See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/330772804

The Gender Identity Scale: Adapting the Gender Unicorn to Measure Gender
Identity

Article in Psychology of Sexual Orientation and Gender Diversity · January 2019

DOI: 10.1037/sgd0000322

CITATIONS READS

28 13,685

2 authors:

Felicity Ho Alexander J Mussap

Monash Health Deakin University
4 PUBLICATIONS 65 CITATIONS 70 PUBLICATIONS 1,717 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Mirror neurons and autism spectrum disorders View project

All content following this page was uploaded by Alexander J Mussap on 09 April 2019.

The user has requested enhancement of the downloaded file.

Psychology of Sexual Orientation and
Gender Diversity
The Gender Identity Scale: Adapting the Gender Unicorn
to Measure Gender Identity
Felicity Ho and Alexander J. Mussap
Online First Publication, January 31, 2019. http://dx.doi.org/10.1037/sgd0000322

CITATION
Ho, F., & Mussap, A. J. (2019, January 31). The Gender Identity Scale: Adapting the Gender Unicorn
to Measure Gender Identity. Psychology of Sexual Orientation and Gender Diversity. Advance
online publication. http://dx.doi.org/10.1037/sgd0000322
 Psychology of Sexual Orientation and Gender Diversity
© 2019 American Psychological Association 2019, Vol. 1, No. 999, 000
2329-0382/19/$12.00 http://dx.doi.org/10.1037/sgd0000322

The Gender Identity Scale: Adapting the Gender Unicorn to Measure

Gender Identity
Felicity Ho and Alexander J. Mussap
Deakin University

The Gender Unicorn is a popular infographic used to educate people about gender diversity. We adapt
the Gender Unicorn for use as a measure of gender identity—the Gender Identity Scale (GIS)—in which
participants report level of identification with each of three genders: female/woman/girl, male/man/boy,
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

and other gender(s). We administer the GIS to a sample of 269 self-identified trans and gender diverse
This document is copyrighted by the American Psychological Association or one of its allied publishers.

adults and use a latent class analysis of responses to identify seven gender classes. We find these classes
to be consistent with participants’ own designations of gender. These results support the utility of the GIS
as a measure of gender that can be used with a gender diverse population.

Public Significance Statement

We adapt the Gender Unicorn to create a measure of gender—the Gender Identity Scale (GIS)—in
which participants are asked to report their level of identification with each of three genders:
female/woman/girl, male/man/boy, other gender(s). We demonstrate that responses to the GIS are
consistent with trans and gender diverse people’s own designations of gender.

Keywords: Gender Unicorn, transgender, gender diversity, gender identity, demographics

Gender identity is an individual’s internal sense of being a woman/ logical Association, 2010). Unnecessary uncertainty is created
female, a man/male, and/or a nonbinary gender (GLAAD, 2015; when the terms used are associated with sex, but gender is sup-
Johnson, Greaves, & Repta, 2009; Pinn, 2003). It is considered to be posedly the topic of discussion (Muehlenhard & Peterson, 2011;
an important facet of a person’s identity (Wood & Eagly, 2009), one Runnels, Tudiver, Doull, & Boscoe, 2014). For example, female
that does not necessarily correspond to the sex they are assigned at and male are often associated with sex and therefore woman and
birth (which is usually based on genital appearance; Tate, Ledbetter, man are the preferred terms to use when assessing gender (Ansara
& Youssef, 2013).
& Hegarty, 2014; Tate et al., 2013). Correct use of language is
Despite the complexity of gender as a construct and the diversity of
therefore extremely important.
gender identities present in the population, many researchers continue
to measure their participants’ gender in a manner that conflates gender The importance of language continues to apply in the use of
with sex by using a single self-report item that offer participants a gendered language to describe groups of people. Using gendered
choice of female and male (Tate et al., 2013; Treharne, 2011; West- language to identify people in ways that they do not identify them-
brook & Saperstein, 2015). Participants whose gender identity does selves is considered misgendering (Ansara & Hegarty, 2012, 2014).
not conform to the woman/man binary may find themselves excluded Misgendering is a sexist practice that delegitimizes people’s own
from such research or their gender misrepresented and/or constrained designation of gender (Ansara & Hegarty, 2014). Forcing people to
to inappropriate gender categories (Treharne, 2011). This could po- endorse a binary option for gender, and consequently describing them
tentially underestimate the effects of gender and gender diversity with those labels, may therefore constitute a form of discriminatory
(Westbrook & Saperstein, 2015). practice through the potential misgendering of participants who do not
identify with the gender binary.
Measuring Gender in Inclusive Ways One attempt to be more inclusive when measuring gender is to
Gender, rather than sex, is the correct term and construct to use add an option for transgender. This single category fails to recog-
when referring to social groupings of people (American Psycho- nize the multitude of genders within it that may contribute to
differences in research outcomes (Budge et al., 2013; Harrison,
Grant, & Herman, 2012). It also fails to recognize those who do
not identify as transgender but as women or men, neither, or with
some other term (Ansara & Hegarty, 2014). Attempts to address
Felicity Ho and Alexander J. Mussap, School of Psychology, Deakin
University.
this by adding a fourth option of “do not identify as female, male,
Correspondence concerning this article should be addressed to Felicity or transgender” (The GenIUSS Group, 2014) may still be prob-
Ho, School of Psychology, Deakin University, 221 Burwood Highway, lematic as people may identify with both transgender and woman
Burwood, Victoria, Australia 3125. E-mail: felicity.ho@deakin.edu.au or man (Ansara & Hegarty, 2014).
1
 2 HO AND MUSSAP

Another option is to expand the selection of gender categories identity line now anchored by woman and man, with genderqueer
and allowing participants to endorse any categories that apply to at the midpoint. The Genderbread Person was updated shortly
them (Grant et al., 2011; Tate et al., 2013; The GenIUSS Group, afterward (Killerman, 2012b) with each construct now represented
2014). However, this approach requires researchers to make as- by two lines rather than one. Gender identity used two lines each
sumptions about the terms used to describe categories and is anchored by nongendered on the left and woman-ness and man-
complicated by terms varying between cultural groups and evolv- ness on the right (with no label between these anchors). This
ing over time (Brown, 2016). Terms that were prevalent in the allowed people to use the woman/man labels independently with-
recent past, such as “transgendered” (Serano, 2007), have faded out being constrained to a gender-binary response. For example,
from use and may even be considered offensive by some within the someone who identified as bigender could mark both lines in
trans and gender diverse community (GLAAD, 2015). Difficulties gender identity highly and someone who did not identify with any
with the nonspecificity of some terms, such as transgender or gender could mark both lines at zero (left). This was unclear in the
genderqueer, remain. Researchers also assume that each partici- previous version where the middle of the scale might be applicable
pant uses endorsed terms in the same way. This also raises ques- to both, or people who did not identify with any gender might feel
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

tions about the meaning of multiple endorsements— both at a excluded. Similarly, two lines anchored agender and feminine/
This document is copyrighted by the American Psychological Association or one of its allied publishers.

