European Child & Adolescent Psychiatry

https://doi.org/10.1007/s00787-019-01319-3

ORIGINAL CONTRIBUTION

Separation anxiety and gender variance in a community sample

of children
Alanna Santarossa1 · A. Natisha Nabbijohn1 · Anna I. R. van der Miesen2 · Diana E. Peragine1 · Doug P. VanderLaan1,3

Received: 21 June 2018 / Accepted: 18 March 2019

© Springer-Verlag GmbH Germany, part of Springer Nature 2019

Abstract
In clinical child and retrospective adult samples, childhood gender variance (GV; i.e., cross-gender behaviour) has been
associated with separation anxiety (SA; i.e., distress related to separation from attachment figures) in males. This study
examined GV and SA in a nonclinical sample of 892 boys and 933 girls aged 6–12 years via parent-reports. Parental factors
(i.e., parenting style, parent–child relationship, willingness to serve as an attachment figure, attitudes towards gender ste-
reotypes in children) were examined as potential moderators. GV predicted SA in boys, even when statistically controlling
for general psychopathology and demographic variables. Authoritative parenting, closeness in the parent–child relation-
ship, willingness to serve as an attachment figure, and liberal attitudes towards gender stereotypes in children moderated the
association between GV and SA in both boys and girls. Thus, SA may be a unique internalizing problem related to GV in
boys in nonclinical samples and influenced by a variety of parental factors.

Keywords  Separation anxiety · Gender variance · Children · Parenting styles · Parental attitudes · Parent–child relationship

Introduction of opposite sex” and “behaves like opposite sex” items from
the Child Behaviour Checklist (CBCL) [4–6], or, less often,
In children, gender variance (GV) is characterized by gen- measures that are psychometrically validated and compre-
der-“atypical” behaviour and interests in domains such as hensive in terms of capturing the various domains of GV [7].
toy, activity, playmate, self-identification, and clothing GV can be considered comparable, although not equiva-
and hairstyle preferences [1]. Prior literature has employed lent, to gender dysphoria (GD) [1, 7]. GD, formerly gender
a variety of measures to examine GV and its correlates in identity disorder (GID), is a clinical diagnosis defined by
nonclinical community child samples. Relatively few studies marked GV that is accompanied by distress related to an
have employed self-report measures designed to gauge chil- incongruence between one’s birth-assigned vs. experienced
dren’s gender expression by asking them to rate, for exam- gender identity [8]. Among children clinic-referred for GD/
ple, their felt gender-“(a)typicality” and contentment with GID, previous research indicated elevated emotional and
gender assignment [2], or their perceived similarity to other behavioural problems on screening measures such as the
boys and girls in domains such as activity, playmate, and CBCL [4] (for review see [9]). A study employing a struc-
appearance preferences [3]. More commonly, GV is meas- tured parent-report diagnostic interview schedule based on
ured via parent-report using, for example, the “wishes to be the Diagnostic and Statistical Manual of Mental Disorders
(DSM) found that 23% experienced externalizing problems
and 37% experienced internalizing problems, with 31% spe-
* Doug P. VanderLaan
[email protected] cifically experiencing anxiety [10].
Similarly, GV in nonclinical community samples of chil-
1
Department of Psychology, University of Toronto dren and youth has been associated with lower psychological
Mississauga, Mississauga, ON L5L 1C6, Canada well-being, as measured by parent-reports [6, 7] and child
2
Department of Child and Adolescent Psychiatry, Center and youth self-reports (e.g., [2, 3, 11]). van Beijsterveldt
of Expertise on Gender Dysphoria, VU University Medical et  al. [6] examined twins where GV (i.e., cross-gender
Center, Amsterdam, The Netherlands
behaviour and wishes) and behavioural and emotional prob-
3
Child and Youth Psychiatry, Centre for Addiction and Mental lems were reported by parents using the CBCL. In boys and
Health, Toronto, ON, Canada

13
Vol.:(0123456789)
 European Child & Adolescent Psychiatry

girls, GV was associated positively with internalizing and Most recently, VanderLaan et al. [14] found elevated
externalizing problems. van der Miesen et al. [7] examined SA among gender-referred birth-assigned male children
both the CBCL cross-sex wishes item and the Gender Iden- using liberal and conservative criteria as in Zucker et al.
tity Questionnaire for Children (GIQC) [12], a more com- [16]. Specifically, 55.8% met the liberal criterion and 5.3%
prehensive measure of GV, to predict clinical-range CBCL met the conservative criterion for SAD; these rates were
problems; the CBCL cross-sex wishes item predicted a significantly greater than the population prevalence rate of
greater rate of clinical-range externalizing problems in SAD in boys. Additionally, a parent-report measure of GV
girls, and increases in GV on the GIQC were associated was associated with elevated SA; however, this association
with increased rates of clinical-range total, internalizing, and appeared to overlap with the general pattern of elevated
externalizing problems in boys. Similarly, using child self- internalizing problems and peer problems (i.e., the GV-SA
reports, GV and pressure to conform to gender norms were association was no longer significant when internalizing
associated with elevated internalizing problems in children problems and peer problems were statistically controlled).
[2], and perceived similarity to children of the opposite- As such, VanderLaan et al.’s [14] findings suggested that GV
birth-sex was associated with poorer social adjustment and is commonly associated with elevated SA in gender-referred
lower global self-esteem [3]. In LGBTQ adolescents and birth-assigned male children, but it also casts possible doubt
young adults, GV was associated with lower psychological on whether SA is uniquely related to GV relative to other
well-being [11]. internalizing problems.
Although numerous studies indicated GD and GV are In addition, retrospective studies of males who exhibited
associated with elevated behavioural and emotional prob- marked femininity as children and androphilia (i.e., sexual
lems in children and youth, a handful of studies spanning attraction to men) as adults are relevant. Such studies are
several decades have raised the possibility that separation relevant because longitudinal research has shown the major-
anxiety (SA) is a unique internalizing problem among chil- ity of children with GD report same-sex attraction in adoles-
dren showing GV. Separation anxiety disorder (SAD) is cence and adulthood [17]. Similarly, in nonclinical samples,
defined as distress related to separation from an attachment childhood GV is associated with same-sex sexual orientation
figure, typically a parent [8], and has a population preva- in adolescence and adulthood [18, 19]. Given that children
lence rate estimated at 4.1% in North America [13], with a who exhibit GD or GV appear more likely to be same-sex
higher prevalence rate in girls (5.64% in girls and 2.56% in attracted later in development, same-sex attracted adults in
boys; for details, see [14]). Two earlier studies focusing on the general population who showed marked GV in child-
SA in specialty child gender identity services reported that hood may provide insight regarding whether GV and SA are
SA is common among birth-assigned males with a clinical associated beyond clinical samples.
diagnosis of GID [15, 16]. Coates and Person [15] reported To date, such retrospective studies have been cross-cul-
that among 25 birth-assigned male children diagnosed with tural in scope and focused on non-Western birth-assigned
GID, 60% also met the diagnostic criteria for SAD. Simi- males who are markedly feminine members of ‘third’ gender
larly, Zucker et al. [16] found that birth-assigned male chil- categories distinct from ‘men’ and ‘women’ as well as on
dren who met the complete diagnostic criteria for GID had Canadian men and women of varying sexual orientations.
elevated levels of SA compared to those who did not meet Vasey et al. [20] reported that members of the Samoan third
the complete diagnostic criteria for GID. Using a liberal gender, fa’afafine, recalled more childhood SA than Samoan
criterion for defining SAD (i.e., parental endorsement of men and women. Similarly, Gómez et al. [21] reported that
“Sometimes” or “Yes” to at least three criteria for diagnosis members of the Mexican Istmo Zapotec third gender, muxes,
of SAD), 64.4% diagnosed with GID met the criteria for recalled more SA in childhood than men, but similar levels
SAD, whereas only 38.1% subthreshold for a GID diagno- of recalled SA relative to women. Likewise, in a Canadian
sis met the criteria for SAD. Thus, the more gender-variant sample, VanderLaan et al. [22] found that gay men who
birth-assigned male children (i.e., those who met criteria for
GID diagnosis) had elevated SA. In another clinical study,
Wallien et al. [10] reported that in a sample of Dutch chil-
Footnote 1 (continued)
dren diagnosed with GID, 5.8% of birth-assigned males and
test to compare Wallien et  al.’s [10] sample to the North American
5.9% of birth-assigned females met the criteria for SAD; this
population estimates of SAD rates among girls (5.64%) and boys
rate of SAD among the Dutch birth-assigned males with (2.56%), respectively [13]. The rate of SAD among the Dutch birth-
GID could also be considered to be elevated if compared to assigned males with GID (5.8%) was marginally elevated compared
the rate for North American boys cited above.1 to North American boys, z = 1.92, two-tailed p = 0.055. This elevation
may similarly suggest SAD is elevated among birth-assigned males
1
  Wallien et al. [10] did not report a control sample to compare rates who experience GID/GD. None of the other Dutch GID group com-
of diagnoses, nor are there data available on the prevalence of SAD parisons to these North American estimates were statistically or mar-
among boys and girls in the Dutch population. We used the binomial ginally significant, all z < 1.23, all two-tailed p > 0.22.

