchange rate? Spring? Chinese New Year? I saw that as example of tossing out possible variables rather than an actual opinion that it DOES vary seasonally (correction welcome!)
As for the debate about whether B&T specifically target Chinese buyers beyond any other company, surely that's a data point that should have an impact on the wild assertions being made?
Not least because if they do target Chinese residents more than any other company, the results - such as they are - would be skewed upwards.
-
Steve Curtis, in reply to
Keith, could you elaborate on what you see as the seasonal effects on name frequency?
Others too are wondering about KN depth of stats knowledge
He is literally saying his level of certainty is unknown + unknown. Which equals, of course: unknown.
Well he ( salmond) certainly played with numbers to make it look better, which is as Prof Lumley says all too common with scientists too.
http://www.statschat.org.nz/2015/07/15/a-modest-proposal/Of course the Herald isnt renowned for its rigour with numbers, as we had every 'data journalist' pulling them apart about Dong Lui s incredible shrinking donation to labour. Perhaps not then.
Even today there's something in the Herald for Emma to deconstruct
Curvaceous women are more sensitive to pain than others, scientists believe
http://www.nzherald.co.nz/health/news/article.cfm?c_id=204&objectid=11483203Which to choose:
Pull it apart as numerical and sexist nonsense or include it in the plot line of your newest book -
David Hood, in reply to
I was reading a really good article the other day by an Iran blogger who had been in jail for many years. When the Iran authorities put him away he was at the forefront of a whole set of blogs engaged in vigorous debate about issues. When he came out it had all faded away replaced by social media likes.
https://medium.com/matter/the-web-we-have-to-save-2eb1fe15a426
-
Sacha, in reply to
Others too are wondering about KN depth of stats knowledge
You're not really commenting in good faith. Improve your attitude, dude.
-
David Hood, in reply to
End of financial year in China? Exchange rate? Spring? Chinese New Year? I saw that as example of tossing out possible variables rather than an actual opinion that it DOES vary seasonally (correction welcome!)
I'll amend what I said to ask "seasonal factors about name frequency that affect the analysis under discussion" because while I freely admit I may be missing something, I'm not seeing how those kind of seasonal factors affect the analysis, as I hadn't seen anyone generalising the Barfoots data to the whole year.
-
Wow, you really hate Keith Ng, eh, Steve Curtis? Not sure you're adding anything to the debate other than repeatedly saying you dislike him and looking a little bit racist.
-
Emma Hart, in reply to
Even today there’s something in the Herald for Emma to deconstruct
Curvaceous women are more sensitive to pain than others, scientists believe
Oh Steve. I know as much about pain as Keith does about stats.
-
Steve Curtis, in reply to
Hes a big boy , doesnt need the nomenklatura to play minder.
-
And the discussion doesn’t need to focus on attacking the people taking part. We can all do better than that.
-
Steve (Curtis), I've blocked you from commenting. Your behaviour is in bad faith, hostile and not acceptable.
You can email me via the link at the bottom of any of my posts to discuss this if you like.
-
Whoa, Emma got presented with some Being a Woman on the Internet in a thread about flawed property market statistics. I think I have now seen everything.
-
I know it's very bad form to talk about someone who's just been told to leave the room, but all this talk about "people who work in everyday statistics" just throws into relief a lot of what I find really troubling about this.
A lot of people are talking about "data" and "ethnic groups", and forgetting that this is coming out of a political intervention on a matter of public policy that actually affects human beings not some abstract collective noun.
Morgan Godfrey did a guest post on The Ruminator today When are numbers racist? that everyone should read and think hard about, if they haven't already.
The point is that data analysis isn’t neutral, the narratives we craft from it even less so.
-
nzlemming, in reply to
Russell, I think that perhaps for Speaker posts it’s a good idea to include a line about the author’s political party affiliations and employment.
+1
-
nzlemming, in reply to
Hes a big boy , doesnt need the nomenklatura to play minder.
You're being a dick, Mr Curtis.
Post your response…
This topic is closed.