conceptual level and from an inferential statistics point of view. masculine represented gender expression, two lines anchored asex
Allowing participants to describe their identity in their own and female-ness/male-ness represented biological sex, and two
words overcomes some of these challenges and ensures, most lines anchored nobody and women/females/femininity/men/males/
importantly, that researchers do not misgender respondents (An- masculinity represented sexual orientation (now labeled sexual
sara & Hegarty, 2014). However, this approach is susceptible to attraction).
the idiosyncratic and changing use of descriptors and forces re- The importance of correct and inclusive language motivated the
searchers to make ad hoc groupings of participants if they wish to creation of the Gender Unicorn to address limitations in the revised
conduct quantitative analyses on responses (Hyde et al., 2014; Riggs, Genderbread Person (Pan & Moore, 2014). Pan and Moore (2014)
Power, & von Doussa, 2016). A list of idiosyncratic descriptors may argued that nongendered is neither the only term nor the preferred
also make it difficult to assess the diversity and representativeness of term (agender) that can anchor each of the gender identity scales.
a sample, which therefore defeats the purpose of collecting that They therefore removed the use of labels to anchor any of the
information (Connelly, 2013). scales. Furthermore, they criticized Killerman (2012b) for incor-
rectly using agender to anchor gender expression. Pan and Moore
also argued that “biological sex” was ambiguous and harmful to
From the Gingerbread Person to the Gender Unicorn
trans people, suggesting that sex assigned at birth was more
In the present article, we investigate the possibility that info- accurate. They dismissed the term asex based on it not being a real
graphics used to educate about gender diversity may serve as the word and meaningless given that everyone has some sex charac-
basis for measuring gender in a manner that is inclusive, not reliant teristics prescribed to them. They also added a third scale to gender
on particular linguistic descriptors, and suitable for use with the as some cultures have genders outside of the male/female binary.
trans and gender diverse people. A strength of using these info- The Gender Unicorn (copies of which are accessible at http://
graphics is their history of development and adoption by the trans www.transstudent.org/gender/) consists of five constructs: gender
and gender diverse community (“Gingerbread Person,” 2011; identity, gender expression, sex assigned at birth, physically at-
Lawson, 2011; Pan & Moore, 2014). The trans and gender diverse tracted to, and emotionally attracted to. Sex assigned at birth is the
community places importance on being involved in the discourse only construct that is categorical with three options: female, male,
and conceptualization of their gender identity (Pan & Moore, and other/intersex. The other four constructs are continuous with
2014). This is exemplified through the creation of the Gender three dimensions each. Each dimension has a nil value at one end
Unicorn to reclaim discourse and conceptualization from someone and an unlabeled maximum value at the other, with no explicit
who was not part of the trans and gender diverse community (Pan anchors provided. Gender identity has the dimensions female/
& Moore, 2014). The most respectful way of measuring gender woman/girl, male/man/boy, and other gender(s). Gender expres-
would therefore be to use materials that the trans and gender sion has the dimensions feminine, masculine, and other. Physically
diverse community as a group has reviewed and accepted as the attracted to and emotionally attracted to both have the dimensions
best way to conceptualize their gender diversity. women, men, and other gender(s).
Early infographics distributed through social media (“Ginger-
bread Person,” 2011; Lawson, 2011) drew upon earlier academic Adapting the Gender Unicorn for Use as a
work (Diamond, 2002) to bring attention to the distinction between
Measurement Tool
the constructs of sex, gender, sexual orientation, and gender ex-
pression. Lawson’s (2011) Gingerbread Person further represented In summary, the Gender Unicorn represents not only a departure
each construct as a continuum. Each construct was represented by from the traditional cis-binary conceptualization of gender, it calls
a line anchored by male and female (or masculine/feminine in the into question the adequacy of describing gender along a continuum
case of gender expression) with the centers of each line labeled anchored by woman and man, and the adequacy of only having two
intersex for sex, genderqueer for gender, bisexual for orientation, dimensions to measure gender. It also provides the basis for repre-
and androgynous for gender expression. senting gender in ways that are not overly dependent on linguistic
The Genderbread Person (Killerman, 2012a) continued this gender descriptors.
structure with several changes. The sex and gender constructs were However, although the Gender Unicorn has been used success-
relabeled biological sex and gender identity, with the gender fully as a teaching resource (Solotke, Sitkin, Schwartz, & Encan-
 GENDER IDENTITY SCALE 3

dela, 2017) its use in research to measure gender requires further H3: A gender class or classes will include participants who
psychometric validation. Therefore, in the present study we con- identify with a combination of dimensions of the GIS.
ducted an evaluation of the Gender Unicorn as a measure of gender
in which respondents use sliding scales to indicate their level of H4: Participants who describe their gender using “binary”
identification with female/woman/girl, male/man/boy, and other labels, such as woman or man” will identify predominantly
gender(s). We named the scale used in this study the Gender with the corresponding GIS dimension (i.e., female/woman/
Identity Scale (GIS) as the scale incorporated only the gender- girl, or male/man/boy respectively) and appear together in
identity component of the Gender Unicorn. Although the Gender respective LCA gender classes.
Unicorn consists of five constructs, only two— gender identity and
sex assigned at birth—were included in the GIS. Physical and H5: Participants who describe their gender using “umbrella”
emotional attraction are concepts related to sexuality rather than labels, such as genderqueer or transgender, will not identify
gender (American Psychological Association, 2015; Pan & Moore, exclusively with a single GIS dimension and will appear
2014), and thus we did not include them in the GIS. We also did across multiple LCA gender classes.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

not include gender expression because, although gender expression

This document is copyrighted by the American Psychological Association or one of its allied publishers.

may be a way of communicating gender, it is not necessarily Method

reflective of a person’s gender identity (American Psychological
Association, 2015; Westbrook & Saperstein, 2015).
We focused on validating the GIS with trans and gender diverse Participants
people because, as noted in the literature review, this population
presents researchers with the greatest challenge in terms of mea- The 269 participants were aged between 18 and 79 years (M ⫽
suring gender and are the most vulnerable to misgendering in 39.40, SD ⫽ 14.45), although 17 participants did not provide age
current research (Ansara & Hegarty, 2012, 2014). We analyzed data. The country of birth for the participants were: Australia (n ⫽
their pattern of responding to the GIS via latent class analysis 121), United States (n ⫽ 84), United Kingdom (n ⫽ 26), Canada
(LCA) to reveal underlying gender identity classes. We expected (n ⫽ 12), New Zealand (n ⫽ 9), Germany (n ⫽ 3), two each from
that some participants will identify with the gender binary and Ireland and Turkey, one each from China, Colombia, Iran, Japan,
therefore the LCA will identify classes of participants that highly Luxemburg, Spain, Trinidad and Tobago, and three participants
identify with only one of female/woman/girl or male/man/boy. If who did not provide that information. In terms of sex assigned at
people do identify with genders outside of the binary (Pan & birth, 150 (56.1%) participants reported being assigned female at
Moore, 2014), then the LCA should also identify a class of birth (AFAB), 113 (42.7%) reported being assigned male at birth
participants that highly identify with only other gender(s). The GIS (AMAB), and 3 (1.1%) reported being intersex/other. This was a
has, however, separate dimensions of gender identity so that those more even AFAB:AMAB ratio (1.3:1) than reported by previous
dimensions may vary independently. This suggests that the LCA studies (typically from 2:1, Hyde et al., 2014, to 3:1, Couch et al.,
should also identify classes of participants that identify with dif- 2007; Ho & Mussap, 2017).
ferent combinations of those dimensions, which includes not iden-
tifying with any gender at all. Measures
We then examined the way that participants linguistically de-
scribe their gender within each of the classes identified by the The GIS created for this study was adapted from the gender
LCA. We did this to determine if the identified classes had a identity and sex assigned at birth components of the Gender Unicorn
meaning that was consistent with participants’ own designations of (Pan & Moore, 2014). The GIS measured gender identity by
gender. There are commonly accepted definitions for many labels asking “To what extent do you identify with the following gen-
(e.g., Barker & Richards, 2015; “Gender Identities,” 2015; Trans- ders?” and used the three dimensions of female/woman/girl, male/
Gender Victoria, 2013), but some definitions can be broad. We man/boy, and other gender(s). The scales in the Gender Unicorn
expected that labels with a currently clear definition (e.g., woman are not labeled or anchored and are represented as a continuous
or man) would consistently appear in a single class. However, line (see Appendix). We anchored each scale in the GIS with labels
umbrella labels such as genderqueer or nonbinary can cover many to provide guidance for the direction of response from Not at all to
different types of identities including, but not limited to, mascu- Very strongly and points were also numerically labeled (0, 10 . . .
line, feminine, neutral, or bigender (TransGender Victoria, 2013). 90, 100) so that responses would more likely be interval-level data
We therefore did not expect umbrella labels to appear exclusively (Davies, 2008). We implemented the scales using sliders to convey
within one class. the concept of continuity from the Gender Unicorn. The sliders
We analyzed patterns of responding to the GIS from trans and were constrained to only allow input on the labeled points. The
gender diverse participants via a LCA to test the following hy- scales were therefore 11-point scales, which allow respondents
potheses: enough options to perceive that they are able to express themselves
adequately without decreasing test–retest reliability (Preston &
H1: Three gender classes will correspond to participants who Colman, 2000). A response was compulsory for each scale which
identify with only one of the three dimensions in the GIS— means there was no missing data in the dataset.
female/woman/girl, male/man/boy, or other gender(s). Participants were also asked “How would you usually describe
your gender?” and a text field was provided for responses. The GIS
H2: One gender class will consist of participants who do not asks for sex assigned at birth using the three mutually exclusive
identify with any of the dimensions of the GIS. categories of female, male, and other/intersex.
 4 HO AND MUSSAP