13
European Child & Adolescent Psychiatry

recalled more childhood GV also recalled significantly and youth. Alanko et al. [31] reported a retrospective study
more childhood SA compared to heterosexual men, but not of Finnish adults contacted through a national twin regis-
compared to women. VanderLaan et al. [22] found no asso- try. Participants reported on their recalled childhood GV
ciation between recalled childhood GV and SA in women; and their parents’ parenting styles, which were examined as
however, another Canadian study by Petterson et al. [23] two factors labelled “coldness” and “over-controlling”. The
found that lesbian women, who also tend to recall elevated study found an association between recalled childhood GV
childhood GV [24], reported the highest level of recalled and adulthood psychopathology (i.e., depression, anxiety,
childhood SA compared to gay men and heterosexual men somatization) that was moderated by over-controlling par-
and women, providing some support for a possible GV-SA enting in both men and women, and moderated by parental
association in girls. Additionally, Petterson et al. [23] factor coldness in men only.
analyzed DSM-based internalizing symptoms in a sample Parental acceptance of cross-gender behaviour and felt
of heterosexual and homosexual men and women and found pressure to conform to gender norms also appears to be
evidence of separate factors for recalled childhood SA vs. important in the psychological well-being of gender-variant
depression and anxiety more generally. Thus, this retrospec- children [2, 32]. van Beusekom et al. [32] examined a sam-
tive literature suggests GV and SA are associated in males ple of Dutch secondary school students who reported, by
beyond clinical settings as well as cross-culturally. The Pet- means of a paper questionnaire, on their level of same-sex
terson et al. [23] study in particular indicates SA may also attraction, level of GV, psychological distress, and perceived
be associated with increased GV in girls, and that SA may parental acceptance. Perceived paternal acceptance moder-
be unique relative to other internalizing problems. ated the association between GV and psychological distress
These clinical and nonclinical, cross-cultural literatures in boys; however, maternal acceptance moderated the associ-
raise the question of what factor(s) might underpin the asso- ation between same-sex attraction and psychological distress
ciation between GV and SA in birth-assigned male chil- in girls [32]. Similarly, Yunger et al. [2] conducted a longitu-
dren—and possibly birth-assigned female children. Given dinal study on children in grades three to seven enrolled in a
that SA is germane to the child–parent relationship, it seems state school. The children self-reported on several measures
plausible that parental factors might play some role. Parental including their gender typicality and felt pressure to conform
factors such as psychological flexibility in parenting, adap- to gender norms from parents, peers and themselves. Felt
tive parenting [25] and satisfaction in the parent–child rela- pressure to conform to gender norms was associated with
tionship [26] have been associated with general psychologi- increased internalizing problems the following year, with the
cal well-being among children and youth. A meta-analysis largest impact being on the more gender-variant youth [2].
reported that higher child anxiety is associated with parental Indeed, stigmatization in general appears to mediate the rela-
rejection and greater parental control, especially low levels tionship between GV and psychological well-being in youth
of autonomy granting and excessive over-involvement, while [11] while it has been suggested that therapeutic strategies
parenting that is warm and low on withdrawal and aversive- centering on increased parental acceptance might be impor-
ness is associated with less child anxiety [27]. Literature tant for decreasing anxiety in youth experiencing GD [33].
reviews on childhood SA in particular have suggested that A parent who provides a secure base facilitates the devel-
parental influences include parental rejection and criticism, opment of secure attachment in childhood [34] and, thus,
overprotective parenting, controlling and intrusive parent- can be regarded as key for a secure attachment to develop.
ing, family conflict, general parental distress, and parental Insecure attachment is correlated with psychopathology in
internalizing problems [28, 29]. Children who exhibit GV children, particularly anxiety [35, 36], and has been nega-
experience elevated mental health risk [2, 6, 11] and often tively associated with gender typicality and gender content-
face additional stressors such as peer rejection and ostra- edness as well as positively associated with felt pressure
cism [30], further highlighting the importance of consid- for gender differentiation [37]. The degree of parent–child
ering parental factors, which might serve as protective or closeness also affects psychological well-being; a negative
risk factors in relation to SA among gender-variant children. parent–child relationship has been associated with elevated
Key parental factors considered here are those implicated in adulthood traits of anxiety and depression [38], parental
the reviews of the child literature on anxiety as well as in closeness has been identified as a protective factor lowering
prior literature relevant to childhood GV and psychological suicidality in boys and girls [39], and greater communica-
well-being: parenting styles, parental gender role attitudes, tion between mothers and sons is associated with reduced
parental willingness to provide a secure base for the child, anxiety in the child [40]. In line with these findings, birth-
and closeness in parent–child relations. assigned females who experience GD receive less social
Regarding parenting styles, childhood GV, and psycho- support compared to controls [41] and, thus, fostering a
logical well-being, insights have been provided by retrospec- positive parent–child relationship characterized by closeness
tive research with adults as well as by studies of children and support has been emphasized as important to promote

13
 European Child & Adolescent Psychiatry

psychological well-being in youth who experience GD [42]. serve as a secure base, or the quality of the parent–child
As such, a secure parental base and a close parent–child rela- relationship. Determining which parental factors allevi-
tionship may ameliorate elevated SA among gender-variant ate or contribute to SA among gender-variant children
children. may provide important insight into improving their psy-
Following from the literature described above, the present chological well-being.
study examined GV and SA in a large, nonclinical sample
of children 6–12 years old via parent-reports to address four
issues: Method

1. Can previous clinical findings be extended to a nonclini- Participants

cal child sample? Presently, the only nonclinical stud-
ies in this area were retrospective studies of nonclinical Participants were recruited via online advertisements (i.e.,
adult samples [20–23]. It has been noted that memory Facebook, Kijiji and Craigslist). In addition, various par-
distortion or recall bias is a limitation of such retrospec- ent and child organizations across Canada were contacted
tive research [43]. Thus, examining a nonclinical sample via email and asked to distribute paper and/or online adver-
of children via parent-reports would provide more clar- tisements to their members (e.g., camps for children, par-
ity about whether GV and SA are associated in nonclini- ent support groups, community centres for children). All
cal child populations. Further, doing so would inform advertisements promoted the study as investigating psycho-
whether elevated SA in birth-assigned males with GD logical well-being and gender expression in children. Par-
might be related more to the general pattern of GV they ticipants were required to be over the age of 18 years, the
exhibit, rather than the distress they experience in rela- primary caregiver of a child aged 6–12 years, and proficient
tion to their gender identity. in English. The online survey was hosted on Qualtrics, a
2. Is the association between SA and GV present in girls? private software company. Prior to completing the online
To date, no clinical studies specifically examined the questionnaire, informed consent was obtained from all indi-
association between SA and GD in birth-assigned vidual participants included in the study in accordance with
females and research has been limited in nonclinical procedures approved by the University of Toronto Research
samples. It is possible the pattern for girls parallels that Ethics Board.
found previously for feminine males [14–16, 20–22] The majority of participants were recruited via Face-
and lesbian women [23] such that SA will be elevated book advertisements, comprising 97.27% of the total sam-
in those who show GV. Alternatively, given that SA ple. Overall, the Facebook advertisement campaign reached
appears to be elevated among feminine males and shows 182,887 people, with 6410 people clicking the advertisement
female-biased population prevalence rates [13], gender- and 2457 people completing the survey. Of the remaining
variant girls—who are more masculine—may show less recruitment methods employed, 33 participants (1.31%)
SA. were recruited via Kijiji or Craigslist, 27 (1.07%) were
3. Is SA a unique internalizing problem associated with recruited by word of mouth and 9 (0.36%) were recruited
GV? VanderLaan et  al. [14] examined this question from an organization or local advertisement. Of these 2526
in a clinical sample of gender-referred birth-assigned participants, 2268 completed the information necessary for
males and reported that SA appeared to be part of a data analysis in the present study (i.e., child’s age, gender,
more general pattern of internalizing problems associ- SA, GV). To focus on SA in a nonclinical sample, the pre-
ated with GV. In contrast, Petterson et al. [23] found sent sample was limited to parent-reports of children who
SA belonged to its own factor, with indicators of other had not received any mental health diagnoses. Additionally,
anxiety or depressive disorders belonging to a separate all duplicate entries for the same child were identified and
factor in a recall study of adults of varying sexual ori- removed by comparing multiple entries from the same IP
entations. Given this distinction, SA might be a unique Address on core variables (e.g., age, gender) to determine
internalizing problem among gender-variant children. whether reports from the same IP Address pertained to dif-
To provide further clarity, this study examined whether ferent children. Doing so resulted in a final sample size of
an association between SA and GV existed after statisti- N = 1825 (933 girls; 892 boys).
cally controlling for internalizing problems.
4. Do parental factors moderate the association between SA Measures
and GV in children? Despite the potential relevance of
parental factors, no research has yet examined childhood Descriptive statistics for the demographic variables, accord-
SA and GV in relation to parenting style, parental atti- ing to gender, are shown in Table 1. Descriptive statistics
tudes towards gender stereotypes, parental willingness to for the focal variables, according to gender, are shown in

