The Relationship between Interpersonal Touch and Attachment Organization
Supervisor: Torun Lindholm
MASTER THESIS, VT 2016
THE RELATIONSHIP BETWEEN INTERPERSONAL TOUCH AND ATTACHMENT ORGANIZATION
The aim of this study was to assess the relationships between interpersonal touch, attitudes towards touch, and attachment organization. using a sample of Amazon Mechanical Turk workers (N = 420) and a battery of tests including a short form of the Experience in Close Relationships (ECR-S), Touch Avoidance Questionnaire (TAQ), and the newly developed Interpersonal Touch Scale (IPTS). In line with previous research, the results revealed that attachment avoidance have significant limiting impacts on type specific touch behaviors. Mediation of touch avoidance could fully account for the relationship between attachment avoidance and several type specific touch behaviors, but not for hitting behavior (which also was correlated with attachment anxiety). No significant target specific relationships were observed except for a moderate correlation between attachment avoidance and lower frequency of partner touch, partly mediated by touch avoidance.
Touch is one of the first senses to develop in the embryo and a vital part of the human experience from the cradle to the grave (Barnett, 1972; Gallace, & Spence, 2010). Not only does it alert us of potential dangers to our health and well-being; it is also a sense of pleasure and interpersonal communication (Finnegan, 2005). As with verbal communication, we all use and interpret tactile stimulation differently (Elfenbein, & Ambady, 2002), giving and receiving with varying levels of comfort in relation to specific others such as partners, children, friends and strangers (Deethardt, &
Hines, 1983; Fromme et al., 1989). The basis for these preferences are most likely developed in infancy when other ways of communication are limited, with the touch of our parents signaling the trust and care that will shape our social behaviors later in life (Harlow, 1958; Takeuchi et al., 2010).
For example, touch aversion among infants have been observed to be significantly higher in those whose mothers did not provide affectionate touch while breastfeeding during first half-year of life (Feldman, Keren, Gross-Rozval, & Tyano, 2010; Field et al., 1994). Deprivation of touch during infancy has also been related to later cognitive (MacLean, 2003) and neurodevelopmental delays (Chugani et al., 2001) which might persist into late adolescence (Beckett et al., 2006).
As adults we use interpersonal touch to express emotions and signal intentions among our peers (Hertenstein et al., 2006), which in turn regulates our felt security and emotional well-being (Field, 2001; Spence, 2002). However, social touch is also highly contextual and various types of touch might mean very different things when applied in different environments (Jones, & Yarbrough, 1985). A firm handshake signals mutual respect in a work meeting while a gentle stroke on the arm would seem odd and possibly inappropriate, vice versa for a romantic evening with a partner. Even the slightest nudge from a stranger have a tendency to elicit powerful emotional experiences with feelings of intrusion or arousal as a consequence. The art and intensity of these reactions seems to be sensitive to the cultural context in which the touch behavior is performed (McDaniel, & Andersen, 1998; Remland, Jones, & Brinkman, 1995). A kiss on the cheek when greeting an opposite sex friend in a public environment might be perceived in vastly different lights in France and Iran (McElroy, &
Vahdat, 2014, May 22; Remland, & Jones, 1988).
Importantly, touch behavior should be distinguished from attitudes towards touch. Most research conducted on attitudes towards touch has focused on gender differences (Andersen, 2005;
Andersen, & Leibowitz, 1978; Stier, & Hall, 1984). A study using the Touch Avoidance Measure (TAM) in a student population found that men, especially those who identified with more traditionally
“masculine” traits, reported a higher rate of same-sex, but not opposite-sex touch avoidance than
females (Martin, & Anderson, 1993). Men have also been shown to score higher in homophobic attitudes, which correlates with same-sex touch avoidance and mediate same-sex touch behavior (Roese et al., 1992). Women tend to report higher levels of opposite-sex touch avoidance than men (Andersen, Andersen, & Lustig, 1987). Noteworthy is that people high in androgynous traits report less same-sex and opposite sex touch avoidance (Crawford, 1994). In romantic relationships the pattern of opposite sex touch avoidance seems to change depending on relational stage. A curvilinear relationship between relational stage (beginning, intermediate, & stable stage) and frequency of touch has been observed in cross-sex couples, with level of touch avoidance corresponding with frequency (Guerrero, & Andersen, 1994). In casual relationships men initiate the touching of women more than the opposite, but vice versa in serious and married couples (Guerrero, & Andersen, 1994).
Comfort with touch has previously been shown to predict willingness to participate in activities involving intimate touch behaviors and seems related to one’s openness to express intimate behavior (Fromme et al., 1989) and personal distance (Andersen, & Sull, 1985). Like aversion towards other forms of intimacy, some interpersonal touch and attitudes towards touch has previously been predicted by attachment organization. The attachment organization of an individual is formed – starting in infancy – by the social interplay between the subject and its primary caregiver (i.e.
attachment figure, see Ainsworth, 1973; Bowlby, 1969). When a child is in distress its attachment system (crying, arms outstretched etc.) is activated and it will seek solace in the caregiver. A caregiver who provides a safe haven of support and comfort for the infant allows for a secure attachment organization to develop. This organization will iterate in other social contexts throughout life with a sense of emotional security in close relationships (Dykas, & Cassidy, 2011). If a caregiver is absent or rejective of the infant, it will learn to suppress the attachment system by becoming emotionally avoidant in close relationships. The individual will learn that expressing emotions leads to abandonment and will thus refrain from doing so in order to avoid emotional pain. If a caregiver shows incongruent or fearful behavior when the infant’s attachment system is triggered it will never achieve any sense of security in adulthood, but instead continue to signal need and anxiety in close relationships. Unlike the avoidantly attached, the anxiously attached individual fears that if they stop expressing negative emotions they will be forgotten and abandoned by their peers. If the caregiver is abusive or neglecting, the risk of the child becoming dissociated, expressing unpredictable, aggressive, and even violent behavior, is high (Dutton, 2000).