Procedure model (it is nonparametric), requiring only independence of ob-

servations between classes (Oberski, 2016). Tein, Coxe, and Cham
The Deakin University Human Research Ethics Committee ap- (2013) noted that the minimum sample size requirements for LCAs
proved this study. Participants for this study were recruited via
is understudied and the risk of underpowered studies is that the
posts in online support forums, social media, and websites. Exam-
number of classes extracted is less than the true number of classes
ples included Gender Diversity Australia and Association of
in the data. The Bayesian information criterion (BIC) was there-
Transgender Professionals support forums, and Gender Queer
fore used for potential solutions to determine the number of classes
Australia website. We placed physical notices in the clubrooms of
to extract.
university lesbian, gay, bisexual, transgender, and intersex clubs
We tested H1, H2, and H3 in terms of whether or not the
and associations, and encouraged peer referral to the study. The
predicted gender classes appeared in the results of the LCA. For
study was advertised as being open to adults (18 years of age or
example, according to H1 the LCA should identify classes of
older) who self-identified as trans or gender diverse. Advertise-
ments directed participants to an online survey that they could “binary” participants (i.e., participants who identify exclusively
complete after reading the plain language statement, providing with the “female/woman/girl” or “male/man/boy” dimensions of
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

consent, and confirming that they were over the age of 18 years the GIS). Failure to do so, particularly given previous research
This document is copyrighted by the American Psychological Association or one of its allied publishers.

and identify as trans or gender diverse. Age and gender identity documenting the prevalence of these binary gender identities in the
were thus the only inclusion criteria. We did not offer an hono- trans and gender diverse community (Riggs & Due, 2013), would
rarium to participants. result in the rejection of H1 and force us to reconsider the GIS
and/or or methodolgy. Similarly, a failure to identify participants
who do not identify with any GIS dimensions (H2) or with a
Analytic Design
combination of several dimensions (H3), would result in the re-
We used R (Version 3.2.5; R Core Team, 2016) with packages jection of these hypotheses. Such an outcome would be at odds
psych (Version 1.6.12; Revelle, 2016) and mclust (Version 5.2; with our current understanding of gender diversity (Pan & Moore,
Fraley, Raftery, Murphy, & Scrucca, 2012) to conduct an LCA of 2014; Riggs & Due, 2013).
participant responses in order to derive classes based on similar We tested H4 and H5 by assigning participants to the following
patterns of responding. LCA does not make assumptions concern- groups based on their response to the open-ended question that
ing the nature of observed variables that are inputted into the asked them to describe their gender in their own words:

Figure 1. Self-reported gender identity of participants plotted in 3-dimensional space with each axis repre-
senting one of the gender dimensions. Numbers represent how many people identify with the gender represented
by that point.
 GENDER IDENTITY SCALE 5

1. “Binary” labels such as male, man, guy, or boy. ments that gave no clear indication of gender such as “I
don’t.”
2. “Binary” labels such as female, woman, or girl.
We then individually compared the prevalence of Groups 1 and
3. Nonbinary but relatively specific labels such as trans-
2 across the classes identified by the LCA using a chi-square test.
femme, transmasc, agender, or neutrois (Gender Spec-
We used a Fisher’s exact test if the chi-square assumption was
trum, n.d.; gqid, 2015).
violated, such as when any of the frequencies were less than five.
4. Umbrella terms such as genderqueer, transgender, non- A nonsignificant result for either of these two groups would
binary (Gender Spectrum, n.d.; gqid, 2015), or state- suggest that the usage of the “binary” label is not dependent upon
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Figure 2. Plots of gender identity by sex assigned at birth. Upper diagonal is for people who were assigned
male at birth. Lower diagonal is for people who were assigned female at birth.
 6 HO AND MUSSAP

the genders identified by the GIS. We would therefore reject H4. mented on the match between current definitions, the classification
We would also reject H4 if the odds ratio for the usage of these by the LCA, and the response to the GIS. This analysis is not part
binary labels were not in favor of the relevant binary classes of a specific hypothesis, but it provides us with rich qualitative
identified by the LCA. data regarding how responses to the GIS relate to current usage of
We tested H5 using a binomial test to determine if the propor- gender labels.
tion of participants using umbrella terms was greater than zero in
more than one of the classes identified by the LCA. If this were not
the case, then these umbrella labels would have a more precise Results
meaning than predicted and we would reject the hypothesis.
Responses to the GIS are plotted as a three-dimensional scat-
If participants responded with more than one gender label, we
terplot with each axis of the plot corresponding to a subscale
assigned them to the first group according to the order of the four
within the GIS (see Figure 1). Note that because of overlapping
groups listed above. For example, if a participant responded “gen-
responses each data point is numbered to indicate how many
derqueer man,” then the use of the man label meant we assigned
participants correspond to that point. Visual inspection of Figure 1
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

them to Group 1. This strategy was had the advantage of increasing

indicates that while each subscale was fully used across partici-
This document is copyrighted by the American Psychological Association or one of its allied publishers.

the frequency of “binary” labels appearing in all the nonbinary

classes identified by the LCA, therefore making the chi-square test pants (i.e., responses to each ranged from 0 to 100) combinations
less likely to pass. It simultaneously reduced the frequency of of gender dimensions (that presumably corresponded to a partici-
umbrella labels, therefore making the binomial test more difficult. pant’s gender identity) were asymmetrical, with no participants
This provided us with greater confidence that any significant identifying very strongly with all three genders. The most common
results were not due to confirmatory bias. Of course, in carrying responses were to identify totally and exclusively with one dimen-
out this strategy we did not wish to imply that resultant group sion of the gender “binary”: female/woman/girl (n ⫽ 52) or
assignment was representative of their gender or should supersede male/man/boy (n ⫽ 22). However, many participants clearly iden-
their own original designation of gender in any way. tified with a nonbinary gender, including participants who identi-
We then conducted a qualitative analysis of the label used by fied with no gender at all and might be categorized as agender, and
each participant from Group 3. We compared their label with those who had a strong sense of gender but one that could not be
known definitions (Gender Spectrum, n.d.; gqid, 2015) and com- located in the “female-male” plane.