13
European Child & Adolescent Psychiatry

Table 1  Descriptive statistics Variable Boys Girls Total t or χ2 p ­valuea

for demographic variables
Age (in years) 0.07 0.94
 M 8.69 8.70 8.69
 SD 1.99 1.94 1.97
 n 892 933 1825
School ­performanceb − 2.90 0.004
 M 3.29 3.38 3.34
 SD 0.61 0.58 0.59
 n 799 838 1637
Annual income, n (%) − 6.83 0.23
 < $23,999 87 (9.8) 87 (9.4) 174 (9.6)
 $24,000–$49,999 161 (18.1) 208 (22.4) 369 (20.3)
 $50,000–$79,999 205 (23.1) 205 (22.1) 410 (22.6)
 $80,000–$124,999 249 (28.0) 234 (25.2) 483 (26.6)
 > $125,000 159 (17.9) 158 (17.0) 317 (17.4)
 Unknown 28 (3.1) 37 (4.0) 65 (3.6)
Parent’s marital status, n (%) 0.07 0.79
 Married/common-law 693 (77.7) 720 (77.2) 1413 (77.4)
 Other 199 (22.3) 213 (22.8) 412 (22.6)
Area type, n (%) 3.18 0.20
 Urban 279 (31.5) 298 (32.0) 577 (31.8)
 Suburban 363 (40.9) 408 (43.9) 771 (42.4)
 Rural 245 (27.6) 224 (24.1) 469 (25.8)
Geographic region, n (%) 2.50 0.64
 Ontario 493 (55.4) 527 (56.7) 1020 (56)
 Quebec 31 (3.5) 25 (2.7) 56 (3.1)
 Eastern Canada 88 (9.9) 93 (10) 181 (9.9)
 Western/Northern Canada 278 (31.3) 285 (30.6) 563 (30.9)
Ethnicity, n (%) 7.17 0.62
 European Canadian 509 (57.8) 503 (54.3) 1012 (56.0)
 South Asian 25 (2.8) 20 (2.2) 45 (2.5)
 East Asian 9 (1.0) 16 (1.7) 25 (1.4)
 African 3 (0.3) 3 (0.3) 6 (0.3)
 Latin 6 (0.7) 9 (1.0) 15 (0.8)
 Caribbean 14 (1.6) 10 (1.1) 24 (1.3)
 Aboriginal 38 (4.3) 50 (5.4) 88 (4.9)
 Arab 4 (0.5) 5 (0.5) 9 (0.5)
 Other 108 (12.3) 115 (12.4) 223 (12.3)
 Multiple 165 (18.7) 195 (21.1) 360 (19.9)
Religion, n (%) 5.50 0.79
 Roman Catholic 172 (19.4) 171 (18.3) 343 (18.8)
 Protestant 57 (6.4) 68 (7.3) 125 (6.9)
 Christian 224 (25.2) 233 (25.0) 457 (25.1)
 Muslim 21 (2.4) 27 (2.9) 48 (2.6)
 Jewish 7 (0.8) 13 (1.4) 20 (1.1)
 Buddhist 6 (0.7) 5 (0.5) 11 (0.6)
 Hindu 6 (0.7) 10 (1.1) 16 (0.9)
 Sikh 3 (0.3) 3 (0.3) 6 (0.3)
 No religion 332 (37.4) 354 (37.9) 686 (37.7)
 Other 60 (6.8) 49 (5.3) 109 (6.0)

M mean, SD standard deviation

a
 Results of independent samples t tests (means) or Chi squared tests (counts) comparing the boys and the
girls
b
 School performance is the average numerical level performance (i.e., D = 1 to A = 4) in math, science,
English, social science, and language studies

13
 European Child & Adolescent Psychiatry

Table 2  Descriptive statistics for focal variables

Variable Boys Girls t value p ­value1

Gender variance Mean (SD) 1.96 (0.37) 2.10 (0.40) − 7.68 < 0.001
n 892 933
Separation anxiety Mean (SD) 2.22 (2.29) 2.43 (2.51) − 1.84 0.07
n 892 933
Authoritative parenting style Mean (SD) 4.23 (0.47) 4.20 (0.48) 1.03 0.31
n 769 816
Authoritarian parenting style Mean (SD) 1.64 (0.44) 1.61 (0.41) 1.28 0.20
n 769 816
Permissive parenting style Mean (SD) 2.02 (0.64) 2.03 (0.64) − 0.21 0.83
n 769 816
Conflict in parent–child relationship Mean (SD) 17.23 (6.28) 17.21 (6.35) 0.07 0.94
n 783 831
Closeness in parent–child relationship Mean (SD) 32.21 (3.00) 32.47 (2.85) − 1.80 0.07
n 783 831
Parental willingness to serve as a secure base Mean (SD) 6.04 (0.66) 6.06 (0.63) − 0.59 0.56
n 787 840
Parental attitudes towards gender stereotypes Mean (SD) 85.73 (8.05) 85.84 (8.00) − 0.29 0.77
n 798 850
Peer problems Mean (SD) 0.19 (0.34) 0.15 (0.29) 2.75 0.006
n 879 906
Internalizing problems Mean (SD) 52.56 (10.58) 51.71 (10.52) 1.67 0.09
n 839 882
Externalizing problems Mean (SD) 50.45 (10.51) 49.27 (10.12) 2.37 0.02
n 839 882

Gender variance = GIQC score, separation anxiety = SAI weighted mean sum score, authoritative parenting style = PSDQ-SF authoritative
subscale, authoritarian parenting style = PSDQ-SF authoritarian subscale, permissive parenting style = PSDQ-SF permissive subscale, conflict
in parent–child relationship = CPRS-SF conflict subscale, closeness in parent–child relationship = CPRS-SF closeness subscale, parental will-
ingness to serve as a secure base = CRPR-SF, parental attitudes towards gender stereotypes = CRSRAS, peer problems = CBCL peer problems
subscale, internalizing problems = CBCL internalizing T score¸ externalizing problems = CBCL externalizing T score

Table 2. Independent samples t tests or Chi squared tests Separation anxiety interview (SAI)
were performed to identify any differences between boys
and girls on the demographic (Table 1) and focal variables The SAI [16] is a parent-report measure designed to
(Table 2). assess traits associated with SAD expressed in the last
6–12 months. It is composed of 21 items (e.g., “Does your
Demographic questions child worry in an unrealistic way about something harmful
happening to you?”) categorized into 9 domains that corre-
Parents were asked to report their marital status, the fam- spond to the DSM criteria for SAD (for details on changes
ily’s annual income, the family’s predominant religious to the SAD diagnostic criteria across different versions of
background, and their current geographical location of resi- the DSM, see VanderLaan et al. [14]). The 21 items were
dence in Canada (i.e., Northern Canada, Eastern Canada, answered on a 3-point scale (0 = never, 1 = sometimes, and
Western Canada, Ontario, and Quebec). Questions regarding 2 = always). A mean score was calculated for each of the
the child included their average school performance (i.e., the nine domains (i.e., for each domain, the item scores were
average of reported grade performance, ranging from a letter summed and divided by the total number of items in that
grade of D [coded as 1] to a letter grade of A [coded as 4], particular domain). Subsequently, a weighted mean sum
in math, science, English, social science, and language stud- score for the entire scale was calculated for each child as
ies, respectively), age (in years), and ethnicity. There was the sum of the mean scores from the nine domains. The
a significant difference between boys and girls for average Cronbach’s alpha value for the current sample was 0.81.
school performance, with girls performing better on average
than boys.