Securely attached individuals has been shown to be more prone to non-verbal communication – including laughter, receptive gaze, facial pleasantness and touch with their partners – than both anxious and avoidant individuals (Tucker, & Anders, 1998). In a study using Bartholomew’s four attachment prototypes (i.e. secure, fearful-avoidant, preoccupied, dismissing: Bartholomew, 1991;
Fraley, 2005) , fearful-avoidant individuals have been observed to engage the least in non-verbal communication and prefer a longer interpersonal distance than both preoccupied and dismissives (Guerrero, 1996). People on the avoidant end of the spectrum are also more reactive to intrusion of their personal space, especially when the intruder is male (Kaitz, Bar-Haim, Lehrer, & Grossman, 2004). In anxiety provoking situations highly avoidant people tend to seek and provide less emotional support and physical comfort than less avoidant, anxious and securely attached individuals (Simpson, Rholes, & Nelligan, 1992).
While the literature on the consequences of touch as a means of non-verbal communication is
rather extensive, the knowledge of the general patterns and frequencies of different types of
interpersonal touch is lacking. We know the why but not the how. One of the most prominent
problems of studying interpersonal touch frequency is inconsistent methodology, with observational
studies being the most common (Stier, & Hall, 1984). Previous studies on interpersonal touch have
most often used observational methods in laboratory and natural environments ( Gallace, & Spence,
2010) . While certainly the most reliable method of collecting data on touch behavior, it can be highly
time consuming when dealing with a large number of participants (Schutte, Malouff, & Adams,
1984). This might be unavoidable when investigating for example how touch behavior changes under
different experimental conditions. In other cases researchers might be more interested in individuals’
general touch behavior, in which a self-report measure could provide a faster albeit reliable measure.
However, at present time few alternative methods of investigation have been empirically tested (e.g.
Jourard, 1966; Suvilehto et al., 2015). Therefore, a newly developed interpersonal touch self-report scale by Sorokowska and Croy (in preparation) was used to pertain the aim of this study; to assess interpersonal touch and attitudes towards touch in relation to attachment organization.
The main research question is whether attachment organization predict certain patterns of touch behavior, and if these potential relationships are mediated by attitudes towards touch. Given the previous research on touch behavior in relation to attachment organization, it is hypothesized that individuals high in attachment avoidance will display less frequent touch behavior in relation to both type and target specific touch. Furthermore, the strength of the negative correlations is assumed to be higher for intimate touch behaviors such as hugs and kisses compared to more casual behaviors such as handshakes. Likewise, it is suggested that the more intimate the relationship the stronger the correlation. Thus, the strength of the negative correlation between attachment avoidance and the frequency of target specific touch behaviors is hypothesized to be higher in relation to partners and youngest children compared to friends and strangers. Opposite of the intuitive notion of anxiously attached individuals seeking attention through physical proximity, it has been shown that they do not display more frequent touch behavior than securely attached (Simpson, Rholes, & Nelligan, 1992).
This should also be evident in the results of this study. Also, anxious and fearful-avoidant attachment organization have previously been demonstrated to predict violent behavior (Dutton, 2000; Orcutt, Garcia, & Pickett, 2005; Wekerle, & Wolfe, 1998) and should therefore be reflected in the type specific touch category ‘hit’. In relation to mediation, studies have found inconsistent results regarding the relationship between touch behavior and attitudes towards touch (Jones, 1986; Jones &
Brown, 1996; Roese et al., 1992). These studies have used many different constructs encompassing a wide range of targets and type specific touch behaviors. It is therefore hypothesized that mediation of attitudes between attachment organization and interpersonal touch will depend on type and target specificity.
Respondents (≥18 y/o) were recruited to participate in a study of “social relations and behavior” via Amazon Mechanical Turk (mTurk) without additional worker requirements. Reward for participation was $0.50 and received after completion of the study. Initial pilot testing showed that the time estimated to complete the study with reliable scores was approximately seven minutes. By a liberal criteria, completion rates under four minutes was deemed to contain an unfavorable amount of random responses and were rejected from further analysis. Based on previous research using mTurk in relation to ECR-S (Oldmeadow, Quinn, & Kowert, 2013; Richman, DeWall, & Wolff, 2015;
Wickham, Reed, & Williamson, 2015) sample size was set to 400, of which a total of N = 420 (44.3%
= 34.77, SD age
= 10.58) were accepted for analysis. Descriptives of demographics showed that 50.2% was citizens of India and the remainder US citizens. Sexual orientation (83%
heterosexual, 5% homosexual, 7% bisexual &, 3% preferred not to say) showed no significant variation in relation to gender specific touch and was not taken into account in further analyses.
Before participation all respondents gave their informed consent in accordance with the ethical guidelines of the Department of Psychology at Stockholm University.
Experiences in Close Relationships – Short form (ECR-S). The ECR-S (Wei, Russell,
Mallinckrodt, & Vogel, 2007) is a 12-item short version of the experiences in close relationships
scale, which is used to assess a general pattern of adult attachment. It consists of two subscales,
avoidance and anxiety, which can be interpreted independently as well as in connection with
Bartholomew’s four attachment prototypes. In the latter case, the data is analyzed in a continuous fashion resulting in a two-dimensional plot with low-high anxiety on one axis and low-high avoidance on the other axis. The four prototypes aligns in a 45° angle across the two dimensional space.
Participants rate how well the statements of each item reflect their typical feelings in romantic relationships. Answers to items were indicated on a partly anchored 7-point visual analog scale (VAS), ranging from 1 (strongly disagree) to 7 (strongly agree). The internal consistency of the two subscales in this sample was acceptable (Cronbach’s α: avoidance = .80; anxiety = .77) and did not differ notably with the exclusion of any item.
Touch Avoidance Questionnaire (TAQ). The TAQ was developed by Ozolins and Sandberg (2009) to assess levels of touch avoidance in relation to attachment organization. It is comprised of five factors (partner, family, same sex, opposite sex, & strangers) which can be interpreted separately or together as a measure of general touch avoidance. As suggested by previous research (Andersen et al., 1987; Deethardt, & Hines, 1983; Fromme et al., 1989), the items reflect subjective feelings of different touch contexts, such as touching a partner, siblings, parents, friends of same or opposite sex, and complete strangers. Answers to the items were indicated on a partly anchored 7-point VAS, ranging from 1 (strongly disagree) to 7 (strongly agree). Based on the factor analysis by Ozolins and Sandberg, items with factor loadings under .60 and double loading items (8 in total) was removed a priori to ensure reliability. Subsequent factor analysis (PCA, direct oblimin) of the remaining 29 items revealed five factors with eigenvalue over one and a scree plot “leveling off” after five factors, replicating previous findings. The Kaiser-Meyer-Olkin measure of sampling adequacy was .83, and Bartlett’s test of sphericity was significant ( χ2
 = 4800.48; p<.00) . The internal consistency of four of the five subscales in this sample was acceptable ( Cronbach’s α: partner = .82; family = .77;
same sex = .81; opposite sex = .86) while alpha for the subgroup strangers (α = .63) was questionable.
