Swingers in Germany: Sociodemography and Event Preferences Assessed from Harvested Web Data

Oliver Maor

doi:10.1007/s10508-025-03198-z

. 2025 Aug 1;54(7):2401–2427. doi: 10.1007/s10508-025-03198-z

Swingers in Germany: Sociodemography and Event Preferences Assessed from Harvested Web Data

Oliver Maor ^1,^✉

PMCID: PMC12457489 PMID: 40751037

Abstract

This exploratory study investigated Germany’s heterosexual swinger community within the consensual non-monogamy spectrum. It used anonymized, verified data harvested in September and October 2023 from Joyclub, the leading online platform in Germany focused on sexual exploration. The first part examined anonymized profiles from self-declared swingers, totaling 22,973 individuals, revealing a mature demographic, with median ages of 44 for women and 46 for men. Despite high regional variations, no clear correlations between swinger prevalence and regional socioeconomic factors were found. Subgroup analysis identified BDSM-affiliated swingers and those involved in open relationships or polyamory, with a higher concentration in urban areas. The second study assessed 14,008 attendees at 82 categorized swinger events, which showed a preference for sex-focused events across all age strata. Contrary to the male-centric perspective, the data revealed events strongly focused on women’s preferences, including those targeting younger male partners. Events also combined sexual activities with social elements and drew participants from greater distances, indicating a strong commitment to the community lifestyle. A prominent presence of solo women was observed, contradicting earlier findings from other countries. Reduced pricing for women did not correlate with higher attendance, questioning the effectiveness of such strategies. The use of verified data and a large sample size provides robust insights into Germany’s swinger community, emphasizing gender participation and regional variations. This study offers an empirical foundation for future research and highlights the importance of local contexts and terminology in studying sexual behavior.

Supplementary Information

The online version contains supplementary material available at 10.1007/s10508-025-03198-z.

Keywords: Sexuality, Consensual non-monogamy, Germany, Swingers, Data harvesting

Introduction

This study aims to build on previous research by exploring the German swinger scene, which differs notably from other contexts in its openness and organization. Utilizing data from the prominent Joyclub (www.joyclub.de) website, this research seeks to provide updated insights into the age, gender, and social backgrounds of swingers by addressing gaps in current understanding. The specific characteristics of the Joyclub platform and the German context provide a novel opportunity for comprehensive data collection. The methodology applied in this study has both strengths and limitations, which will be discussed throughout the paper.

Swingers, as a subgroup of consensual non-monogamy (CNM), represent a unique dimension of intimate relationships. CNM broadly encompasses arrangements where individuals in a romantic relationship mutually consent to engage in sexual, romantic, or intimate relations with additional partners. This includes swinging, open relationships, and polyamory, as outlined by Balzarini and Muise (2020) and others.

Despite some media interest in consensual non-monogamy, limited quantitative data on the prevalence and acceptance of non-monogamous relationships are available for Germany. As Mayer (2024) highlighted, many studies focused on questions framed within a monogamy-centered perspective. This leaves gaps in understanding CNM practices because they are not adequately captured by traditional questions about extradyadic sexual encounters.

According to a study by Haversath et al. (2017), which did not distinguish between consensual and non-consensual non-monogamy, 7% of respondents (8% of men, 6% of women) reported engaging in sexual activity “outside their current relationship” (p. 548). The GeSiD study, conducted between 2018 and 2019, provided the first comprehensive data on sexual behavior in Germany (Matthiesen et al., 2021). In the course of that study, it was noted that 5% of women aged 18–35, 3% of women aged 36–75, and 9% of men across both age groups engaged in sexual activity outside their permanent relationship. Among them, 15% of women and 6% of men did so with their partner’s consent (Institut für Sexualforschung, Sexualmedizin und Forensische Psychiatrie, Universitätsklinikum Hamburg-Eppendorf [UKE], 2020, p. 28). This means that 0.6% of the heterosexual adults in Germany living in permanent partnerships engage in CNM. By contrast, North American CNM studies report prevalence rates of 4–9.7% for couples and 2.6–5.6% for the general adult population (Conley et al., 2013; Fairbrother et al., 2019; Levine et al., 2018; Rubel & Burleigh, 2020; Rubin et al., 2014).

Lehmiller (2020) found that, in the USA, 32.6% of respondents, all in monogamous relationships, imagined engaging in CNM. In Germany, a commercial survey by ElitePartner (PE Digital, 2023) reported lower figures, with 23% of men and 11% of women indicating they could imagine living in an open relationship. Lack of time for more than one partner, emotional concerns like potential jealousy, fear of being left alone by the partner, and the danger of falling in love were stated as frequent barriers. Religious concerns were not mentioned in the survey, aligning with Wolf’s (2008) finding that Germany is one of the most secular countries in the world. The ElitePartner survey should be interpreted with some caution, as it is not an academic study.

Definition of Swinging

Swinging specifically involves couples who exchange partners temporarily, often at social events designed for this purpose (Balzarini & Muise, 2020; Wood et al., 2021). This practice is distinguished by its focus on sexual activities, rather than emotional or romantic involvement with external partners. Swingers typically maintain a primary emotional commitment to their partner while engaging in sexual activities with others, often in social settings (Wood et al., 2021).

Whereas earlier literature, such as Jenks (1998) and Vaillancourt (2006), focused predominantly on married couples engaging in swinging for sexual purposes, contemporary definitions have become more inclusive. Swinging is now seen as part of a broader spectrum of CNM relationships, where emotional and sexual exclusivity is not mandatory (Conley et al., 2018; Moors et al., 2013). This flexibility allows for overlapping identities and practices, such as “swolly” individuals who identify as both swingers and polyamorous (Moors et al., 2013).

