Tag Archives: census

Preliminary thoughts: making recommendations regarding sex status for the 2016 census

The Australian Bureau of Statistics (ABS) has been planning for our 2016 census for a while now. Recently they announced information sessions relating to changes that might appear in the 2016 census because they want submissions from organisations and individuals containing feedback and recommendations. Having a strong research interest in the representation of persons I wanted to write a submission of my own, specifically addressing some of the issues relating to the representation of sex. Even if it doesn’t lead to changes in the next census, any submission of mine will be accessible to them during any future conversations.

While I’m not too optimistic anything will change, as there are many interested parties relying on the field staying exactly the same for easy comparison with previous years’ results, there is a real possibility I can make some difference. It also feels a little overwhelming planning to write a formal report to government based on my PhD research, so I may be asking various people for advice before May next year, when submissions close.

The following is a brief outline of some of the concerns I want to include. I’d love any feedback you might have, and if you know of anyone with an interest in this topic (you probably do!) I’d appreciate you forwarding the link to them =)

Reviewing the sex standard

The ABS have begun to review its sex standard. My understanding is that this is a separate project to the 2016 census, but will inform its practices. They state

This standard provides a basis for the ABS to collect statistics relating to people but also provides a standard for other organisations to collect data about sex in surveys and administrative collections. (‘REVIEW OF THE SEX STANDARD / POTENTIAL NEW GENDER STANDARD‘, 14 September 2012)

I think this is an important move! Not only will it mean the category will be consistent across all ABS data (they do more than just the census), but its intention for use within other organisations, if widely adopted, will make data comparisons more accurate.

However, this makes it a lot more important for the ABS to get it right – a difficult project with so many stakeholders involved.

The current standard

Details can be read here. To summarise, the ABS acknowledge a trend in substituting ‘gender’ for ‘sex’, but discourage that practice because they believe they are different terms.

Sex refers to a person’s biological sex and associated physical characteristics. Gender refers to the way a person self identifies and presents them self to the community, such as their name, outward appearance, mannerisms, and dress. A person’s sex is not necessarily consistent with their gender. (‘REVIEW OF THE SEX STANDARD / POTENTIAL NEW GENDER STANDARD‘, 14 September 2012)

The ABS currently define sex the following ways:

5. Sex is defined as the biological distinction between male and female. Where there is an inconsistency between anatomical and chromosomal characteristics, sex is based on anatomical characteristics. (‘Underlying concepts‘, 28 March 2006)


8. Physical biology is the criterion used to classify persons into the categories ‘male’ and ‘female’. This criterion is physical rather than genetic to cover the exceptional cases such as sex change operations. In the vast majority of cases the physical and genetic categories are identical. (‘Classification and coding‘, 28 March 2006)

So, in short, sex is always defined as a biological attribute, except in those cases where it is defined as a physical one.

While I partly wrote the previous sentence as a joke, I do see the justification for this definition. The problem is, however, that this complex set of ‘exceptional cases such as sex change operations’ is not clearly defined. At what stage of sexual reassignment would a person qualify as one sex and not the other? Which of the many conditions that might blur someone’s sex status count them as female or male based on their physical attributes as opposed to biological ones? What conditions of self-identification, if any, qualify a person to legitimately categorise themselves as the sex they identify with under this system?

This presents a problem with this definition of sex being used as a standard. But not only is the demarcation unclear; any standard demarcation presents a problem for practical use as different studies of populations are usually after any one of a number of understandings of sex or gender. For example, the ABS state ‘gender is frequently used (for example in psychological or sociological discussions) to refer to cultural or social differences as opposed to biological ones’ (‘Underlying concepts‘, 28 March 2006). In such cases, it would be preferable for sex to be defined as relating to identity and lived experience rather than strict biological status. However, sociologically useful definitions such as this would be problematic for medical studies looking at, say, the prevalence of diseases or conditions compared to chromosomal sex.

I’m still thinking this all through, but my current understanding is that census and other ABS data tend to be used mostly for studies of populations and this leads me to believe a sociological definition would be the most appropriate option for them to adopt. This will certainly present a problem for any statistical data being used for studies that require a strict biological definition of sex, of course. Perhaps one solution would be to propose two separate definitions of sex, and for the ABS to adopt one of them for use across all their statistical projects as the default. (I return to this discussion below.)