However, the internal consistency for all items was acceptable (total α = .84) and did not differ notably with the exclusion of any item.
Interpersonal Touch Scale (IPTS). The IPTS is a newly developed measure to assess interpersonal touch (Sorokowska, & Croy, in preparation). It was here presented in a matrix consisting of six targets (partner, youngest child, female friend, male friend, female stranger & male stranger) and seven type specific touch behaviors (casual contact, handshake, embrace, stroke, kiss, hit, & hug).
There were two segments to the IPTS with the first consisting of control items such as “are you currently involved in a romantic relationship?”, “do you work in physical proximity to other people?”
and “what is the age of your youngest child (if you have children)?”. In the second and main segment, respondents indicated what type of touch they had engaged in with each of the specific targets during the last week. Type of touch was illustrated with stylized images displayed in the left-most column followed by multiple choice radio buttons for each target in the consecutive columns. A factor analysis (PAF, direct oblimin) was conducted using all 42 type/target items (e.g. hug/partner, stroke/male friend, hit/female stranger, etc.). The analysis yielded 13 factors with eigenvalue over one, and a scree plot “leveling off” after six factors. The Kaiser-Meyer-Olkin measure of sampling adequacy was .78, and Bartlett’s test of sphericity was significant ( χ2
 = 6539.12; p<.00). The internal consistency of the total number of items in this sample was acceptable (Cronbach’s α = .78) with target specific subscale consistency varying depending on specificity of target ( Cronbach’s α:
partner = .85; youngest child = .86; female friend = .72; male friend = .69; female stranger = .56; male
stranger .46). All type specific subscales yielded unacceptable alpha levels. Given these results it is
postulated that the scale reliably measures six target specific factors consisting of seven type specific
touch behaviors. After an initial overall correlational analysis, the specific touch behavior “hit” was
excluded from the total and target specific factors since it was shown to correlate in the opposite
direction with the other type specific touch behaviors.
The survey was completed using Qualtrics web-based platform. The order of the measures were identical for each respondent, starting with a demographic inquiry followed by a battery of tests. All items of the ECR-S and the TAQ were randomized for each respondent, while the structure of the IPTS remained fixed across-the-board. Answers to both the ECR-S and the TAQ items were indicated using a VAS comprising of a continuous 1 to 7 range with two decimals (hidden from the respondent’s view). The response indicator was located at mid-point by default. This approach has previously been suggested to increase response accuracy, task engagement and reduce satisficing in trained panelists by demanding more cognitive effort from respondents ( Reips, & Funke, 2008;
Roster, Lucianetti, & Albaum, 2015 ).
While the VAS might encourage more thoughtful responses from participants, the task also becomes more time consuming leading to a higher rate of missing data and break-offs (Couper, Tourangeau, Conrad, & Singer, 2006; Funke, Reips, & Thomas, 2011). To counteract missing data, forced responses were required before completion of each survey segment. The main part of the IPTS used a matrix of multiple choice responses which indicated type and target specific interpersonal touch. To counter the potential problem of uniform answers due to high cognitive effort and confusion when using an internet-based matrix design (Gräf, 2002), the target specification was iterated for each type of touch. At the end of the survey the respondents received an automatically generated code for validation of mTurk HIT completion, which also ensured full participation before payment.
All analyses were made using SPSS for Windows, version 23. The descriptives for the ECR-S constructs of total sample were compared to previous results (Fraley, 2005; Fraley, Heffernan, Vicary,
& Brumbaugh, 2011). No significant difference between the samples was found.
Zero-order correlations of ECR-S subscales (anxiety, avoidance), TAQ total and subscales (partner, family, same sex, opposite sex, strangers) and IPTS total, target- (partner, youngest child, female friend, male friend, female stranger, male stranger) and type (casual contact, handshake, embrace, stroke, kiss, hit, hug) specific subscales for total sample were then calculated (see Tables 2, 3 & 4). Significant correlations were extracted in the next step into a series of regression analyses based on the work of Picardi, Toni and Caroppo (2005), resulting in three sets of six to seven regression models. In each regression model, gender, age, country, avoidance and anxiety were used as independent variables, and for each separate model the TAQ total and subscales, the IPTS total, type and target specific subscales were used as dependent variables. In the IPTS partner (n = 338) and IPTS youngest child (n = 219) regression models, cases were selected on the basis of the respondents answers to the IPTS control questions. Before the final construction of the regression models, the Mahalanobis’ distance was computed to search for multivariate outliers (Tabachnick, & Fidell, 2013).
Four outliers were identified and removed from further analyses.
After establishing the significant regression coefficients, path analyses were conducted in order to assess potential mediation of touch avoidance between attachment organization and interpersonal touch (Figure 1). A four-step procedure of regressions (Kenny, Kashy, & Bolger, 1998;
Lindholm, 2008) was used to (1) test path c with the independent variable attachment organization as
a predictor of the dependent variable interpersonal touch, (2) test path a with attachment organization
as predictor of the mediating variable touch avoidance, (3) test path b with both attachment
organization and touch avoidance as predictors of the dependent variable interpersonal touch and, (4)
test path c′ with attachment organization as a predictor of interpersonal touch controlling for touch
avoidance. Step 3 and 4 was calculated in the same regression equation and the indirect effect of path
ab was defined as the product of ( path a × path b). An interactive calculation tool for a z-distributed
modified Sobel test was used to test the significance of the indirect path (Preacher, & Leonardelli,
2001). Post hoc Holm-Bonferroni corrections of the p-values for the standardized regression
coefficients in each of the three sets of models – as well as for the path analyses – were conducted
using an Excel calculator developed by Gaetano (2013).
Figure 1. Path model of the relationship between attachment style, touch avoidance, and interpersonal touch.
Mean scores and standard deviations of ECR-S subscales, TAQ total and subscales, IPTS total and type/target specific subscales for total sample are presented in Table 1. A two-way MANOVA with country and gender as independent variables revealed that there was a significant difference in attachment avoidance and anxiety between the US and the Indian sample (Wilks’ Λ = .94, F[5, 415] = 13.59, p < .00), but not between the genders (Wilks’ Λ = 1, F[5, 415] = .89, p = .41). No significant interaction effect was observed (Wilks’ Λ = 1, F[5, 415] = .41, p = .68).
