In the following section, the psychometric properties of the two subscales SMCL are described.
SMCL Submission Scale
Factor Structure and Reliability
An initial principal component analysis revealed a clear single-factor structure, with the first factor accounting for 29 % of the scale variance. In addition, the scree test criterion also indicated a clear break between the first and the second factor (Cattell, 1978). The data contained sufficient shared variance for factor analysis (Kaiser–Meyer–Olkin criterion, KMO = .85). The reliability of the 24-item scale of Cronbach’s alpha = .96 was sufficient and all items had statistically significant (p < .01) corrected item total correlations (M = .47, SD = .10). Similarly, the mean factor loading of all 24 items (Table 1) onto the first factor was .53 (SD = .12).
To additionally reduce the items to a smaller set of independent composite variables, the six non-trivial factors with an eigenvalue higher than 1 were rotated to a varimax solution. These six factors accounted for 64 % of the variance, whereas each factor clustered a subset of items that were interpretable and content-relevant (Table 1): The strongest factor, accounting for 15.22 % of the total variance, was associated with items related to dominance, such as humiliating not only the partner but also with others, or restricting his or her psychological needs. The second factor (explaining 11.26 % of the variance) referred to items assessing the pleasure related to the use of toys that can be used to inflict pain, such as clamps, plugs, or wax. The third factor, accounting for 10.48 % of the total variance, was comprised items assessing rather soft SM play, including soft bondage behavior such as not only blindfolding and restricting the partner, giving commands, but also having “hard” intercourse. Even though the latter item does not seem to be a rather soft item, it still does rather match with the other items in this category compared to other factors. The three items related to beatings loaded onto the fourth factor, sharing 9.95 % of the total variance. For the last two factors, which accounted for 9.59 and 7.57 % of the variance, the items related to breath deprivation and bodily fluids could be, respectively, reflected. Only one item, clawing
sub, showed ambiguous factor loadings (i.e., coefficients not above .5) on three factors. Cronbach’s alpha coefficients for the six dimensions were calculated as a measure for reliability and revealed satisfying results (sub-scale domination: .84, sub-scale toys: .80, sub-scale soft play: 74, sub-scale beatings: .78, sub-scale breath: .80, and sub-scale body fluids: .63). The lower body fluids reliability could be traced back to the comparably low number of responses to the feces
Pleasure Gain and Engagement in Submissive Behavior
As the engagement in submissive practices was ordinal-scaled and could not be summed up, the relation between pleasure gain and the engagement in certain behaviors was analyzed item-wise, using Spearman correlations. For every item, participants who reported a higher pleasure gain also reported a higher engagement in the corresponding behavior, indicated by masturbation fantasies or real-life experiences (Mean Spearman correlation coefficient = .61, SD = .11; all ps < .001).
For both groups of submissives and switches, differences in pleasure gain between female and male participants were calculated for the total score of the scale. Due to unequal numbers of female and male participants in the groups, Mann–Whitney U tests were conducted and revealed neither a statistically significant differences between female and male participants in the group of submissives (Z = 1.85, p = .064, r = .12) nor in the group of switches (Z = .59, p = .557, r = .05).
Differences in Pleasure Gain Across Groups for Submissive Practices
A Kruskal–Wallis test for differences in pleasure gain submissive practices across the four groups revealed a significant main effect of group, χ
2(3) = 409.56, p < .001, η
= .64. Post-hoc calculated Bonferroni-corrected Mann–Whitney U tests for multiple comparisons demonstrated medium to large effect sizes for differences across all four groups (Table 2): whereas participants from the group of submissives reported the highest pleasure gain, followed by participants from the group of switches, participants from the group of dominants showed an even lower pleasure gain than participants with a conventional interest.
SMCL Dominance Scale
Factor Structure and Reliability
A principal component analysis was carried out on the data from the groups of dominants and switches. In line with the scree test criterion, the initial unrotated factor solution favored a single-factor structure, with the first factor accounting for 29 % and a clear break after the first factor. The result of the KMO measure was .84. For the 24-item scale, Cronbach’s alpha was .89. All items had significant corrected item total correlation (M = .47, SD = .10; all p < .01). Likewise, all items except the item feces
dom had sufficient factor loadings (Table 3) onto the first factor (M = .60, SD = .14).
To additionally combine the items into smaller subsets, all non-trivial factors with an eigenvalue >1 were subsequently varimax rotated (Table 3). The six factors, accounting for 64 % of the total variance, confirmed a similar conceptual distinctiveness than the varimax solution from the submission items. The six factors contributed to the differentiation of various preferences of sadistic behavior. The first factor (explained variance of 20.31 %) comprised almost the same items as the first factor from the Submission scale and was therefore also associated with SM dominance. The second factor, which accounted for 13.2 % of the variance, was likewise associated with the use of toys such as plugs or clamps in the play with the partner. The third orthogonal factor covered the three items related to beatings and accounted for 11.36 % of the variance. Another three items were grouped by the fourth factor, explaining 9.71 % of the variance and including the three items associated with breath reduction. The fifth (8.96 % explained variance) and sixth (7.57 % variance) factors included two items each and subsumed soft SM play behavior as well as the two body fluids items feces
dom and urinating
dom. The three items tying up
dom, and rough intercourse
dom could not be associated with one distinctive factor, but seemed to be part of different forms of SM preferences. Cronbach’s alpha coefficients for the six dimensions were the following: domination .90, toys .74, beatings .92, breath .78, soft play .84, and body fluids .60. As for the submission items, the lower reliability of body fluids could be traced back to the comparably lower number of responses to the feces
Pleasure Gain and Frequency of Sadomasochistic Behavior
The relation between pleasure gain and the ordinally scaled engagement in sadistic practices was analyzed item-wise, using Spearman correlations. As for the submission scale items, participants who described a higher pleasure gain from dominant behaviors also reported a higher engagement in the respective behaviors (Mean Spearman correlation coefficient = .55, SD = .21; all ps < .001).
Mann–Whitney U tests were conducted for differences in pleasure gain between female and male participants from the groups of dominants and switches. There was neither a statistically significant differences between female and male participants in the group of dominants (Z = .47, p = .640, r = .04) nor in the group of switches (Z = .56, p = .576, r = .04).
Differences in Pleasure Gain across Groups for Dominant Practices
There was a significant difference in pleasure gain from dominant practices across the four groups, Kruskal–Wallis test: χ
2(3) = 338.58, p < .001, η
= .52. To account for individual group differences, Bonferroni-corrected Mann–Whitney U tests were calculated and revealed medium to large effect sizes for differences across the four groups of participants (Table 4): contrary to the submissive items, participants from the group of dominants reported the highest overall pleasure gain, whereas participants from the group of submissives reported least pleasure gain. Participants from the group of switches reported the second highest pleasure gain, whereas participants from the conventional group scored higher than the group of submissives.
Additionally, paired sample t tests were calculated for differences between the submission and the dominance total score within groups. Significant differences between the two scores were obtained in the two groups of dominants and submissives, dominants: t(132) = 27.61, p < .001, d
z = 2.40; submissives: t(227) = 34.30, p < .001, d
z = 2.13, and confirmed clear preferences for one type of SM behavior over the other. No significant differences were found for the other two groups, indicating no consistent preferences, switches: t(153) < 1, d
z = .01; conventional group: t(128) < 1, d
z = .05.