Sex status is digitised using the following coding:

0 Not stated/Inadequately described
1 Male
2 Female

They add

If it is necessary to include the category ‘Intersex or Indeterminate’, it should be allocated the code ’3′. (‘Classification and coding‘, 28 March 2006).

However, clarifying how sex data is collected in the census, the ABS write

There is no non-response for the Sex (SEXP) data item because missing values are imputed. If a form was received but there was no response to the question, Sex was imputed using other information on the form, such as name, relationship or number of children. If this process was not successful then sex was allocated randomly. Clerical intervention during processing was also required where both Male and Female responses were marked. These various types of imputations occurred for 2.2% of all persons. (‘Sex (SEXP)‘, 24 October 2012)

This explains that census data does not use the zero code – it is strictly a binary field where deviation or non-completion of the field is responded to with an assumed or random assignment of code 1 or 2. (This is the first time I’ve been able to confirm Organisation Intersex International Australia’s claim about random assignment.) Other problems aside, this actually means that the census is not actually complying with its publicised standard for sex. Any non-response or non-standard response to this questions should, by their own rules, be counted as ’0 Not stated/Inadequately described’, not assigned to one of the other options as they see fit. What is the point of the standard if it is used inconsistently?

Imputed or random assignment?

One other issue I have with the practice described in the above quote is the use of the word ‘impute’ to describe sexual assignment. It is one thing for the record of a person who elsewhere describes their relationship to another person as a ‘father’ to be assigned ‘male’, but an entirely different situation for another person to be assigned ‘male’ because their name is, say, ‘John’. From the description I think it’s safe to assume both fall under ABS’s practice of ‘imputing’. My problem is that the word suggests more of a logical calculation (such as determining a person’s age based on their birth date) than a assumption based on social experiences.

This may seem like an insignificant distinction, but i feel it’s actually important to be clear about the potentially flawed assumptions at play here, and another word may be more appropriate.


Perhaps my biggest concern about sex status on the census, and all forms in general really, is the issue of clarity. It is rarely clear what is being asked so the respondent is left to make their own assumptions. I also feel the Australian census deals with this issue rather poorly. Various parts of the ABS documentation talk about the tendency for people to conflate the terms gender and sex and the confusion this leads to, and they conclude it is important to be clear about the standard in order to make it consistent across various surveys and collections. This is an admirable theoretical stance to take, but their execution of this does not appear consistent with their aim when the distinction is not even mentioned.

Screenshot of the 2011 census form asking 'Is the person male or female?'

The above image is all we are presented with. There is no occurrence of the word ‘sex’ anywhere on the form. (‘Female’ and ‘male’ are sex terms, not gendered terms, but I suspect the vast majority of people aren’t certain about this.) If you would like a clarification of the question (‘I’m female bodied but have presented as a male for the past fifteen years. What response are they looking for here?’) you would need to go to their website and hope you can make sense of the documentation – the same documentation that I’ve been looking closely through for the past few weeks and still feel a little confused about. The last time the question was helpfully labeled ‘Sex’ was back in 1991.

The ABS test the wording and presentation of new questions on real people before incorporating them into the census, so apparently nobody tested was confused enough by the question as presented to warrant a revision. However, I think this depends heavily on the people being tested.

If there was just one improvement I would like to see (partly because it’s so simple that it actually has a chance of being incorporated) would be for the census to replace the question ‘Is the person male or female?’ with something like ‘What is the biological sex of the person?’. This would of course open them up to complications where some citizens are supposed to declare their physical sex (see discussion above), but at least the question makes sense to the general population who are regularly exposed to inconsistent use of terms related to sex and gender. (Thanks a lot, Facebook et al.)

Including gender as an additional field

Part of the review of the sex standard looks at ‘whether there are grounds to develop a new gender standard to complement the revised sex standard’ (‘REVIEW OF THE SEX STANDARD / POTENTIAL NEW GENDER STANDARD‘, 14 September 2012). A strong, initial response of mine is to think, ‘Really? Sex status is confusing and complicated enough here as it is and you’re wondering whether it’s a good idea to also ask respondents for their gender?’ However, I think that view may be a bit too critical.

