Obsessions Across Two Cultures: A Comparison of Belgian and Turkish Non-clinical Samples

Abstract

There is a growing interest in the role of culture in Obsessive-Compulsive Disorder, yet cultural studies to date have suffered from methodological limitations and lack a clear theoretical framework. In the current study, we adopted a rigorous methodological approach, and a clear cultural psychological framework. We compared the structure and frequency of obsessions in non-clinical samples ( N = 706) from Belgium, a Western culture, and Turkey, a non-Western cultural context. Obsessions were measured by a newly compiled instrument that included a broad range of obsessions. Cross-cultural equivalence of the structure of obsessions was assessed both in the pooled data, and in each culture separately. At an abstract level, we found a two-factor structure that was cross-culturally invariant, and that fit both cultures equally well. These two types of obsessions each corresponded with a different model of agency. Compared to the Turkish sample, the Belgian sample reported more obsessions that can be understood from a disjoint (independent) model of agency as frequently found in Western cultures, whereas the Turkish sample, compared to their Belgian counterparts, reported more obsessions that can be tied to a conjoint (shared) model of agency as frequently found in non-Western cultural contexts. Differences in the prevalent types of obsessions were systematic and interpretable, therefore. In addition to the cross-culturally equivalent two-factor structure, we found culture-specific factor solutions; these solutions point to cultural differences in the experience of obsessions that have yet to be fully understood. In the Discussion, we outline future directions of the research on culture and obsessions.

ONVAR: A simultaneous component analysis approach for disentangling outlying and non-outlying variables

“…We start by looking at how often ONVar and LBCM were able to find the correct partition of variables into outlying and non-outlying ones 4 , 5 . In Table 1, we see that, for ONVar, this was the case for 90% of the simulated data sets, while LBCM had a success rate of 88%.…”

“…There were only two datasets out of 2880 simulated ones, for which ONVar found the correct number of outlying variables, but not the correct partition. 5 We also investigated in a pilot study focusing on the two blocks settings how sensitive ONVar is to incorrect specification of the Bernoulli parameter in the first step of the algorithm. To this end, we ran the algorithm twice, once setting this parameter to 1/3, and once setting it correctly to #outlying/(#outlying+#non-outlying).…”

“…Such data consist of multiple object-by-variable blocks, which all share the variable mode. In behavioral studies, the blocks might represent, for example, different experimental groups [2] or countries [3,4,5] to which the participants belong. In metabolomics, studies on the expression of multiple genes in a few participant groups (e.g., groups that are vaccinated with different vaccines; [6]) would yield such multi-block multivariate data as well.…”

Detecting outlying variables in multigroup data: A comparison of different loading similarity coefficients

“…For a first example, one can think of studies that inspect the expression of multiple genes in a few participant groups (e.g., groups that are vaccinated with different vaccines 3 ). A second example is given by cross‐cultural studies in which inhabitants of different countries fill in the same questionnaire 4 . A third example that we will focus on in this paper is sensory profiling data, where a number of assessors rate different food samples with respect to a set of sensory attributes 5,6 …”

“…In psychometrics, after the SCA analysis is conducted, researchers often execute follow‐up analyses in which they investigate whether the groups differ with respect to their mean scores on the obtained SCA constructs 4 . Alternatively, they inspect how the obtained SCA components relate to other measured variables by computing correlations between the component scores and these covariates 9 .…”