Should the census be compulsory?

There’s been a lot of discussion about the ABS decision to retain names and addresses in the Census until 2020 rather than deleting them more rapidly as in the past. Although the details differ, there’s been a dispute of this kind before every Census I can recall. Rather than debate the details, I’d like to think about the question: should the Census be compulsory, and if not what kind of requirement should there be?

I can think of three main reasons why we (Australians/the government/the ABS) might plausibly want to make Census participation compulsory. I’m not fully convinced by any of them

(i) Because we can. Historically, government agencies of all kinds have relied on compulsion simply because they are established by law. For example, public servants used to be banned from going on strike and teachers were shipped from one school to another at a whim. The military still works this way: the need for immediate obedience to orders in battle is taken to justify the use of this power in all contexts. Almost certainly, this contributes to high quite rates

(ii) The idea that a census is better than a sample, and that this superiority depends on counting everybody. The suspicions of the apocryphal critic who decried conclusions about the whole population derived from a sample of few thousand “and worse than that, a random sample” live on in most of us to some extent. But in fact the census can never be a complete count – around 2 per cent of the population is missed. Given that those missed are likely to be fairly atypical, the effect on aggregate estimates is probably comparable in magnitude to the standard error for a sample of, say, 10000.

(iii) The legal role of the Census in determining various kinds of allocations, most obviously electoral boundaries. Even if we were confident in a statistical sense, no one would be happy with the use of a sample of votes rather than a complete count in an election, and it makes intuitive sense to apply the same standard to a count used to determine boundaries.

Against this, the fact of compulsion creates costs and raises the risk that people will supply incorrect data. More fundamentally, the less compulsion we have the better.

Given all of that, my preferred policy would be to require everyone to complete the census form, but to allow an opt-out request, asking for the information supplied to be deleted except for the basic facts required to establish an accurate population count. My guess is that the opt-out rate would be very low, and that the 5-yearly fuss would go away.

As an aside, this is, pretty much, the situation we have with “compulsory” voting. You have to turn up and have your name crossed off, but there’s no compulsion to write anything on the ballot paper. So, simple laziness doesn’t give you an opt-out, but any genuine objection can be accommodated. This is, among other things, why I favor optional preferential voting.

71 thoughts on “Should the census be compulsory?

  1. @Ivor

    Anyone who could fold the Senate voting paper into a paper plane and get it to fly across the room would be acclaimed as an aeronautical engineering genius. It would be well worth any fine a court might impose.

  2. @Bruce Bradbury

    The US has an issue with the representativeness of its census, because the collectors of the completed forms are reluctant to go into the dangerous areas to pick up them up.

  3. @Jim Rose

    Getting by and thriving are different, the latter is clearly preferable. It seems pretty obvious that a census could assist planning and result in better services at a lower cost. If it isn’t doing or can’t at a reasonable cost I’m happy to drop it. However, if you believe your government is out to get you, eliminating or hamstringing government seems to me a particularly moronic response. You need to improve your government, or possibly, get a psychologist.

  4. Ivor :
    Crazy ….
    The AEC states:

    According to section 268 of the Commonwealth Electoral Act (1918), a vote is informal if:

    and there are seven conditions.
    If the AEC says “a vote is informal” then this means there are “informal votes”.
    The Section is titled….

    “Informal voting”

    and yet we have someone arguing here there is no informal votes !!!!!?????

    The AEC does not have the authority to change a law made by Parliament just by putting something on their website.

    Whatever title the AEC uses on its website, the title of section 268 in the Act is not ‘Informal votes’ or ‘Informal voting’ but ‘Informal ballot papers’.

    Section 268 is too long to quote all of it here, but it’s easily found online at comlaw.gov.au, and here’s part of it (my highlighting added):

    (1) A ballot paper shall … be informal if:

    (b) subject to sections 268A and 269, in a Senate election, it has no vote indicated on it, …
    (c) in a House of Representatives election, it has no vote indicated on it …

    There’s nothing in that section, or anywhere else in the Act, about ‘informal votes’, and the idea of something described as being ‘no vote’ also being described as an ‘informal vote’ is obviously a contradiction in terms.

    The question ‘Why does the AEC, on its website, use the expression “Informal votes” when under the law there is no such thing?’ is an excellent one, best directed to the AEC.

