Census 2021

[rutabowa]

rutabowa This thread pretty much covers the same topic if you want to read up on it.

Why is it claimed that a sample is often more accurate than a census?

When learning the course of sampling, I meet the following two statements: 1) Sampling error leads to mostly variability, nonsampling errors lead to bias. 2) Because of nonsampling error, a sampl...

stats.stackexchange.com

I mean 3 of the 5 the answers in that link say that a census will most likely be more accurate than a survey (except in very specific cirumstances). The top voted answer says "So apart from quite unusual circumstances I would say a census is going to be more accurate than a sample." The other two answers point to possible lack of care in some censuses due to the assumption that everyone is covered so no effort needs to be put in to engaging. But as I said, in this specific census we are talking about there are community engagement programmes to try to mitigate this.

And NONE of the answers go so far as you have done and said "a statistical sample is always better than a census and should always replace them". they make points about efficiency, which of course I agree with and that is why the census only happen every 10 years and most ONS studies are done using sampling (which as i said they need the census data for to do accurately)

 

[kenny g]

Fair enough. I don't have a great deal of faith in the effectiveness of community engagement programmes and reckon the census misses out significant swathes of the population. I am not aware of ONS doing "deep surveys" In particular areas to see if normal processes are missing out substantial numbers or sectors but I may be wrong. I am thinking of the people who sleep in shop basements, beds in sheds, or the bunk beds in annexes on family houses etc etc. Also warehouses with bunk beds where staff live illegally. Some London Boroughs have loads of these kinds of accommodation.

 

[rutabowa]

For sure there will be people missed... I do think, though, that those people who you mentioned would also be missed by a statistical sample (how would the statisticians know that they exist when planning the sample?), and at least with the census there is slightly more of a chance at representing them as there are people going out to every address.

 

[SpookyFrank]

I get that, but the circumstance under which the ONS has decided to proceed with this census are new.
Elderly and vulnerable shielding folk who might have had support from family members in the past are on their own atm and the previous MO of fieldworkers going into houses to act as an amanuensis is inoperable.

And as a result the data produced will be a pile of steaming shit. Comparing like for like with prior censuses, which is half the point of the whole excercise, will be impossible.

 

[hash tag]

From this morning's walk

 

[Orang Utan]

was gonna tick no religion but might now vote Jewish so I get more holidays. I’m doing this right, aren’t I?

 

[hash tag]

Imagine

Nothing to kill or die for
And no religion, too
Imagine all the people
Living life in peace... You...

 

[rutabowa]

And as a result the data produced will be a pile of steaming shit. Comparing like for like with prior censuses, which is half the point of the whole excercise, will be impossible.

Were the census people allowed to go inside people's houses to help fill in forms in 2011 then? It is definitely not allowed this year, but the reason given isn't covid, it is just personal safety.

 

[planetgeli]

was gonna tick no religion but might now vote Jewish so I get more holidays. I’m doing this right, aren’t I?

Yes. Add 'French' somewhere for even more. You'll barely need to work again.

 

[Orang Utan]

Imagine

Nothing to kill or die for
And no religion, too
Imagine all the people
Living life in peace... You...

Written by a man who had a whole apartment hired below his to store all of his and Ono's fur coats

 

[SpookyFrank]

From this morning's walkView attachment 257540

They clearly don't give a shit about who is not religious or the number of places in catholic schools wouldn't outstrip the number of catholic children by a factor of about 3:1.

 

[muscovyduck]

I filled this in now. Got tripped up a bit on some of the education questions. For nationality I went with "how would I describe myself if I was in a foreign country/how would they describe me". Went with no religion, am culturally attached to one but I want to see its downfall

 

[planetgeli]

For nationality I went with "how would I describe myself if I was in a foreign country/how would they describe me".

Ah, but which foreign country? In the Philippines we had a couple talking to us who got really offended because they insisted we must be Dutch and we were trying to tell them we were from the UK.