Furthermore, swinging differs from polyamory, which emphasizes emotional and romantic commitments to multiple partners, and open relationships, which generally focus on sexual relations outside the primary couple with minimal emotional involvement (Balzarini et al., 2017; Baumgartner, 2020).

The swinging subculture itself is diverse, with varying degrees of involvement and practices among individuals and couples (Venn, 2015). This diversity is echoed in the broader postmodern understanding of subcultures, which emphasizes fluidity, lifestyle choices, and varying degrees of commitment (Frank, 2019; Kimberly & Hans, 2017).

In summary, swingers within the CNM framework are best defined as individuals or couples who, with mutual consent, engage in sexual activities with others, typically within social environments. This practice is distinct from other forms of CNM due to its focus on sexual rather than emotional or romantic connections with external partners.

Methodological Challenges

Research on swingers has long faced challenges such as regional sampling biases and access issues (Jenks, 1998; Niekamp et al., 2013). Vaynman (2024) criticized the lack of comprehensive methods and theory, and Rubin et al. (2014) noted that techniques like snowball sampling are prone to bias. To address these issues, researchers have adopted more advanced methods, such as online surveys (Platteau et al., 2016) and data harvesting from a UK-based swinger website (Haywood, 2022). Data harvesting has the particular advantage that the result quality does not rely on response rates and that participation is not restricted to persons belonging to the same peer groups, ensuring more comprehensive and unbiased data collection. By using data harvesting, this study leverages these advantages.

As critically described by Frank (2019) in detail, previous research on swingers has sometimes encountered resistance within the academic community, with accusations of bias (Conley et al., 2018), and a tendency to focus on negative aspects such as STI risks (e.g., Dukers-Muijrers et al., 2010; Friedman et al., 2017; Niekamp et al., 2013), whereas in the long term, it was established that swingers did correspond to the normal clinic population in terms of STI prevalence (Dukers-Muijrers et al., 2017). Anderson (2010) saw the so-called monogamy script as widely viewed as authoritatively compelling in the research context. In this study, I maintain a strictly neutral position on swinging, focusing on statistical outcomes without excessive interpretation.

Earlier Research Results on Size and Demographics of the Swinger Scene

As a consequence of the described methodological challenges, not much is yet known about the demographic composition of the group of individuals who, using one of the existing definitions, can be considered swingers.

Regarding social background, Jenks (1998) reported the assumption, which persisted until the 1990s, that the typical American swinger was a “white, middle to upper middle-class person in his or her late 30 s who is fairly conventional in all ways” (p. 510). Sheff and Hammers (2011) reached similar conclusions. O’Byrne and Watts (2011) found that swingers in their studies were in their 40 s and more likely to be educated and economically well off. Haywood (2022) calculated an average age of 43 for females and 45 for males who provided feedback on swinger events. Cerwenka et al. (2020) reported with regard to Germany that partner swapping and swinging are most prevalent among individuals aged 45–60, with 12% of men and 7% of women reporting current or past experiences with swinging, compared to just 3% for both sexes aged 18–30. These studies therefore suggest an ostensible profile of a typical swinger.

In contrast, research on polyamorous individuals has yielded different results. In particular, Moors et al. (2021), referring to the USA, found that previous engagement in polyamory was more likely among individuals with lower levels of education. Balzarini et al. (2019), diverging from Haupert et al. (2017), reported an association of polyamorous persons with lower income levels. This indicates that findings related to specific subgroups cannot be considered representative of the CNM population as a whole.

Contextual Background: the German Swinger Scene

In this exploratory research, I aim to overcome some constraints in accessing swingers by focusing on the well-organized German swinger scene. In Germany, swinging is not handled as a clandestine affair. This contrasts sharply with societies where more rigid attitudes on sexuality prevail, notably the UK, as reported by Haywood (2022).

Swinging is legally permissible in Germany under general laws concerning consent and age. Swinger clubs, often licensed as restaurants, operate without facing allegations of immorality, as evidenced by several rulings from German courts (Bundesverwaltungsgericht, 2002, 2022; Rheinland-Pfalz, 2021). These decisions underscore the principle of individual sexual autonomy and lack of societal harm in such activities. Participation in swinging does not result in legal disadvantages, such as job loss due to extramarital relations (Mayer, 2013), nor does it lead to family law repercussions based on assumptions of an unwritten public order, as seen previously in Spain (Navarro, 2017). Swingers’ rights to privacy and personal freedom in their sexual lives are respected, aligning with broader societal and legal norms. The German media, also tabloids (Selig, 2023; Witte, 2023), tend to portray swinger clubs neutrally or positively, without invoking public outcry or sensationalism. This contributes to a normalized view of swinging activities within society. Swinging activities and clubs in Germany operate openly, without the need for secrecy. This is exemplified by the visibility of clubs such as “Steinenhaus” in Hattingen, which is even recognized in public transport networks by naming a bus stop after it (Verkehrsgesellschaft Ennepe-Ruhr mbH & Verkehrsverbund Rhein-Ruhr AöR, 2020). Joyclub, the major online platform for the swinger community, also reflects this openness.

Therefore, the German context offers an opportunity for transparent and open research into the swinger community, contrasting with the more clandestine approaches in other regions.

Research Aims

By providing updated demographic data and insights into the German swinger scene, this research seeks to offer a more nuanced understanding of swingers, test existing stereotypes, and lay a foundation for future studies. In parallel to Niekamp et al. (2013), I focused on swinger website use as well as attendance at physical swinger events. The aim was to reassess assumptions about the typical age of swingers, gender composition, age differences in couples, and, based on regional indicators, the social positioning of the phenomenon in terms of income, occupational status, and migration background. I also examined whether swinging is primarily a practice of couples or includes significant numbers of solo participants, and whether participation by men and women occurs at similar levels, as observed in earlier studies of CNM (Rubin et al., 2014). These aims were addressed through two distinct but thematically connected studies.