  5. @J-D

    ‘Why does the AEC, on its website, use the expression “Informal votes” when under the law there is no such thing?’

    Perhaps because the public thinks there is such a thing as an informal vote – understandably, because politicians and the media use the term all the time – and one of the roles of the AEC website is to communicate to the public how to make a vote that will be counted as one of the votes that affects the outcome of the election they are voting in.

  6. ‘Why does the AEC, on its website, use the expression “Informal votes” when under the law there is no such thing?’

    Maybe if you directed your question to the AEC they’d be able to tell you?

  7. @J-D

    You are just trolling.

    The only way people vote is by ballot papers. An informal ballot paper is automatically an informal vote. A formal ballot paper is automatically a formal vote.

    The number of ballot papers (formal and informal) equals the number of votes (formal and informal).

    Any extra ballot papers that are not votes are either discarded or spoilt. It would be quite reasonable for someone to hand their ballot paper back saying that they have no knowledge of any of the candidates. The AEC would not force them to mark it or threaten prosecution.

    You can meet all your responsibilities by getting your name marked off and dealing with your papers in one of several ways – some will create a vote, some will not. Either way, you will never be in a situation of appearing not to have voted – if you have your name marked off (even by an impersonator).

    Your posts have been opportunistic and inconsistent. You want to use vague uncorroborated AEC advice on one point [above], but then you disparage AEC official website advice at another.

    Considering ballot papers is a diversion because no one has questioned “compulsory ballot papering”.

    If assertions the Australia has “compulsory voting” is the issue, then “compulsory ballot papering” is only relevant if ballot papers automatically equal votes.

    I think it is best to rely on AEC documents – not trolling inconsistent confusion.

    In Australia it is compulsory to register on the Roll and get your name marked off at a polling booth.

    It may be compulsory to receive papers and deal with them reasonably (maybe creating a vote, maybe not).

    But that is all.

    There is only a very restricted category of people who are liable for a penalty for not voting and this cannot apply to anyone who has had their name crossed off and dealt with their papers in reasonable manner.

  8. The simple “sample” versus “complete count” discussion misses the fact that in almost all practical cases, you have to subset the data to get at the population you are actually interested in, so a sample of size 10000 quickly dwindles to a sample of size 100 if I am interested in studying middle-aged white males living in the country. To give a not-quite-so-silly example, the transport agencies all use the journey-to-work information/questions in the census to get a pretty detailed idea about commuting flows. This only works because of the (near) complete count: a statistical sample of origin->destination flows would be useless because your trip matrix would be way too sparse.

    In general, there are many benefits of the (near) complete count approach over the statistical sample approach. I’m not sure if they outweigh the costs, but they do exist, and will be lost with any switch to a sampling approach.

  9. There is so much misinformation in this debate it is appalling.
    Names and addresses have always been retained in the 18 months after the census while they process it, as it may be needed for checking purposes. Then they destroyed them after 18 months. The retention of name and address for 4 years is not much of a change. The significant change since the 2016 Census has been the linking of name and address from the Census to name and address on certain administrative records during that window when name and address was kept. It is the linkage of census name and address data to death records that has enabled us for the first time to accurately estimate Indigenous life expectancy. http://www.theage.com.au/federal-politics/political-news/census-without-your-names-we-wont-know-how-long-you-live-say-the-experts-20160731-gqhj3c.html
    Retaining name and address for 4 years rather than 18 months will enable this estimate to be even more accurate.

    On the longitudinal linkage of a 5% sample of people in the Census, this linkage is done through a statistical linkage key. It is not done through keeping the name and address of the people in the 5% sample for ever and a day.

    The ABS are extremely careful in their handling of identifying information from the Census. There has been no example of privacy being breached with Census data.

    So you should have no hesitation about including your name and address in the Census.

    On the question of compulsion, experience has shown that just saying it is compulsory, even if you don’t fine people for not completing, does lead to much higher response rates, and therefore the Census is more representative of the total population, particularly at the small area level.

  10. I will happily do the census. Like the vast majority of the population, I’m far too insignificant for my information to be of interest to anyone, other than when aggregated with everyone elses.

  11. Of interest, most people are quite happy with electronic banking, submitting tax electronically to the ATO and host of other computer based records held be government but apparently concerned with ABS data collection. On a side note the ABS collects a lot of business pricing information for the PPI which seems of greater commercial sensitivity.

