Construction and Validation of the Touch Experiences and Attitudes Questionnaire (TEAQ): A Self-report Measure to Determine Attitudes Toward and Experiences of Positive Touch

The 117-item draft TEAQ was completed by 867 participants. Of these participants, 249 had > 5% of their data missing and were excluded. All participants included in the analysis had less than 1.2% missing data. For the whole dataset, only 0.3% of data was missing once the 249 participants had been excluded.

Of the 618 participant responses included in this study, 440 were female and 178 were male. Mean age ± standard deviation was 26.9 ± 9.3 years. Age, sex, and socio-economic demographics of the sample are presented in Table 1. The majority of participants were either students (58%), or in full-time employment (30%). Most participants were in a relationship (60%, including 26% married/cohabitating), 37% were single, 2% were separated/divorced, and 0.2% widowed. Additionally, the majority of participants had no children (82%).

The authors assert that all procedures contributing to all work reported in this article complies with the ethical standards of the relevant national and institutional committees on human experimentation and with the Helsinki Declaration of 1975, as revised in 2008. All participants gave informed consent. The study was approved by the University of Manchester research ethics committee. All members of the Faculty of Medical and Human Sciences at the University of Manchester received an emailed invitation to participate including undergraduates, postgraduates, academics, clerical and general staff, allowing a reasonably diverse age range. The study was also advertised through an online social networking site.

Questionnaire items were produced via discussion with professional psychologists and psychiatrists. The main types of interpersonal positive touch were identified as hugs, kisses, skin–skin, and hair–skin contact. The main circumstances in which interpersonal positive touch occurs were identified as greeting, consoling, intimacy, and childhood contact. Circumstances relating to non-interpersonal positive touch were identified as self-care, including personal grooming behaviours, touch with animals, and touching fabrics. Questionnaire items were systematically generated to cover the main types of touch and circumstances in which touch occurs. For each circumstance and touch type, 3 items were generated where possible: one to determine how often an individual experiences that touch, one to determine the attitude an individual has towards receiving that touch, and one to determine the attitude an individual has of giving that touch to someone else. Some general questions about attitude to and experience of touch were also included, such as: “I am put off by physical familiarity” and “My life lacks physical affection.” Items consisted of statements and a 5-point Likert scale of agreement, with “Disagree strongly” being the point furthest on the left and “Agree strongly” being the point furthest on the right. The following values were assigned to responses: 1 = “Disagree strongly”, 2 = “Disagree a little”, 3 = “Neither agree nor disagree”, 4 = “Agree a little”, 5 = “Agree strongly”, apart from negatively worded items which were reverse scored.

The terms ‘partner’, ‘boyfriend/girlfriend’, ‘husband/wife’ were avoided and instead any questions about intimate touch referred to ‘someone you are close to/fond of/know intimately’. Questions about childhood touch were limited to the amount of various forms of touch they recalled receiving and not their attitudes, as the latter was considered harder to recall and interpret. 117 items were generated. Item order was randomized using a random number generator. The item order remained the same for all participants.

Participants completed the TEAQ online alongside some general questions to obtain demographic data. This investigation was carried out online, allowing anonymity, wider dissemination, and minimal influence of embarrassment, social conformity, and pressure to participate. The first sample was for item reduction and selection.

PCA, rather than exploratory factor analysis was carried out on the initial dataset from the 117 item touch questionnaire, as we wanted to use a data reduction technique, to reduce the number of variables in the questionnaire while retaining as much information as possible, rather than simply identifying the latent constructs underlying the questionnaire variables (Fabrigar et al. 1999). PCA with direct oblimin rotation was carried out using SPSS Version 16 (SPSS Inc., Chicago, IL). Covariance rather than correlations analysis was used because covariance analysis is less influenced by variation in the distribution of scores between items on a 5-point Likert scale (Field 2005; Tinsley and Tinsley 1987). Missing values were excluded pairwise. Responses for negatively phrased items were reverse scored, so all item scores reflected greater touch experience or positive attitude. The number of components extracted was determined using Cattell’s scree test (Cattell 1966).

A correlation matrix was used to exclude redundant items correlating significantly with > 80% of other items, or correlating r > 0.8 with another item. Items with measures of sampling adequacy (MSA) < 0.6 were removed, as were items with rescaled communalities < 0.3, indicating these items explained only a small proportion of shared variance ( Field 2005). Stevens (1992) suggests component loadings > 0.4 should be considered of interest. Any items with component loadings < 0.4 for all components were removed, as were any items loading similarly on two components. Reliability analysis was carried out for each component, for which all items belonging to each component were given an equal weighting. Items which did not significantly increase Cronbach’s α were removed. For the remainder of this manuscript, subscale scores have been calculated by reverse scoring the items indicated with an R in Table 2, then calculating a mean score for each subscale.