The specific aim of Study 1 was to identify structural characteristics of the swinger scene in Germany based on verifiable indicators: age distributions, gender composition, couple and solo structures, and regional density patterns. These indicators were examined in relation to social variables such as income level, occupational sector, and migration background to assess the broader social positioning of the phenomenon. A follow-up data collection extended this analysis to subgroups defined by self-declared interest in BDSM, open relationships, and polyamory.

Study 2 examined publicly announced swinger events to capture how swinging manifests in physical space. It aimed to establish the age structure of admitted visitors, differences between solo and couple admissions, and the event types associated with different age and gender profiles. Additional attention was given to pricing structures, gender-based access rules, and their effects on participation. The spatial distribution and reach of events were analyzed using residence data of registered visitors. Long-distance attendance and its correlation with event pricing were examined to approximate the commitment and travel behavior associated with different event types.

Study 1: Localized Swinger Profiles

Method

Data from residents of selected counties of Germany who could be identified as swingers were harvested from the Joyclub (www.joyclub.de) online community, a platform dedicated to recreational intimacy, and were analyzed. The data retrieved, which consisted of age, gender, and the residence data at the county level, were anonymized from the onset. In the case of couples, both ages of the partners were collected and saved as one case.

Process of Data Selection from the Joyclub Website

The data selection process involved a series of structured steps, including the identification of the platform-based data source, the application of specific inclusion criteria, and the construction of the final data set.

Data Source: Joyclub Website

The Joyclub platform enjoys visibility, in part, due to mentions in the national media landscape (Ponath, 2022; Süddeutsche Zeitung, 2012). The veracity of the claim that the number of active member profiles amounts to upwards of five million remains unverified. According to traffic data from SimilarWeb (n.d.), which provides estimates based on a variety of data sources, Joyclub.de was ranked as the 66th most visited website in Germany in August 2024, placing it alongside major platforms such as Lidl.de (ranked 64th), Microsoft.com (ranked 65th), and Vodafone.de (ranked 68th). This high ranking underscores Joyclub’s prominence in Germany’s online landscape, indicating substantial user engagement and reinforcing the platform’s significance for research into online swinger communities. Since its establishment as a significant player in 2012, the website has fostered a distinctive business model primarily centered around user authenticity and quality of experience, abstaining from the utilization of paid advertising. This approach appears to have contributed to its robust financial health in 2022, marked by an appreciable increase in sales and a solid equity ratio (Ackermann & Noack, 2024).

Central to the business model is a rigorous mechanism aimed at mitigating the presence of fake profiles. Comprehensive verification processes, including video and on-site validations requiring official identification, are employed to uphold the integrity of user profiles (Joyclub, n.d.-a). The platform’s operational practices suggest a prioritization of authentic and quality user experiences, although this has not been formally evaluated in scientific studies.

German court rulings have clearly established that it is illegal for platform providers to employ fake profiles on dating platforms, with severe penalties for violations (Landgericht Berlin, 2022; Landgericht Flensburg, 2022). Attempts by platforms to justify such conduct through terms and conditions have been dismissed. Legal protections do not guarantee compliance, but the high likelihood of detecting violations, combined with the threat of penalties and reputational damage, enhances the credibility of Joyclub’s assertion that it does not utilize fake profiles. Additionally, platforms using fake profiles often follow a revenue model that charges per interaction (Verbraucherzentrale Bayern e.V., 2017), which is not the case with Joyclub. Details are provided in part B.1–2 of the Supplementary Material.

Joyclub membership at its foundational level is complimentary, but there exist options for upgrades. For instance, a premium membership entails entitlements such as a minimum 10% discount on entry fees for events facilitated by participating organizers once registered through Joyclub (Joyclub, n.d.-c). Joyclub profile configurations allow for pseudonymous nicknames, incorporation of images, disclosure of sexual inclinations, and textual inputs structured within the platform’s framework. By default, profile contents are accessible to all Joyclub registrants. A dedicated search feature facilitates member discovery based on variables such as gender, age, locale, and sexual predilections (Joyclub, n.d.-d).

Structural and Terminological Basis for Profile Classification

In compliance with ethical standards and to minimize bias, the terminology in this research reflects the demographic composition of the user base of Joyclub. An analysis of the Joyclub profiles relating to entries across Germany indicated a predominant orientation toward heterosexual and bisexual preferences, with 0.21% of the profiles showing identification with exclusive same-gender preferences. Details are provided in the data package for this study (Maor, 2025a, 2025b). Although Joyclub allows for registrations indicating “diverse” and multiple gender identities, a manual, hands-on review of search and event pages showed their minimal presence within the scope of this study.

Accordingly, the term couple is used to denote pairings of women and men, mirroring the predominant user demographic and accurately representing the heterosexual population of the study. Profiles on Joyclub not categorized as belonging to a couple are referred to as solo in this research. This designation highlights the type of profile rather than relational status, adhering to the privacy considerations integral to the research design.

Selection of Representative Counties

To limit the effort of data collection, not all suitable profile data from Germany were called up. Therefore, a representative selection of counties had been made. The selection also enabled structural alignment between the two studies, as the verification of Study 1 through independently gathered data in Study 2 required a consistent regional framework.

Germany is divided into states, intermediary regions, counties, and independent cities. Berlin and Hamburg, as city-states, operate as singular entities, without further subdivision. In line with the European Union’s Nomenclature of Territorial Units for Statistics (NUTS), counties and autonomous cities typically correspond to the NUTS-3 classification (European Commission: Eurostat, 2022). City-states function at the NUTS-2 level, bypassing the NUTS-3 designation (Statistisches Bundesamt, n.d.).

For simplicity, the term counties will refer to all administrative regions included in this study, encompassing counties, independent cities, and the city-states of Berlin and Hamburg.