    In terms of compulsory I think John overstates the response rate if it wasn’t. I really can’t envisage such a high turnout on a lengthy survey if you simply don’t have to do it. I think the problem of not making it compulsory is that there are likely to by highly correlated demographic characteristics in the group that does not respond to produce biased representations.

    If you don’t want to be counted then great, just abstain from completing it. But then maybe voting shouldn’t be compulsory either if your relevance to social planning and research isn’t required.

  12. I formulated the question ‘Why does the AEC, on its website, use the expression “Informal votes” when under the law there is no such thing?’??

    Beethoven wrote in response to my comment:

    Perhaps because the public thinks there is such a thing as an informal vote – understandably, because politicians and the media use the term all the time – and one of the roles of the AEC website is to communicate to the public …

    That’s a very plausible guess. It makes sense that the AEC would adopt a loose informal way of expressing itself in order to discharge its responsibility to communicate with the public effectively, even though that loose informal expression is not legally correct; but if that is what the AEC is doing, it does not convert a form of expression which is not legally correct into one which is legally correct.

    zoot wrote in response to my comment:

    Maybe if you directed your question to the AEC they’d be able to tell you?

    That was my suggestion too, when I wrote that the question was ‘best directed to the AEC’.

    Ivor wrote in response to my comment:

    The only way people vote is by ballot papers.

    That’s the truth, but not the whole truth. People vote by marking their votes on ballot papers. Section 239 of the Act explains in detail how ballot papers must be marked to vote in Senate elections, and section 240 explains in detail how ballot papers must be marked to vote in House of Representatives elections. Those are legal definitions of voting (for Commonwealth elections); there are no others. People who don’t mark their ballot papers haven’t voted.

    Ivor wrote further:

    It would be quite reasonable for someone to hand their ballot paper back saying that they have no knowledge of any of the candidates. The AEC would not force them to mark it or threaten prosecution.

    I recall a case in which Australian academics Eugene Kamenka and Alice Erh-Soon Tay were asked by the AEC (in accordance with the procedures laid down in the Act) to explain their failure to vote in an election. Their explanation was that they had been out of the country during the campaign and had not had the opportunity to follow it, so that they were not sufficiently well informed to vote. This was accepted as a ‘valid and sufficient reason’ for the purposes of section 245 of the Act.

    There are two things to note about this. The first is that even if not knowing how to vote is a valid and sufficient reason for not voting, it doesn’t convert not voting into voting. Eugene Kamenka and Alice Erh-Soon Tay did not argue that they had voted; they acknowledged that they had not voted, but argued that they had a valid and sufficient reason for not voting. People who haven’t marked their ballot papers haven’t voted.

    The second point is that Divisional Returning Officers are authorised under the Act to decide whether people’s reasons for not voting are ‘valid and sufficient’, but this authorisation is not extended to all the staff at every polling place on election day. If you try to avoid marking your ballot papers at a polling place on election day, the staff there are not authorised to accept your reason for doing so (unless the Divisional Returning Officer happens to be present at that particular moment). According to what I was told by the AEC when I asked this question, what the staff at the polling place are supposed to do if you try to leave without having marked your ballot papers and put them in the ballot box is to explain to you how you are required to vote. Ivor tells us they wouldn’t, but the AEC has told me that they are supposed to. On this question, I believe the AEC more than I believe Ivor.

    Ivor wrote further:

    You can meet all your responsibilities by getting your name marked off and dealing with your papers in one of several ways – some will create a vote, some will not.

    People who deal with their ballot papers without voting have not discharged their duty under section 245 of the Act to vote.

    Ivor wrote further:

    Your posts have been opportunistic and inconsistent. You want to use vague uncorroborated AEC advice on one point [above], but then you disparage AEC official website advice at another.

    I believe the specific answer the AEC gave me to a specific question more than I believe Ivor; but I believe the specific text of an Act of Parliament more than I believe a general reference on the AEC website.

    Ivor wrote further:

    It may be compulsory to receive papers and deal with them reasonably (maybe creating a vote, maybe not).

    There’s nothing in the Act, and nothing on the AEC website, about reasonable dealing with ballot papers. I don’t know where Ivor’s getting that from. The Act says specifically what voters are required to do with their ballot papers: mark votes on them and put them in ballot boxes. Why Ivor does not want to accept what the Act says I don’t know.