13
European Child & Adolescent Psychiatry

Gender identity questionnaire for children (GIQC) Parenting styles and dimensions questionnaire, short‑form
(PSDQ‑SF)
The GIQC [12] is a parent-report measure designed to
assess GV in children. It involves rating 16 items (e.g., The PSDQ-SF [48] is a 32-item parent-report scale com-
“The child plays with girl-type dolls, such as ‘Barbie’”) prised of 3 subscales designed to assess authoritative par-
on a 5-point scale, with lower scores being associated with enting (15 items; e.g., “I give comfort and understanding
more GV and higher scores with more gender conformity. when my child is upset”), authoritarian parenting (12 items;
Items were scored separately for boys and girls. For ease e.g., “I use physical punishment as a way of disciplining my
of interpretation, we reversed the GIQC scores for analysis child”) and permissive parenting (5 items; e.g., “I state pun-
such that higher scores indicated more GV. As per Johnson ishments to my child and do not actually do them”), respec-
et al. [12], items 8 and 16 of the GIQC were excluded, and tively. The items were answered on a 5-point scale (1 = never
a mean score was calculated based on the remaining 14 to 5 = always). A mean score was calculated for each of the
items. The Cronbach’s alpha for the current sample was three subscales. The Cronbach’s alpha values for the current
0.78 for parent-reports of boys and 0.68 for parent-reports sample were 0.86 for the authoritative subscale, 0.81 for the
of girls. There was a significant difference between boys authoritarian subscale, and 0.72 for the permissive subscale.
and girls for mean GIQC score, with girls scoring higher
(i.e., more gender-variant) than boys. Child‑rearing sex‑role attitude scale, adapted version
(CRSRAS)

The child–parent relationship scale, short‑form (CPRS‑SF) The CRSRAS [49] is a 19-item (e.g., “I would buy my son a
doll”) parent-report scale designed to assess parental beliefs
The CPRS-SF [44, 45] is a parent-report measure designed about gender stereotypic and counter-stereotypic behaviour
to assess the child–parent relationship, divided into two in children measured on a 5-point scale (1 = strongly disa-
subscales: (1) closeness (7 items; e.g., “My child is gree to 5 = strongly agree). A sum score was calculated for
uncomfortable with physical affection or touch from me”), the 19 items, with lower scores indicating more traditional
and (2) conflict (8 items; e.g., “My child’s feelings toward attitudes towards gendered behaviour in children. The Cron-
me can be unpredictable or can change suddenly”). The bach’s alpha value for the current sample was 0.84.
two subscales have been found to have a relatively low
correlation, suggesting they each assess unique aspects Child behaviour checklist (CBCL)
of the parent–child relationship [44]. Maternal and pater-
nal reports are relatively stable over the course of early The CBCL [4] is a parent-report questionnaire used to
childhood [44]. The CPRS-SF has a 5-point response scale measure general psychopathology among children aged
(1 = definitely does not apply to 5 = definitely applies). A 6–18 years. Items are rated on a 3-point Likert scale (0 = not
sum score was calculated for the items on the closeness true, 1 = somewhat or sometimes true, and 2 = very true or
and conflict subscales, respectively. Cronbach’s alphas for often true). The CBCL consists of 113 items and has two
the current sample were 0.73 for the closeness subscale broad-band factors: (1) internalizing problems (i.e., emo-
and 0.84 for the conflict subscale. tional problems) and (2) externalizing problems (i.e., behav-
ioural problems). Cronbach’s alpha values for the internal-
izing and externalizing subscales are reported as 0.90 and
The child rearing practices report, short‑form (CRPR‑SF) 0.94, respectively [50]. For each subscale, T scores were
calculated by converting the raw scores into standardized
The CRPR-SF [46, 47] is a parent-report measure designed scores through comparison with nonclinical samples of the
to assess whether the parent provides a secure base for the relevant age group. A peer problems subscale score was gen-
child by engaging in behaviours that lend themselves to the erated using three items (i.e., items 25, 38, and 48) from the
child forming a secure attachment. This scale is composed CBCL, not already included in the internalizing and exter-
of ten items (e.g., “I encourage my child to talk about his/ nalizing scales [51]. Item 25 states “Doesn’t get along with
her troubles”) measured on a 7-point scale (1 = most unde- other kids”, item 38 states “Gets teased a lot”, and item 48
scriptive for me to 7 = most descriptive for me). A mean states “Not liked by other kids”. A sum score for these three
score was calculated, with higher scores indicating more items was generated to serve as the peer problem subscale.
of a willingness to serve as an attachment figure for the The Cronbach’s alpha value for this subscale in the present
child. The Cronbach’s alpha value for the current sample sample was 0.68. On average, boys showed significantly
was 0.74. more peer problems than girls on this measure (Table 2).

13
 European Child & Adolescent Psychiatry

Statistical analyses associated with SA, peer problems, internalizing problems,

and externalizing problems; however, GV was significantly
Correlation analyses assessed zero-order associations among negatively associated with age and annual income. In girls
study variables. Spearman’s rho was used for correlations only, GV was significantly positively associated with age;
where one or both of the variables were rank-order and however, GV was significantly negatively associated with
point-biserial correlation coefficients were used when one average school performance.
variable was dichotomous and the other variable was con-
tinuous. Pearson’s r was used for all other correlations. The Separation anxiety and gender variance in boys
association between scores on the GIQC and the SAI dif- and girls
fered by gender (see details below). Thus, subsequent mul-
tiple regression analyses were conducted separately for boys A multiple regression analysis was conducted for the entire
and girls. Several variables that were significantly associ- sample, with SA as the outcome variable. Step 1 included
ated with the SAI and/or have been shown to be associated the demographic control variables (i.e., age, marital status,
with SA in previous studies (i.e., age, annual income, aver- school performance, and annual income), peer problems,
age school performance, marital status, and CBCL scores) internalizing problems, and externalizing problems as pre-
were included in the regression models as control variables. dictor variables. Step 2 included GV and the interaction term
Parental variables and the GIQC were entered next, consti- between GV and gender. Gender moderated the associa-
tuting a main effect block. Interaction effects between the tion between GV and SA, B = − 0.15, SE = .05, ß = − 0.06,
parental measures and the GIQC were sequentially added p < 0.01. Thus, further regression analyses predicting SA
to the model and removed, subsequent to the control vari- were conducted for boys and girls separately.
able and main effect blocks. This method of investigating
the interaction effects was employed because the interaction Multiple regression analyses for boys
terms were highly correlated with one another due to the
necessary inclusion of the GIQC score in the calculation of Table 4 shows the results of the multiple regression analyses
each interaction term; thus, entering the interaction terms with SA as the outcome variable for boys. Step 1 included
into separate models avoided the issue of multicollinearity. the demographic control variables (i.e., age, marital status,
All continuous predictor variables were standardized prior school performance, and annual income), peer problems,
to analyses, separately by gender. In the case of missing data, internalizing problems, and externalizing problems as pre-
the data were excluded pair wise. The critical alpha value dictor variables. Average school performance, age, marital
was set at 0.05 for all the analyses. status, peer problems, and internalizing problems were sig-
nificant predictors in the model. Boys scored higher for SA
when they performed more poorly in school, were younger,
Results when their parent had a marital status of “other”, when they
had more peer problems, and when they had more internal-
Correlation analyses izing problems.
Step 2 included the parental factors (i.e., parenting styles,
Correlations between demographic variables, peer problems, closeness in the parent–child relationship, conflict in the
internalizing problems, externalizing problems, parental fac- parent–child relationship, parental willingness to serve as
tors, GV, and SA for boys and girls, respectively, are shown a secure base, and parental attitudes towards gender stereo-
in Table 3. The main correlations of interest were those per- types) and GV as predictor variables. Authoritative parent-
taining to the focal variables of GV and SA. In both boys ing style, permissive parenting style, and GV were signif-
and girls, SA was significantly positively associated with icant predictors in the model. Boys scored higher for SA
authoritarian parenting style, permissive parenting style, when their parent had a more authoritative parenting style,
conflict in the parent–child relationship, peer problems, a more permissive parenting style, and reported more GV
internalizing problems, and externalizing problems. In both in their child.
boys and girls, SA was significantly negatively associated Step 3 examined interactions between GV and each of
with age, annual income, marital status, average school per- the parental factors. Each interaction term was entered on
formance, and closeness in the parent–child relationship. In its own to avoid multicollinearity. There was a significant
girls only, SA was significantly negatively associated with interaction effect for GV by authoritative parenting style, GV
parental willingness to serve as a secure base. by permissive parenting style, GV by closeness in the par-
In both boys and girls, GV was significantly positively ent–child relationship, GV by parental willingness to serve
associated with parental attitudes towards gender ste- as a secure base, and GV by parental attitudes towards gen-
reotypes. In boys only, GV was significantly positively der stereotypes. Boys who were more gender-variant had

13
 Table 3  Correlations among study variables for boys (above diagonal) and girls (below diagonal)
Variables 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

1. Gender variance – 0.16*** − 0.09** −0.10** − 0.04 0.02 0.01 0.01 − 0.01 0.04 − 0.05 0.02 0.22*** 0.18*** 0.12** 0.16***
2. Separation anxi- − 0.04 – − 0.16*** − 0.16*** − 0.14*** − 0.17*** 0.00 0.08* 0.20*** 0.25*** − 0.08* − 0.04 0.04 0.32*** 0.54*** 0.37***
ety
3. Age 0.07* − 0.11*** – 0.03 0.00 − 0.11** − 0.04 0.037 − 0.03 − 0.02 − 0.10** − 0.03 0.00 0.09** 0.04 − 0.11**
4. Annual i­ncomea − 0.03 − 0.11*** 0.02 – 0.37*** 0.21*** 0.09* − 0.00 − 0.09** − 0.04 0.07* − 0.04 0.11** − 0.16*** − 0.10** 0.12**
5. Marital status − 0.04 − 0.08* − 0.10** 0.37*** – 0.05 0.01 0.12** .01 − 0.01 0.02 − 0.10** − 0.04 − 0.07* − 0.09* − 0.05
European Child & Adolescent Psychiatry