Touch Avoidance and Attachment Organization
Table 2 summarizes the results of the regression analyses exploring the relationships between the independent variables gender, age, country and ECR-S subscales (avoidance, anxiety) in relation to the dependent variables TAQ total and subscales (partner, family, same sex, opposite sex, &
strangers). Each model contains F-statistics, R, R 2
, R 2
adjusted, zero-order correlation, the standardized regression coefficient ( β) and the semi-partial correlation squared. The latter displays the unique contribution of the independent variable on the variance of the dependent variable (e.g. how much R2
would decrease if that variable was removed from the regression equation).
In the TAQ total, partner, family and stranger regression models the independent variable attachment avoidance was the only significant regression coefficient, showing (in order of appearance left to right) moderate, strong and weak negative correlations with the dependent variables. In the TAQ same sex regression model the independent variables attachment avoidance and gender both showed significant albeit weak negative correlations with the dependent variable, while only gender showed a significant weak correlation in the TAQ opposite sex regression model.
Target Specific Interpersonal Touch and Attachment Organization
Table 3 summarizes the results of the regression analyses exploring the relationships between the independent variables gender, age, country and ECR-S subscales in relation to the dependent variables IPTS total and target specific subscales (partner, youngest child, female friend, male friend, female stranger, & male stranger).
In the IPTS total regression model the independent variables attachment avoidance and
country showed significant but weak negative correlations with the dependent variable, while only
attachment avoidance was significant and moderately negatively correlated in the IPTS partner
regression model. No significant correlations were observed in the IPTS youngest child regression
model. Age was the only significant coefficient in the IPTS female friend regression model with a
weak negative correlation with the dependent variable. In the IPTS male friend regression model the
independent variable gender showed a significant moderate negative correlation and the independent
variable country a significant but weak negative correlation with the dependent variable. No
significant correlations were observed in the IPTS female friend regression model, and only the
7Table 1. Mean scores and standard deviations of ECR-S subscales, TAQ total and subscales, IPTS total and type/target specific subscales for total sample. ECR-SM (SD) TAQM (SD) IPTS targetM (SD) IPTS typeM (SD) Avoidance2.84 (1.07) Total3.61 (0.74) Total10.41 (4.98)Casual 2.19 (1.32) Anxiety3.63 (1.21) Partner2.87 (1.13) Partner3.60 (2.27) Handshake2.03 (1.55) Family 4.04 (1.26) Youngest child2.08 (2.33) Embrace 1.76 (1.02) Same sex 3.44 (1.10) Female friend1.85 (1.80) Stroke 1.33 (0.94) Opposite sex 3.56 (1.28) Male friend1.93 (1.72) Kiss 1.34 (0.75) Stranger 4.14 (1.22) Female stranger0.69 (1.04) Hit 0.54 (0.74) Male stranger0.80 (0.98) Hug1.77 (1.03)
8Table2.MultipleregressionanalyseswithTAQtotalorsubscalescoresasdependentvariable,andgender,age,country,ECR-Ssubscalescoresasindependentvariables.Note: p-values of the standardized regression coefficients adjusted with a post hoc Holm-Bonferroni correction. TotalPartnerFamily Same sex Opposite sex Stranger F35.33 (3, 419) 108.81 (3, 419) 10.60 (1, 419) 19.33 (3, 419) 13.42 (2, 419) 11.95 (3, 419) R.45 .66 .16 .35 .25 .28 R2.20 .44 .03 .12 .06 .08 R2 Adjusted .20 .44 .02 .12 .06 .07 r βsr2r βsr2r βsr2r βsr2r βsr2r βsr2 Gender (0=♀, 1=♂) .21 .18 .03 -.22-.23.05 Age-.15-.09.01 Country (0=India, 1=USA) .11 -.03.00 -.10.02 .00 Avoidance.45 .42 .15 .66 .62 .31 .16 .16 .02 .29 .23 .04 .10 .12 .01 .23 .15 .01 Anxiety.23 .05 .00 .35 .09 .01 .20 .09 .01 .23 .14 .01 Bold ▪ p ≤ .05; underscore ▪ p ≤ .01; r ▪ zero-order correlation; β ▪ standardized regression coefficient; sr2 ▪ squared semipartial correlation.
9Table3.MultipleregressionwithIPTStotalortargetspecificsubscalescoresasdependentvariable,andgender,age,country,ECR-Ssubscalescoresasindependentvariables. Note: p-values of the standardized regression coefficients adjusted with a post hoc Holm-Bonferroni correction. TotalPartnerYoungest childFemale friendMale friendFemale strangerMale stranger F10.54 (3, 419) 29.00 (2, 337) 7.45 (2, 218) 10.39 (2, 419) 23.97 (5, 419) 4.99 (3, 419) 6.81 (3, 419) R.27 .35 .25 .22 .47 .19 .22 R2.07 .12 .07 .05 .23 .04 .05 R2 Adjusted .06 .12 .06 .04 .22 .03 .04 r βsr2r βsr2r βsr2r βsr2r βsr2r βsr2r βsr2 Gender (0=♀, 1=♂)-.10-.12.01 .41 .35 .12 .15 .13 .02. Age-.11-.11.01 -.18-.20.01 -.22-.13.01 -.12-.08.00 -.13-.13.02 Country (0=India, 1= USA) -.16-.17.03 -.27-.16.02 Avoidance-.15-.20.04 -.35-.36.09 -.24-.20.01 -.10.00 .00 .13 .07 .00 Anxiety-.12.04 .00 -.18-.10.00 -.16.06 .00 .15 .10 .01 .10 .05 .01 Bold ▪ p ≤ .05; underscore ▪ p ≤ .01; r ▪ zero-order correlation; β ▪ standardized regression coefficient; sr2 ▪ squared semipartial correlation.