Figure 3. Bayesian information criterion (BIC) for mclust models versus the number of classes extracted for
the latent class analysis. Each line and symbol represent a different parameterization of the covariance matrix,
where each letter describes the volume, shape, and orientation of the covariance structure. I ⫽ identity matrix;
E ⫽ equal; V ⫽ variable. Parameterizations ending with II therefore have spherical distributions, those ending
with EI or VI have diagonal distributions, and those ending with E or V have ellipsoidal distributions. See the
online article for the color version of this figure.
 GENDER IDENTITY SCALE 7
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Figure 4. Classifications from latent class analysis with covariances of components superimposed. See the
online article for the color version of this figure.

Responses were separated out and plotted according to sex for each response point. Groups were numbered as shown in Table
assigned at birth and, as shown in Figure 2, the identities of 1 with descriptive labels based on the scale means for each
AMAB participants appear to cluster toward a binary female/ classification as shown in Figure 5.
woman/girl identity. AFAB participants appear to have a greater The names of the classes identified by the LCA in Table 1 are
range of gender identities with more participants identifying mod- descriptive of the GIS dimensions with which participants in that
erately or highly with a gender that aligns with their sex assigned class tended to identify. It is important to note that not everyone in
at birth or with other. There were 30% more AMAB participants each class described their gender using the name of that class.
than AFAB participants, but more than twice as many participants Therefore, the names are not representative of the genders in that
identified as a binary female/woman/girl gender compared to a class and it would misgendering to say, for example, that the
binary male/man/boy gender. GIS_man class was a group of men. It would be more accurate to
An LCA was then used to identify classes of participants on the say that the GIS_man class was a group of participants who tended
basis of their pattern of responding to the three gender identity to highly identify with the male/man/boy dimension of the GIS and
questions contained in the GIS. Figure 3 shows that the best model low with all the other dimensions.
according to the Bayesian information criterion (BIC) is a seven- The GIS_woman, GIS_man, and GIS_other classes appear to be
class solution with covariances that have equal volume, equal made up of participants who identify highly with only a single GIS
shape, and varying orientation. Figure 4 shows the classification dimension. The GIS_(woman) ⫹ (other), GIS_woman ⫹
 8 HO AND MUSSAP

Table 1 differ according to class, ␹2(1, N ⫽ 145) ⬍ 0.01, p ⫽ .96, OR ⫽

Descriptive Classifications With Total Number of Assigned- 1.36.
Female-at-Birth (AFAB), Assigned-Male-at-Birth (AMAB), and An investigation into the usage of labels by participants in the
Intersex Participants GIS_woman ⫹ (other) class found that only two participants used
woman-type labels without any modifiers and, out of those two,
Class n AFAB n AMAB n intersex n total only one participant used a woman-type label (“female”) as their
1. (⫹) GIS_woman 7 126 2 135 sole description of gender. All the other seven participants used a
2. (Œ) GIS_man 58 0 1 59 trans or transgender qualification such as “transgender woman,”
3. (e) GIS_other 18 3 0 21 and one participant described their gender as “woman or trans
4. () GIS_(woman) ⫹ (other) 9 8 0 17 woman.”
5. (x) GIS_man ⫹ other 18 0 0 18
6. () GIS_woman ⫹ (other) 0 11 0 11 To test if the proportion of participants using umbrella terms
7. (Œ) GIS_none 5 3 0 8 was significantly greater than zero, we used an expected propor-
tion of .001 for the binomial tests as the tests cannot fail with a true
Note. GIS ⫽ Gender Identity Scale. Symbols correspond to the classes in
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Figure 3. value of zero. The results of the binomial tests (see Table 3)
This document is copyrighted by the American Psychological Association or one of its allied publishers.

showed that, in all but the GIS_man class, the proportion of

participants using umbrella terms was significantly greater than
.001.
(other), and GIS_man ⫹ other categories appear to be made of Participants used nonbinary but specific labels (see Table 4) in
participants who identify with a combination of dimensions. ways that were consistent with current definitions (gqid, 2015;
The GIS_none category appears to be made up of participants Micah, 2016; ShineSA, 2017). One participant in the GIS_woman
who have a very low identification with all the dimensions, class identified as a lesbian, which is a sexual orientation rather
although the standard error for the male/man/boy dimension is than a gender identity. It does have implications regarding
notably large for that class. gender, however, which are consistent with their responses to
We assigned participants to groups based on their textual response the GIS. The LCA potentially misclassified three participants in
for the description of their gender, as detailed in the analytic design. the GIS_woman class, who identified as bigender or nonbinary
The resultant contingency table (see Table 2) allowed us to combine transmasculine, as their identification with the female/woman/girl
columns to create further contingency tables to compare how many dimension appears to be low compared to other participants in that
participants used a man-type label versus how many did not, and how class. Their gender descriptions are consistent with their responses
many participants used a woman-type label versus how many did not. to the GIS but reflect an identity that is different to woman (gqid,
A full list of the labels that participants used (see Figure 6) shows that 2015). The LCA also potentially misclassified two participants in
more participants used nonbinary or umbrella labels than accounted the GIS_man class who identified as agender. Their identification
for in the contingency table. Their use of another term meant that we with the male/man/boy dimensions appeared to be low compared
allocated them to one of the binary groups to maintain independence to other participants in that class. Their gender descriptions are
between groups. also consistent with their responses to the GIS but reflect an
Table 2 shows that four classes had no participants use a identity that is different to man (gqid, 2015).
man-type label and only one participant in the GIS_none class used
a man-type label. A generalization of the Fisher’s exact test with
all the classes showed that the proportion of participants using Discussion
the man-type label significantly differed by class (p ⬍ .001). A The results confirmed that the gender identity dimensions of the
Fisher’s exact test comparing the GIS_man and GIS_none class Gender Unicorn can be operationalized in the form of a GIS and
showed that participants in the GIS_man class were signifi- used to measure gender in a trans and gender diverse population.
cantly more likely to use the man-type label (p ⬍ .001, odds Analyses of patterns of responses to the GIS revealed seven gender
ratio [OR] ⫽ 42.88). A chi-square test with Yates’ continuity classes. There were classes in which there was high identifica-
correction comparing the GIS_man and GIS_man ⫹ (other) tion with only of one each of the three dimensions. There were
class showed that participants in the GIS_man class were sig- also classes corresponding to identification with more than one
nificantly more likely to use the man-type label, ␹2(1, N ⫽ dimension— combinations with the dimension for other gender(s)
75) ⫽ 13.64, p ⬍ .001, OR ⫽ 9.63. in particular. This suggests that the other gender(s) dimension is
Table 2 shows that four classes had no participants use a woman- important and that ignoring this dimension would lead to incorrect
type label and only four participants in the GIS_(woman) ⫹ (other) conclusions about participants’ genders.
class used a woman-type label. A generalization of the Fisher’s exact There was also a class corresponding to low identification with
test with all the classes showed that the proportion of participants all of the gender dimensions. This suggests that some participants
using the woman-type label significantly differed by class (p ⬍ would not be able to record their gender correctly using older
.001). A Fisher’s exact test comparing the GIS_woman and systems that represent man and woman as opposites on a single
GIS_(woman) ⫹ (other) class showed that participants in the continuum. The genders of some participants therefore do vary
GIS_woman class were significantly more likely to use the independently on the different dimensions. The multidimensional
woman-type label (p ⬍ .001, OR ⫽ 11.77). A chi-square test system of measuring gender therefore appears to be useful.
with Yates’ continuity correction comparing the GIS_woman Unsurprisingly, participants used binary labels of gender clearly
and GIS_woman ⫹ (other) class showed that the proportion of that were consistent with their responses to the GIS. The meaning
participants using a woman-type label did not significantly of the trans- prefix or the transgender label was less clear. For the
 GENDER IDENTITY SCALE 9
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Figure 5. Plots of mean Gender Identity Scale (GIS) dimension scores for each category identified by the latent
class analysis. Error bars are for the standard error of the mean. For each plot, M ⫽ male/man/boy dimension;
W ⫽ female/woman/girl dimension; O ⫽ other gender(s) dimension.

