Hard News: Mo' Indexing
75 Responses
First ←Older Page 1 2 3 Newer→ Last
-
You hope... The data may have been weighted at that stage, or they may have done the weighting themselves, I wouldn't be prepared to put money on it though
according to my on-tap stats guru, that would be a fairly fundamental mistake. i.e one you'd not find among decent analysts.
unless, of course, you did it deliberately.
-
"New Zealand ranking second only to the United States for cannabis use provides further evidence that our country is descending into the world's gutters," Mr Sabin said.
And you solution, Mr Sabin, is to copy USA's drug policy?
http://methcon.co.nz/press/28may08.pdf -
Descending into the world's gutters one All Black at a time.
The only time I've ended up in the gutter is when I drank too much.
It's the same for almost every one else I know.
-
(how do you measure the socio-economic status of Mechanics Bay, where no one lives?)
I would consider Mechanics Bay to be the triangle of land between Beach Road, The Strand and the harbour, and there are heaps of new apartments and townhouses there.
Yeah.
-
I spoke too soon...
Drug survey shows NZ slipping into 'world's gutters' - expert
http://www.nzherald.co.nz/section/1/story.cfm?c_id=1&objectid=10519516 -
The reporting surrounding drug survey once again shows the perils of "ranked x in the world" type statements. Very often the "world" is the OECD, a smaller number of countries, or simply whatever countries data could be found for.
-
according to my on-tap stats guru, that would be a fairly fundamental mistake. i.e one you'd not find among decent analysts
Yup. Now for bonus points, do you think you can find a statistical correlation between newspaper stories about studies and fundamental mistakes?
What kind of carrot/stick/magic wand would be needed to get all news stories regarding such studies to include a reference to where one can obtain the source document.
And while I'm at it, if they could be forced to say what they mean by average and if they mean Mean then include the standard deviation.
ah, sweet, sweet fantasyland
-
An interesting aspect of the NZDep is how it is quantified. There are 9 dimensions taken into account - people aged 18-64 receiving a means tested benefit, people living in households with income below an income threshold, people not living in own home, people aged <65 living in a single parent family, people aged 18-64 unemployed, people aged 18-64 without any qualifications, people living in households below a bedroom occupancy threshold, people with no access to a telephone, people with no access to a car. I think some people find it surprising that there are people out there living without broadband, let alone no telephone...
The ethnic breakdown of deprivation is also intersting, with an under-representation of European and Other groups and an over-representation of Maori and Pacific people in the most deprived deciles. And it means something - life expectancy reduces / mortality rates increase with increasing deprivation (and then independent of deprivation mortality rates are higher than for non-Maori) - that is there is a socioeconomic gradient in mortality.
In these maps they seem to have gone for quintiles rather than deciles, so the red parts are areas that are the most deprived 20% of the country and the pale pink parts are areas that are the least deprived 20% of the country. But it is, of course, relative. So it's not like the red parts get redder and pale pink parts become paler as inequality grows. Yet socioeconomic inequality increased over the 80's and 90's and so did inequalities in mortality. The late 90's and early 2000's saw a slowing of the increase in relative inequality. Let's hope that's a trend that continues.
-
Whatever people think of Anderton, he does listen to evidence.
Maybe with drugs but not with fisheries.
You hope... The data may have been weighted at that stage, or they may have done the weighting themselves, I wouldn't be prepared to put money on it though
Wait? What are they weighting? (see what I did there?)
You usually over-sample parts of a population in order to achieve a large enough number of that part to do later analysis on. We do it all the time here at...the place that I work.
Why would you over-sample and then weight the results as well? (I suppose, he says answering his own question, if your over-sampling results in a very skewed sample)
-
The only time I've ended up in the gutter is when I drank too much.
Well that's just good citizenship. A decent rain will wash that away, whereas someone's doorstep, it stays there for weeks.
I spoke too soon...
ODT and NZ Herald. Yup, no great surprise. One pretending to be a big city paper, when it's not. The other seemingly doing the reverse.
-
I won't pretend to understand the statistical geekery at anything beyond surface level. Oversampling gives them big enough sub-samples to separately do statistically valid analysis of Pacific and Maori populations. No idea how they subequently adjust the total sample to acknowledge that they've skewed it.
However, the methodology seems solid for the survey as a whole so I'm pretty confident they won't have made fundamental stuff ups - other than seeing mental health through a clinical lens and following the WHO's dodgy understanding of disability.
Much like distorted interpretations of drug usage by those with an interest in stoking fear...
-
Interesting to see that Ponsonby still has a pocket of "most deprived". Deprived of "what"? This season's Issey Miyake?