10Table4.MultipleregressionwithIPTStotalortypespecificsubscalescoresasdependentvariable,andgender,age,country,ECR-Ssubscalescoresasindependentvariables. Note: p-values of the standardized regression coefficients adjusted with a post hoc Holm-Bonferroni correction. Casual HandshakeEmbrace Stroke Kiss Hit Hug F (5, 419) 14.64 (1, 419) 14.42 (3, 419) 13.38 (1, 419) 7.68 (3, 419) 7.72 (2, 419) 34.28 (5, 419) 8.82 (1, 419) R.18 .31 .18 .23 .19 .54 .14 R2 .03 .10 .03 .05 .04 .29 .02 R2 Adjusted .03 .09 .03 .05 .03 .28 .02 r βsr2 r βsr2 r βsr2 r βsr2 r βsr2 r βsr2 r βsr2 Gender (0=♀, 1=♂).19 .14 .01 .13 .02 .00 Age-.18-.13.01 -.10-.06.00 -.14.01 .00 Country (0=India, 1= USA) -.24-.19.03 -.22-.20.03 -.12-.14.02 -.46-.38.13 Avoidance-.18-.18.03 -.18-.18.03 -.12-.14.02 .29 .14 .01 -.14-.14.02 Anxiety.11 -.04.00 .37 .21 .03 Bold ▪ p ≤ .05; underscore ▪ p ≤ .01; r ▪ zero-order correlation; β ▪ standardized regression coefficient; sr2 ▪ squared semipartial correlation.
independent variable gender showed a weak negative correlation with the dependent variable in the IPTS female friend regression model.
Type Specific Interpersonal Touch and Attachment Organization
Table 4 summarizes the results of the regression analyses exploring the relationships between the independent variables gender, age, country and ECR-S subscales in relation to the dependent variables IPTS type specific subscales (casual contact, handshake, embrace, stroke, kiss, hit, & hug).
In the IPTS casual contact, embrace and hug regression models the independent variable attachment avoidance was the only significant regression coefficient, showing weak negative correlations with the dependent variables. In the IPTS handshake regression model the independent variables gender and country both showed significant albeit weak negative correlations with the dependent variable, while only country showed a significant weak correlation in the IPTS stroke regression model. In the IPTS kiss regression model the independent variables country and attachment avoidance showed significant but weak negative correlations with the dependent variable. In the IPTS hit regression model the independent variable country showed a significant moderate correlation with the dependent variable, while the independent variables attachment avoidance and attachment anxiety showed significant but weak positive correlations.
Table 5. Standardized regression coefficients (
β ) from the path analyses of the relations between ECR-S subscales
(avoidance, anxiety), TAQ total and subscale (partner), IPTS total, target specific subscale (partner), and type specific subscales (casual contact, embrace, kiss, hit, hug). Note 1: significance of path ab was estimated using a modified Sobel test. Note 2: p-values adjusted with a post hoc Holm-Bonferroni correction.
ECR-S - TAQ - IPTS Path c Path a Path ab Path b Path c
Avoid. - Total - Total -.15 .45 -.09 -.21 -.06
Avoid. - Partner - Partner -.36 .67 -.17 -.25 -.18
Avoid. - Total - Casual contact -.18 .45 -.09 -.21 -.09
Avoid. - Total - Embrace -.18 .45 -.08 -.18 -.10
Avoid. - Total - Kiss -.12 .45 -.00
Avoid. - Total - Hit .16 .42 .02
Anx. .30 .06
Avoid. - Total - Hug -.14 .45 -.11 -.24 -.04
Bold ▪ p ≤ .05; underscore ▪ p ≤ .01
Mediation of Touch Avoidance between Interpersonal Touch and Attachment Organization
Table 5 summarizes the results of the path analyses investigating the significant relationships between the independent variables ECR-S subscales and the dependent variables IPTS total, type and target specific subscales when controlling for the mediating variables TAQ total and subscales. Each path model contains standardized regression coefficients for path a, path b, path c and path c ′, as well as the product of path a times path b resulting in path ab (tested for significance with a modified Sobel test).
The relationships between attachment avoidance and IPTS total, casual contact, embrace and
hug were mediated by TAQ total. The relationship between attachment avoidance and IPTS partner
was only partly mediated by TAQ partner. No relationship between TAQ total and IPTS hit was
observed, neither was there any significant relationship between TAQ total and IPTS kiss.
The aim of this study was to assess the relationships between interpersonal touch, attitudes towards touch and attachment organization. Specifically, it sought to investigate how type and target specific touch behaviors correlated with attachment organization, and whether these relationships were mediated by attitudes towards touch. The a priori hypothesis was that attachment avoidance would predict a less frequent overall touch behavior, with increasing strength of the negative correlations the more intimate the type and target of touch. Given the literature on attachment organization and aggressive behaviors (Dutton, 2000; Orcutt, Garcia, & Pickett, 2005; Wekerle, & Wolfe, 1998) , attachment anxiety were hypothesized to show a positive correlation with violent touch behavior (i.e.
The result suggests that people reporting attachment avoidance have a clear tendency to also hold aversive attitudes towards touch (Table 2). In line with the main hypothesis, partner touch avoidance showed the strongest correlation, but surprisingly a very weak relation to family touch avoidance. Following the literature, same sex touch avoidance was more common among males and opposite sex touch avoidance among females. Attachment avoidance had a significant negative relationship with same sex, but not opposite sex touch avoidance. An explanation for this would be interesting to further investigate, as it seems counter intuitive given the strong correlation between attachment avoidance and partner touch avoidance (also accounting the heterosexual homogeneity of the sample). Stranger touch avoidance had the weakest significant correlation of the regression models, but highlight the fact that attachment avoidance seem to affect, at least to some degree, most attitudes towards touch. While attachment anxiety showed some zero-order correlations with touch avoidance, the relationship disappeared when put into the regression models, indicating that the correlation was due to a covariance with attachment avoidance (i.e. fearful-avoidant attachment, see Ainsworth, 1973; Bartholomew, 1991).
Target specific touch behavior showed few significant correlations in relation to attachment organization after the conservative Holm-Bonferroni correction was applied (Table 3). Total frequency of touch was weakly related to attachment avoidance, while partner touch showed a moderate relationship. The touching of female friends seemed to slightly decrease with age and males were more likely to touch both male friends and to some extent male strangers, thus being incongruent with their attitude towards same sex touch. Country shared some of the variance with gender on the touching of male friends – with the Indian sample being more touching than the US sample – which was also reflected in the IPTS total regression model. This could be explained by a higher degree of gender segregation in India (Desai, & Andrist, 2010) as previously discussed by Williams and Willis (1978). Another explanation for the cultural gender differences in touch behavior could be that men shake hands more than women (Kendon, & Ferber, 1973) and so with higher frequency on average in India, as seen in Table 4.