First of all, there could actually be great social value to collecting such information. I’d love to see a breakdown of what genders the population of Australia identify as – even while I have difficulty working out how I personally feel about my own gender identity and would take forever to formulate a response of my own. Such questions could assist in highlighting gender diversity and, in turn, help individuals feel more comfortable about an identity that may traditionally be seen as ‘abnormal’. This could be great for research in the social sciences, too!

Of course, it’s difficult to envision a system that would be practical. First of all, there may be an urge to limit responses to a predefined set which may only serve to further alienate anyone who does not neatly fit within the available options (see: othering) and could never hope to encompass the vast diversity of gender identity (see: Yay genderform!).

The only way I could see it working in both a practical and socially responsible way is if gender was an optional, blank text field (see: ‘Disalienation: Why Gender is a Text Field on Diaspora‘, Sarah Mei, 26 November 2010). (Perhaps a brief note on the form about common responses might help transition the nation to this additional field, but it would need to be done delicately so as not to evoke a sense of othering in those who don’t identify with any of the examples.) Yes, the results would be a little chaotic, and some respondents would certainly write in ‘Jedi’, but I think for many studies of gender this chaos would actually be preferable and valuable.

It’s also important to note that including a gender field might actually serve as a solution to the problem of standards addressed earlier, where sociological and medical research both tend to require different interpretations of ‘sex status’. Sex status could be defined as biological and, while there would certainly still be problems with this limited notion, gender status could serve as a place to declare important information relating to your sex that has, until now, never been collected by the census (‘male identified intersex’*, ‘trans man’, etc.). The ABS, and other organisations who choose to adopt their standards, could choose which field (or both) to use when constructing surveys, depending on what sort of data they are looking to collect.

While interesting, this whole idea would likely not be adopted for the 2016 census without support from organisations who would use such data. I’ll need to talk to other researchers to try and, if they see it as a resource of potential value, rally some support and suggestions for my proposal. Of course, it’s also possible for me to recommend this as the standard for the ABS to adopt even if they don’t choose to use it within their census. My understanding is that the sex and gender standard is technically a separate project of theirs that simply informs the census. The formalisation of a standard with more flexibility of representation would be preferable even if its not present in the 2016 census.

Third sex category

Another option ABS has put on the table is to include a new ‘intersex or indeterminate’ category option for sex. Because of the problems of categorical sex assignment (it makes data inaccurate while simultaneously delegitimising experiences of, for example, intersex people by removing them from the official record of the country’s population) this may be something worth considering. However, including a ‘third sex’ option can introduce its own problems, as outlined by Gina Wilson from OII Australia.

I haven’t yet thought much about this ‘third sex’ problem, but I’ll certainly need to. If introduced, there needs to be a comfortable balance struck between being sensitive to all individuals completing such forms and collecting data that is most socially useful. I just don’t know where that balance might be yet.

Questioning the ‘essential’ nature of sex declaration

And it appears I’ve left one of the biggest questions for the end: why ask for a person’s sex status to begin with? To nicely frame one side of this discussion, I’d like to quote Gina Wilson who wrote

Indeed in an equal society there is no reason for sex designators to be included in the vast preponderance of documentation. The real necessity right now is only for census like information to be gathered so marginalized peoples can be identified and resources can be allocated.

If we were not the subject of discrimination and unfair treatment, only our lovers (and then not always) and our medical practitioner would have a legitimate reason to enquire about our sex parts.

The government and society needs to get their noses out of our underpants. (‘Gina Wilson writes on a third sex at Gay Star News‘, Gina Wilson, 13 July 2012)

Asserting an opposing view, the ABS states of its inclusion of the sex field that

Information gained from questions on sex is essential for the estimation of the resident population in each of the states, territories and local areas, which are required by legislation for electoral purposes and the distribution of government funds. (‘Topics – Recommended for Retention, 5 November 2012)

It’s difficult to assess ABS’s view that sex status is ‘essential’ from the documentation I have been looking at. There are vague references to ‘population estimates’, but I can’t see how sex status could inform such things unless by ‘population estimates’ you mean ‘population projections’ and you’re looking at populations of smaller towns and rural areas, which could actually produce information of public value. (Or if by ‘population estimates’ you simply mean ‘estimates of the current number of males and females’, which by definition it is integral to but we are now back where we began, wondering how such data could possibly be of use.) It also mentions it needs population data for reporting to COAG – but not what this is used for.