  13. From what I am reading in the responses it appears that the Australian census form is quite different from Canada’s (see URLs at bottom of message).

    Canada has a two level census , a long-form and a short-form. Both were compulsory, up to 2006.

    We went from a compulsory long-form census to a voluntary long-form census when the late and unlamented Harper government was elected. despite cries of rage and anguish from everyone from major retailers and food service companies, and financial organizations to municipalities, school boards and academic researchers. Oh, the head of Statistics Canada resigned. The short-form remained compulsory.

    As Ronald Brak put it I thought we had done away with them so decision making wouldn’t be contaminated with information.

    Certainly the Harper Conservatives did not want facts to interfere with their agenda.
    I should mention that Harper was a good friend of Tony Abbot.

    We have returned to a compulsory long-form for the 2016 Census (new government) this year. I believe the long-form was sent to 25% of the population. Name and address were required on long-form, just a request for an e-mail address on the short form.

    I have no idea how long names are kept by Statistics Canada. No one, except, possibly, a few paranoid libertarians seemed worried about the name issue.

    Making the long-form voluntary led to a very serious drop in response rate, from 93.5% to 77%, severely hampering decision making and making fact-based decision-making close to impossible in some cases. Well, actually impossible in some cases. … the data agency did not publish community-level data from the 2011 long-form survey for approximately 1,100 communities because of unacceptably low response rates. This is in comparison to fewer than 160 communities whose data were not released as a result of data quality issues in 2006. http://www.cbc.ca/news/politics/mandatory-census-mail-out-1.3557511

    I’d suggest keeping the census compulsory.

    Long form https://www12.statcan.gc.ca/nhs-enm/2016/ref/questionnaires/questions-eng.cfm

    Short form https://www12.statcan.gc.ca/census-recensement/2016/ref/questionnaires/questions-eng.cfm

  14. I think, once the dust settles, it will be the manner of the announcements surrounding the retention of certain data that is the main issue. The issue of cyber-privacy is still brewing, but it is a vexed one, for capitalist forces have tempted most of us to enter into (necessary) agreements in order to purchase a product or a service. There is a perfectly legal cyber-arms-race between those who want wholesale access to micro-data and linkage chains spanning entire cyber-lives and actual lifetimes, and those who resent it and work hard to block, restrict, curtail and corrupt the ongoing hoovering up of all our online footprints. Then there is the subterranean world of illicit micro-data collection and remote surveillance. That’s a whole other story.

    The reality is that we are all prisoners of our own device (fn1), i.e. our desire to use services and the potential of the online world is what shines a light upon our otherwise private real lives. Short of being a hermit or a falsely created virtual person with real birth certificate on file, etc, we can’t really shut ourselves out of this world, we leave footprints with every mouse-click and button press.

    fn1: Is that “Hotel California?” Rings a bell.

  15. So no one has commented on #censusfail yet? Reading the tweets has been very funny. But it’s a complete fiasco.

    JQ has somehow predicted the future with ‘Should the census be compulsory?’ plus ‘For his own self-respect, Turnbull should quit’. Australia has been running national censuses for over a hundred years and as far as I know, this is the first one that’s been stuffed up. But no-one will quit except some medium level people at the ABS, I guess. Total incompetence – but it’s like we don’t expect anymore from government now.

  16. I just wish they’d stop lying when they say no one has access or can gain access to the data.

    They can’t even get their system working to collect the data in the first place. Donald Oats’ comment points a finger in the direction of what it is all actually about: The lesson they try to teach is that we are there to comply, not think, while pointedly refusing a good reason as to why this might be.

  17. There are a couple of issues here.

    1. There was a lot of anti-government hysteria and propaganda pushed before the census. If people are against democratic representative government, it raises the question of what type of government they do want. Do they want corporate fascist government (for example)? Of course, our democratic representative government is imperfect as is our economic system. I have made a number of criticisms of our political economy system on this blog site. However, weakening the level of democracy we already have and starving government of funds and data will not aid social democratic objectives. It will aid corporate capitalist objectives though.