For the selection of counties that are representative for Germany as a whole, principal component analysis was conducted on official data from German counties to identify 24 representative counties that reflect the country’s demographic and structural characteristics.

To identify a representative sample of German counties, a principal component analysis (PCA) was conducted using SPSS 29.0.1.1. Input variables included structural data from all 400 German counties, comprising average disposable income, population density, total population, and age distribution across seven age strata (from 30 to 75 years).

The suitability of the data for factor analysis was confirmed prior to performing PCA. The extraction was conducted using the correlation matrix. Components with eigenvalues exceeding 1.0 were retained (Kaiser criterion) and no rotation was applied, as the primary goal was dimensional reduction for sampling purposes rather than interpretation of underlying factors.

The PCA yielded three principal components with eigenvalues greater than 1, collectively explaining 80.57% of the total variance (Component 1: 47.51%, Component 2: 21.60%, Component 3: 11.46%). The first component strongly correlated with population density (0.800) and younger age groups (30–34 years: 0.913; 35–39 years: 0.805) positively, while negatively correlating with older age groups (55–59 years: − 0.877; 60–64 years: − 0.907; 65–75 years: − 0.823). The second component showed strong associations with average income (0.767) and middle-aged groups (45–49 years: 0.884; 50–54 years: 0.705). The third component exhibited strong correlation with the 40–44 age group (0.716) (see Table S1 in the Supplementary Material).

Based on these PCA results, 24 counties were selected that represented diverse combinations of factor scores across all three components. Counties near the median values of the component scores (e.g., Ilm-Kreis, Frankfurt (Oder), Düren, Traunstein) were included to ensure balanced representation across the sociodemographic spectrum. The selection deliberately also included economic outliers: the wealthiest (Starnberg and Hochtaunuskreis) and the poorest counties in Germany (Gelsenkirchen and Duisburg), measured by average available income. For East Germany, Potsdam-Mittelmark was included as an economic outlier (highest income in East Germany). Leipzig and Halle were selected as population density outliers in East Germany. This selection strategy allowed for examining whether extreme sociodemographic characteristics correlated with swinger demographics.

Assessment of the Demographic Data of the Representative Counties

At a 95% confidence level, this selection represents a cross-section of the national demographic and occupational landscape across the examined variables, namely average disposable income, population density, total population, and age distribution across seven age strata, as well as workforce participation in the service sector and in occupations in sectors related to mathematics, information technology, natural sciences, and technology, as defined in German statistical classifications (MINT), and divorce rates in 2021, normalized per 100,000 residents. For population density, I used a population-weighted national average as a benchmark rather than the simple mean of county-level densities, which would overrepresent sparsely populated regions, and not reflect the average environment in terms of population density which the inhabitants were living in. The 24 counties identified in this manner are referred to as representative counties.

Spearman bivariate correlation analysis, accompanied by bootstrap calculations, was applied to the sociodemographic data of the representative counties, excluding age-related variables, and including data on the quota of inhabitants having an immigration background, to identify correlations between the variables of these data. The assumption of normality was not met; Spearman’s rho correlation analysis is a nonparametric test, which is robust against nonnormal distributions.

Selection Criteria with Respect to Profiles

The search function on the Joyclub website displays results with profile nicknames, ages (or, in the case of couples, both ages), and the county of residence (Joyclub, n.d.-d). For mixed-gender couples, the woman’s age appears first. Each search result links to the full profile page.

The search function includes filters that allow the display of profiles matching specific parameters. For this study, I identified swingers based on users’ self-identification as swingers in their profiles. I applied filters to include only profiles that were verified by Joyclub through their verification routines. Additionally, I used only profiles with a profile picture, as one of my inclusion criteria. In the following discussions, individuals with profiles meeting these criteria (verified profile, bearing a profile picture, and self-identification as swingers) will be referred to as identified swingers.

Collection of Swinger Data

Using these filters, I searched for profiles within a radius that extended beyond the maximum distances possible within each county. Searches were conducted separately for men, women, and couples across the 24 representative counties. In the case of Leipzig, both the city and the surrounding county were treated as a single unit, as Joyclub displays both under “Leipzig” in the search results. The county data for Leipzig were adjusted accordingly.

Only the results pages were accessed, and not the profile pages themselves. This was sufficient to meet the research aims, because the results page output can be filtered to show only men, women, or couple profiles, and show the relevant data, namely the age and the county of residence.

In this way, I harvested the search results as raw data. From these, I extracted the age or ages and ascertained that the place of residence of each profile entry referred to the target county. Ultimately, I inserted these extracted data, pertaining to the age, the residence county, and the gender information, into the corpus. In case of couple profiles, I retained the link between the man and the woman in the respective couples. I performed these steps described using Python scripts, which are contained in the data package for this study (Maor, 2025a, 2025b). After random testing for accuracy of the results provided by the scripts, I deleted the raw data.

In addition, I conducted the same search and harvesting with regard to the four cities in Germany with more than one million inhabitants, referred to as big cities: Berlin, Munich, Hamburg, and Cologne.

Assessment of Duplicate Profiles

Because the extrapolation of the overall number of identified swingers and the interpretation of demographic distributions rely on profile-based data, it was necessary to assess whether—and to what extent—profile duplication could affect the results. I therefore estimated the prevalence and structural impact of duplicate profiles across the data set.

Two duplication scenarios must be distinguished. The first involves a person maintaining a solo profile and also partaking in a couple profile, which is allowed under Joyclub’s rules (solo-couple duplication). The second pertains to profiles of the same type—two solo profiles of the same gender or two couple profiles—managed by the same individual or couple (same-type duplication).