Cattell’s scree test identified 6 components. A total of 60 items failed inclusion criteria: 40 had low communality scores, 7 loaded < 0.4 on all components, 6 had low MSA scores, 3 did not increase Cronbach’s α, 1 loaded similarly on two components, 1 significantly correlated with over 80% of the other questionnaire items and 2 pairs of items correlated with each other > 0.8, so one item from each pair was deleted. This left a total of 57 items explaining 56% of the variance. The component structure identified is presented in Table 2. A copy of the TEAQ with scoring instructions can be found in “Appendix 1”. Copyright of the TEAQ remains with the authors. The range of Cronbach’s α for the components was 0.78–0.92, suggesting one dimension per component and internal consistency.

Of the components identified, touch with others concerned 5 components and one component (component 4) concerned attitude to self-care (5 items). The largest component in terms of variance explained, contained 11 items and was termed Friends and Family Touch (FFT) because it loaded on items about amount and liking of giving and receiving affectionate touch with friends and family. This appears to be a general component since it correlated with other component scores r = 0.50–0.52, except attitude to self-care where r = 0.36. The FFT component contains both attitudinal and amount measures, but as these all loaded on the same component, it appears these measures cannot be separated for this component. Two components (2 and 3) concerned amount of touch, respectively in intimate relationships (14 items) and in childhood (9 items), and two (5 and 6) concerned affective attitude to touch with others, respectively in intimate relationships (13 items) and unfamiliar touch (5 items). Component names are shown in Table 2.

As shown in Table 3, component scores correlated significantly with each other (p < 0.001). For this reason, the solution obtained from direct oblimin, oblique rotation was accepted, rather than an orthogonally rotated solution (Tabachnick and Fidell 2013). Attitude to self-care correlated least with other components and friends and family touch correlated most. The strongest correlation was between current intimate touch and attitude to intimate touch (r = 0.58). The weakest correlation was between attitude to self-care and attitude to unfamiliar touch (r = 0.12).

The sample for this study was recruited in the same way as that described for Study 1. The questionnaire battery, including the 57-item TEAQ was completed by 817 participants. Of these participants, 113 were excluded due to > 5% missing data, leaving 704 participants, 73.7% of which were female. Mean ± standard deviation age was 27.4 ± 9.6 years. As shown in Table 1, socio-economic demographics were similar to the study 1 sample.

The replication sample completed the reduced TEAQ online and CFA was carried out using AMOS statistical software (Amos™ 7; SPSS Inc.). The criteria used to determine goodness of model fit were a Root Mean Square Error of Approximation (RMSEA) < 0.06 with a narrow confidence interval, an RMSEA probability value > 0.5, a Comparative Fit Index (CFI) > 0.95, a low Akaike’s Information Criterion (AIC), a standardised Root Mean Square Residual (RMR) < 0.05 and a Tucker-Lewis Index (TLI) > 0.95 (Byrne 2001).

This study was approved by Liverpool John Moores Research Ethics Committee. Participants were recruited from Liverpool John Moores University. Both students and staff were invited to take part. In total, 210 participants took part in this study. Any participants with any missing responses for any questionnaire were excluded to minimize inaccuracies caused by missing values (Tabachnick and Fidell 2007), resulting in the exclusion of 9 participants. The final study sample consisted of 201 participants of which 79.6% were female. Participants were aged 18–81 years old, mean ± standard deviation age was 25.5 ± 10.1 years. As shown in Table 1, socio-economic demographics were similar to Studies 1 and 2, although the sample consisted of more students and fewer participants who were working full-time.

Participants completed the TEAQ as part of an online questionnaire battery. The questionnaire battery additionally included some general questions to obtain demographic data. To investigate the concurrent and discriminant validity of the TEAQ compared to previously published physical touch questionnaires, three additional physical touch questionnaires were included in the questionnaire battery. The 16-item Familial Touch Orientation (FTO) Scale (Gladney and Barker 1979) is a measure of positive physical touch experiences as a child. A low score on this scale represents a high frequency of positive touch experiences during childhood. The Social Touch Questionnaire (Wilhelm et al. 2001) is a 20-item questionnaire focusing on attitudes to physical touch. A low score on this questionnaire represents a more positive attitude towards social touch.