test of binary label usage, we prioritized the “binary” labels to therefore used the same labels for different purposes and with
create the test groups. The GIS_woman ⫹ (other) class showed, different implications. These labels are therefore difficult to inter-
however, that the trans- prefix or the transgender label could be an pret without additional information but the responses to the GIS
important part of their identity (Levitt & Ippolito, 2014), which helps to clarify those meanings.
may be explicitly different to being only a man or woman. The use Usage of other nonumbrella labels appeared to be consistent
of these labels in the GIS_man and GIS_woman class alternatively with responses to the GIS. Usage of these terms occurred even
shows that participants may use these labels in recognition of their within the GIS_man and GIS_woman class. This demonstrates the
history and/or experiences rather than their identity being any continuous nature and lack of boundaries between genders as
different to a man or woman (Levitt & Ippolito, 2014). Participants people’s identities move away from the gender binary. People may
 10 HO AND MUSSAP

Table 2 descriptions. Several participants provided responses that would

Contingency Table of Participants With Rows Categorized by not be possible to classify based on the written descriptions alone.
the Latent Class Analysis Identified Classes and Columns The results suggest that it would be erroneous to believe that not
Categorized by Assigned Group Based on the Terms in the providing a specific label is indicative of similar gender identities.
Label Used by Participant to Describe Their Gender Identity If gender was a grouping variable, without the GIS researchers
may improperly allocate these participants or exclude their valu-
Class Man Woman Umbrella Specific able input from the research.
GIS_woman 0 105 14 15 Our ability to identify gender classes from GIS responses may
GIS_man 49 0 7 1 provide future researchers with the ability to explore where bound-
GIS_other 0 0 6 15 aries between genders currently lie and how they may change with
GIS_(woman) ⫹ (other) 0 4 4 9 time and across cultures (West & Zimmermn, 1987). A large study
GIS_man ⫹ (other) 7 0 3 8
GIS_woman ⫹ (other) 0 8 1 2 may be useful in providing guidance for potential “cut-offs” be-
GIS_none 1 0 4 3 tween groups. This would allow groups to have common bound-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

aries between different studies, which would enable easier com-

Note. GIS ⫽ Gender Identity Scale. Column labels are not indicative of
This document is copyrighted by the American Psychological Association or one of its allied publishers.

the participant’s own designations of gender and are only for the analysis parison and validation of research. Researchers would therefore
of language as described in the text. not need to conduct their own classification analyses, which are
data driven and which may result in different groups that are
still have a leaning toward one side of the binary without totally dependent upon the sample. That is not to imply that researchers
relating to it and may use terms such demi-boy and demi-girl could not conduct an LCA on their own data, for example, how-
(Barker & Richards, 2015) or transmasculine and transfeminine ever they would need to acknowledge the limitations of doing so.
(ShineSA, 2017) to describe these identities. These genders rep- Of course, researchers must consider the implications of assign-
resent people who do not identify completely with a binary gender ing participants to groups that may be incongruent with their
(gqid, 2015). There is otherwise no hard line or definition that a self-designated label, (e.g., placing someone who identifies as a
person crosses where they must identify as a man or woman versus man into a group in which all the other members identify as
demi- or trans-. This reinforces the difficulty of classifications via genderqueer). It is important to recognize the context and limita-
labels and the importance of not using the LCA class names as tions of the classification and to remember that self-designated
designations of gender. This also reinforces the usefulness of the identities are not invalid—they are only inadequate for the purpose
GIS in clarifying the meaning of labels. of group comparisons. Researchers therefore need to be careful not
Similarly, people who have a neutral gender may identify as to misgender participants by naming groups with labels that mis-
neutrois or neutral (Micah, 2016), but there is overlap where gender their members, (e.g., “the masculine genderqueer group”).
people could identify as neutrois versus demi-. The overlap means It is for this reason that we labeled the classes in this study more
that people may use different labels but respond in the same way descriptively rather than using specific gender identities. We re-
to the GIS, or they may use the same label but respond differently inforce that the names of the classes are not the gender of the
to the GIS. Manual groupings based on these labels may therefore participants within that class but are representative of their re-
be erroneous and the GIS can again provide clarity. sponses to the GIS.
Participants used umbrella terms across most of the classes. This The purpose for collection and classification is another impor-
highlights the difficulty of grouping participants based on umbrella tant consideration. If the aim is descriptive demographic informa-
terms such as transgender or genderqueer. The GIS (and the LCA) tion, then classification may not be necessary—a 3D plot may be
is able to distinguish between different types of genderqueer and sufficient in illustrating the range of genders present and their
nonbinary genders. The GIS therefore respects the great diversity relative representation. Similarly, if the self-designated label is
of gender identities, but still maintains clarity without forcing important then classification according to the GIS may also be
participants to utilize language with which they may not be com- unnecessary. For example, if a study were to explore how partic-
fortable. ipants with different identities experience stigma differently, re-
Inconsistent use of gender terminology was particularly evident searchers would need to identify whether the focus is on the
with participants who did not identify with any gender at all. communicative use and consequences of the labels themselves or
Agender is potentially an ambiguous term as it may also mean a whether gender identity, and potentially gender expression, ac-
neutral gender rather than an absence of gender (Micah, 2016) but cording to the GIS is more appropriate.
some community groups do define it solely as having no gender
(ShineSA, 2017). An identity of agender boy would appear to be
Limitations/Considerations
contradictory using the latter definition, yet it would be reasonable
under the former definition— even if it still does not provide a The formatting of the survey may have contributed to a misin-
clear idea of exactly how someone identifies. The GIS provides the terpretation of the independence of the gender dimensions. We
ability to categorize participants independently of label definitions. numbered the dimensions from 0 to 100 to help reinforce that it
Even though language evolves over time and individuals create was a continuous construct, rather than categorical. We intended
new labels to identify with (Brown, 2016), the GIS would still for the numbers to reinforce the labels, which ranged from 0 (not
allow consistent reporting, classification, and identification of the at all) and 100 (very strongly). Participants may have interpreted
diversity and similarities in gender identities captured. these as overall percentages however, and therefore adjusted their
The GIS also provides a meaningful way to identify gender overall identity such that all their responses added up to 100. This
when participants are unable or unwilling to provide linguistic may introduce ambiguity for some responses. Participants who
 GENDER IDENTITY SCALE 11

GIS_man GIS_woman
Male / man / guy (31) Trans guy/man/male (14) Female / woman (71) Trans/transgender
FTM (7) Transmasculine (5) Transgender / trans (7) woman/female/girl (29)
Non-binary (4) Agender (4) MTF (3) Fluid / Genderfluid (3)
Trans / transgender (3) Masculine (2) Genderqueer (3) Bigender (3)
Demiboy (1) Genderqueer (1) Mostly female (2) Crossdressing (2)
Demiflux (1) Queer (1) Non-binary (2) Female and trans (1)
… of/with trans experience (3) Feminine (1) Transsexual (1)
“Male with transsexual history” (1) Diverse (1) Trans guy (1)
Neutral (1) Transmasculine (1)
GIS_other
In transition (1) Complicated (1)
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Genderqueer (8) Non-binary (9)

This document is copyrighted by the American Psychological Association or one of its allied publishers.