Theoretical frameworks of online identity management (Bullingham & Vasconcelos, 2013), contextual identity representation (Hogan, 2010), and research on identity construction in anonymous online environments (Zhao et al., 2008) suggest that solo-couple duplicates should not be interpreted as deceptive. Rather, such profiles reflect different social contexts in which swingers operate, allowing them to present themselves both independently and as part of a couple, depending on the situation. The same rationale applies to cases in which a person maintains a solo profile and, for example, a second profile as an artist—something commonly done by authors or photographers.

To accurately estimate the prevalence of duplicate profiles and to understand their impact on the findings, I developed two Python scripts to calculate both solo-couple duplication and same-type duplication rates. These scripts compared real data from the representative county set (proven to be representative for Germany) against randomly generated duplication patterns using Monte Carlo simulations. The simulation parameters were informed by data independently collected in Study 2. For solo-couple duplications, county-specific age distribution data from official sources (see Part H of the Supplementary Material) further refined the estimates.

The scripts computed the divergence between the real data set and the synthetic duplication scenarios, accounting for variables such as age, gender, and location. A third Python script applied a bootstrap procedure with 10,000 iterations to assess the robustness of the solo-couple duplication estimates. For same-type duplications, the bootstrap was integrated into the respective script. All scripts, along with the detailed methodology, are included in the data package accompanying this study (Maor, 2025a), and a complete explanation is provided in Sections B.6 and B.7 of the Supplementary Material.

Data Analysis

The following analyses were based on the previously described data corpus, which combined profile-level information on identified swingers with structural county-level indicators. The swinger data set contained age-related information for solo and couple profiles, enabling the calculation of gender-specific median ages and density measures. The county-level structural data included all variables used for county selection and representativity assessment—namely, average disposable income, population density, total population, age distribution across seven strata, divorce rate, and workforce participation in both the service sector and in MINT-related occupations. In addition, the quota of residents with a migration background was included for use in correlation analyses.

The analyses described below were carried out to characterize the swinger population, to assess structural correlations with demographic patterns, and to explore potential clustering of counties based on their socioeconomic environments. Each step is described in detail in the subsections that follow.

Descriptive Data Analysis

From the collected data, I calculated the median ages of men, women, solo men, solo women, and the median age difference in couples, together with the standard deviations.

In a further step, I calculated a variable that will be referred to as swinger density, representing the number of identified swingers normalized against 100,000 inhabitants of the county. In the same way, with the aim of detecting possible gender differences, I calculated density rates for women, men, solo women, and solo men.

Correlations Assessment

The correlations between the aggregated age data of identified swingers, the age difference in identified swinger couples, the swinger density rates, and the county structural data were examined. A bivariate Spearman’s rho correlation analysis, accompanied by 1000 bootstrap calculations, was used as the data did not meet the assumption of normality and this nonparametric test is robust against nonnormal distributions. Statistical power was calculated post hoc using the software G*POWER, Release 3.1.9.7 (Faul et al., 2009), and in all cases reached the typical threshold of 80%.

Cluster Analysis

To enable a more nuanced interpretation of sociodemographic influences on swinger profiles, I conducted a cluster analysis. This approach was chosen for two main reasons. First, analyzing correlations between swinger characteristics and each sociodemographic variable individually would not adequately capture how these factors interact within integrated structural contexts. Second, the presence of multicollinearity among structural variables—identified through the correlation analysis of county-level data—made it difficult to interpret their independent effects in isolation.

To therefore allow better interpretation when analyzing data at the county level, I used IBM SPSS software in version 29.0.1.1 to identify four clusters from the 24 representative counties based on the structural county data, excluding those related to age. These clusters will be referred to as sociodemographic clusters. Using the corpus, in which I had amended the structural data of the county of residence to each case, the clustering was normalized against the numbers of swinger cases, thus reflecting the environments in which the identified swingers live. Therefore, the cluster membership of the county of residence can be considered a variable broadly reflecting the personal living environment of the persons studied. Details of this clustering process are provided in a file within the data package for this study (Maor, 2025a). Then, the mean and median swinger data for each of the clusters were calculated.

This calculation was replicated for the data pertaining to each of the big cities (Berlin, Cologne, Hamburg, and Munich). The objective was to uncover any potential variations attributable to different urban contexts.

Analysis of Subgroups

In August 2024, I collected data on identified swingers who declared BDSM as an interest or who indicated open relationship or polyamory as their relationship status. I used the corresponding filter criteria and otherwise followed the same methodology as described above.

Extrapolation of Swinger Prevalence to a Germany-Wide Total

The findings on solo-couple duplication and the representativeness of Study 1 data enable an approximation of the total number of persons in Germany meeting the identified swinger criteria. This figure has also been estimated.

Results

A total of 9325 women and 13,648 men were recorded as identified swingers, of which 6,077 each were in couple profiles. This means that the ages and places of residence of 3248 solo women and of 7571 solo men were included in the data pertaining to the representative counties and the big cities. From the 24 representative counties, data of 3135 identified swinger women, with 917 solo women, and of 5256 identified swinger men, with 3038 solo men, were harvested. Unless otherwise specified, the following data will be based on the data collected from the representative counties.

Age of Identified Swingers in the Sample

For identified swinger women, the median age was found to be 44 years (95% CI [43, 44]), and for men, the median was 46 years (CI [46, 47]), as visualized in the accompanying histograms, which illustrate the age distribution of identified swinger women and men, respectively; see Fig. 1 regarding women and Fig. 2 regarding men.

Fig. 1 — Age distribution of women with verified Joyclub profiles (including a picture) who self-declare as swingers, from 24 German counties selected for their sociodemographic representativeness. The vertical line represents the median age (44.0 years)

Fig. 2 — Age distribution of men with verified Joyclub profiles (including a picture) who self-declare as swingers, from 24 German counties selected for their sociodemographic representativeness. The vertical line represents the median age (46.0 years)

Kurtosis for the age of women was − 0.518, and skewness was 0.98. For men, these values were − 0.599 and 0.002, respectively. They suggest a relatively symmetrical distribution around the mean, as shown by the bell-shaped curves overlaid on the histograms. The kurtosis for the age difference within couples was 2.182, paired with a skewness of 0.493, pointing to a heavier tail distribution with a slight tendency toward larger age differences where the man is older.