"You know (they didn't) London, UK."

"OK. So that makes you what religion? Buddhist, right?"

True conversation. We are not the centre of the world we like to think we are.

 

[muscovyduck]

Ah, but which foreign country? In Indonesia we had a couple talking to us who got really offended because they insisted we must be Dutch and we were trying to tell them we were from the UK.

"You know (they didn't) London, UK."

"OK. So that makes you what religion? Buddhist, right?"

True conversation. We are not the centre of the world we like to think we are.

Yes I was gonna elaborate but on my mobile. Mainly thinking of where my family's from and how they describe me 'back home', and places where people don't care about the difference between say English and British or Northern Irish and Irish which is most the world tbh

 

[Kevbad the Bad]

Imagine

Nothing to kill or die for
And no religion, too
Imagine all the people
Living life in peace... You...

Imagine there's no census
It isn't hard to do
no questionnaires or surveys
and no electoral register too
Imagine 95% of the people
with nothing much to do
on the 21st March 2021.

 

[kenny g]

For sure there will be people missed... I do think, though, that those people who you mentioned would also be missed by a statistical sample (how would the statisticians know that they exist when planning the sample?), and at least with the census there is slightly more of a chance at representing them as there are people going out to every address.

The point is they are not going to every address in an effective manner. It is at most bods moonlighting from their council jobs knocking on doors asking for forms to be filled out which they probably will be but excluding the bunch of bunk beds in the garage in the garden. An effective sample would be choosing one area and using satellite/ aerial images along with infrared from helicopters, as well as visits to shop basements/ warehosues and actually drilling down to who is sleeping overnight in one ward in one London Borough. I reckon you would find fuck loads more people than you get on census returns.

 

[steveo87]

Unfortunately, Mrs o87 got to it before I could, so it was filled out legitimately.....

 

[kenny g]

Unfortunately, Mrs o87 got to it before I could, so it was filled out legitimately.....

My wife gets a separate one for her and the children which is possible and a good compromise.

 

[belboid]

The point is they are not going to every address in an effective manner. It is at most bods moonlighting from their council jobs knocking on doors asking for forms to be filled out which they probably will be but excluding the bunch of bunk beds in the garage in the garden. An effective sample would be choosing one area and using satellite/ aerial images along with infrared from helicopters, as well as visits to shop basements/ warehosues and actually drilling down to who is sleeping overnight in one ward in one London Borough. I reckon you would find fuck loads more people than you get on census returns.

Balls.

the census workers (who, as noted throughout the thread if you havent been paying any attention) are temp workers specially hiried (badly) through an agency. They will, as they have done every decade, go around every house that hasn't returned it and push them toward doing so, answering queries as they go. At that point they will (probably) remind returners that they should make mention of everyone in the house that night, including sofa surfers, people in overcrowded conditions etc etc - and they should probably mention that no non-anonymised data is shared with other government departments for benefits, HMRC etc etc (ie, they wont grass if the filler is working unlawfully or living with someone when they declare they are single.) A method such as the one you suggest would, at the very best, merely reveal who lived in one London borough, but it wouldn't really do that effectively as, without giving the officers massively intrusive powers, the amount people would fib would still be as high. And if they had those powers, then those exploiting such people would get wind and just bloody move them for a night or two.

As to travellers, the sites are pretty well known, they aren't in hiding. So they will be visited, pushed and helped to complete, but, if they dont (and it is well known that such marginalised groups are the least likely to complete, although that 4% also includes a broad range of lazy wankers and bizarre objectees) crude numbers will at least be taken. Not perfect, but better than any other method tried. It also shows up where further research efforts are required. something a narrower sample wouldn't (or wouldnt anything like as well). Very small communities are very badly served by such sampling, you might get an idea of how many such small communities there are, but not actually what or where they are.

 

[belboid]

My wife gets a separate one for her and the children which is possible and a good compromise.

you mean she just omits your details from the one form.