Two spirited (1) Androgynous (1)

Femme (2) Genderfluid (2)
FTM (1) Confused (1)
Trans (2) Queer (2)
FAAB (1) Mixed (1)
Genderless (2) Neuter / Neutrois (2)
Genderflux (1) Demigender (1)
Demiflux (1) Agender (1)
Demigirl (1)
GIS_(woman)+(other) “Woman who is trans” (1)
Genderqueer (4) Genderfluid (3) “Woman of transsexual experience” (1)
Transgender / Trans (3) Non-binary (3) “Gender is how you relate the world to emotional self” (1)
Female (3) Femme (2) “Born in the wrong body” (1)
Transfeminine (2) Feminine (1) “Gender is a true sense of one's heart mind body and soul”
Transwoman (1) Demigirl (1) (1)
Agender (1) “An array of emotions” (1)

“I dont (sic) like talking about gender in general” GIS_man+(other)

(1) Genderqueer (5) Non-binary (5)
“From early childhood i (sic) have identified as a Male / man (2) Trans man/boy (2)
snail as they are intersex by nature” (1) Trans masculine (3) Non-binary boy/boi (1)
Transsexual male (1)
GIS_woman+(other) GIS_none
Trans/transgender Female/Woman (2) Agender (4) Genderqueer (2)
woman/female (7) Transfemme (1) Non-binary (2) Genderfluid (1)
Demisexual (1) Genderfluid (1) Null gender (1) Male (1)
Genderqueer (1) “A bit of both” (1)
Androgynous (1) “I don’t” (1)

Figure 6. Summary of terms that participants used to describe their gender and grouped according to latent
class analysis class. Participants may have used more than one term each.

have the same gender may respond differently to the GIS based on tify highly with that gender. An alternative is that the participant
differences in their belief that the scores need to sum to 100. identifies as almost genderless and they identify highly with being
If participants view gender as having a fixed quantity (of 100%), genderless. This is a misinterpretation of the other gender(s) di-
then they may also misinterpret the other gender(s) dimension. For mension as a filler category, but this dimension represents a gender
example, examining the response of “genderqueer/neutrois/agender” that exists outside of the male/female spectrum.
where the participant has responded 0 to male/man/boy, 10 to The creators of the Gender Unicorn did not intend these
female/woman/girl, and 90 to other gender(s), there are two pos- limitations, as evidenced by an example they provide where
sible explanations or reasons for that response. The first is that the multiple dimensions are marked highly. The dimensions on the
participant interprets neutrois and agender to mean a neutral original infographic are not numbered nor anchored with labels.
gender that exists outside of the traditional binary, and they iden- We added numbers and anchors in line with best practice for
 12 HO AND MUSSAP

Table 3 assigned at birth (Conron, Landers, Reisner, & Sell, 2014), which
Results of a Binomial Test to Determine if the Proportion of means that some trans or gender diverse participants may not have
Umbrella Term Usage is Greater Than .001 for Each Class been accounted for. More importantly however, they used gender
to define the “normative” sample, where normative were the
Class n true n trials Proportion p value

GIS_woman 15 134 .11 ⬍.001

GIS_man 1 57 .02 .06 Table 4
GIS_other 15 21 .71 ⬍.001 Responses to Gender Identity Scale (GIS) and Descriptions of
GIS_(woman) ⫹ (other) 9 17 .53 ⬍.001
GIS_man ⫹ (other) 8 18 .44 ⬍.001 Gender for Participants Who Used Nonbinary But Relatively
GIS_woman ⫹ (other) 1 11 .09 .01 Specific Labels for Descriptions of Their Gender
GIS_none 3 8 .38 ⬍.001
Man Woman Other Description
Note. GIS ⫽ Gender Identity Scale. A “true” for the binomial test is
someone who only used an umbrella term. People who used an umbrella GIS_woman
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

term in combination with a specific term (e.g. genderqueer woman) were 0 100 0 Transgender lesbian
This document is copyrighted by the American Psychological Association or one of its allied publishers.

not counted as “true.” 50 50 0 Bigender

30 90 0 Feminine (tomboyish)
20 80 0 Two spirited
80 40 0 Nonbinary transmasculine
psychological instruments (Preston & Colman, 2000). Our use 20 80 0 Gender fluid, mostly female but
of these particular numbers, however, may be confounding. It residual maleness
would be advisable, in the future, to utilize a straightforward 10 90 0 Mostly female
0 70 0 Androgynous
numbering scheme with a different number of points, such as 0 50 50 0 Bigender
to 8, so that the implication of summing to 100% is not present. 30 70 0 Crossdressing
0 100 0 Elegantly feminine
It may also be useful to explicitly state that participants may 20 80 0 Mostly female
identify highly on more than one dimension, or very low on all 40 60 0 Crossdressing
the dimensions. A short explanation of the other gender(s)
dimension may also be helpful in reducing confusion. GIS_man
The method for querying gender may have been confusing and a 40 10 30 Agender
80 20 0 Transmasculine
potential confound for participants who identify as genderfluid. The 70 20 10 Transmasculine
query did not specify a timeframe and our implementation of each 90 0 20 Masculine
dimension accepted only a single response. Future versions of the 30 10 50 Demiflux or agender
80 0 10 Transmasculine
scale should therefore specify a time point, such as “right now.” 100 10 0 Transmasculine
Future research could also explore the change and diversity of gen-
derfluid identities by allowing participants to designate a range that GIS_other
they identify with. 0 0 80 Genderless or agender
Another limitation of this study is that we only tested the GIS on 10 10 100 Neuter
0 60 40 Nonbinary demigender femme
participants who self-identified as trans or gender diverse. The defi- 0 10 90 Genderqueer/neutrois/agender
nition of cisgender suggests that cisgender people have total and full 10 10 80 Genderqueer/genderless
identification with only the gender that they were assigned at birth 0 60 60 Demigirl
(ACON, 2017). There is some evidence, however, that gender vari- GIS_(woman) ⫹ (other)
ance exists among the (statistically) “normative” population (Joel, 10 60 50 Transfeminine, genderqueer
Tarrasch, Berman, Mukamel, & Ziv, 2014), which calls into question 10 80 30 Nobinary femme
the meaning of the GIS scores if a cisgender person and a trans or 60 50 50 Some days I feel a lot more feminine
than other days
gender diverse person both report the same scores.
10 60 70 Agender/genderfluid/demigirl
A potential explanation might be that their gender identity is the
same but that the difference with self-identification is due to prefer- GIS_man ⫹ (other)
ences with labels, knowledge of terminology, readiness to self- 80 0 100 Transmasculine
identify, or a lack of exploration of their own gender identity. Non- 50 0 70 Transmasculine nonbinary
binary people may, for example identify as cisgender (M. J. Barker & 50 10 70 Neutral
Richards, 2015) and some people, who might arguably have a trans or GIS_woman ⫹ (other)
gender diverse identity, may struggle with being “trans enough” 30 80 50 A bit of both
(Langer, 2011). This poses a philosophical question about the defini- 40 80 70 Transfemme
tion of cisgender, transgender, or gender diverse, and the relationship
between respect for self-determination and research categorisation. It GIS_none
also further reinforces the difficulties of working with labels. 0 0 0 Nope! (Agender/null gender)
0 0 0 Agender
Limitations in the study by Joel et al. (2014) might mean that the 0 0 0 Agender
findings of diversity in “normative” individuals were inflated. 0 0 0 Agender, nonbinary
First, their demographic measure enquired about sex rather than Note. Man ⫽ male/man/boy dimension; Woman ⫽ the female/woman/girl
sex assigned at birth. Sex can be interpreted differently by trans or dimension; Other ⫽ other gender(s) dimension; GIS ⫽ Gender Identity Scale.
gender diverse people and is not necessarily equivalent to sex Participants Are Grouped by the Assigned Class from the latent class analysis.
 GENDER IDENTITY SCALE 13