Focusing on the age difference across couples, see Fig. 3, the 5th percentile indicated an age difference of -5 years (95% CI [− 5, − 4]), where negative values signify that the woman in the couple is older. Conversely, at the 95th percentile, the age difference was + 13 years (CI [+ 12, + 13]), where positive values indicate that the man in the couple is older. These percentile values show the range of age differences observed and variability in the age differences among couples within the study sample.

The median estimates from the distribution show moderate precision, with standard errors ranging from 0.01 to 0.50 across age percentiles, derived from 10,000 bootstrap samples. This level of precision affirms the reliability of the age characteristics in the sampled population. In the bootstrap exercise, weighted averages were calculated by normalizing weights to preserve the actual sample size, aligning the sum of weights with the number of observations. This method upholds the sample’s representativeness.

County-Related Assessment of Ages

For women, the range of median ages in the individual counties included in the sample was from 41.0 years (n = 812, SD = 9.86; Leipzig) to 51.0 years (n = 24, SD = 12.38; Lüchow-Dannenberg). For men, the values ranged from 43.0 years (n = 233, SD = 9.24; Halle [Saale]) to 53.0 years (n = 31, SD = 11.13; Lüchow-Dannenberg).

The median age difference among couples was two years in 9 counties and three years in 12 counties. In one county, an age difference of one year was found (n = 24, SD = 7.42; Sömmerda); in another county, the median age difference was four years (n = 87, SD = 5.54; Hochtaunuskreis). The highest median age difference found in any county was five years (n = 28, SD = 3.56; Starnberg). The detailed aggregated county-related age data are contained in Table S2 in the Supplementary Material.

Correlations of Swinger Density in Clusters

The sociodemographic clusters, identified through the analysis described in the Methods section, are presented in Table S3, with detailed features shown in Table S4, both in the Supplementary Material. Clusters 1 and 2 contain rural settings and mid-sized cities, with cluster 1 showing a higher average disposable income and a higher proportion of residents with a migration background than cluster 2. Cluster 3 comprises cities with comparatively low wealth, and the two counties in cluster 4 are semiurban and show a high average income.

The clusters distinctly represented the typologies of various counties in Germany, but the numbers of identified swingers within these clusters, normalized against 100,000 inhabitants, showed variation. No clear pattern or regularity emerged in this context. The numbers fluctuated within the clusters and did not follow a geographical alignment, as also clearly visible in Figs. 4 and 5.

Fig. 4 — Number of identified swingers (verified Joyclub profiles with a picture, self-declaring as swingers) per 100,000 inhabitants across German counties. Circle sizes indicate the ratio of identified swingers; shades of gray represent county population density. County boundary data

© GeoBasis-DE/BKG (2023) (data modified). Population density source: Statistisches Bundesamt (2022). Licenses for both datasets: dl-de/by-2-0 (FITKO, n.d.). A location key is provided in Supplementary Material Figure S3

Fig. 5 — Number of persons maintaining a verified Joyclub profile with a picture, self-declaring as swingers (identified swingers), per 100,000 inhabitants in relation to the population density (inhabitants per km²) in 24 representative counties selected for their sociodemographic representativeness of Germany, and the cities of Berlin, Cologne, Hamburg, and Munich

In the ordinary wealthier counties, the swinger density ranged from 82.9 in Ostalbkreis to 380.5 in Ulm. In the rural to semiurban areas, the lowest density was 80.7 in Prignitz, with the highest density in Leipzig (city and county) at 246.2, closely followed by Düren at 244.1. Among the low-wealth cities, Gelsenkirchen, the German county with the lowest available per capita income, registered the lowest density at 147.2, whereas Duisburg, with the second lowest per capita income in the country, recorded a density of 171.6. In the wealthy county subset, Starnberg had a density of 81.9 and Hochtaunuskreis identified 125.3 swingers per 100,000 inhabitants. The figures for major cities also varied significantly, ranging from 106.2 swingers per 100,000 inhabitants in Berlin to 298.6 in Cologne, the latter being the second highest value observed. The number of identified solo women swingers per 100,000 inhabitants also varied and ranged from 2.3 in Altmarkkreis Salzwedel to 36.7 in Cologne. The density of solo men varied wider and ranged from 24.8 in Lüchow-Dannenberg County to 151.2 in Ulm. In Ostalbkreis County, 58.4% of all identified swinger women were solo women, whereas in Altmarkkreis Salzwedel, the corresponding share amounted to only 7.0%, with the lowest value detected.

Detailed information, including gender-specific data and distinctions between all swingers and solos, is available in Table S5 in the Supplementary Material.

Correlations Within Swinger and Sociodemographic Data

As detailed in Table S6 in the Supplementary Material, significant correlations were found within the sociodemographic county data. These include relationships between population density and migration background rate, between tertiary and MINT employment ratios, and between migration background rate and average income. Knowing these correlations is essential for interpreting correlations between swinger data these sociodemographic variables. No correlation was found with the number of divorces per 100,000 inhabitants.

The analysis of correlations between swinger densities, age data, and sociodemographic data yielded significant correlations, as shown in Table 1. Again, no significant correlations were found with the divorce rate. The complete correlation matrix is contained in Table S7 in the Supplementary Material.

Table 1.