The final physical touch questionnaire included, the Questionnaire on Physical Contact Experience (QPCE) (Cochrane 1990) is an 8-item questionnaire which asks participants to rate on a four-point scale from “None” to “A lot”, how much good, bad and ‘other’ physical contact they experienced during childhood and at present. Two further items ask whether participants believe they were loved as a child and at present. The QPCE produces two dichotomous variables, one for childhood physical contact experience and another for present physical contact experience. Satisfactory physical contact experience, represented by a score of 1, is defined as the experience of substantial good physical contact with no substantial bad physical contact. If these conditions are not met, a score of 0, representing unsatisfactory physical contact experience is given.

To investigate the predictive validity of the TEAQ, the 28-item short form of the Childhood Trauma Questionnaire (CTQ-SF) (Bernstein et al. 2003) was included, as the TEAQ ChT subscale was expected to be negatively predictive of childhood trauma. The CTQ contains five items per subscale, with three additional minimisation questions. The five subscales of this questionnaire relate to childhood sexual, physical and emotional abuse and physical and emotional neglect. Analysis of the factor structure of the childhood trauma questionnaire has identified that using the total score, excluding the 3 minimisation items, is a good fit of the data and a valid use of the questionnaire (Spinhoven et al. 2014), therefore a total CTQ score was calculated and included the analysis.

The 6-item Social Support Questionnaire (SSQ6) (Sarason et al. 1987) was additionally included, as TEAQ subscales relating to current interpersonal touch experience in particular were expected to predict current social support. The SSQ6 was altered slightly. The first question: ‘‘Whom can you really count on to distract you from your worries when you feel under stress?” was replaced with a similar item from the original Social Support Questionnaire (SSQ) (Sarason et al. 1983): “Whom can you really count on to be dependable when you need help?”. It was felt that this item was easier to understand, more general, and more relevant, allowing a wider range of situations to be covered. The SSQ6 was also simplified by asking participants to rate the number of people they could depend on for each circumstance as “None” “A Few” or “Lots” rather than asking participants to name the people they could depend on in each circumstance. As with the original SSQ6, participants rated both the number of people they could depend on and their satisfaction with the support they currently experienced for each item, allowing two scores to be produced; number of social supports (SSQN) and satisfaction with social support (SSQS).

This investigation was carried out online using Qualtrics^® software (Qualtics, Provo, UT), allowing anonymity, wider dissemination and minimal influence of embarrassment, social conformity and pressure to participate. Demographic questions were always completed first, but the remaining questionnaires were completed in a randomised order, except for the FTO, which was always completed last. This is because the FTO defines childhood as birth to 10 years of age, but a cut-off age for childhood for the other two childhood touch measures, the QPCE and the TEAQ ChT scales was deliberately left unspecified (Cochrane 1990). Therefore, to ensure consistency between participants, the FTO was always completed last. After completing these questionnaires, participants were thanked for their participation and provided with a debriefing sheet.

All questionnaires were scored according to authors’ instructions and all data analyses were carried out using IBM^® SPSS^® version 23. Recruitment was not targeted towards participants who had experienced childhood trauma. As such, the majority of participants had very low scores on the CTQ subscales, so this data was not normally distributed. For this reason, bootstrapping was used for multiple regression analysis and Spearman’s rho correlation coefficients are reported.

For all measures used in this study, Cronbach’s alpha was good. The scale with the lowest Cronbach’s alpha was the CTQ physical neglect scale (α = .77) and the highest being for the CTQ sexual abuse scale (α = .96). For the TEAQ subscales, the current intimate touch subscale had the highest Cronbach’s alpha (α = .93) and the attitude to self-care subscale had the lowest Cronbach’s alpha (α = .81).

Cronbach’s alpha for the STQ was .88. interestingly, item number 20 “I like petting animals” had an extremely low correlation with the total STQ score (r = .06). This is the only item relating to touch with animals. STQ item number 12, “As a child, I was often cuddled by family members (e.g., parents, siblings)”, the only item relating to physical touch during childhood and item number 18, “If I had the means, I would get weekly professional massages” also correlated < .3 with total STQ score (r = .25 and .26 respectively), suggesting these items are not strongly related to the underlying construct the STQ is measuring. As the factor structure of the STQ is unknown, this is not necessarily surprising.