One problem with NZDep is that it makes some assumptions about "deprivation" that may apply for the bulk of mainstream suburban NZ, but don't work well for truly urban lifestyles. Lack of access to a car, and renting rather than owning one's home, both contribute to "deprivation". Thus places like Wellington's CBD have a surprisingly high deprivation index, despite a fairly affluent population. Living without a car in the outer suburbs would be a nightmare, but when urban amenities are at your doorstep and there's decent public transport, a car-free life is a liberation rather than a deprivation.
Also, I'm not sure whether they normalise income for age and family circumstances. A single twenty-something might be earning less than the average wage and still have a high disposable income. Thus the NZDep index can be quite useful if you're comparing families from two different suburbs, but when you compare that to people with very different needs, some apparent anomalies arise.
-
Why would you over-sample and then weight the results as well?
I'm no stats person, but I'd imagine you oversample (say) Maori, so that if you take out the Maori data and talk about it by itself, you have meaningful numbers - ie, not 10 people.
When the data is dealt with as a whole, you weight it so your oversampling doesn't distort your picture of 'NZ as a whole'.
-
What Kyle said, I think.
-
This deprivation atlas is a little counter intuative.
Parts of Riccarton are at 7 & 9, sure. Bryndwr 2 ,Fendalton 3 & Deans Bush 1, OK, but Mona Vale 4?
Mona Vale is populated by palaces and is rated the same as Burnham Military Camp?I must have missed what they're measuring.
-
ODT and NZ Herald. Yup, no great surprise. One pretending to be a big city paper, when it's not. The other seemingly doing the reverse.
They're both the same NZPA story, which is just a rewrite of Sabin's press release. Welcome to modern news reporting.
-
Thus the NZDep index can be quite useful if you're comparing families from two different suburbs, but when you compare that to people with very different needs, some apparent anomalies arise.
Excellent point Tom.
You see a similar "ghetto" in the Auckland CBD, especially along the Nelson-Hobson ridge, where there are a lot of students living in tiny apartments with no car parking.
-
When the data is dealt with as a whole, you weight it so your oversampling doesn't distort your picture of 'NZ as a whole'.
yeah, that's what I tried to write in a rather glib way.
Having just checked our stuff, what you do is weight the over-sampled sub-populations with the inverse of the probability of their selection.
So for example: if Pakeha have a 1/1000 chance of selection thier weighting is 1000 and if you over-sample Pasifika at 2/1000 then their weight is 500.
-
I must have missed what they're measuring.
Me too...having spent much time in areas in places like Jawa Tengah where families live in one room with a dirt floor, where 60% have no access to clean running water and although the filthy stream running down next to your road floods every year for a week or two but you need to stay in the room because that's all you have, I find the application of the word deprived to folks that don't have a car or access to a phone or are forced to live on a benefit, as relative as it might be, rather bizarre.
-
A single twenty-something might be earning less than the average wage and still have a high disposable income
And that goes double for students, most of whom are living on under $15k a year and still have a reasonable amount of disposable income, by dint of flatting in large groups. There's just no comparison, even between people of the same age, where one is a student flatting and the other has family to support or a kid.
Actually, I'd love to see a comparison of that deprivation index to average family size; I've heard anecdotally that number of children is one of the best indicators of poverty/deprivation.
-
Drug survey shows NZ slipping into 'world's gutters' - expert
But at least we're happy ...
15th in the world (eight places above Australia!) according to the latest University of Michigan study of global happiness.
Zimbabwe is the least happy country in the world.
Cause for optimism: Northern Ireland is no.5 on the happiness league table, and Colombia is no.3.
And today's utterly reckless correlation is: global happiness is rising at the same time as global drug consumption. Thank you, there are press releases on the table by the door ...
-
All bets are off, of course, under a National government. Whatever people think of Anderton, he does listen to evidence. I don't think the same can be said of National's would-be drug tsar, Jacqui Dean.
Anderton and Dean offer the same failled prohibitionist "solution" to drug taking.
Perhaps National will farm drug czaring out to their minor coalition leader Rodney Hide, who unlike Jim Anderton has shown ability to actually hear what people tell him.
-
I’d have thought the happiest country in the world would be the one that doesn’t bother itself with these infuriating surveys…
-
Denmark is the happiest, and also has the lowest Gini Coefficient (measure of inequality) in the world. Perhaps they're related?
-
Mind you, living in a country where this was the norm would make me happy too. Only a matter of time before Aotearoa catches on, surely.
Post your response…
This topic is closed.