*** *** **
6. School perfor- − 0.07* − 0.12*** − 0.09** 0.17*** 0.14 – 0.11** − 0.08* − 0.09* − 0.19 0.10 0.10** 0.04 − 0.19*** − 0.13*** − 0.21***
mance
7. Authoritative 0.03 0.00 − 0.05 0.06 0.04 0.11** – − 0.42*** − 0.28*** − 0.38*** 0.50*** 0.49*** 0.32*** − 0.11** − 0.17*** − 0.23***
parenting style
8. Authoritarian − 0.03 0.16*** − 0.09* − 0.06 0.05 − 0.09* − 0.39*** – 0.55*** 0.56*** − 0.38*** − 0.62*** − 0.25*** 0.16*** 0.23*** 0.39***
parenting style
9. Permissive par- − 0.02 0.18*** − 0.05 − 0.12*** − 0.01 − 0.13*** − 0.18*** 0.48*** – 0.42*** − 0.27*** − 0.35*** − 0.17*** 0.16*** 0.22*** 0.33***
enting style
10. Conflict in par- − 0.02 0.28*** − 0.07* − 0.053 − 0.03 − 0.09* − 0.28*** 0.51*** 0.39*** – − 0.49*** − 0.61*** − 0.13*** 0.29*** 0.39*** 0.68***
ent–child relation-
ship
11. Closeness in − 0.07* − 0.10** − 0.06 0.031 − 0.01 0.14*** 0.46*** − 0.34*** − 0.28*** − 0.41*** – 0.50*** 0.20*** − 0.24** − 0.25** − 0.32**
parent–child rela-
tionship
12. Parental willing- − 0.02 − 0.14*** 0.07* − 0.055 − 0.05 0.09* 0.43*** − 0.58*** − 0.31*** − 0.64*** 0.48*** – 0.25*** − 0.21*** − 0.21*** − 0.39***
ness to serve as a
secure base
13. Parental 0.11** − 0.04 − 0.01 0.10** − 0.05 0.03 0.33*** − 0.23*** − 0.13*** − 0.09* 0.24*** 0.20*** – 0.00 0.03 − 0.01
attitudes towards
gender stereotypes
14. Peer problems 0.01 0.33*** 0.10** − 0.08* − 0.03 − 0.20*** − 0.08* 0.17** 0.15*** 0.26*** − 0.19*** − 0.22*** − 0.10** – 0.46*** 0.43***
15. Internalizing − 0.02 0.62*** 0.02 − 0.06 − 0.05 − 0.11** − 0.10** 0.26*** 0.22*** 0.44*** − 0.24*** − 0.29*** − 0.02 0.42*** – 0.55***
problems
16. Externalizing − 0.03 0.41*** − 0.10** − 0.12*** − 0.08* − 0.20*** − 0.20*** 0.38*** 0.30*** 0.63*** − 0.31*** − 0.45*** − 0.03 0.43*** 0.61*** –
problems
a
 Spearman’s rho used as correlation coefficient
*p < 0.05, **p < 0.01, and ***p < 0.001
Gender variance = GIQC score, separation anxiety = SAI weighted mean sum score, authoritative parenting style = PSDQ-SF authoritative subscale, authoritarian parenting style = PSDQ-
SF authoritarian subscale, permissive parenting style = PSDQ-SF permissive subscale, conflict in parent–child relationship = CPRS-SF conflict  subscale, closeness in parent–child relation-
ship = CPRS-SF closeness subscale, parental willingness to serve as a secure Base = CRPR-SF, parental attitudes towards gender stereotypes = CRSRAS, peer problems = CBCL peer problems
subscale, internalizing problems = CBCL internalizing T score¸ externalizing problems = CBCL externalizing T score

13
 European Child & Adolescent Psychiatry

lower levels of SA when they had a more authoritative par- Separation anxiety and internalizing problems
ent, more closeness in the parent–child relationship, a parent
who was more willing to serve as a secure base, and a parent This study was the first to suggest that SA is associated with
with more liberal attitudes towards gender stereotypes in GV among boys above and beyond internalizing problems
children. more generally. As one would expect given SA is a type of
internalizing problem, CBCL internalizing problems were
Multiple regression analyses for girls positively associated with SA in both boys and girls; how-
ever, the association between GV and SA in boys remained
Table 4 shows the multiple regression analyses with SA as significant even when statistically controlling for CBCL
the outcome variable for girls. Step 1 included demographic internalizing problems—as well as when statistically con-
control variables (i.e., age, marital status, school perfor- trolling for CBCL externalizing and peer problems. These
mance, and annual income), peer problems, internalizing findings contrast with those of VanderLaan et al. [14], who
problems, and externalizing problems as predictors. Age, were not able to distinguish SA from the more general pat-
internalizing problems, and peer problems were significant tern of elevated internalizing problems in a clinical sample
predictors in the model. Girls scored higher for SA when of GD-referred birth-assigned male children. One possible
they were younger, when they had more internalizing prob- reason for the discrepancy in findings is that the sample
lems, and when they had more peer problems. size in the VanderLaan et al. [14] study was relatively small
Step 2 included the parental factors (i.e., parenting styles, (n = 360) compared with that of the current study (n = 892).
closeness in the parent–child relationship, conflict in the It is possible, then, that the lack of significant findings
parent–child relationship, parental willingness to serve as regarding a link between GV and SA in VanderLaan et al.’s
a secure base, and parental attitudes towards gender stereo- [14] clinical sample when controlling for other internaliz-
types) and GV as predictor variables. None of these vari- ing problems might have been due to insufficient statistical
ables were significant predictors. power. Another consideration is that children in nonclinical
Step 3 examined interactions between GV and each of samples likely experience fewer internalizing problems than
the parental factors. Each interaction term was entered on its children in clinical samples by virtue of the different recruit-
own to avoid multicollinearity. There were significant inter- ment methods. As such, among nonclinical as compared to
action effects for GV by authoritative parenting style, GV by clinical samples, SA might “stand out” more readily from
closeness in the parent–child relationship, GV by parental the background internalizing profile. If so, one might expect
willingness to serve as a secure base, and GV by parental that among birth-assigned males clinic-referred for GD, SA
attitudes towards gender stereotypes. Girls who were more could often still be evident among those cases who show few
gender-variant had lower levels of SA when they had a more internalizing problems in other domains.
authoritative parent, more closeness in the parent–child rela-
tionship, a parent who was more willing to serve as a secure A role for feminine gender role expression
base, and a parent with more liberal attitudes towards gender in influencing separation anxiety?
stereotypes in children.
The present study suggests SA is a unique internalizing
problem associated with GV and SA in boys. One possible
Discussion explanation for this association is based on the observa-
tion that SA is a more female-typical characteristic. Indeed,
The present study found an association between GV and SA SAD is more common among girls than boys [13], and
in a nonclinical sample of boys, but not among a nonclinical although there was no significant difference in levels of SA
sample of girls. This finding regarding boys is consistent between boys and girls in the present sample, there was a
with cross-cultural studies that found positive associations trend towards more SA in girls. Gender-variant boys display
between recalled childhood GV and SA in nonclinical sam- more feminine-typed traits (e.g., feminine play preferences)
ples of feminine androphilic males from Canada [22], Samoa and, thus, may possess aspects of femininity that predispose
[20], and the Istmo Zapotec region of Mexico [21]. As such, them to SA. It remains unclear, however, whether feminine
it seems unlikely the findings of these retrospective stud- traits do in fact underlie SA and, if so, which feminine traits
ies were due to memory distortion. In addition, the GV-SA and why.
association in the current nonclinical sample of boys is con- The present data concerning girls challenge, to some
sistent with reports of elevated levels of SA among clinical extent, the hypothesis that childhood SA is associated with
samples of GD-referred birth-assigned males [14–16]. The femininity. Gender-variant girls adopt more masculine-
consistency in findings across studies suggests this associa- typed traits. As such, this hypothesis would lead one to
tion is not limited to clinical populations. expect them to experience less SA. Yet, this study found no

13
 Table 4  Multiple linear regression predicting separation anxiety in boys and girls
Boys Girls
B SE ß t value p value B SE ß t value p value

Step 1: control variables

 Age (in years) − 0.45 0.07 − 0.20 − 6.20 <.001 − 0.36 0.08 − 0.14 − 4.72 < 0.001
 Average school performance − 0.25 0.08 − 0.10 − 3.14 0.002 − 0.11 0.09 − 0.04 − 1.33 0.19
European Child & Adolescent Psychiatry

