cut the cost of producing population statistics. - WTF ?
It's once every 10 years FFS, How on earth can it cost THAT much, that we need this to save it ? Are the DUP delivering the forms by unicorn ?
Mobile phone data could be used to gather information on people's workplaces for future censuses as the government tries to cut the cost of producing population statistics. The census is crucial for understanding the UK's population and has been run around every ten years since 1801. Residents are legally required to respond …
Agree.
Also don't under estimate the value of having literally hundreds of years to understand the gaps and omissions in your data. Something you have to do all over again if you change source or collect method. Now as an data enhancement proposition some of these datasets are interesting but not as core data replacement.
The census costs about half a billion quid, so it's not cheap. It's certainly good value and isn't going away any time soon. Where the ONS have struggled is efficiently producing more fine-grained statistics between those ten year snapshots. We also know that, while useful, the Census has large known gaps and probably plenty of unknown gaps.
Most importantly trying to collect more information from the census is expensive and difficult - you can't just "add a question" to find something out. It's got to to be developed, tested, trialled and controlled and there's only so much you can get from a self-reported answer to a question once every ten years.
Gathering new, proxy data to derive population statistics cheaply, checking their quality against the census (and other data sets) is only a good thing.
>Gathering new, proxy data to derive population statistics cheaply, checking their quality against the census (and other data sets) is only a good thing.
But, in the end, those proxies will only have value if they can be verified - by the census.
Proxies are useful, but only when you've got fixed, reliable points to relate to. If our only information is from assumptions, based on 3rd-party data (collected for some other, probably-commercial reason), extrapolated across the nation - then we'll know...fuck all.
Are the DUP delivering the forms by unicorn ?
Possibly, but they will only be doing that in NI, where the Census is run by NISRA (Northern Ireland Statistics and Research Agency). The Census in Scotland is run by National Records Scotland. ONS runs only the Census for England and Wales.
Those inflated numbers from the mobile data vs the consensus could be in part from the amount of illegal immigrants in those areas. They have mobile phones but are not on any register.
Often saw multiple gang masters picking up obvious groups of illegals in beat up vans to take them to work and back again from where I was located in the Croydon Home office area for a while. Maybe mobile phone data will finally show how bad that problem is.
Aways some social libtard social justice warrior who jumps to the wrong conclusion.
Mixed race family here with some of my family being immigrants myself. You are barking up the wrong tree.
Unless you have lived in the area and see things with your own eyes, you just wouldn't understand. But yeah, anyone who mentions illegal immigrants is a bigoted dailymail reader to some people. /ShakeHead
How did you guess they are illegal?
Based on my own experience of Croydon Home Office, the people behind the counters would be difficult to be perceived as legal once they exit Luna House. They barely speak English, are near illiterate and type with two fingers at ~ 5-10 words a minute. The one dealing with the SWMBO permanent residence application 12 years ago was definitely in that category. We were very close to ask him to get his manager to assign a translation expert from whatever he was talking into English.
So if you observed them at that location, my guess would be Luna House (Home Office) employees.
Are they having a larf?
My commute is all of 30 seconds from the house to the office at the bottom of the garden. Oh, and my phone is connected via WiFi so nothing changes.
This is fine for the poor sods having to stand for more than an hour on Southern Railway services (if they bother to run them that is) but many people I know work within one cell tower distance of their home.
Can we have a new unit of measurment?
How many 'cell towers' did you use on your commute to work?
The ONS are quite good at privacy, and being statisticians they're also quite good at knowing what is actually anonymous and what isn't. One of the delays in getting census data out is caused by double checking that the anonymisation can't be easily reversed. They also only report at geographic granularity where they can be reasonably sure of not identifying one. The academic paper on the construction of 2001 output areas is quite a good read if you're into that sort of thing.
http://eprints.whiterose.ac.uk/5003/1/05-2.pdf
Cell phone companies on the other hand seem quite unaware. There was another paper about five years ago showing how to de-anonymise anonymous cell phone data with almost no effort at all.