Correlations of swinger-related data and sociodemographic county data in the representative set

Value	Average available income in the county in Euro		Quota of persons employed in the tertiary sector^a		Quota of persons employed in the MINT^b sector^a		Population density^c		Proportion of inhabitants with a migration background in %
Value	r_s(23)	p	r_s(23)	p	r_s(23)	p	r_s(23)	p	r_s(23)	p
Median age of men^d	.46 [.04, .78]	.025	.11 [− .34, .54]	.597	.05 [− .38, .45]	.833	− .20 [− .61, .23]	.354	.13 [− .30, .58]	.556
Median age difference in couples^{d, e}	.35 [− .07, .69]	.093	.41^g [− .13, .74]	.045	.41^g [− .03, .72]	.045	.35 [− .07, .69]	.093	.39 [− .06, .69]	.057
Swinger density^{d, f}	− .14 [− .54, .32]	.511	.17 [− .26, .57]	.416	.16 [− .30, .57]	.445	.46 [.09, .72]	.023	.33 [− .10, .70]	.113
Density of solo swinger women^{d, f}	− .17 [− .56, .33]	.426	.33 [− .12, .71]	.116	.27 [− .17, .66]	.200	.65 [.33, .80]	< .001	.39 [− .07, .73]	.063
Density of solo swinger men^{d, f}	− .18 [− .57, .30]	.391	.02 [− .36, .43]	.920	.29 [− .19, .69]	.169	.60 [.24, .82]	.002	.44 [.04, .75]	.029
Density of swinger men^{d, f}	− .16 [− .55, .31]	.463	.14 [− .29, .53]	.517	.15 [− .32, .60]	.485	.46 [.10, .72]	.025	.31 [− .10, .66]	.144

Open in a new tab

r_s values represent Spearman’s rho correlations. Only correlations between age and density variables with sociodemographic variables are shown. Values in square brackets indicate the 95% confidence intervals for each correlation. Bias-corrected and accelerated confidence intervals were calculated using bootstrapping with 1000 iterations

^aIn relation to all employees, in percent

^bThe acronym “MINT” relates to sectors related to mathematics, information technology, natural sciences, and technology (as defined in German statistical classifications)

^cInhabitants per km²

^dThe figure relates to swingers identified by a verified Joyclub profile with a picture, self-declaring as swingers

^ePositive values indicate that the man is older

^fRates are normalized by 100,000 inhabitants in the county of residence

^gAlthough the correlation is statistically significant at p < .05, the bootstrapped 95% confidence interval includes zero, indicating that the correlation may not be robust. This suggests that any observed correlation should be interpreted with caution

The findings indicate interdependencies among swinger demographics and certain sociodemographic factors. Generally, population density correlate with swinger density rates, and age data correlate with income figures and the factors correlated to it, involving the migration quota and, with the caveat that bootstrapping results require careful interpretation, the occupation structure. The moderate positive correlation between swinger men density and the proportion of inhabitants with a migration quota forms an exception in this regard.

Subgroup Analysis

In the counties of the representative dataset, the density—defined as the number of swingers per 100,000 inhabitants—of individuals expressing interest in BDSM reached 37.4. The same dataset revealed a density of 22.1 for swingers who declared an open relationship as their relationship status and 4.5 for those identifying with polyamory. These subgroups exhibited a higher prevalence in major cities compared to other counties. For instance, Berlin recorded a density of 52.5 for BDSM swingers, 34.2 for open relationship swingers, and 8.6 for polyamorous swingers. Cologne displayed the highest density, with 78.1 BDSM swingers. Additional details can be found in Table S13 in the Supplementary Material.

Significant correlations of the subgroup data, collected in August 2024, with the main dataset from 2023 were found regarding median ages, identified persons normalized against the respective county population, and variations of such densities among counties and between subgroups. The number of BDSM swingers per 100,000 inhabitants and the quota of persons employed in the MINT sector in percent correlate with r_s(23) = 0.59, p = 0.003, 95%CI[0.28, 0.81]. In addition, some correlations found within the structural data from the main dataset were also identified in the subgroup sets. More detailed results are reported in Tables S7 to S12 in the Supplementary Material.

Use of Duplicate Profiles

The analysis revealed an estimated number of 1,734 solo-couple duplications (95% CI[1561, 1908]) in counties including big cities, potentially affecting 610 solo women (CI[526, 700)] and 1124 solo men profiles (CI[994, 1261]). This represents 16.0% (CI[14.4, 17.6]) of the total 10,819 solo profiles in the dataset including the big cities. When excluding big cities, the estimated solo-couple duplications drop to 748 individuals (CI[673, 820]), affecting 248 solo women (CI[215, 283]) and 501 solo men profiles (CI[435, 571]), representing 18.9% (CI[17.0, 20.7]) of the 3,955 solo profiles in the dataset without big cities.

Moderate to strong positive Spearman correlations were observed between these duplications and the population size (including big cities: r_s(27) = 0.92, p < 0.001, 95% CI[0.83, 0.97]; excluding big cities: r_s(23) = 0.43, p = 0.023, CI[0.04, 0.71]), the population density (including big cities: r_s(27) = 0.50, p = 0.007, CI[0.12, 0.77]; excluding big cities: r_s(23) = 0.62, p < 0.001, CI[0.28, 0.83]), and the swinger density values (including big cities: r_s(27) = 0.51, p = 0.006, CI[0.15, 0.77]; excluding big cities: r_s(23) = 0.59, p = 0.001, CI[0.25, 0.81]), with all 95% CI values being based on 1,000 bootstrap samples.

Spearman analyses of the data adjusted by the duplications did not reveal other correlations than the non-adjusted data. The correlation of non-adjusted data with population density remained consistent after adjustment, with the result for the adjusted data being situated within the confidence interval of that based on unadjusted data (see part B.7 of the Supplementary Material for details).

In the representative dataset, the likelihood of same-type duplicate profiles across all categories is minimal, with estimated duplications affecting between 0.9% and 1.1% of profiles. These low percentages suggest that the prevalence of same-type duplicate profiles is not significant enough to introduce bias or inaccuracies in the analysis. The reliability of the predictive model was further validated by including the larger cities, where the model predicted overlaps with near-perfect accuracy. Therefore, the data can be considered highly reliable in this respect, and the observed differences between expected and actual duplications are well within the acceptable range.