 Marital status − 0.46 0.17 − 0.08 − 2.71 0.007 − 0.33 0.17 − 0.06 − 1.90 0.06
 Annual income 0.00 0.00 − 0.01 − 0.35 0.72 0.00 0.00 − 0.04 − 1.49 0.14
 Peer problems 0.21 0.10 0.08 2.07 0.04 0.27 0.10 0.09 2.71 0.007
 Internalizing problems 0.10 0.01 0.48 12.30 < 0.001 0.14 0.01 0.58 15.54 < 0.001
 Externalizing problems 0.01 0.01 0.03 0.84 0.40 0.00 0.01 0.00 − 0.09 0.93
Step 2: parental variables and gender variance
 Authoritative parenting style 0.21 0.09 0.09 2.36 0.02 0.17 0.09 0.07 1.90 0.06
 Authoritarian parenting style − 0.15 0.10 − 0.07 − 1.47 0.14 − 0.03 0.10 − 0.01 − 0.27 0.79
 Permissive parenting style 0.30 0.09 0.13 3.44 0.001 0.11 0.09 0.04 1.24 0.21
 Conflict in parent–child relationship 0.20 0.13 0.08 1.61 0.11 0.09 0.12 0.03 0.76 0.45
 Closeness in parent–child relationship 0.06 0.10 0.03 0.65 0.52 0.08 0.09 0.03 0.85 0.40
 Parental willingness to serve as a secure base 0.21 0.11 0.09 1.90 0.06 0.15 0.12 0.06 1.40 0.16
 Parental attitudes towards gender stereotypes − 0.07 .08 − 0.03 − 0.85 0.40 − 0.15 0.08 − 0.06 − 1.92 0.06
 Gender variance 0.18 0.07 0.08 2.43 0.02 − 0.04 0.07 − 0.02 − 0.52 0.60
Step 3: ­interactionsa
 Gender variance by authoritative parenting style − 3.30 0.92 − 1.45 − 3.60 0.001 − 1.83 0.80 − 0.72 − 2.29 0.02
 Gender variance by authoritarian parenting style 0.06 0.42 0.03 0.15 0.88 0.22 0.39 0.09 0.57 0.57
 Gender variance by permissive parenting style − 1.03 0.51 − 0.45 − 2.04 0.04 0.02 0.40 0.01 0.05 0.96
 Gender variance by conflict in parent–child relationship 0.03 0.47 0.01 0.05 0.96 0.28 0.43 0.10 0.64 0.53
 Gender variance by closeness in parent–child relationship − 9.87 1.85 − 4.28 − 5.33 < 0.001 − 3.41 0.85 − 1.34 − 4.01 < 0.001
 Gender variance by parental willingness to serve as a secure base − 3.95 .90 − 1.71 − 4.39 < 0.001 − 1.48 0.67 − 0.59 − 2.19 0.03
 Gender variance by parental attitudes towards gender stereotypes − 9.49 1.23 − 4.16 − 7.73 < 0.001 − 2.87 0.79 − 1.14 − 3.66 < 0.001
a
 Interaction terms were entered into the model individually to avoid multicollinearity
Gender variance = GIQC score, separation anxiety = SAI weighted mean sum score, authoritative parenting style = PSDQ-SF authoritative subscale, authoritarian parenting style = PSDQ-
SF authoritarian subscale, permissive parenting style = PSDQ-SF permissive subscale, conflict in parent–child relationship = CPRS-SF conflict  subscale, closeness in parent–child relation-
ship = CPRS-SF closeness subscale, parental willingness to serve as a secure base = CRPR-SF, parental attitudes towards gender stereotypes = CRSRAS, peer problems = CBCL peer problems
subscale, internalizing problems = CBCL internalizing T score¸ externalizing problems = CBCL externalizing T score

13
 European Child & Adolescent Psychiatry

significant association between SA and GV in girls. The lack The association of GV and SA appeared to be moderated
of a GV-SA association in girls is somewhat inconsistent by several parental factors. These effects were found in both
with the study by Petterson et al. [23], who found that lesbi- boys and girls, suggesting that although girls do not differ
ans—who generally reported higher levels of GV in child- in their levels of SA purely as a function of their degree of
hood—had elevated levels of SA. That said, Petterson et al. gender (a)typicality, gender expression in conjunction with
[23] did not find correlations between recalled childhood SA parental factors is relevant to SA in girls. For gender-variant
and GV among lesbians at the level of individual differences. boys and girls, warm parenting styles (i.e., authoritative for
The majority of data on this issue, then, indicate a lack of boys and girls; permissive for boys) appeared to be a pro-
an association between gender expression and SA in girls. tective factor in relation to SA. Alanko et al. [31] found
Thus, it seems that for any hypotheses regarding the role of that adults who recalled more childhood GV and more cold
feminine childhood gender role expression in influencing and over-controlling parents experienced more adulthood
SA to be tenable, they would have to explain why gender psychopathology. Interestingly, cold and over-controlling
role expression should be relevant for boys but not for girls. parents (i.e., an authoritarian parenting style) were not a
Alternatively, highly gender-typical boys appear to be dis- SA risk factor for gender-variant children in the current
tinct from girls and gender-variant boys. Levels of masculin- study. Instead, risk for SA among gender-variant children
ity typical of boys may be associated with less SA, possibly appeared to be related more so to the relative absence of
due to one or more factors. For example, gender-typical boys parenting style that is warm (i.e., authoritative and permis-
may receive differential parental and caregiver treatment sive for boys) or warm and controlling (i.e., authoritative
based upon their gender and gender expression. In contrast, for boys and girls). That said, Alanko et al. [31] examined
all girls may be perceived as feminine and treated accord- the moderating influence of parenting styles on the associa-
ingly regardless of gender expression while gender-variant tion between childhood GV and adulthood psychopathology;
boys may be perceived as distinct from other boys if they thus, different patterns might exist for childhood vs. adult-
do not exhibit masculine-typed traits. Gender-variant boys hood psychological adjustment. Also, it may be that SA is
may then be treated by parents in a more feminine-typical to some extent unique from other forms of psychopathology
manner, resulting in female-typical emotional expression and, therefore, may be impacted by parenting styles in a
and SA. Also, gender-typical boys may be encouraged more unique manner.
so to seek comfort beyond their attachment figure, leading Closeness in the parent–child relationship also appeared
to greater emotional ease when separated from attachment to protect gender-variant children in regards to SA. How-
figures. Future research should evaluate these possibilities. ever, the lack of main effects of closeness in the parent–child
relationship on boys’ and girls’ levels of SA was inconsist-
Separation anxiety and parent‑related factors ent with previous literature suggesting that children who
have a closer relationship with their parents have improved
A number of parental factors were examined to assess psychological well-being [39, 40]. It might be the case that
their potential influence on SA in boys and girls. In terms gender-variant boys and girls are particularly sensitive to
of main effects, most parenting variables were not signifi- and in need of a close parental figure to provide a sense of
cantly related to SA. The exceptions were authoritative and emotional support. Fostering closeness in the parent–child
permissive parenting styles (i.e., warm parenting styles), relationship is a method that has already been recommended
which were associated with elevated SA among boys. Due to increase the psychological well-being of gender-variant
to the correlational nature of these data, the direction of children [42], and the results of the current study further
the relationship between warm parenting styles and SA in suggest that encouraging parent–child closeness may be a
boys is unclear. It may be that boys who experience distress particularly important strategy for reducing SA among gen-
upon separation from their parents elicit warm and support- der-variant children.
ive parental reactions. On the other hand, warm parental For gender-variant boys and girls, parental willingness
behaviour might contribute to SA in boys (e.g., by making to serve as an attachment figure appeared to be a protec-
proximity normative and, thus, instances of separation unfa- tive factor in relation to SA. Gender-variant children may,
miliar and distressing). In girls, however, parenting styles therefore, be particularly sensitive to the attachment environ-
and SA showed no associations. One explanation for this ment. Parents who are more willing to serve as an attach-
gender difference could relate to findings that girls typi- ment figure can be inferred to be more likely to foster secure
cally receive more warm parenting while boys are shown attachment; thus, the underlying construct driving the effect
less affection [52]. As such, perhaps boys do tend to receive may in fact be a secure attachment style. These results are
warm responses from their parents more so only when show- consistent with previous studies that found insecure attach-
ing psychological distress. ment is associated with anxiety [35, 36] and gender atypical-
ity in children [37].

