Figure I heard last night was 10 million. Last Census had a 98 percent completion rate.
I read they need at least 7 million responses for it to be statistically significant. Last I read they were only around 5 million.
Sent from my SM-G900I using AULRO mobile app
Figure I heard last night was 10 million. Last Census had a 98 percent completion rate.
I'm assuming they will need close to 100% count for electoral division work, for example NT population being border line for two senate seats(?) in the past (guaranteed now?). Also Census resets or recalibrates data trends determined by intercensal admininstrative data ie population growth from birth and death records, migration cards etc.
After that, if they are running over budget because of the system failure and catch up, it may become a question of how many of the forms they will process to provide the social data.
I believe it was the 1986(?) census when the ABS had a budget problem, they got all the forms (solves electoral division concerns), but ran out of money for the processing for social data. NT and ACT(?) forms were 100% processed, while only 50% of state forms were processed. Weightings would be applied.
In response to 1986(?) ABS got a budget increase so it wouldn't happen again. I expect parliament would come up with the funds if this were to be the case again...but you can never tell.
700 jobs have been axed from the ABS so that should delay processing, I imagine.
Sent from my SM-G900I using AULRO mobile app
What constitutes a response though? Is it a premises, or a person appearing on a form? If it's 5 million people have appeared on a form, then they weren't even a quarter of the way there, but if it's 10 million premises, then it could be 99% done.
A response should be for a household - but as you suggest, it would be easy, and, in a way, correct, to count individuals. Or somewhere in between, households plus individual responses (separate individual form etc).
Unless clearly specified, you would not know what was being referred to.
John
John
JDNSW
1986 110 County 3.9 diesel
1970 2a 109 2.25 petrol
http://mobile.abc.net.au/news/2016-0...s-bias/7761848
Good explanation, I think.
Sent from my SM-G900I using AULRO mobile app
Those who bothered to record anything on the night might have some chance of filling in the form correctly, but in the absence of a form they wouldn't have known exactly what to record. There would also be a proportion of respondents who didn't record anything.
ABS have no way of knowing who did and who didn't so they have to assume that information given is from recall.
Lawyers often argue that recall is unreliable. I don't know what Scientists think, but I guess that it would be something similar.
I know that it's nit picking but how rough a rough idea do the ABS want?
Cheers, Billy.
Keeping it simple is complicated.
| Search AULRO.com ONLY! |
Search All the Web! |
|---|
|
|
|
Bookmarks