 

[rutabowa]

An effective sample would be choosing one area and using satellite/ aerial images along with infrared from helicopters, as well as visits to shop basements/ warehosues and actually drilling down to who is sleeping overnight in one ward in one London Borough.

That sounds really intrusive to me compared to the census, I don't think it would even be legal. Has anyone seriously suggested that as a possibility? Are you really sure you would prefer that level of surveillance?

 

[xenon]

And as a result the data produced will be a pile of steaming shit. Comparing like for like with prior censuses, which is half the point of the whole excercise, will be impossible.

I agree they should’ve postponed it until next year. but this idea that all the elderly or otherwise shielding people, are totally Internet illiterate, have no one who can help them is a bit over the top.

 

[kenny g]

Balls.

the census workers (who, as noted throughout the thread if you havent been paying any attention) are temp workers specially hiried (badly) through an agency. They will, as they have done every decade, go around every house that hasn't returned it and push them toward doing so, answering queries as they go. At that point they will (probably) remind returners that they should make mention of everyone in the house that night, including sofa surfers, people in overcrowded conditions etc etc - and they should probably mention that no non-anonymised data is shared with other government departments for benefits, HMRC etc etc (ie, they wont grass if the filler is working unlawfully or living with someone when they declare they are single.) A method such as the one you suggest would, at the very best, merely reveal who lived in one London borough, but it wouldn't really do that effectively as, without giving the officers massively intrusive powers, the amount people would fib would still be as high. And if they had those powers, then those exploiting such people would get wind and just bloody move them for a night or two.

As to travellers, the sites are pretty well known, they aren't in hiding. So they will be visited, pushed and helped to complete, but, if they dont (and it is well known that such marginalised groups are the least likely to complete, although that 4% also includes a broad range of lazy wankers and bizarre objectees) crude numbers will at least be taken. Not perfect, but better than any other method tried. It also shows up where further research efforts are required. something a narrower sample wouldn't (or wouldnt anything like as well). Very small communities are very badly served by such sampling, you might get an idea of how many such small communities there are, but not actually what or where they are.

Bollocks and balderdash!

Ignoring your "you haven't been paying attention" tones, you appear to have little or no understanding of statistical sampling. The evidence of one ward of one Borough could be used to extrapolate to other wards in other Boroughs.

Using IR to identify beds in sheds has already been used in local authorities for planning enforcement and EPC purposes. Checking shop basements/ warehouses or in fact people's garages for residents would not be an especially intrusive power and is already available to local authority enforcement teams.

 

[xenon]

I haven’t had my letter with the code on yet. Unless I have already accidentally shredded it.

 

[rutabowa]

It is at most bods moonlighting from their council jobs knocking on doors

This is just 100% completely wrong.

 

[rutabowa]

you appear to have little or no understanding of statistical sampling.

The only you link you provided with information about statistical sampling says that census data is more accurate than statistical sampling (in all but a few unusual situations).

 

[kabbes]

We just did ours and it was “do you have a medical condition?” that caused the most uncertainty in this house of middle-aged people.

 

[kenny g]

This is just 100% completely wrong.

OK, give me evidence why it is 100% wrong. Fool.

 

[kenny g]

We just did ours and it was “do you have a medical condition?” that caused the most uncertainty in this house of middle-aged people.

Nobby stiles?

 

[belboid]

Bollocks and balderdash!

Ignoring your "you haven't been paying attention" tones, you appear to have little or no understanding of statistical sampling. The evidence of one ward of one Borough could be used to extrapolate to other wards in other Boroughs.

Using IR to identify beds in sheds has already been used in local authorities for planning enforcement and EPC purposes. Checking shop basements/ warehouses or in fact people's garages for residents would not be an especially intrusive power and is already available to local authority enforcement teams.

giggles

Of course things can be extrapolated from one ward, but they dont do see very well. See every election opinion poll ever carried out