13
European Child & Adolescent Psychiatry

Lastly, liberal parental attitudes towards gender stereo- objective measures of psychological well-being and from
types in children appeared to protect against SA in gender- considering the potentially respective influences of moth-
variant children. This finding is consistent with previous lit- ers and fathers. Additionally, the current data lack clinically
erature suggesting that parental acceptance of cross-gender diagnostic information and, therefore, whether they gener-
behaviour is important for the psychological well-being of alize to clinical settings is equivocal. It is also important to
gender-variant children, particularly with regard to internal- note that due to the correlational nature of the present data,
izing problems [2, 32]. Moreover, the results provide fur- it is not possible to determine causality. Thus, it is unclear
ther evidence for the potential effectiveness of interventions whether GV leads to SA, SA leads to GV, or they develop
focused on increasing parental acceptance for children with simultaneously in boys due to some factor(s) not considered
high levels of GV, which have previously been suggested to here. In this regard, longitudinal studies would be helpful.
decrease anxiety [33].

Clinical implications
Summary
This study provides potentially important information for
In nonclinical populations, SA appears to be linked to GV
clinicians. Recognizing that SA can be a unique internaliz-
among boys, but not girls. The GV-SA association in boys
ing problem for gender-variant birth-assigned male children
exists even when controlling for internalizing problems,
may allow clinicians to be more attentive to and, thus, better
suggesting some aspect of SA is uniquely associated with
identify symptomology of SA in this population. As such,
GV. Several parental variables were identified as potential
psychological well-being may be improved via increased
protective factors against SA for gender-variant boys and
clinical attention to possible SA in particular, as opposed to
girls. These include authoritative parenting, willingness to
a more general focus on internalizing problems. Addition-
serve as an attachment figure, closeness in the parent–child
ally, the present study identified several parental variables
relationship, and liberal attitudes towards gender stereotypes
that may serve as protective factors in relation to SA in both
in children. Thus, screening for SA may be important and
gender-variant birth-assigned males and females. For gen-
several parental factors may be relevant to alleviating SA in
der-variant children experiencing SA, psychological well-
gender-variant children.
being might be promoted by encouraging their parents to
focus on providing a secure base for the child, using a warm Acknowledgements  We thank an anonymous reviewer for comments
yet controlling parenting approach, encouraging a close rela- on a prior draft. AS and ANN were supported by Social Sciences
tionship with the child, and encouraging the parent to hold and Humanities Research Council (SSHRC) of Canada Undergradu-
liberal attitudes towards GV. Further research is needed, ate Research Awards. This research was funded by a University of
Toronto Mississauga Research and Scholarly Activity Fund Award and
however, to determine the utility of these approaches. a SSHRC Insight Development Grant to DPV.

Limitations Compliance with ethical standards 

Regarding the use of parent-reports, parents are more likely Conflict of interest  The authors declare they have no conflict of inter-
to respond with positive self-evaluations and evaluations est.
of their children [53]. Parents willing to complete a sur-
Research involving human participants  All the procedures performed
vey about their child’s psychological well-being may differ in studies involving human participants were in accordance with the
from those who were not in ways relevant to the study out- ethical standards of the institutional and/or national research committee
comes (e.g., they may be more concerned about their child’s and with the 1964 Helsinki Declaration and its later amendments or
mental health). Also, this study relied primarily on mater- comparable ethical standards. This article does not contain any studies
with animals performed by any of the authors.
nal reports. Some research has suggested that mothers and
fathers have different impacts. For example, in their study Informed consent  Informed consent was obtained from all individual
of adults, Alanko et al. [31] found that the parenting styles participants included in the study.
of parents who were of the same sex as the participants had
a greater moderating influence on the association between
recalled childhood GV and psychopathology in adulthood.
Similarly, van Beusekom et al. [32] found that same sex, References
compared to opposite sex, parental acceptance had a more
1. Adelson SL (2012) Practice parameter on gay, lesbian, or bisexual
profound impact on the association between GV and psycho-
sexual orientation, gender nonconformity, and gender discordance
logical distress and social anxiety in youth. As such, future in children and adolescents. J Am Acad Child Adolesc Psychiatry
research might benefit from considering potentially more 51:957–974. https​://doi.org/10.1016/j.jaac.2012.07.004

13
 European Child & Adolescent Psychiatry

2. Yunger JL, Carver PR, Perry DG (2004) Does gender identity Psychiatry 47:1413–1423. https ​ : //doi.org/10.1007/s1050​
influence children’s psychological well-being? Dev Psychol 8-009-9517-3
40:572–582. https​://doi.org/10.1037/0012-1649.40.4.572 18. Li G, Kung KTF, Hines M (2017) Childhood gender-typed
3. Martin CL, Andrews NC, England DE, Zosuls K, Ruble DN behavior and adolescent sexual orientation: a longitudinal
(2017) A dual identity approach for conceptualizing and measur- population-based study. Dev Psychol 53:764–777. https​://doi.
ing children’s gender identity. Child Dev 88:167–182. https​://doi. org/10.1037/dev00​00281​
org/10.1111/cdev.12568​ 19. Steensma TD, van der Ende J, Verhulst FC, Cohen-Kettenis PT
4. Achenbach TM (1991) Manual for the child behavior check- (2013) Gender variance in childhood and sexual orientation in
list/4–18 and 1991 profile. University of Vermont, Department adulthood: a prospective study. J Sex Med 10:2723–2733. https​
of Psychiatry, Burlington ://doi.org/10.1111/j.1743-6109.2012.02701​.x
5. Strang JF, Kenworthy L, Dominska A, Sokoloff J, Kenealy LE, 20. Vasey PL, VanderLaan DP, Gothreau LM, Bartlett NH (2011)
Berl M, Walsh K, Menvielle E, Slesaransky-Poe G, Kim KE, Traits of separation anxiety in childhood: a retrospective study
Luong-Tran C, Meagher H, Wallace GL (2014) Increased gen- of Samoan men, women, and fa’afafine. Arch Sex Behav
der variance in autism spectrum disorders and attention deficit 40:511–517. https​://doi.org/10.1007/s1050​8-009-9589-0
hyperactivity disorder. Arch Sex Behav 43:1525–1533. https​:// 21. Gómez FR, Semenyna SW, Vasey PL (2017) Recalled separa-
doi.org/10.1007/s1050​8-014-0285-3 tion anxiety in childhood in Istmo Zapotec men, women, and
6. van Beijsterveldt CEM, Hudziak JJ, Boomsma DI (2006) Genetic muxes. Arch Sex Behav 46:109–117. https​://doi.org/10.1007/
and environmental influences on cross-gender behavior and rela- s1050​8-016-0917-x
tion to behavior problems: a study of Dutch twins at ages 7 and 22. VanderLaan DP, Gothreau L, Bartlett NH, Vasey PL (2011)
10 years. Arch Sex Behav 35:647–658. https​://doi.org/10.1007/ Recalled separation anxiety and gender atypicality in child-
s1050​8-006-9072-0 hood: a study of  Canadian heterosexual and homosexual
7. van der Miesen AIR, Nabbijohn AN, Santarossa A, VanderLaan men and women. Arch Sex Behav 40:1233–1240. https​://doi.
DP (2018) Behavioral and emotional problems in gender-noncon- org/10.1007/s1050​8-010-9695-z
forming children: a Canadian community-based study. J Am Acad 23. Petterson LJ, VanderLaan DP, Vasey PL (2017) Sex, sexual ori-
Child Adolesc Psychiatry 57:491–499. https​://doi.org/10.1016/j. entation, gender atypicality, and indicators of depression and
jaac.2018.03.015 anxiety in childhood and adulthood. Arch Sex Behav 46:1383–
8. American Psychiatric Association (2013) Diagnostic and statisti- 1392. https​://doi.org/10.1007/s1050​8-016-0690-x
cal manual of mental disorders, 5th edn. American Psychiatric 24. Bailey JM, Zucker KJ (1995) Childhood sex-typed behav-
Publishing, Arlington ior and sexual orientation: a conceptual analysis and
9. Zucker KJ, Wood H, VanderLaan DP (2014) Models of psycho- quantitative review. Dev Psychol 31:43–55. https ​ : //doi.
pathology in children and adolescents with gender dysphoria. In: org/10.1037/0012-1649.31.1.43
Kreukels BPC, Steensma TD, de Vries ALC (eds) Gender dyspho- 25. Brassell AA, Rosenberg E, Parent J, Rough JN, Fondacaro K,
ria and disorders of sex development. Focus on sexuality research. Seehuus M (2016) Parent’s psychological flexibility: associations
Springer, Boston, pp 171–192 with parenting and child psychosocial well-being. J Contextual
10. Wallien MSC, Swaab H, Cohen-Kettenis PT (2007) Psychiat- Behav Sci 5:111–120. https​://doi.org/10.1016/j.jcbs.2016.03.001
ric comorbidity among children with gender identity disorder. J 26. Videon T (2005) Parent-child relations and children’s psychologi-
Am Acad Child Adolesc Psychiatry 46:1307–1314. https​://doi. cal well-being: do dads matter? J Fam Issues 26:55–78. https​://
org/10.1097/chi.0b013​e3181​37384​8 doi.org/10.1177/01925​13X04​27026​2
11. Baams L, Beek T, Hille H, Zevenbergen F, Bos HMW (2013) 27. McLeod BD, Wood JJ, Weisz JR (2007) Examining the asso-
Gender nonconformity, perceived stigmatization, and psychologi- ciation between parenting and childhood anxiety: a meta-anal-
cal well-being in Dutch sexual minority youth and young adults: ysis. Clin Psychol Rev 27:155–172. https​://doi.org/10.1016/j.
a mediation analysis. Arch Sex Behav 42:765–773. https​://doi. cpr.2006.09.002
org/10.1007/s1050​8-012-0055-z 28. Suveg C, Aschenbrand SG, Kendall PC (2005) Separation anxi-
12. Johnson LL, Bradley SJ, Birkenfeld-Adams AS, Kuksis MA, ety disorder, panic disorder, and school refusal. Child Adolesc
Maing DM, Mitchell JN, Zucker KJ (2004) A parent-report gender Psychiatr Clin N Am 14:773–795. https​://doi.org/10.1016/j.
identity questionnaire for children. Arch Sex Behav 33:105–116. chc.2005.05.005
https​://doi.org/10.1023/B:ASEB.00000​14325​.68094​.f3 29. Victor AM, Bernstein GA (2009) Anxiety disorders and post-
13. Shear K, Jin R, Ruscio AM, Walters EE, Kessler RC (2006) Preva- traumatic stress disorder update. Psychiatr Clin N Am 32:57–69.
lence and correlates of estimated DSM-IV child and adult separa- https​://doi.org/10.1016/j.psc.2008.11.004
tion anxiety disorder in the national comorbidity survey replica- 30. Wallien MSC, Veenstra R, Kreukels BPC, Cohen-Kettenis PT
tion. Am J Psychiatry 163:1074–1083. https​://doi.org/10.1176/ (2010) Peer group status of gender dysphoric children: a sociomet-
ajp.2006.163.6.1074 ric study. Arch Sex Behav 39:553–560. https​://doi.org/10.1007/
14. VanderLaan DP, Santarossa A, Nabbijohn AN, Wood H, Owen- s1050​8-009-9517-3
Anderson A, Zucker KJ (2018) Separation anxiety among birth- 31. Alanko K, Santtila P, Harlaar N, Witting K, Varjonen M, Jern
assigned male children in a specialty gender identity service. Eur P, Johansson A, von der Pahlen B, Sandnabba NK (2008) The
Child Adolesc Psychiatry 27:89–98. https:​ //doi.org/10.1007/s0078​ association between childhood gender atypical behavior and adult
7-017-1018-7 psychiatric symptoms is moderated by parenting style. Sex Roles
15. Coates S, Person ES (1985) Extreme boyhood femininity: iso- 58:837–847. https​://doi.org/10.1007/s1119​9-008-9395-5
lated behavior or pervasive disorder? J Am Acad Child Psychiatry 32. van Beusekom G, Bos HMW, Overbeek G, Sandfort TGM (2015)
24:702–709. https​://doi.org/10.1016/S0002​-7138(10)60113​-6 Same-sex attraction, gender nonconformity, and mental health: the
16. Zucker KJ, Bradley SJ, Sullivan CBL (1996) Traits of separation protective role of parental acceptance. Psychol Sex Orientat Gend
anxiety in boys with gender identity disorder. J Am Acad Child Divers 2:307–312. https​://doi.org/10.1037/sgd00​00118​
Adolesc Psychiatry 35:791–798. https​://doi.org/10.1097/00004​ 33. Rosenberg M, Jellinek MS (2002) Children with gender iden-
583-19960​6000-00019​ tity issues and their parents in individual and group treatment.
17. Wallien MSC, Cohen-Kettenis PT (2008) Psychosexual out- J Am Acad Child Adolesc Psychiatry 41:619–621. https​://doi.
come of gender-dysphoric children. J Am Acad Child Adolesc org/10.1097/00004​583-20020​5000-00020​