I think this is the paper on how to identify people from seemingly anonymous cell phone data.
Unique in the Crowd: The privacy bounds of human mobility
Additionally, I found the following relevant papers:
but the implications is that they were not even aware of being under this creepy surveillance.
De-anonymization of a data is a criminal offence in the most recent data protection legislation. It will be entertaining to observe the fallout if someone files a challenge.
>De-anonymization of a data is a criminal offence in the most recent data protection legislation.
The courts will have to decide just what an "anonymous dataset" is and also what (the process of) "De-anonymization" is, as for example it would seem that raw mobile phone records are not anonymized data, given what can be directly inferred from the data.
No the UK has the opposite system.
It forces them to live somewhere awful (the channel islands) and fly into London on monday/friday. that way they are only present in the UK for 3days/week and so 156 days/year so no tax. Interestingly this even works if you are a cabinet minister
The ONS have no chance.
It was widely reported that the 2011 census returns for some areas of London were as low as 60%. (A 97% return being the target such that extrapolation to 100% was made with a degree of certainty thereafter.)
So why do the ONS use that as the benchmark for 2011? Especially for the areas quoted in the article.
What's wrong with using much better sources, eg The Land Registry, the local electoral rolls, local and government benefits, school registers, ( hell, even the kids must give their fingerprints before they can buy an overpriced "meal"), DVLC records, the NHS DNA database, and no doubt many more.
Ok, all the above may miss the homeless guys (always men aint' it?) on the street but they are unlikely to have a mobile phone or commute much anyways.
The ONS has since been beavering away to figure out how to use administrative data to produce population and household statistics that can be used as an alternative to the census.
This data might all be very interesting and helpful, but there is no way it can be miraculously converted into anything that looks like the data derived from the conventional census. It might be a very useful adjunct to the paper census; I fail to see how it could ever be an alternative to it, unless the ONS suddenly puts its hand up and says "we have been collecting all the wrong data since, er, for ever".
If mobile phone movements correlate with the census findings, it could be a useful fill-in between census dates. That assumes nothing else changes, but you could sample the population each year to keep the indicators calibrated.
Vastly oversimplified example - Say 10K people say they commute from Reading to London daily, and that matches ticket sales, but you find 11k phones make that trip. For Croydon it might be 20k people/22k phones. Yay, we have a correlation - 1.1 phones per head. If the total number of phones goes up from 33k to 50k next year, you can just re-survey Reading and extrapolate.
If they enacted this nationwide employment statistics would sit at about 400% of the nation employed.
If this went live then the census would conclude that my wife (who stopped work to look after out two youngsters) works at:
Tesco,
The Library
Playmania (local soft play centre) and,
Our local park
All places she goes to on average twice a week or more, normally travels to during commuting hours (when she's taking mk1 to school) and once there spends a significant amount of time, during "normal working hours".
Their logic needs to go further than factoring in night shifts.
With gig economy, zero hours contarcts, things get complex.
e.g. real life case (no names obv), someone I know has 3 jobs, all zero hours, varying hours per week, additionally, one of those is a cleaning job and the locations to be cleaned can vary (no guarantee it's always same office block, hotel, etc.)
That will be quite a challenge. As @ rmason said, how to distinguish that from his wife's weekly non work travels
Have fun with all the people who turn off their mobiles a lot of the time, so not "on" at home, and only turn them on when out and about if required (the classic emergency use only mobile phone scenario) - not everyone is (yet!) welded to their mobile phone (plus those additional "bit on the side" phones that will be unlikely to be used anywhere near someones home)
My phone has two SIMs on two different networks, one for work, one is mine - does this mean I am two people when the census-by-mobile counts people? If I've also got my LTE tablet with me am I three people?
I really don't think this is going to be accurate enough to use for a census. You'd need to follow it up with more accurate measures to correct for errors, so you might as well do it properly in the first place. The census is supposed to be a count of people, not a rough guess.