participants who identified as man or woman and not as transgen- processes: Facilitative and avoidant coping throughout gender transition-
der or other. Many people who might be conceptualized as trans or ing. The Counseling Psychologist, 41, 601– 647. http://dx.doi.org/10
gender diverse do not self-identify in that way and only identify as .1177/0011000011432753
man or woman (Ansara & Hegarty, 2014), which was also re- Cisnormativity. (2011). Gingerbread person. Retrieved from http://
flected in the terminology used by participants in this present cisnormativity.tumblr.com/post/9652156368/gingerbread-person-this-
research of the GIS. Reliance on that self-identification may also is-an-anthropomorphic
Connelly, L. M. (2013). Demographic data in research studies. Medsurg
have included many trans or gender diverse participants, which
Nursing, 22, 269 –270.
would increase the diversity in the “normative” sample. They also
Conron, K., Landers, S., Reisner, S., & Sell, R. (2014). Sex and Gender in
operationalised gender identity as a frequency of how often some- the US Health Surveillance System: A Call to Action. American Journal
one thought of themself as a particular gender. Frequency and Of Public Health, 104, 970 –976. http://dx.doi.org/10.2105/ajph.2013
intensity are different components in affect (Schimmack & Diener, .301831
1997), so frequency in thinking about ones gender may be different Couch, M., Pitts, M., Mulcare, H., Croy, S., Mitchell, A., & Patel, S.
(albeit related) to enquiring about level of identity. Having pro- (2007). tranZnation: A report on the health and wellbeing of transgen-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

vided evidence for validity of the GIS within the TGD population, der people in Australia and New Zealand. Melbourne: Australian Re-
This document is copyrighted by the American Psychological Association or one of its allied publishers.

future research could therefore replicate the study by Joel et al. search Centre in Sex, Health & Society (ARCSHS). Retrieved from
while addressing some of the limitations. Importantly, the GIS http://arrow.latrobe.edu.au:8080/vital/access/manager/Repository/
allows gender to be conceptualized in a way that acknowledges latrobe:23974
diversity as requested by Joel et al. Davies, R. S. (2008). Designing a response scale to improve average group
response reliability. Evaluation and Research in Education, 21, 134 –
146. http://dx.doi.org/10.1080/09500790802152209
Conclusion Diamond, M. (2002). Sex and gender are different: Sexual identity and
gender identity are different. Clinical Child Psychology and Psychiatry,
The results of our study highlight the potential of the GIS as a
7, 320 –334. http://dx.doi.org/10.1177/1359104502007003002
measure of gender identity that allows researchers to collect gen-
Fraley, C., Raftery, A. E., Murphy, T. B., & Scrucca, L. (2012). mclust
der identity information from participants in a manner that is not Version 4 for R: Normal mixture modeling for model-based clustering,
reliant on labels and that does not misgender them. Researchers classification, and density estimation (Tech. Rep. No. 597). Seattle:
can present the diversity and range of gender identities of partic- University of Washington.
ipants without needing to summarize lists of self-designated labels. Gender identities. (2015). Gender Identities. Retrieved from http://gender
There is also the potential to use the GIS in classifying participants .wikia.com/wiki/Category:Gender_Identities
for quantitative research, but further research would need to be Gender Spectrum. (n.d.). Understanding gender. Retrieved from https://
conducted if a standard set of classifications were desired. How- www.genderspectrum.org/quick-links/understanding-gender/
ever, there was also evidence that the numerical anchors of the GLAAD. (2015). GLAAD media reference guide: Transgender issues.
scale used in this study and the interpretation of the “other gen- Retrieved from http://www.glaad.org/reference/transgender
der(s) scale” introduces potential confounds, but these could be gqid. (2015). Genderqueer and non-binary identities & terminology. Re-
addressed with some straightforward changes in subsequent iter- trieved from http://genderqueerid.com/gq-terms
ations of the GIS. Grant, J. M., Mottet, L. A., Tanis, J., Harrison, J., Herman, J. L., &
Keisling, M. (2011). Injustice at every turn: A report of the National
Transgender Discrimination Survey. Washington, DC: National Center
References for Transgender Equality and National Gay and Lesbian Task Force.
Harrison, J., Grant, J., & Herman, J. (2012). A gender not listed here:
American Psychological Association. (2010). Publication manual of the Genderqueers, gender rebels, and otherwise in the National Transgender
American Psychological Association (6th ed.). Washington, DC: Author.
Discrimination Survey. LGBTQ Public Policy Journal at the Harvard
American Psychological Association. (2015). Guidelines for psychological
Kennedy School, 2, 11–24. Retrieved from http://escholarship.org/uc/
practice with transgender and gender nonconforming people. American
item/2zj46213.pdf
Psychologist, 70, 832– 864. http://dx.doi.org/10.1037/a0039906
Ho, F., & Mussap, A. J. (2017). Transgender Mental Health in Australia:
Ansara, Y. G., & Hegarty, P. (2012). Cisgenderism in psychology: Patholo-
Satisfaction with Practitioners and the Standards of Care. Australian
gizing and misgendering children from 1999 to 2008. Psychology and
Psychologist, 52, 209 –218.
Sexuality, 3, 137–160. http://dx.doi.org/10.1080/19419899.2011.576696
Hyde, Z., Doherty, M., Tilley, P. J. M., McCaul, K., Rooney, R., & Jancey,
Ansara, Y. G., & Hegarty, P. (2014). Methodologies of misgendering:
Recommendations for reducing cisgenderism in psychological research. J. (2014). The first Australian national trans mental health study:
Feminism & Psychology, 24, 259 –270. http://dx.doi.org/10.1177/ Summary of results. Perth, Australia: School of Public Health, Curtin
0959353514526217 University. Retrieved from https://www.beyondblue.org.au/docs/
Barker, M. J., & Richards, C. (2015). Further genders. In C. Richards & default-source/research-project-files/bw0274.pdf
M. J. Barker (Eds.), The Palgrave handbook of the psychology of Joel, D., Tarrasch, R., Berman, Z., Mukamel, M., & Ziv, E. (2014).
sexuality and gender (pp. 166 –182). London, UK: Palgrave Macmillan. Queering gender: studying gender identity in ‘normative’ individuals.
http://dx.doi.org/10.1057/9781137345899_11 Psychology & Sexuality, 5, 291–321. http://dx.doi.org/10.1080/
Brown, L. B. (2016). What’s in a name? Examining the creation and use of 19419899.2013.830640
sexual orientation and gender identity labels. Issues in Religion and Johnson, J. L., Greaves, L., & Repta, R. (2009). Better science with sex and
Psychotherapy, 37, 23–32. Retrieved from https://ojs.lib.byu.edu/spc/ gender: Facilitating the use of a sex and gender-based analysis in health
index.php/IssuesInReligionAndPsychotherapy/article/download/36729/ research. International Journal for Equity in Health, 8, 14. http://dx.doi
34463 .org/10.1186/1475-9276-8-14
Budge, S. L., Katz-Wise, S. L., Tebbe, E. N., Howard, K. A. S., Schneider, Killerman, S. (2012a). The genderbread person. Retrieved from http://
C. L., & Rodriguez, A. (2013). Transgender emotional and coping itspronouncedmetrosexual.com/2012/01/the-genderbread-person/
 14 HO AND MUSSAP