13
European Child & Adolescent Psychiatry

34. Connors ME (2011) Attachment theory: a “secure base” for psy- 44. Driscoll K, Pianta RC (2011) Mothers’ and fathers’ perceptions
chotherapy integration. J Psychother Integr 21:348–362. https​:// of conflict and closeness in parent-child relationships during early
doi.org/10.1037/a0025​460 childhood. J Early Child Infant Psychol 7:1–24
35. Colonnesi C, Draijer EM, Stams GJJM, van der Bruggen CO, 45. Pianta RC (1995) Child-parent relationship scale. University of
Bögels SM, Noom MJ (2011) The relation between insecure Virginia Press, Charlottesville
attachment and child anxiety: a meta-analytic review. J Clin 46. Block JH (1965) The child-rearing practices report: A technique
Child Adolesc Psychol 40:630–645. https:​ //doi.org/10.1080/15374​ for evaluating parental socialization orientations. University of
416.2011.58162​3 California, Institute of Human Development, Berkeley
36. Zeanah CH, Keyes A, Settles L (2003) Attachment relationship 47. Kerns KA, Klepac L, Cole A (1996) Peer relationships
experiences and childhood psychopathology. Ann N Y Acad Sci and preadolescents’ perceptions of security in the child-
1008:22–30. https​://doi.org/10.1196/annal​s.1301.003 mother relationship. Dev Psychol 32:457–466. https ​ : //doi.
37. Cooper PJ, Pauletti RE, Tobin DD, Menon M, Menon M, Spatta org/10.1037/0012-1649.32.3.457
BC, Hodges EVE, Perry DG (2013) Mother-child attachment and 48. Robinson CC, Mandleco B, Olsen SF, Hart CH (1995) Authori-
gender identity in preadolescence. Sex Roles 69:618–631. https​:// tative, authoritarian, and permissive parenting practices: devel-
doi.org/10.1007/s1119​9-013-0310-3 opment of a new measure. Psychol Rep 77:819–830. https​://doi.
38. Bradford AB, Burningham KL, Sandberg JG, Johnson LN (2017) org/10.2466/pr0.1995.77.3.819
The association between the parent-child relationship and symp- 49. Freeman NK (2007) Preschoolers’ perceptions of gender appro-
toms of anxiety and depression: the roles of attachment and per- priate toys and their parents’ beliefs about genderized behav-
ceived spouse attachment behaviors. J Marital Fam Ther 43:291– iors: Miscommunication, mixed messages, or hidden truths?
307. https​://doi.org/10.1111/jmft.12190​ Early Child Educ J 34:357–366. https​://doi.org/10.1007/s1064​
39. Liu R (2005) Parent-youth closeness and youth’s suicidal idea- 3-006-0123-x
tion. Youth Soc 37:145–175. https:​ //doi.org/10.1177/004411​ 8X04​ 50. Achenbach TM, Rescorla LA (2001) Manual for the ASEBA
27229​0 school-age forms and profiles. University of Vermont, Research
40. Matthewson M, Smith RB, Montgomery I (2012) Does the parent- Center for Children, Youth, and Families. Burlington, VT
child relationship contribute to children’s and parents’ anxiety? J 51. Cohen-Kettenis PT, Owen A, Kaijser VG, Bradley SJ, Zucker
Relatsh Res 3:1–9. https​://doi.org/10.1017/jrr.2012.2 KJ (2003) Demographic characteristics, social competence, and
41. Davey A, Bouman WP, Arcelus J, Meyer C (2014) Social support behavior problems in children with gender identity disorder: a
and psychological well-being in gender dysphoria: a comparison cross-national, cross-clinic comparative analysis. J Abnorm Child
of patients with matched controls. J Sex Med 11:2976–2985. https​ Psychol 31:41–53. https​://doi.org/10.1023/A:10217​69215​342
://doi.org/10.1111/jsm.12681​ 52. Zahn-Waxler C, Shirtcliff EA, Marceau K (2008) Disorders of
42. Chin E (2016) The role of family therapy in children and ado- childhood and adolescence: gender and psychopathology. Annu
lescents with gender dysphoria and psychiatric problems. J Am Rev Clin Psychol 4:275–303. https​: //doi.org/10.1146/annur​
Acad Child Adolesc Psychiatry 55:S80. https​://doi.org/10.1016/j. ev.clinp​sy.3.02280​6.09135​8
jaac.2016.07.752 53. Seifer R, Sameroff A, Dickstein S, Schiller M, Hayden LC (2004)
43. van den Bergh O, Walentynowicz M (2016) Accuracy and bias in Your own children are special: clues to the sources of reporting
retrospective symptom reporting. Curr Opin Psychiatry 29:302– bias in temperament assessments. Infant Behav Dev 27:323–341.
308. https​://doi.org/10.1097/YCO.00000​00000​00026​7 https​://doi.org/10.1016/j.infbe​h.2003.12.005

13