The best justification I have found is the statements that ‘[sex] is a core cross-tabulation for practically all social statistical topics such as employment, education, and health’ (‘REVIEW OF THE SEX STANDARD / POTENTIAL NEW GENDER STANDARD‘, 14 September 2012), and that ‘[i]nformation on sex is essential for most socio-demographic analysis of Census data’ (‘Topics – Recommended for Retention, 5 November 2012). So, in short, sociological analyses and imposed government requirements in turn require sex information to be collected on the census, but there is no mention of why these external organisations want it. I don’t doubt there are reasons, and perhaps many good ones, but they are entirely hidden from the discussion behind the declaration that it is ‘essential’.

One issue I have with mandatory declaration here is that I suspect that government organisations (such as COAG, who the ABS actually state as an example) and sociological researchers all too easily fall back on treating biological sex as important field for ‘cross-tabulation’ because of tradition rather than questioning whether it is actually relevant in many of the situations it is introduced, and what the social implications of treating sex status with such importance might be. But without transparency regarding requests for its inclusion and retention, I just can’t participate in that conversation. (I can talk with people within the ABS about this, though. I’m only explaining how the data is not easy for us to access.)

In short, I largely side myself with the sentiment behind Gina Wilson’s quote above. However, I should be clear that I don’t entirely dismiss the idea that sex status can be a useful field within the census. (For example, I strongly support the ability for same sex couples to be counted in the census. This is a socially important demographic to publicise.) The important thing, I feel, is to question our deeply held assumptions about the importance of sex status and determine how, exactly, and if at all, it is actually benefiting us when we include it.

It should be a considered inclusion, not a habitual one.

Where to from here?

I have a lot of thesis writing to do in the near future which is not entirely related to this ABS submission, so it won’t be an easy task. I also feel a little intimidated by the idea of trying to submit a document to a government organisation that, if done correctly, could actually introduce a positive change. I’m definitely going to need some help on this, and to talk to others about it in order to receive occasional reminders that I’m not wasting my time. Now that I’ve written out some of my early ideas, doing a call out for people to talk to about this, to get advice, and to just generally talk it up would probably be the next step.

First, I might talk to some social sciences and gender studies researchers on campus to throw a few ideas at them. It’d be great to find people who will state that research would benefit from future census data, if some of my proposed recommendations are taken on board.

Actually, I’m attending the awgsa conference that’s being held on my campus this week, so this would be a good opportunity to ask some people I already know to recommend others to talk to during the event.

I will also get in touch with someone from OII Australia. My proposal would have a direct impact on intersex individuals so it would be kinda inappropriate for me to submit something without asking them for advice. Also, they have people who have already done some work looking into the census and other systems of population categorisations so their experience would be invaluable to this project – and my thesis research overall.

I’m also considering planning a semi-formal presentation/discussion at UNSW one evening if I find enough people are interested in this project. It’s not difficult to book a room; it’s just difficult to find a time people are available!

Lastly, but certainly not least, I need to get in touch with someone from the ABS who knows something about the history of conversations about sex status. I get the impression they’re very happy to have people meet with them to get clarification on things in order to prepare a submission, but I feel I should get some advice from others first so I’m more prepared. (Nevertheless, I can not leave this too long.)

Submissions are due by May 31 2013. By around that time I need to have completed drafts of four of my thesis chapters (out of six!). It’s going to be rough, but I also feel it’s important to get this done. Even if my recommendations are not incorporated into the 2016 census, I need to vocalise them so they will be available for reference in future discussions around possible revisions.

. . .

If you’ve made it this far through my post you’re no doubt thinking, ‘Yeah, this is really important and I wonder how I can help!’ Please, please feel free to get in touch with me. I’m more than happy to make some time to discuss many of the issues and concepts I’m quickly glancing over here, and that’s when I’m not currently working on a project relating to them.

