Towards a Statistical Standard for Gender

A submission to Statistics New Zealand

Statistics New Zealand’s 2015 Statistical Standard for Gender, which sought to standardise collection of gender information both by Stats NZ and by other New Zealand organisations, was received poorly by the transgender community. The top level of the classification is “male”, “female”, and “gender diverse”, however the second level placed both transgender men and transgender women under gender diverse and not under male and female respectively. This was seen as ignoring the responses given to the consultation that they carried out before hand, and the system has not been widely adopted.

Five years later Stats NZ is carrying out a new consultation for the gender and also the sex standards, which ends today August 13 2020. Given the result of the previous round I was reluctant to participate myself, however I’ve changed my mind and written the following answers to their questions. If you are reading this before 5pm local time you can do the same at this link. The original text is quoted.

Submission questions

Gender by default principle

We propose that the ‘gender by default’ principle is adopted in the updated standard. This is an approach that defaults to the use of gender data as opposed to sex at birth. Collection of sex at birth information should be viewed as an exception.

In most cases a person’s gender – their social and personal identity – is most relevant for policy making and research rather than their sex at birth. Gender based analysis is used in a range of areas, from income equality to health and education. Recent guidance has recommended that in most cases when sex or gender information is required, gender is most relevant to collect.


…but with a significant caveat. As noted in the consultation document gender and sex are often conflated, however this should not be assumed to be undesirable or something to be avoided. The language given here implies a conflation instead between sex and sex assigned at birth (ASAB), which is a much worse situation. In particular it conflicts with other official documents, implying e.g. that a transgender person changing their “sex” on their passport is in some way illegitimate or fraudulent, and in general gives support to the contention that transgender people are “really” their assigned sex. Instead, consider making sex and gender synonyms, with gender preferred, or preserving sex as a separate question while explicitly acknowledging that transgender people’s sex can match their gender.

‘Gender’ concept definition

We propose the following gender definition:

‘Gender refers to a person’s social and personal identity as male, female, or another gender such as non-binary. Gender may include how a person describes themselves (’gender identity’), and/or the gender a person publicly expresses (‘gender expression’) in their daily life. A person’s current gender may differ from the sex recorded at their birth and may differ from what is indicated on their current legal documents. A person’s gender may change over time. Some people may not identify with any gender.’

For details see:


…although I would change “or another gender such as non-binary” to “a non-binary gender, or no gender at all”. It makes as much sense to list “non-binary” as an example of a gender as it would “binary.”

Another gender

We propose ‘another gender’ as the category for classifying responses as opposed to ‘gender diverse’. This both renames that classification (which is currently known as ‘gender diverse’) and limits it to those who specify their gender as ‘another gender’.

In the New Zealand context, the term ‘gender diverse’ is often used and understood as an umbrella term, similar to the terms trans or transgender. However, some trans people may not use the term and not all gender diverse people may identify as trans (Oliphant, 2018).

This consistent approach also avoids ascribing an umbrella term that may not be a good fit for some respondents. It will also more clearly indicate that data reported in this third ‘another gender’ category does not represent all transgender people (as many will have selected male or female responses) and avoid the confusion created under the current ‘gender diverse’ classification.

For details see:


Under the existing system “gender diverse” attempts to capture all transgender people, including (binary) trans people who would be much more likely to pick “male” or “female” and find the idea that they are (or have to be) a third gender offensive. Limiting the third category to “another gender” is superior, although I don’t know if this is the best name.

Two-step method

We propose use of the two-step method in the updated standard.

This involves asking a question about sex at birth, combined with a question on gender. The two-step approach is considered best practice for use in population representative data collections, where reflecting the transgender population is required. It is also the approach implemented by Statistics Canada in some of their surveys.

For details see:

Strongly disagree

While this method of identifying transgender people for statistical purposes is superior to forcing all trans people into a third gender grouping, it is substantially inferior to simply asking something along the lines of “are you transgender? (Yes/No).”

The consultation document does not favour this approach, saying:

A transgender status measure may be a good approach in some settings, particularly in targeted surveys where researchers have greater confidence the question wording is well understood and inclusive enough for the sample population (for example, LGBTQI+ community-specific surveys).

At present, we are less certain that this approach would capture information with adequate coverage of the diverse range of identities in the transgender population in New Zealand. Therefore, we are not endorsing it as an approach for representative surveys.

Among the reasons for this are:

  • There is less empirical research supporting the effectiveness of this approach to meet information needs.
  • The terminology used by trans people to describe themselves can vary widely, particularly across cultures, ages, and other factors. This approach maybe better suited when modified to a specific community, where there is more certainty or consensus on terminology to adequately reflect the population of interest.
  • Designing a question that adequately identifies the transgender population while being well understood by the population as a whole is a challenge–there has been less testing to support the effectiveness of this approach.

However there are several issues with this analysis. In particular:

Overall even if more research has been done on asking transgender people for their ASAB, it is still not to be preferred to asking people directly if they are trans.

‘Sex at birth’ concept

We propose introducing a specific definition and question module based on ‘sex at birth’ for use in surveys, used solely in the two-step method (where identifying transgender populations is required).

Sex at birth refers to the sex recorded at a person’s birth (e.g. recorded on their original birth certificate).

For details see:


As mentioned earlier, I do not favour this method for identifying the transgender population, and fear that a sex at birth concept may be inappropriately conflated with sex in general.

Further, while the text of this section states that sex at birth would only be used to identify transgender people, other material including the text of the “gender by default” section of this consultation form implies that other use for this data exists. However, this is never expanded upon, but needs to be carefully considered and then explained.

There is a common belief in certain circles that for the purpose of medical treatment transgender people should be considered as their assigned sex, on the grounds that they are still “biologically” that sex. This is incorrect, and instead the relevance of therapies, examinations, and interventions associated with sex in cisgender individuals vary substantially. For example transgender women on Hormone Replacement Therapy (HRT) usually develop breasts and therefore require screening for breast cancer, while because of an overlap between some common HRT drugs and drugs used to test prostate cancer this latter male-associated cancer is a much lower concern. Conversely trans women generally have no need for pap smears to detect cervical cancer even though rigid IT systems may attempt to book appointments for them, while trans men often do have need for this but are unable to make bookings with the same systems. Instead of recording an explicit assigned sex, medical systems need to be more flexible and make fewer gender-based assumptions.

Additionally, and as noted in the consultation document, only providing male and female options for assigned sex at birth fails to take into account that there are people who are not assigned either.

In sum the creation of a sex assigned at birth standard seems unnecessarily invasive without providing sufficient useful information.

Intersex information needs

We propose adoption of an intersex variation question where intersex population data is required.

Where intersex population data is required, international best practice is to use a separate question asking whether a person was born with an intersex variation.

For details see:


I have no additional comment on this section; I lack sufficient knowledge of this particular issue.

Further information you would like to share

Is there any other information you would like to share to assist us in the review of these standards? 4000 characters maximum

All gender, sex, transgender or intersex status, and related information should be treated with extreme care, confidentiality, and respect. Statistics New Zealand should recommend only collecting the information that is absolutely necessary for a particular purpose, but it should also minimise the creation of additional statistical standards, particularly for sex assigned at birth, for the same reason. Failing to create a statistical standard for assigned sex would not prevent such a question being asked improperly or maliciously, but would prevent Stats NZ appearing to endorse it.

File under: