OnPoint: Because Statistical Rigour
77 Responses
First ←Older Page 1 2 3 4 Newer→ Last
-
Chris Waugh, in reply to
At the recent Autism NZ conference, Tony Attwood mentioned that he now reckons that 100% of kids on the autism spectrum experience bullying.
I read an op-ed piece (i.e., to be taken with an appropriate grain of salt) discussing the case of a mildly autistic boy down in Shenzhen who had been refused entry to his local school simply because the staff and parents of the other children were scared of him. Some parents had even organised a petition to keep him out of school. The author then argued that overseas (国外/guówài - frequently code for 'in Western, developed countries') kids like him would be accepted into the mainstream system. Sigh.
-
Sacha, in reply to
It’s odd how the papers are telling us national standards are misleading and destructive – but also drawing wild conclusions from the same standards data.
Keith has a post with a refresher lesson for us on correlation and significance.
-
Of course, this ongoing tinkering with the state education system, as with hospitals, is intended to ultimately lead to private academies (and private health care) and the dismantling of the state systems. The majority in this government believe in that idealogically ... get the state out of everything. NS is part of this steady erosion by trying to show private and fee paying schools as superior.
-
Russell Brown, in reply to
Keith has a post with a refresher lesson for us on correlation and significance.
I've just realised that Jonathan Milne is actually the author of both the story about the failure of national standards in Britain and the "big class sizes are awesome" news story.
-
I listened yesterday to the 4th Reith lecture given in Wellington by Niall Ferguson, an economic historian. He was loud in his praise of current English education, and in particular private and charter schools.
I don't think he appreciated the strength of the opposition to his ideas. -
Martin Lindberg, in reply to
I've just realised that Jonathan Milne is actually the author of both the story about the failure of national standards in Britain and the "big class sizes are awesome" news story.
I noticed that yesterday and I thought that was kinda weird.
-
I listened yesterday to the 4th Reith lecture given in Wellington by Niall Ferguson...
Ferguson has been the least impressive Rieth lecturer I can remember.
-
Let me put a question to you. With all these caveats, what *is* the right way to use the data?
Correlate this data against other data.
You suggest your readers are smart enough to gain the right insight from this data. They are not. *Nobody* can get the right insight from this data because it's not there.
People are smart enough to use this in conjunction with other sources of data.
... why publish the data at all?
Because publishing this data, no matter how flawed, allows readers to expand on their existing data sets and gain a better understanding of the school system.
-
Worth noting the latest school newsletter from my daughter's primary school notes that they, like all the central Wellington primary schools have, in fact, submitted their NCEA details to MinEd (as they're required to), and the DomPost's accusation they are covering up their numbers is an outright lie.
-
Bart Janssen, in reply to
Because publishing this data, no matter how flawed, allows readers to expand on their existing data sets and gain a better understanding of the school system.
Which might be remotely plausible IF they had not drawn a false and misleading correlation between two sets of data that statistically have no significant relationship and used that false correlation as a headline to sell papers.
Publishing the data, stupid but defensible.
Drawing graphs and printing false conclusions as headlines ... are you really defending that Angus?
-
Russell Brown, in reply to
People are smart enough to use this in conjunction with other sources of data.
And yet, not smart enough, apparently, to understand decile rankings.
Because publishing this data, no matter how flawed, allows readers to expand on their existing data sets and gain a better understanding of the school system.
I'm not sure how an unmoderated set of standards that aren't synchronised with the actual primary school curriculum will really increase anyone's understanding of the school system -- especially if people are reading the information in the newspaper.
If they read the front page of the Weekend Herald, they'd have been told that our children are "failing" at reading. But is that it? Or is it more likely that the big disparity in the numbers achieving over the standard in writing (17.19%) and reading (35.49%) is that the reading and writing standards are out whack? Even without the bung journalism, I'm struggling to see where it tells me five eighths of fuck all.
-
mic weevil, in reply to
I'm struggling to see where it tells me five eighths of fuck all.
You know the standard measure for fuck all in metrics & decimals now, right?
-
Angus Robertson, in reply to
Drawing graphs and printing false conclusions as headlines ... are you really defending that Angus?
No. Just the data.
-
Sacha, in reply to
The data is widely regarded as garbage, so good luck with that.
-
Parata gets a D for the NS and an E for her Chch reorganisation of schools, and those are very generous grades.
-
Steve Parks, in reply to
I’ve just realised that Jonathan Milne is actually the author of both the story about the failure of national standards in Britain and the “big class sizes are awesome” news story.
In an update to this post, the Dim-Post notes Milne's explanation for the 'class sizes' angle.
Update: HoS editor Jonathan Milne called me to clarify that (a) they ran the analysis with and without the special schools, and acknowledged that the trend is less-pronounced without them, but still there – which is why they angled on it, and (b) there’s commentary and analysis of the data and results in the hard copy that hasn’t made it online yet.
-
linger, in reply to
The data is uninterpretable, because the numbers used do not express a single known quantity. To put it briefly: the data is rubbish. Too much noise, too little signal.
You defend publishing the data :: you defend publishing rubbish.
Any attempt to “analyse” such data cannot possibly be any better than interpreting chicken entrails, an utter waste of your writers’ (and then your readers’) time before you even begin.
Worse than that, it does actual harm, because the act of publication makes the numbers seem meaningful.Add to that the complete lack of any mathematical skills on display in this “analysis”, as evidenced by
- the wild leap to claim that there is a “trend”, based on a statistically non-significant level of correlation,
- between two sets of numbers both of which are at best only indirect proxies for what you want them to measure,
- a correlation which even if it were significant still should not be claimed to demonstrate any causal relationship,
… and … I’m actually angry that something so uninformative, so devoid of solid foundation in fact, got published as “news”. -
Brent Jackson, in reply to
Well said linger.
The really annoying thing is that there seems to be nothing that can be done to stop such reporting debacles from happening.
-
Sacha, in reply to
the act of publication makes the numbers seem meaningful
Exactly as intended. Well done, 'free' press.
-
@Brent: Of course, media faith in statistics as a magic device to create information out of pure noise isn’t limited to newspapers – as seen in CSI and similar shows where “we used an algorithm to enhance the resolution of an image reflected in the victim’s eye, captured on a crappy security camera”.
-
And do the Education Forum and private school lobby have any role in all this, beyond simple advertising revenue?
-
Steve Barnes, in reply to
Any attempt to “analyse” such data cannot possibly be any better than interpreting chicken entrails,
So. Next week are we to expect the Min of Ed to be sending representatives around
schools with chickens and long knives as an improvement on their current random collection of disparate data? -
Islander, in reply to
Why not with gagged&bound politicians & long knives?
Their entrails must be much more reliable for prognistication.
Yours fondly&feathery
-A. Chook -
Ian Dalziel, in reply to
Organ grinders..
Why not with gagged & bound politicians & long knives?
Their entrails must be much more reliable for prognosticationHaruspicy relies heavily on the liver for augury, imagine divining the future from Muldoon's liver, just prior to a snap election!
-
Steve Barnes, in reply to
imagine divining the future from Muldoon’s liver, just prior to a snap election!
Hmm, tastes just like chicken.
ETA. Goes well with cigars, cheers Keri. ;-)
Post your response…
This topic is closed.