    The hysteria about the sensitivity and security of census data is just that – hysteria. Most people give far more data about themselves and their family to farcebook, tattletitter and indeed inadvertently to their internet service provider who recorded every site they have visited (even before government required it). All these sites are also hackable. Every business site people use from the their bank to their shoe store is also hackable. Every piece of information they leave on their own PC or put through their keyboard as keystrokes is hackable. But suddenly the one thing they worry about is census. Really? How much sensitive data is on the census? Sweet FA as far as I can tell.

    2. Are we approaching the limits of usefulness of the internet? I raise this question now we know about the DNS (denial of service) attacks on the census site. While DNS attacks on their own do cannot hijack data, it appears we are approaching the limits of the usefulness of the internet in some regards. It appears that sensitive state processes (voting and census come to mind) really cannot be trusted to the internet. Answer, keep the processes manual. Heck, it might even allow us to keep more people employed. Worried about the pulp and paper costs? Use hemp farms for fibre production. I mean use hemp varieties with low THC production.

    Why make the assumption that everything has to be computerised? Computerisation like all processes has benefits and costs. There are also benefits as well as costs to keeping systems manual. Of course, manual data can be lost too. But remember. “To err is human but to really stuff up it takes computer.” Computers automate mistakes as well as automating correct processes and its a lot easier to hijack gigs of data online than to hijack a truckload of forms. Hijacking physical data requires physical proximity and the solving of real logistical problems.

    The theory of data security must include the theory of “hybrid” security setups which include using both physically secure processes and electronically secure processes in concert. For some applications, to entirely switch to electronic data at all steps and thus to electronic security only is not always wise.

  18. I have a long comment in moderation. Here’s a short version of the main points.

    There is no data in the census that is sensitive. The whole fuss about this is a storm in a teacup. The census asks your name, your address, your religion, your job, your earnings and whether you worked last week. Stuff like that. Whether you realise it or not all this information about you is pretty much all in the public domain all the time anyway. The other point is that many commercial websites hold much more data about you. How secure is that data across the board? Honestly, it is not secure at all. The corporates already possess and have already shared or leaked all your data anyway.

    On the one hand there is this paranoia about agencies of democratic government having some very basic data. On the other hand, people blithely give away much more and much more sensitive data to businesses and corporations all the time. I mean to those undemocratic entities out to exploit you, manipulate you and cross-trade all your data. People really are worried about the wrong thing. The propaganda against the legitimate and necessary operations of social democratic government is another deliberate diversion from the real dangers of corporate control stalking our society.

  19. jrkrideau :
    From what I am reading in the responses it appears that the Australian census form is quite different from Canada’s (see URLs at bottom of message).

    They seem similar to me. What significant differences are you supposing?

  20. @Val

    Australia has been running national censuses for over a hundred years and as far as I know, this is the first one that’s been stuffed up.

    It was also the first one to be done online. That might have had something to do with it. When has any big IT project, in any walk of life, not “been stuffed up”?

  21. With my rather discursive (rambling?) long post still in moderation, I will add this.

    Why do we make the assumption that everything has to be computerised? Computerisation like all processes has benefits and costs. There are also benefits as well as costs in keeping some systems manual. Of course, manual data can be lost too. But remember, “To err is human but to really stuff things up takes a computer.” Computers rapidly automate mistakes as well as automating correct processes. Also, it is a lot easier to pinch gigs of online data than to pinch a truckload or warehouse of forms. Pinching physical assets including physically stored data requires physical proximity and the solving of real logistical problems.

    An overall theory of data security must include the theory of “hybrid” security setups which employ physically secure processes and electronically secure processes in concert. For some applications it is not always wise to switch entirely to electronic data at all steps and thus restrict the system to the limits of electronic security only. The census is a good example. It would be possible to maintain paper form collection, then use data entry into a mainframe not connected to the internet. Research of the census data could then proceed via government user requests into some sort of statistical analysis system which would run batch jobs on the unconnected mainframe. There is no genuine need for real-time queries of or internet connection to census data, even for government analysts.

    The extra costs of such a system would not be important. Indeed, extra employment would be generated by the manual process components. If extra automation creates data vulnerability and more unemployment, then the savings and benefits overall are largely or even completely illusory. Costs and risks will be shifted to other parts of the economy and onto ordinary citizens. The mania for total automation and cost-cutting is the real problem not the compulsory nature of the census and the collecting of names.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s