A robust hierarchical regression analysis based on 1000 bootstrap samples was carried out using the total CTQ score. All assumptions for this analysis, as described by Field (2013) were met. Although there were no issues with multicollinearity, the familial touch score correlated highly with the TEAQ childhood touch score (r = − .77). To understand the predictive validity of these two measures, these measures were entered into the regression model separately, to understand whether one measure explained significantly more of the variance in total CTQ score than the other. TEAQ childhood touch was entered into the model first and significantly predicted CTQ total score (F_1,199 = 128.30, p < .001), explaining 39.2% of the variance. Addition of FTO total score did not explain significantly more of the variance in CTQ total score (ΔR² = .01, ΔF_1,198 = 2.28, p = .133).

When FTO was entered into the model first, FTO significantly predicted total CTQ score, explaining 28.5% of the variance (F_1,199 = 79.21, p < .001), but addition of TEAQ ChT score explained significantly more of the variance than FTO score alone (ΔR²= .11, ΔF_1,198 = 37.62, p < .001) and when TEAQ ChT score was added, FTO score was no longer significantly predictive of CTQ total score (β = 0.13, p = .141). The predictive validity of TEAQ ChT for CTQ total score appears to be greater than FTO score. Addition of QPCE childhood score to this model explained significantly more of the variance (ΔR² = .08, ΔF_1,197 = 31.03, p < .001), this is likely due to QPCE childhood score taking into account negative as well as positive touch experienced during childhood, so explains some of the variance not explained by either FTO score or TEAQ ChT score. However, when QPCE childhood was entered into the model first, it was significantly predictive of total CTQ score, explaining 35.3% of the variance (F_1,199 = 108.627, p < .001), but addition of TEAQ ChT explained significantly more of the variance (ΔR²= 0.120, ΔF_1,198 = 44.991, p < .001). This highlights the discriminant validity of the TEAQ ChT scale in that the variance explained by all three childhood touch measures is not entirely conflated. Addition of all adult touch measures did not explain significantly more of the variance than the childhood touch measures alone (ΔR² = .03, ΔF_7,190 = 1.59, p = .142).

Overall, the model with the 3 childhood touch predictors explained 48.1% (R² = .48) of the variance in total CTQ score with adjusted R² being very similar (adjusted R² = .47), suggesting the model to be generalizable. Results are presented in Table 7.

To understand the relationship between TEAQ ChT score and childhood trauma, a robust hierarchical regression analysis based on 1000 bootstrap samples was carried out to determine which childhood trauma factor was most predictive of TEAQ ChT score. We had no a priori hypotheses about which childhood trauma factors we expected to be most predictive of TEAQ ChT score, so all predictors were entered into the model simultaneously. All assumptions for this analysis, as described by Field (2013) were met.

Overall, the model significantly predicted TEAQ ChT score (F_5,195 = 60.78, p < .001), explaining 60.9% (R² = .61) of the variance, with adjusted R² being similar (adjusted R² = .60), suggesting the model to be generalizable. Childhood emotional neglect was significantly negatively predictive of TEAQ ChT (β = − .85, p = .001). The other childhood trauma factors were not significantly predictive of TEAQ ChT score. Results are presented in Table 8.

A hierarchical robust regression analysis based on 1000 bootstrap samples was carried out using satisfaction with social support as the outcome variable. All assumptions for this analysis, as described by Field (2013) were met. To investigate the predictive validity of the TEAQ, all TEAQ factor scores were entered into the model, then it was determined whether the touch scores from the other questionnaire measures explained significantly more of the remaining variance. If not, then the other questionnaire measures would not be more predictive of satisfaction with social support than the TEAQ. It was expected that out of the TEAQ measures, the factors relating to current touch experience would be more predictive of satisfaction with social support than the factors relating to attitude to physical touch and that all factors relating to current attitudes and experiences of physical touch would be more predictive of satisfaction with social support than the childhood touch measure. For these reasons, the two factors relating to current touch experience, FFT and CIT were added to the model in the first block, followed by the TEAQ attitude factors, ASC, AIT and AUT in the second block. The third block contained the TEAQ ChT factor. The remaining touch measures were then added. Out of these measures, it was again predicted those relating to current touch experience would be more predictive of satisfaction with social support than those relating to childhood touch experiences, so the two factors relating to current touch experience, QPCE present an STQ were entered in the fourth block followed by QPCE childhood and FTO total in the fifth block. All assumptions for this analysis, as described by A. Field (2013) were met.

