-I fielded the variety of experimental measures with

Analyzing experimental YouGov UK to their online panel representative of the UK

general population.
survey data of new -I expect there to be statistically significant differences
between the test measures and the controls. However, I
measures of sexuality and suspect that the difference will be more complicated among

gender identity in the Measures of sexual orientation and gender the gender identity measures.
-I test for significant differences by examining the margin
of error in distribution of identities and analyzing the
United Kingdom. results of logistic regression models.

Introduction
-Less than 3% of research on LGBTQ+ political issues
identity are best constructed when LGBTQ+ Table 1 Sexuality Test Measures by Sexual Orientation Category

Asexual
Test 1 Test 2 Test 3 Control Total

people are included in the design process.

(little or no sexual
1.6% 0.8% 1.7% 1.0%
-
involves a researcher surveying or including an openly attraction to anyone) (32) (14) (29) (75)

LGBTQ+ person. Bisexual

(sexual attraction to 3.8% 3.5% 2.6% 2.4%
-Some cite a lack of data availability for the disparity in more than one
gender) (76) (62) (44)
-
(182)

research, but I argue that data availability is a symptom Gay or

of a deeper systemic problem causing queer exclusion in lesbian
(homosexual/attracted
3.4% 3.0% 2.9%
-
2.3%
(69) (53) (50) (172)
political science: cis-heteronormativity. to people of the same
gender)

Cis-heteronormativity and Survey Measures

-Cis-heteronormativity precludes the very study of
In the U.K., the most effective In the U.K., the most effective Pansexual
(sexual attraction to
people regardless of
their gender)
1.0%
(21)
1.0%
(17)
1.3%
(22)
-
0.8%
(60)

measure of sexual orientation measure of gender identity is

Queer
LGBTQ+ people by centering straight and cisgender (fluidly attracted to
people of different 0.5% 0.2% 0.4% 0.3%
paradigms. genders in a way not
defined by other
labels)
(11) (3) (7)
-
(21)

-In research methods, cis-heteronormativity can present

itself by homogenizing queer people and experiences. In
surveys, this manifests by the absence of measures that is an inclusive list with an inclusive list with a follow- Straight
(heterosexual/sexuall
y attracted to people
of the opposite sex)
84.5%
(1696)
83.6%
(1483)
83.1%
(1413)
85.9%
(1684)
84.3%
(6276)

count and code LGBTQ+ people OR by using poorly

definitions: up question on trans identity:
LGB
(Lesbian, gay, 10.3% 8.5% 8.9% 6.6% 1.7%
constructed measures of sexuality and gender identity. bisexual, denoted by
the coding schema
used in the control)
(209) (149) (152) (130) (130)
- Examples of cis-heteronormative measures are the Other 0.3% (7) 1.1% (20) 0.7% (21) 1.3% (25) 0.9% (64)
Understanding Society Survey’s sexuality measure, Don’t Know 1.0% (21) 1.4% (25) 1.2% (21) 2.0% (39) 1.4% (106)
3.6% (73) 5.5% (97) 6.1% (103) 4.2% (82) 4.8% (355)
which only allows respondents to choose between Prefer not to say
100.0% 100.0% 100.0% 100.0% 100.0%
Total
identifying as “straight” or as “Lesbian, gay or
Which one of the following, if any, best What is your gender?
(2006) (1774) (1701) (1960) (7441)
Chi Square: 602.267 p< 0.001 n=7441
bisexual.” This homogenizes the variety of sexualities as
MOE= +/-1.14%
only being interesting in relation to heterosexuality. The
American National Election Study’s gender measure is describes your sexual orientation? Table 2 Binary Logistic Regression for Sexuality Measures on predicting Straight or
another example, where respondents can choose
between “male,” “female,” and “other.” Here, diverse 1. Gender-nonconforming Not Straight Identity

gender identities are obscured and deemed too

uninteresting to count properly.
1. Asexual (little or no sexual attraction to 2. Man Constant
beta (s.e.)
-2.561 (0.091)
Exp(B)
0.077***

-Cis-heteronormative measures persist because we

understandably recycle measures to save time and
anyone) 3. Nonbinary Test 1 0.468 (0.117) 1.596***

money. Additionally, nearly all measures of gender and

sexuality were constructed without input or engagement
2. Bisexual (sexual attraction to more than 4. Queer Test 2

Test 3
0.264 (0.125)

0.332 (0.125)
1.301*

1.393**
with LGBTQ+ people. The presence of cis-
heteronormative measures could lead to response biases
one gender) 5. Woman pseudo r-square: 0.005*** p= >0.05* >0.01** >0.001***

from LGBTQ+ people and other data quality problems.

-In a series of focus group interviews conducted with
3. Gay or lesbian (homosexual/attracted to 6. Other Table 3 Gender Identity Test Measures by Cis or Not with Uncertainty
Cisgender Not Cisgender Uncertain Total
LGBTQ+ people in the U.K. I discovered that many people of the same gender) 7. Don’t know Test 1
93.2%
(1555)
2.0%
(33)
4.9%
(81)
100%
(1669)
LGBTQ+ people stop taking a survey when they
encounter a poorly constructed gender identity or 4. Pansexual (sexual attraction to people 8. Prefer not to say Test 2
94.0%
(1590)
1.4%
(24)
4.6%
(78)
100%
(1692)
sexuality measure, or they answer dishonestly. 94.6% 1.8% 3.6% 100%
-In these focus group interviews, I also asked LGBTQ+ regardless of their gender) Test 3
(1576)
93.6%
(30)
2.3%
(60)
4.0%
(1666)
100%
people to evaluate current measures of gender identity
and sexuality, and to make suggestions about how 5. Queer (fluidly attracted to people of Would you consider yourself trans? Test 4
(1605)
95.2%
(40)
2.0%
(69)
2.8%
(1714)
100%
gender identity and sexuality measures ought to be Test 5
constructed. different genders in a way not defined by Test 6
(1645)
94.0%
(34)
1.4%
(49)
4.6%
(1728)
100%
(1571) (23) (77) (1671)
Experimenting with Sexuality and Gender Identity other labels) 1. Yes Control
95.6% 0.7% 3.7% 100%
-This poster evaluates a variety of potential measures of (1591) (11) (62) (1664)

gender and sexuality that were constructed from 6. Straight (heterosexual/sexually attracted to 2. No Total
94.3% 1.7% 4.0% 100%
(11133) (195) (476) (11804)
previous focus group interviews.
-There were three experimental sexuality measures people of the opposite sex) 3. Don’t know Chi Square: 32.032 p= 0.001 n=11804

7. Other 4. Prefer not to say

compared to one control. MOE= +/- 0.90%

-There were six experimental gender identity compared Table 4 Binary Logistic Regression for Gender Identity Measures on predicting
to one control…(cont. on opposite side)
8. Don’t know Cisgender or Not Cisgender Identity

beta (s.e.) Exp(B)

Scan
the QR
9. Prefer not to say Constant
Test 1
-4.974 (0.303)
1.121 (0.350)
0.007***
3.069***

code for Test 2 0.781 (0.366) 2.183*

an online Test 3 1.013 (0.354) 2.753**
version of
this poster Nate Roundy, PhD Candidate Test 4

Test 5
1.282 (0.342)

1.095 (0.349)
3.605***

2.989**
and to read
the draft Department of Politics and International Studies Test 6 0.750 (0.303) 0.007***

paper. pseudo r-square: 0.012** p= >0.05* >0.01** >0.001***