The frequencies of type specific touch behaviors seem to be in part dependent on attachment
organization. Casual contact as well as more intimate behaviors such as embracing, kissing and
hugging was shown to be negatively related to attachment avoidance. Stroking was not related to
attachment organization, but more common among the Indian sample. Likewise, hitting seemed to be
dependent on culture, following previous global statistical inquiries (World Health Organization,
2013), but further discussions on cultural differences of violent behavior is beyond the scope of this
study (see Walker, 1999). Hitting was also related to both anxious and avoidant attachment
organization indicating that those with a fearful attachment organization were more prone to hitting
behavior. Still, attachment anxiety was statistically a more accurate predictor of hitting, as shown by
Orcutt, Garcia and Pickett (2005) . While no attempts were here made to clarify the target of the
hitting behavior, it is noteworthy that anxiously attached females involved in relationships with
avoidantly attached males have been shown to report a higher rate of bidirectional domestic partner
violence ( Doumas, Pearson, Elgin, & McKinley, 2008) . Interpersonal permutation like this might not
only provide an explanation for the occurrence and frequency of violent behavior, but for social touch
overall. Further studies on interpersonal touch in relation to attachment organization should thus take into account the target of touch and ideally their attachment organization as well.
The mediating analyses revealed that most of the significant relations between attachment organization and interpersonal touch was mediated by attitudes towards touch (Table 5). This in contrast to a study by Jones (1996) which employed several touch avoidance measures that showed that attitudes towards touch did not predict self-recorded touch behavior. The relationship between attachment avoidance and total frequency of touch, alongside with the subscales casual, embrace and hug were all fully mediated by touch avoidant attitudes, while attachment avoidance and partner touch was only partly mediated by touch avoidance. The latter relationship suggests that individuals high in attachment avoidance do not necessarily need to hold an aversive attitude against the reciprocal touch of their partner/s to display a less frequent touching behavior. However, the indirect path from attachment avoidance to partner touch through partner touch avoidance was the strongest correlation observed. The frequency of kissing showed no relationship with touch avoidant attitudes which excluded any potential mediation. The direct relationship with attachment avoidance could possibly be explained by kissing being an intimate behavior of affection primarily practiced in close relationships (neither the US nor India are known to kiss as a greeting, see Martin, & Chaney, 2012).
Hitting was also unrelated to touch avoidance, and so was attachment anxiety. From this can be concluded, with caution, that the hitting behavior reported by this sample was not driven by aversive attitudes towards touch, but generated in part by attachment related issues.
All the self-report measures used in this study, except the newly developed interpersonal touch scale (IPTS) have previously been successfully applied in attachment and attitudes towards touch related studies ( Fraley et al., 2011; Guerrero, 1996; Kaitz et al., 2004; Simpson, Rholes, &
Nelligan, 1992; Tucker, & Anders, 1998). The study of interpersonal touch frequencies however, has been limited by inconsistent methodologies, with costly and time consuming observational methods being the most prevalent strategy. The IPTS was developed by Sorokowska and Croy (in preparation) to counter this by offering a reliable self-report measure of retrospective touch behavior. One design problem was observed, such as a proportion of subjects had indicated touch of partner or youngest child even though they had not answered “yes” on the control questions ( “are you currently involved in a romantic relationship?”, & “do you have children?”) . This could be due to a misunderstanding of the target youngest child as meaning any young child, or the arbitrary definition of a romantic relationship. Using a more advanced survey design, such as customized matrices depending on the answers to the control questions, could eradicate this confusion, but with some concern. While this problem was in the present study solved by excluding all respondents who answered “no” on the control questions in the regression models of IPTS partner and youngest child, it raises some suspicion to the reliability of the results. However, the factor analysis, alpha levels and the fit of the results suggests that this would only prove a minimal difference. Thus, the present study has given evidence to the reliability of the measure by uncovering replicated trends in relation to attachment organization and attitudes towards touch.
In order to test the validity of the measure, it seems reasonable to, in a future study, combine
the IPTS with a control group conducting a one week self-recording of interpersonal touch and
compare the results. Self-recording allows for a more detailed account of touch behavior, since the
subject notes every instance of actual touch rather than drawing from memory. To match the
self-recording measure, an adjustment to the IPTS is required which would serve as a possible
improvement to the accuracy of the instrument. One suggestion based on the grid scale by Schutte,
Malouff and Adams (1984) is that subjects indicate the frequency of type-to-target touch by entering a
number – corresponding to Likert-type description (i.e. 1 = never – 7 = very frequently) – into an
appointed check box. The instructions must clearly inform participants that they are to enter the
number with respect to the time spent with the specific target. For example, a long lost friend whom
the subject sees twice a year but touches frequently during that time would result in a “7”, while; a
co-worker whom the subject sees every day but rarely touches would result in a “2”. This approach
would require that the target of touch would be more person specific than in the present design (e.g.
subject chooses a particular male friend as target, rather than indicating the touching of male friends in general). Naturally if such a study was conducted, the self-record group would also have to indicate the approximate time spent with each target of touch (or non-touch).
Strengthening the reliability of the IPTS by utilizing and elaborating on these improvements in future research would be a great benefit to the study of interpersonal touch. Whether it be to uncover and minimize the destructive impact of violent behavior or shedding light on the possible gains of affectionate social touch, a standardized, accessible and effective instrument is much needed.
While the results of the IPTS in the present study showed many interesting correlations with attachment organization and attitudes towards touch, future research on this topic should focus on the causal direction of the touching behavior. For instance, what impact would decreased interpersonal touch have on an individual’s attachment organization? Because in a world where we are increasingly divided by computer screens, limiting our tactile communication to mere taps on a plastic keyboard, knowledge about how this affects our oldest and probably most social sensory system is of great importance.
Ainsworth, M. D. S. (1973). The development of infant-mother attachment. In B. Cardwell & H. Ricciuti (Eds.), Review of Child Development Research (Vol. 3, pp. 1-94) Chicago: University of Chicago Press.
Andersen, P. A. (2005). The touch avoidance measure. In Manusov, V. L. (Ed.) The Sourcebook of Nonverbal Measures:
Going Beyond Words (p. 57-65). Mahwah, NJ: Lawrence Erlbaum.
Andersen, J. F., Andersen, P. A., & Lustig, M. W. (1987). Opposite sex touch avoidance: a national replication and
extension. Journal of Nonverbal Behavior, 11, 89–109.