Killerman, S. (2012b). The genderbread person v2.0. Retrieved from Runnels, V., Tudiver, S., Doull, M., & Boscoe, M. (2014). The challenges of
http://itspronouncedmetrosexual.com/2012/03/the-genderbread-person- including sex/gender analysis in systematic reviews: A qualitative survey.
v2-0/ Systematic Reviews, 3, 33. http://dx.doi.org/10.1186/2046-4053-3-33
Langer, S. (2011). Gender (Dis)Agreement: A Dialogue on the Clinical Schimmack, U., & Diener, E. (1997). Affect intensity: Separating intensity
Implications of Gendered Language. Journal Of Gay & Lesbian Mental and frequency in repeatedly measured affect. Journal Of Personality
Health, 15, 300 –307. http://dx.doi.org/10.1080/19359705.2011.581194 And Social Psychology, 73, 1313–1329. http://dx.doi.org/10.1037/0022-
Lawson, B. (2011). Genderbread man. Retrieved from http://brucel.tumblr 3514.73.6.1313
.com/post/7303294235/genderbread-man Serano, J. (2007). Whipping girl: A transsexual woman on sexism and the
Levitt, H. M., & Ippolito, M. R. (2014). Being transgender: The experience scapegoating of femininity. Emeryville, CA: Seal Press.
of transgender identity development. Journal of Homosexuality, 61,
ShineSA. (2017). Trans wellbeing definitions. Retrieved from https://www
1727–1758. http://dx.doi.org/10.1080/00918369.2014.951262
.shinesa.org.au/media/2017/03/Trans-Wellbeing-Definitions.pdf
Micah. (2016). What is neutrois? Retrieved from http://neutrois.com/what-
Solotke, M., Sitkin, N. A., Schwartz, M. L., & Encandela, J. A. (2017).
is-neutrois/
Twelve tips for incorporating and teaching sexual and gender minority
Muehlenhard, C. L., & Peterson, Z. D. (2011). Distinguishing Between Sex
health in medical school curricula. Medical Teacher. Advance online
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

and Gender: History, Current Conceptualizations, and Implications. Sex

Roles: A Journal of Research, 64, 791– 803. http://dx.doi.org/10.1007/ publication. http://dx.doi.org/10.1080/0142159X.2017.1407867
This document is copyrighted by the American Psychological Association or one of its allied publishers.

s11199-011-9932-5 Tate, C. C., Ledbetter, J. N., & Youssef, C. P. (2013). A two-question

Oberski, D. (2016). Mixture models: Latent Profile and latent class anal- method for assessing gender categories in the social and medical sci-
ysis. In J. Robertson & M. Kaptein (Eds.), Modern statistical methods ences. Journal of Sex Research, 50, 767–776. http://dx.doi.org/10.1080/
for HCI (pp. 275–287). Cham, Switzerland: Springer International Pub- 00224499.2012.690110
lishing. http://dx.doi.org/10.1007/978-3-319-26633-6_12 Tein, J.-Y., Coxe, S., & Cham, H. (2013). Statistical power to detect the
Pan, L., & Moore, A. (2014). The Gender Unicorn. Retrieved from correct number of classes in latent profile analysis. Structural Equa-
http://www.transstudent.org/gender tion Modeling, 20, 640 – 657. http://dx.doi.org/10.1080/10705511
Pinn, V. W. (2003). Sex and gender factors in medical studies: Implications .2013.824781
for health and clinical practice. Journal of the American Medical Asso- The GenIUSS Group. (2014). Best practices for asking questions to iden-
ciation, 289, 397– 400. http://dx.doi.org/10.1001/jama.289.4.397 tify transgender and other gender minority respondents on population-
Preston, C. C., & Colman, A. M. (2000). Optimal number of response based surveys. Los Angeles, CA: The Williams Institute. Retrieved from
categories in rating scales: Reliability, validity, discriminating power, http://williamsinstitute.law.ucla.edu/wp-content/uploads/geniuss-report-
and respondent preferences. Acta Psychologica, 104, 1–15. http://dx.doi sep-2014.pdf
.org/10.1016/S0001-6918(99)00050-5 TransGender Victoria. (2013). Definitions. Retrieved from http://www
R Core Team. (2016). R: A language and environment for statistical
.transgendervictoria.com/about/definitions
computing. Vienna, Austria: R Foundation for Statistical Computing.
Treharne, G. J. (2011). Questioning sex/gender and sexuality: Reflections
Retrieved from https://www.r-project.org/
on recruitment and stratification. Gay & Lesbian Issues and Psychology
Revelle, W. (2016). psych: Procedures for psychological, psychometric,
Review, 7, 132–154.
and personality research. Evanston, IL: Northwestern University. Re-
trieved from https://cran.r-project.org/package⫽psych West, C., & Zimmermn, D. H. (1987). Doing Gender. Gender & Society,
Riggs, D. W., & Due, C. (2013). Gender identity Australia: The healthcare 1, 125–151. http://dx.doi.org/10.1177/0891243287001002002
experiences of people whose gender identity differs from that expected of Westbrook, L., & Saperstein, A. (2015). New categories are not enough:
their natally assigned sex. Adelaide, Australia: Flinders University. Rethinking the measurement of sex and gender in social surveys. Gender
Riggs, D. W., Power, J., & von Doussa, H. (2016). Parenting and Austra- & Society, 29, 534 –560. http://dx.doi.org/10.1177/0891243215584758
lian trans and gender diverse people: An exploratory survey. Interna- Wood, W., & Eagly, A. H. (2009). Gender identity. In M. R. Leary & R. H.
tional Journal of Transgenderism, 17, 59 – 65. http://dx.doi.org/10.1080/ Hoyle (Eds.), Handbook of individual differences in social behavior (pp.
15532739.2016.1149539 109 –125). New York, NY: Guilford Press.

(Appendix follows)
 This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

View publication stats

Appendix
GENDER IDENTITY SCALE

The Gender Identity Scale

Received March 29, 2018

Revision received October 29, 2018
Accepted October 31, 2018 䡲
15