The two TEAQ factors relating to current touch, FFT and CIT were entered into the model first and explained a significant amount of the variance in satisfaction with social support (F_2,198 = 25.99, p < .001), explaining 20.8% of the variance. Addition of the TEAQ measures relating to attitudes to physical touch, ASC, AIT and AUT, explained significantly more of the variance in satisfaction with social support (ΔR² = .06, ΔF_3,195 = 5.17, p = .002). Addition of the TEAQ ChT measure explained significantly more of the variance (ΔR² = .02, ΔF_1,194 = 4.19, p = .042). Addition of the two remaining current touch measures QPCE present and STQ, explained significantly more of the variance (ΔR² = .03, ΔF_2,192 = 4.54, p = .012). Addition of the remaining childhood touch measures, QPCE childhood and FTO, did not explain significantly more of the variance (ΔR² = .01, ΔF_2,190 = 1.22, p = .298).

As the addition of the remaining childhood touch measures, QPCE childhood and FTO, did not explain significantly more of the variance, these measures were removed from the model. The final results for this analysis are presented in Table 9.

Overall, the final model which included all 6 TEAQ factors and the QPCE present and STQ measures explained 31.4% (R² = .31) of the variance in satisfaction with social support with adjusted R² being similar (adjusted R² = .29), suggesting the model to be generalizable. The final model identified the TEAQ CIT subscale as most predictive of satisfaction with social support (β = .51, p = .001), followed by the TEAQ AIT subscale (β = − .37, p = .002), followed by the STQ (β = − .25, p = .040). As expected, low levels of current intimate touch predicted lower satisfaction with social support. Additionally, those rating intimate touch more positively, in terms of TEAQ AIT score and those with a more negative attitude to social touch in general in terms of STQ score, with more anxiety and avoidance of touch situations, had lower satisfaction with social support.

Before excluding participants due to missing data, the sample consisted of 1509 participants, of which 73.2% were female. Mean ± standard deviation age was 27.0 ± 9.6 years. Further demographic information for the sample is presented in Table 1.

For all analyses, IBM^® SPSS^® version 23 was used. For the analysis of the effect of gender on TEAQ responses, independent samples t tests were used. Due to the very large sample size, Cohen’s d effect sizes were also calculated and effect sizes greater than .20 taken to be effects of interest (Cohen’s d: 0.2—small effect size; 0.5—medium effect size; 0.8—large effect size).

For the analysis of marital status, most participants could be categorized as single, in a relationship or married/cohabiting (Table 1), with similar numbers of participants per group. One-way ANOVAs were used to compare these three groups for the 6 TEAQ subscales. When Levene’s test was significant, meaning the variance between groups was unequal, Welch’s F has been reported with Games-Howell post hoc test results. When variance between groups was equal, as determined by the Levene’s test, Hochberg’s GT2 post hoc tests were used. For the ANOVA results, effect size (r) was calculated by calculating eta squared, then deriving the square root of this number (Field 2013). Recommended cut-offs for r are > .10 = small effect, > .30 = medium effect and > .50 = large effect. Thus, effect sizes < .10 were not deemed of interest, so post hoc results are not reported. For post hoc tests, Cohen’s d effect sizes were calculated.

For the independent samples t tests investigating gender effects and the one-way ANOVAs investigating the effect of marital status, p values are reported uncorrected for repeated measures. However, as there are 6 TEAQ subscales, a Bonferroni correction was applied and only p values < .0083 deemed significant.

Pearson’s correlation coefficients (r) were calculated to investigate the relationship of each TEAQ subscale with age. Due to the sample size, very small correlation coefficients were significant, so only those greater than .3, explaining at least 9% of the variance were deemed of importance. Due to Hall and Veccia (1990) identifying an influence of gender on the relationship between age and touch behaviours, correlations were additionally determined for each gender separately. For all analyses, normality was assumed due to the central limit theorem.

The participant reporting their gender as ‘other’ was not included in the analysis, as a single response did not enable any group comparisons and so the analysis compared responses of males to females. The sample for the analysis of gender effects consisted of 1106 females and 401 males. As shown in Fig. 1, independent samples t tests revealed the TEAQ FFT score for females (M = 3.58, SD = 0.96) was significantly greater than males (M = 3.11, SD = 0.88), with a medium effect size (t_768.748 = 8.89, p < .001, Cohen’s d = 0.50). TEAQ CIT scores were significantly higher for females (M = 3.48, SD = 0.95) than males (M = 3.19, SD = 0.97), with a small effect size (t₁₅₀₅ = 5.19, p < .001, Cohen’s d = 0.30). TEAQ ChT scores were significantly higher for females (M = 3.82, SD = 0.99) than males (M = 3.44, SD = 0.85) with a small effect size (t_810.827 = 7.38, p < .001, Cohen’s d = 0.40). TEAQ ASC scores were significantly higher for females (M = 3.69, SD = 0.90) than males (M = 2.52, SD = 0.89) with a large effect size (t₁₅₀₅ = 22.45, p < .001, Cohen’s d = 1.31). TEAQ AIT scores for males (M = 4.18, SD = 0.64) and females (M = 4.25, SD = 0.70) were comparable (t_770.458 = 1.90, p = .057, Cohen’s d = 0.11). Finally, TEAQ AUT scores were significantly greater for males (M = 3.29, SD = 0.89) than females (M = 3.07, SD = 0.90) with a small effect size (t₁₅₀₅ = 4.06, p < .001, Cohen’s d = 0.24).