If you have any thoughts or would simply like to read over a draft once I have something written, let me know. If you’d like to actually help draft the submission, definitely let me know. And if you disagree with something I’ve said here or have concerns about my approach, it’s probably most important that you get in touch with me (I like to think I respond quite well to criticism).

You can email me at andrew@exhipigeonist.net, or simply indicate that you’d like to have a discussion if you happen to see me around. And if you happen to know of anyone else who may be interested in this project, please forward them to this post.

In the mean time, I’ll likely post any updates here on my blog. Stay tuned, and thanks for reading =)

* A gender field may not appear to be the most appropriate place to declare intersex status, but it could be helpful here. And it sure beats writing it in as your religion.

Pre-writing history

There’s an old saying that history is written by the victors – or winners, or some other variant of this. This relates to Orwell’s ‘He who controls the present, controls the past. He who controls the past, controls the future’. Nietzsche also offers a nice alternative: ‘Memory says, “I did that.” Pride replies, “I could not have done that.” Eventually, memory yields’. After a very brief search I couldn’t find where this sentiment may have originated.

I find it interesting but I’d never really thought about it much until I heard a variation on this last year that went something like,

It’s not the winners that write history; it’s the authors.

(I can’t for the life of me remember where I heard this – possibly on a podcast – but hopefully it’s hidden somewhere in my complex note system. I’ll get back to you.)

I quite liked the idea behind this revision. There are individuals – authors and others – who hold the most power over collective memory so it’s a bit limited to suggest the entire group ‘the winners’ are the ones who ‘write history’. This got me thinking about these ideas in relation to my own work.

I had been looking at the representation of populations and reading about Organisation Intersex International Australia‘s campaign calling for a revision to the binary sex representation of individuals in the census. The problem is that as many as one one percent of individuals who may not easily fit within the binary understanding of sex are delegitimised by not being counted. (And are technically committing perjury if they follow the rules.) In addition to more immediate social effects, ABS’s decision means that intersex individuals, among others, are being written out of the official historical record of our country’s population.

Any categorical limitations introduced in order to simplify census fields can make large demographics suddenly invisible.

Do you know how many intersex individuals were in Australia according to the last census? None. Since it became national in 1911? None.

But this is only one source of historical documentation, you might say, you could probably find that kind of information elsewhere. You would be correct, but the same could be said about the other quotes above.

The point is that this is a more general phenomenon that can be seen in social media systems or any other digital spaces.  Developers predetermine the rules of identity performance and social engagement, implementing systems that can disproportionately block access from certain demographics and deactivate and remove accounts it doesn’t feel comply with its aims. Which individuals may have their content archived and the strict set of rules that govern the structure of data decide what is present in memory for later access.

With this in mind I offer the following as a further revision:

System designers and programmers write history – in advance.

This isn’t a problem in itself. And we’re kinda stuck with it actually. We just have to be vigilant, as a society, if we are to identify and prevent social problems that result from an over reliance on technologies that archive according to strict, limited categorisations.

Percentages of Facebook users in Australia

In the comments of my previous post I mentioned the benefits of working out the percentage of people who are on Facebook by location, as this would help support an argument about the ethics of excluding people from such spaces. (If more people are on there, the social cost of opting out becomes greater.) I felt inspired to do a quick calculation of the numbers. Prepare yourself for an onslaught of numbers and tables!

Calculating Australia’s current population

We last held a census on August 9 2011 and the Australian Bureau of Statistics (ABS) website is a great resource for demographic information. This page helpfully displays census data for the Greater Sydney area, New South Wales and Australia overall, which I will be referring to here.

However, it has been a little over a year since the census and the population will have increased. To address this I will be increasing the values found on that page by 1.4%, the estimated rate of national population increase in 2011 from the previous year as stated on this page. This won’t make my results exact (states and cities increase at different rates, and the past year’s increase may be different to 2011′s), but I think this is close enough for my purposes.

Estimates of the current Australian population by location:

Location Population
Sydney 4,453,157
New South Wales 7,014,505
Australia 21,808,825

Factoring in age restrictions

Facebook doesn’t (officially) allow users under the age of 13. Many ignore this, and some are even assisted by their parents in  setting up an account. For this brief study I’m going to pretend this (possibly significant?) number of underage users does not exist.