Andersen, P. A., & Leibowitz, K. (1978). The development and nature of the construct touch avoidance. Environmental Psychology and Nonverbal Behavior, 3, 89-106.
Andersen, P. A., & Sull, K. K. (1985). Out of touch, out of reach: Tactile predispositions as predictors of interpersonal distance. Western Journal of Communication (includes Communication Reports), 49, 57-72.
Bartholomew, K., & Horowitz, L. M. (1991). Attachment styles among young adults: a test of a four-category model.
Journal of Personality and Social Psychology, 61, 226.
Barnett, K. (1972). A theoretical construct of the concepts of touch as they relate to nursing. Nursing Research, 21, 102-109.
Beckett, C., Maughan, B., Rutter, M., Castle, J., Colvert, E., Groothues, C., et al (2006). Do the effects of early severe deprivation on cognition persist into early adolescence? Findings from the English and Romanian adoptees study.
Child Development, 77, 696–711.
Bowlby J. (1969). Attachment. Attachment and loss: Vol. 1. Loss. New York: Basic Books.
Desai, S., & Andrist, L. (2010). Gender scripts and age at marriage in India. Demography, 47, 667-687.
Chugani, H. T., Behen, M. E., Muzik, O., Juhasz, C., Nagy, F., & Chugani, D. C. (2001). Local brain functional activity following early deprivation: A study of postinstitutionalized Romanian orphans. Neuroimage, 14, 1290–1301.
Doumas, D. M., Pearson, C. L., Elgin, J. E., & McKinley, L. L. (2008). Adult Attachment as a Risk Factor for Intimate Partner Violence The “Mispairing” of Partners' Attachment Styles. Journal of Interpersonal Violence, 23, 616-634.
Dykas, M. J., & Cassidy, J. (2011). Attachment and the processing of social information across the life span: theory and evidence. Psychological Bulletin, 137, 19.
Couper, M. P., Tourangeau, R., Conrad, F. G., & Singer, E. (2006). Evaluating the effectiveness of visual analog scales a
web experiment. Social Science Computer Review, 24, 227-245.
Crawford, C. B. (1994). Effects of sex and sex roles on avoidance of same-and opposite-sex touch. Perceptual and Motor Skills, 79, 107-112.
Deethardt, J., & Hines, D. (1983). Tactile communication and personality differences. Journal of Nonverbal Behavior, 8(2), 143–156.
Dutton, D. G. (2000). Witnessing parental violence as a traumatic experience shaping the abusive personality. Journal of Aggression, Maltreatment & Trauma, 3, 59-67.
Elfenbein, H. A., & Ambady, N. (2002). On the universality and cultural specificity of emotion recognition: a meta-analysis.
Psychological Bulletin, 128, 203.
Field, T. (2010). Touch for socioemotional and physical well-being: A review.Developmental Review, 30, 367-383.
Field, T., Soliday, B., Lasko, D., Gonzalez, N., & Valdeon, C. (1994). Touching in infant, toddler, and preschool nurseries.
Early Child Development and Care, 98, 113-120.
Finnegan, R. (2005). Tactile communication. In C. Classen (Red.), The Book of Touch (s. 18-25). Oxford: Berg.
Feldman, R., Keren, M., Gross-Rozval, O., & Tyano, S. (2004). Mother–child touch patterns in infant feeding disorders:
relation to maternal, child, and environmental factors. Journal of the American Academy of Child & Adolescent Psychiatry, 43, 1089-1097.
Ferber, S. G., Feldman, R., & Makhoul, I. R. (2008). The development of maternal touch across the first year of life. Early Human Development, 84, 363-370.
Fraley, R. C. (2005, July 23). Information on the experiences in close relationships-revised (ECR-R) adult attachment questionnaire. Retrieved from http://internal.psychology.illinois.edu/~rcfraley/measures/ecrr.htm
Fraley, R. C., Heffernan, M. E., Vicary, A. M., & Brumbaugh, C. C. (2011). The experiences in close
relationships—Relationship Structures Questionnaire: A method for assessing attachment orientations across Relationships. Psychological Assessment, 23, 615.
Fromme, D. K., Jaynes, W. E., Taylor, D.K., Hanold, E. G., Daniell, J., Rountree, J. R. & Fromme, M. L. (1989). Nonverbal
behavior and attitudes toward touch. Journal of Nonverbal Behavior 13, 3–14.
Funke, F., Reips, U. D., & Thomas, R. K. (2011). Sliders for the smart: Type of rating scale on the web interacts with educational level. Social Science Computer Review, 29, 221-231.
Gallace, A., & Spence, C. (2010). The science of interpersonal touch: an overview. Neuroscience & Biobehavioral Reviews, 34, 246-259.
Gaetano J. (2013, May). Holm-Bonferroni sequential correction: An EXCEL calculator (1.2) [Microsoft Excel workbook].
Retrieved from https://www.researchgate.net/publication/242331583_Holm-Bonferroni_Sequential_Correction_
Gräf, L. (2002). Assessing internet questionnaires: The online pretest lab.In Batinic, B., Reips, U. & Bosnjak, M. (eds.) Online Social Sciences (p. 49-68). Seattle, WA: Hogrefe & Huber.
Guerrero, L. K. (1996). Attachment-style differences in intimacy and involvement: A test of the four-category model.
Communications Monographs, 63, 269-292.
Guerrero, L. K., & Andersen, P. A. (1991). The waxing and waning of relational intimacy: Touch as a function of relational stage, gender and touch avoidance. Journal of Social and Personal Relationships, 8, 147-165.
Guerrero, L. K., & Andersen, P. A. (1994). Patterns of matching and initiation: Touch behavior and touch avoidance across romantic relationship stages. Journal of Nonverbal Behavior, 18, 137-153.
Harlow, H.F. (1958). The nature of love. American Psychologist, 13, 673–685.
Hertenstein, M. J., Keltner, D., App, B., Bulleit, B. A., & Jaskolka, A. R. (2006). Touch communicates distinct emotions.
Emotion, 6, 528.
Jones, S. E. (1986). Sex differences in touch communication. Western Journal of Communication (includes Communication Reports), 50, 227-241.
Jones, S. E., & Brown, B. C. (1996). Touch attitudes and behaviors, recollections of early childhood touch, and social self-confidence. Journal of Nonverbal Behavior, 20, 147-163.