One-way ANOVAs were used to determine if there was any effect of marital status on TEAQ score. Total sample size was 1472 for this analysis (Males = 395, Females = 1075, ‘Other’ = 1, not stated = 1). There were 568 single participants, 528 participants in a relationship and 376 married/cohabiting participants.

Following Bonferroni correction (uncorrected p values are reported), no significant effect of marital status on TEAQ FFT (F_2,1469 = 3.96, p = .019, r = 0.07) or TEAQ AUT was identified (F_2,1469 = 0.95, p = .389, r = 0.04). A significant effect of marital status on TEAQ ChT was identified, just surviving correction for multiple comparisons, however, the effect size was not substantial (F_2,882.351 = 4.83, p = .008, r = 0.08). A significant effect of marital status on TEAQ ASC was also identified and survived correction for multiple comparisons, however the effect size was not substantial (F_2,1469 = 5.53, p = .004, r = 0.09).

As shown in Fig. 2, a significant, large effect of TEAQ CIT was identified, (F_2,874.198 = 302.60, p < .001, r = 0.53). Games-Howell post hoc analysis revealed all groups to be significantly different. TEAQ CIT score was significantly higher for those in a relationship (M = 3.93, SD = 0.67) compared to those who were single (M = 2.80, SD = 0.86) with a large effect size (t_1058.592 = 23.24, p < .001, Cohen’s d = 1.46) and those married/cohabiting (M = 3.65, SD = 0.89) with a small effect size (t_657.638 = 5.28, p < .001, Cohen’s d = 0.37). Those who were married/cohabiting had significantly greater TEAQ CIT scores than those who were single with a large effect size (t_784.212 = 14.44, p < .001, Cohen’s d = 0.97).

A small, but significant effect of marital status on TEAQ AIT was identified and survived correction for multiple comparisons (F_2,877.854 = 50.61, p < .001, r = 0.24). Games-Howell post hoc analysis revealed a significant difference between all groups. Those in a relationship had a significantly more positive attitude to intimate touch (M = 4.44, SD = 0.54) than those who were single (M = 4.07, SD = 0.73) with a medium effect size (t_1040.038 = 9.75, p < .001, Cohen’s d = 0.58) and those who were married/cohabiting (M = 4.20, SD = 0.71) with a small effect size (t_664.776 = 5.53, p < .001, Cohen’s d = 0.39). Those who were married/cohabiting had a significantly greater TEAQ AIT score than those who were single with a non-substantial effect size (t_818.978 = 2.84, p = .013, Cohen’s d = 0.19).

Eight participants did not report their age, so were excluded from the analysis, leaving a sample size of 1501 (1100 females, 398 males, 1 ‘Other’, 2 not stated). Pearson’s r for all factors was < .3, so no correlations of interest were identified (TEAQ FFT: r = − .03, p = .211; TEAQ CIT: r = − .13, p < .001; TEAQ ChT: r = − .21, p < .001; TEAQ ASC: r = − .09, p < .001; TEAQ AIT: r = − .09, p < .001; TEAQ AUT: r = − .004, p = .877). Analysis of females only still identified no correlation coefficients > .3 (TEAQ FFT: r < .001, p = .995; TEAQ CIT: r = − .14, p < .001; TEAQ ChT: r = − .21, p < .001; TEAQ ASC: r = − .09, p = .003; TEAQ AIT: r = − .11, p < .001; TEAQ AUT: r = .02, p = .521). Similarly, for males only, no correlation coefficients > .3 were identified (TEAQ FFT: r = − .09, p = .062; TEAQ CIT: r = − .10, p = .050; TEAQ ChT: r = − .19, p < .001; TEAQ ASC: r = − .09, p = .088; TEAQ AIT: r = − .06, p = .253; TEAQ AUT: r = − .08, p = .102).