The problem, then, is removing people under 13 from my population results table above so I can compare it with Facebook data. The census statistics page helpfully breaks down the population by age, but includes ranges ’10-14′ and ’15-19′ rather than having a convenient break between ages 12 and 13. To address this I have used the 2011 totals by location and removed the entire population of under 10s and 3/5 of the 10-14 demographic, and then increased these values by 1.4% as above to estimate the current population.

Revised estimates of current populations to include only those over 13:

Location Population (13+)
Sydney 3,762,165
New South Wales 5,930,528
Australia 18,440,93

Getting data from Facebook

One thing I love about Facebook is that its advertising page allows you to find interesting demographic data about its users. You need to be logged in first (I used my testing account), but all you need to do is go to the ‘Advertise on Facebook’ page, put in a URL (anything – it doesn’t matter), and then play around with the ‘Choose your audience’ section that appears. For example, if you wanted to work out how many Australian accounts have no declared ‘gender’ (yet another example of interchangeability of terms), you select ‘Australia’ as the location and ‘All’ under ‘Gender’ and it displays the audience size to the right (11,624,680 – this and following values retrieved 24 September 2012). Subtract from this the number of ‘Men’ (5,336,740) and ‘Women’ (6,091,320) users and you find that 196,620 (1.69%) Australian accounts have not succumbed to Facebook’s insistence, since 2008, that they choose a sex/gender. (It’s impossible for new accounts to opt out now, of course.)

However, there are issues with trusting this data. First of all, it isn’t clear how accurate the values displayed are, and – discounting a highly unlikely set of chance results – all values are rounded to 10. But more than that the results rely entirely on the accuracy of user entered data.

The 11.6 million ‘Australia’ Facebook accounts counted here include, among those from legitimate users,

  • Accounts of deceased persons
  • Abandoned accounts
  • Accounts from those who have moved away and not updated the location on their profile
  • Additional accounts from those who have multiple accounts
  • Accounts from those who are under 13 (which presents a problem for my calculation)
  • Fake accounts (like mine)
  • Accounts made by others for non-humans such as groups, brands, non-human animals, children (who are human but not operating their own account), etc.
  • Closed accounts that are still archived (I have no way of determining whether these are part of the number)
  • Anything else … ?

I’ll call those ‘non-legitimate accounts’ for the purposes of this discussion.

In addition to this, the total number does not include Australian users who

  • Do not declare their location (for example, I calculated that 5.11% ‘Australia’ Facebook users have not declared a state – it’s unclear how many users choose not to declare a country*)
  • Declare another (fictional or real) location

To an extent, items on these two lists cancel each other out, but it’s difficult to argue whether, say, the active number of Australians on Facebook is higher or lower than the 11.6M stated by the advertising page. (Any thoughts?)

For my purposes here I’m simply going to use the data given to me by Facebook, though I freely admit the issues raised here make any conclusions or results problematic.

Population of Australians on Facebook

By refining my Facebook advertising results by choosing the city ‘Sydney’, the state ‘NSW’, and the country ‘Australia’, and comparing them with the results of my earlier tables, I get the following results:

Location Estimated population (13+) Declared Facebook population Percentage on Facebook
Sydney 3,762,165 2,669,540 60%
New South Wales 5,930,528 3.792.440 54.1%
Australia 18,440,933 11,624,680 53.3%

It makes sense that Sydney has a higher percentage of Facebook users than the rest of the state and country, but I’m actually surprised it’s quite that high as I assumed fewer users would go the extra trouble of declaring a city, let alone their state, on their profile. I also don’t know what the earlier Facebook profile interface was like, but I assume it wouldn’t have categorised locations as well as it currently does so many older accounts may say ‘Sydney’ (text field), or one of many locations that are encapsulated within it, rather than ‘Sydney’ as a tagged, searchable category option. Therefore I suspect many may be missing from this 60% total, though that number may be countered by the number of non-legitimate accounts.


In short, what we’ve seen here is how one may use available data from Facebook and the ABS to determine a non-reliable percentage of Facebook users by location.