Jones, S. E., & Yarbrough, A. E. (1985). A naturalistic study of the meanings of touch. Communications Monographs, 52, 19-56.
Kaitz, M., Bar-Haim, Y., Lehrer, M., & Grossman, E. (2004). Adult attachment style and interpersonal distance. Attachment
& Human Development, 6, 285-304.
Kendon, A., & Ferber, A. (1973). A description of some human greetings. Comparative Ecology and Behaviour of Primates, 591, 668.
Kenny, D. A., Kashy, D. A., & Bolger, N. (1998). Data analysis in social psychology. In D. Gilbert, S. Fiske, & G. Lindzey
(Eds.), The Handbook of Social Psychology (p. 233–265). Boston: McGraw-Hill.
Lindholm, T. (2008). Validity in judgments of high-and low-accurate witnesses of own and other ethnic groups. Legal and
Criminological Psychology, 13, 107-121.
MacLean, K. (2003). The impact of institutionalization on child development. Development and Psychopathology, 15,
Martin, M. M., & Anderson, C. M. (1993). Psychological and biological differences in touch avoidance. Communication Research Reports, 10, 141-147.
Martin, J.S. & Chaney, L.H. (2012). Global business etiquette: a guide to international communication and customs. (2. ed.) Santa Barbara, Calif.: Praeger.
McDaniel, E., & Andersen, P. A. (1998). International patterns of interpersonal tactile communication: A field study.
Journal of Nonverbal Behavior, 22, 59-75.
McElroy, D., & Vahdat, A. (2014, May 22). Iranian actress Leila Hatami faces public flogging. The Telegraph. Retreived From http://www.telegraph.co.uk/news/worldnews/middleeast/iran/10850212/Iranian-actress-Leila-Hatami- faces-public-flogging.html
Oldmeadow, J. A., Quinn, S., & Kowert, R. (2013). Attachment style, social skills, and Facebook use amongst adults.
Computers in Human Behavior, 29, 1142-1149.
Ozolins, A., & Sandberg, C. (2009). Development of a multifactor scale measuring the psychological dimensions of touch
avoidance. International Journal of Psychology: A Biopsychosocial Approach, 33-56
Orcutt, H. K., Garcia, M., & Pickett, S. M. (2005). Female-perpetrated intimate partner violence and romantic attachment style in a college student sample.Violence and victims, 20, 287-302.
Picardi, A., Toni, A., & Caroppo, E. (2005). Stability of alexithymia and its relationships with the ‘big five’factors,
temperament, character, and attachment style. Psychotherapy and Psychosomatics, 74, 371-378.
Preacher, K. J., & Leonardelli, G. J. (2001, March). Calculation for the Sobel test: An interactive calculation tool for mediation tests. Retrieved from http://quantpsy.org/sobel/sobel.htm
Remland, M. S., & Jones, T. S. (1988). Cultural and sex differences in touch avoidance. Perceptual and Motor Skills, 67, 544-546.
Richman, S. B., DeWall, C. N., & Wolff, M. N. (2015). Avoiding affection, avoiding altruism: Why is avoidant attachment related to less helping?. Personality and Individual Differences, 76, 193-197.
Reips, U. D., & Funke, F. (2008). Interval-level measurement with visual analogue scales in Internet-based research: VAS Generator. Behavior Research Methods, 40, 699-704.
Remland, M. S., & Jones, T. S. (1988). Cultural and sex differences in touch avoidance. Perceptual and Motor Skills, 67, 544-546.
Remland, M. S., Jones, T. S., & Brinkman, H. (1995). Interpersonal distance, body orientation, and touch: Effects of culture, gender, and age. The Journal of Social Psychology, 135, 281-297.
Roese, N. J., Olson, J. M., Borenstein, M. N., Martin, A., & Shores, A. L. (1992). Same-sex touching behavior: The moderating role of homophobic attitudes. Journal of Nonverbal Behavior, 16, 249-259.
Roster, C. A., Lucianetti, L., & Albaum, G. (2015). Exploring Slider vs. Categorical Response Formats in Web-Based Surveys. Journal of Research Practice, 1, 1.
Schutte, N. S., Malouff, J. M., & Adams, C. J. (1988). A self-report measure of touching behavior. The Journal of Social Psychology, 128, 597-604.
Simpson, J. A., Rholes, W. S., & Nelligan, J. S. (1992). Support seeking and support giving within couples in an
anxiety-provoking situation: The role of attachment styles. Journal of Personality and Social Psychology, 62, 434.
Spence, C., 2002. The ICI report on the secret of the senses. The Communication Group, London.
Suvilehto, J. T., Glerean, E., Dunbar, R. I., Hari, R., & Nummenmaa, L. (2015). Topography of social touching depends on emotional bonds between humans. Proceedings of the National Academy of Sciences, 201519231
Stier, D. S., & Hall, J. A. (1984). Gender differences in touch: An empirical and theoretical review. Journal of Personality and Social Psychology, 47, 440.
Tucker, J. S., & Anders, S. L. (1998). Adult attachment style and nonverbal closeness in dating couples. Journal of Nonverbal Behavior, 22, 109-124.
Tabachnick, B. G. & Fidell, L. S. (2013). Using multivariate statistics. (6th ed., International ed.) Boston, MA: Pearson Walker, L. E. (1999). Psychology and domestic violence around the world. American Psychologist, 54, 21.
Wei, M., Russell, D. W., Mallinckrodt, B., & Vogel, D. L. (2007). The Experiences in Close Relationship Scale (ECR)-short form: Reliability, validity, and factor structure. Journal of Personality Assessment, 88(2), 187-204.
Wickham, R. E., Reed, D. E., & Williamson, R. E. (2015). Establishing the psychometric properties of the Self and
Perceived-partner Authenticity in Relationships Scale-Short Form (AIRS-SF): Measurement invariance, reliability, and incremental validity. Personality and Individual Differences, 77, 62-67.
Wekerle, C., & Wolfe, D. A. (1998). The role of child maltreatment and attachment style in adolescent relationship violence.
Development and Psychopathology, 10, 571-586.
Williams, S. J., & Willis, F. N. (1978). Interpersonal touch among preschool children at play. The Psychological Record, 28, 501.
World Health Organization. (2013). Global and regional estimates of violence against women: prevalence and health effects of intimate partner violence and non-partner sexual violence. World Health Organization.