This and similar studies could be used in arguments regarding the coerciveness of Facebook and how, because of the high prevalence of use within a population, Facebook and similar systems, despite their status as private enterprises who ‘should be able to do whatever they want because it’s their system’, may actually have a responsibility in making their systems accessible to all by removing barriers such as ‘real identity’ requirements. Because such a high percentage of our social engagement now occurs online, and much of it on Facebook, those who are barred from entry lose the ability to meaningfully engage with their communities, both geographic and virtual, to a fair extent. When systems impose marginalisation and society silences diverse voices because they adopt such systems for everyday interaction, ideas stagnate and we lose as a civilisation.

… And other things I won’t go into now.

Of course, if anyone is actually tempted to use my results or method in making these or other arguments, be prepared for others to dispute the numbers. There’s a wide range of uncertainty within the Facebook data that I’ve only just begun to address.

But that’s not necessarily a bad thing. As someone who advocates for profile systems making all fields optional, it makes me happy to recognise the major limitations of this data =)


* If I could be bothered, I could work out the populations of every country, add them together and subtract that from the number of total Facebook users, and then attribute a percentage of this number (relating to the ratio of declared Australian users) to the Australian population … but that could take a while.

Is limited representation bad for advertisers?

I’m writing a chapter about the representation of individual persons this month so I’ve been thinking more than usual about the justifications for limiting profile options.

One common reason for, say, giving users of your system a limited set of predefined options to choose from (drop-down menus, toggle switch, ticky boxes, etc) is that such strict categorisation allows for easy collection of specific data. You can more easily compare users. And often – for example, if you’re using a free corporate social media service – this categorisation ties right into the advertising side of the business; they offer targeted advertising services to other businesses wanting web traffic in order to remain financially viable.

I was looking at Facebook’s ad targeting page this week, which outlines the process of targeting specific demographics and later seeing detailed metrics that help you review your approach. (Check it out, it’s interesting!)

Another concept I’ve been looking at a lot this week is that of ‘Big Data’. I had some trouble defining it so I evaluated other attempts . . . which weren’t as helpful as I had hoped. It appears Big Data is one of those new buzz words that everyone is using and defining differently in relation to their own work or theoretical context.

However, the MIKE2.0 site offers an interesting understanding that emphasises the complexity of data, rather than its size, as being the definitive property of Big Data. This appeals to me and my approach because it highlights the confusion and potential for new understandings that this phenomenon introduces. To complement this, I also like Mac Slocum’s description of social data as “an oracle waiting for a question” in this context. There is just so much data out there and the new problems we face are less to do with accessing data than trying to ask the right questions to find out something new and exciting.

And this is where we return to Facebook. I’ve always thought the best argument for limited representation (gender/sex is always a good example, but other categories are just as relevant here) is that it helps with their advertisers. However, we have technology that can produce rich datasets and give us more details about individuals, so wouldn’t allowing a greater range of representation actually improve targeted advertising? Companies can include or exclude demographics and then better review the effectiveness of their advertising in much more detail.

One example that Facebook gives is a success story where a wedding photography business targeted women aged 24-30 who were recently engaged. Of course this works*, but it would be even more successful if they could look at and target finer demographics within this quite broad dataset. Imagine if Facebook simply allowed for an ‘other’ gender/sex option – even this simple change would help many advertisers not pay to target ‘those crazy, politically correct hippies who probably wouldn’t want my products anyway’, and give many more advertisers the ability to more easily target this specialised group.

Big Data is complicated, but it enables us to uncover interesting and important details. And where corporations such as Facebook may benefit from expanding its user options, government funded projects like the census who still enforce a male/female binary and ignore non-standard religions – thus making us all miss out on exploring and better understanding our diverse population now and in the future, which is an important aspect of its purpose – really have no excuse.

Admittedly, this is the very early stage of an idea so it’s not completely thought out, but it’s the first time I’ve been able to see Facebook’s binary gender/sex field as being detrimental to their business and thought this was important enough to share.

* As much as any online advertising works, at least. I’m always surprised to hear that people click on advertising links, because it’s counters my own practices so strongly. But I’m happy these people are out there, keeping the Internet alive!