Germany-Wide Total Prevalence of Identified Swingers

Extrapolating values from the representative dataset and factoring in estimated solo-couple duplications, 128,557 individuals (95% CI[127,361, 129,836)] with 48,567 women (CI[47,960, 49,105]) and 79,990 men (CI[78,811, 81,101]), all aged between 18–74, are assumed to meet the identified swinger criteria in Germany (self-declared swingers using verified Joyclub profiles with a profile picture). This corresponds to 0.21% of the German population aged 18–74. For women, the value amounts to 0.16%, and for men to 0.25%. Details of this calculation are reported in part C of the Supplementary Material.

Discussion

The median ages of identified swinger men (46 years) and women (44 years) in Germany slightly exceed those reported in the UK by Haywood (2022), in Dutch STI clinics by Dukers-Muijrers et al. (2010), and those found by O’Byrne and Watts (2011) for visitors of a Canadian swinger club. A detailed comparative analysis of swinger demographics across these countries is outside the scope of this study.

The higher median age of men almost reflects the median age difference in couples, which is three years. This finding contradicts a possible stereotype of midlife men predominantly seeking much younger partners in swinging contexts. The correlation of men’s median age and the median age difference in couples with the wealth of their residential counties, whereas such correlation could not be found for women, might suggest that women participate in swinging, and are accompanied by their typically older partners.

In contrast to earlier suggestions to concentrate on couples in swinger research (Jenks, 1998; Vaillancourt, 2006), a considerable number of solo profiles were observed among identified swingers. This prevalence of solo profiles, particularly among men, indicates that participation in swinging activities extends beyond partnered contexts. In Germany, the prevalence of solo women profiles does not justify the label “unicorns” for unattached swinger women, a term Haywood (2022) used for the UK and Bergstrand and Sinski (2010) applied to the USA context. Nevertheless, the relatively lower representation of women in the dataset remains unexplained by the current data. In particular, the observation that the number of solo swinger women strongly correlated with population density, whereas the corresponding value for solo men was weaker, warrants further investigation to understand the motivations underlying these dynamics.

The findings contradict the portrayal by Friedman et al. (2017) of swinging as an activity predominantly associated with individuals in less affluent circumstances. The presence of swingers in wealthier counties challenges the notion that swingers are exclusively from impoverished or marginalized backgrounds. Furthermore, the lack of significant correlations between swinger density and factors such as divorce rates, employment in MINT occupations, or the proportion of residents with a migration background—except for a correlation with the latter in the case of solo men—further questions assertions by Jenks (1998). Contrary to Jenks’s view that swingers are typically not members of minority groups and generally come from wealthier or more educated backgrounds, this study’s findings suggest a more complex and diverse demographic. These results align with the earlier research by Rubin et al. (2014) and Levine et al. (2018), who defied assumptions about correlations between CNM prevalence and ethnicity, education, or income, while Balzarini et al. (2019) found that the latter two factors were relevant with regard to polyamorists.

Rates of tertiary sector and MINT employment correlate with larger age differences in couples, but they do not impact swinger density. This macro-level observation does not reflect the micro-level based findings of Balzarini et al. (2019) for polyamorists in the USA, who were reported to work significantly more often in information technology professions.

On a macroscopic level, some regional similarities in swinger density suggest the influence of very local cultural factors. For example, the similar swinger densities observed in Cologne and adjacent Düren County, and the relative homogeneity of age and swinger density data in Leipzig and neighboring Halle (Saale), support this idea of regional influence. Any precise prediction of swinger density based on sociodemographic factors, if at all possible, would thus prove to be challenging.

In essence, the results do not confirm that the relationship between socioeconomic factors and swinger demographics aligns with the traditional parameters used in earlier studies. Cross-cultural comparisons or those between different types of CNM practices might be challenging without a viable theoretical foundation, which yet has to be developed, and all comparisons require careful consideration of the level of analysis on which the results were generated.

Some correlations found in the BDSM and open relationship swinger subgroups, particularly regarding the county income, suggest that swinger demographics are influenced by latent factors that could not be covered in this study due to its design-related limitations. Future research should explore individual motivations and the cultural influences across regions to better understand the interaction between related factors.

The intersections among swingers and BDSM, open relationship, and polyamory subgroups in Germany require careful interpretation. Providing related data is optional for profile users. Therefore, strategic choices in identity presentation rather than comprehensive descriptions of individuals’ sexual practices are likely represented. The seemingly lower prevalence of open relationship and polyamory identifiers among swingers requires particularly careful interpretation. As noted by Haupert et al. (2017) and Cardoso et al. (2021), terms like open relationship and polyamory carry different connotations across cultural contexts. Schmidt (2023) identified a narrower definition of the German term Polyamorie compared to the English term polyamory, which may influence how individuals categorize their relationship. Rubel and Burleigh (2020) highlight the varied understandings of polyamory in the USA, with only 38.6% of participants at all showing some basic understanding of the term. Therefore, different interpretations may complicate demographic analysis based on self-identification, in particular across different languages.

Specifically, the particularly low prevalence of polyamory among swingers may simply reflect differences in terminology. Venn’s (2015, pp. 266–267) concept of “Cliquen-Swinging” refers to a pattern where swingers limit their sexual activities to a close-knit group or “clique” of trusted partners. These cliques often involve stable and repeated interactions, which may resemble “polyamory” according to some definitions and from an outside perspective, but participants may not identify with this term.

A theoretical limitation of this Study 1 concerns the prevalence of potential duplicate profiles. The possibility of same-type duplications cannot be dismissed, but the estimates show low values. Given the relatively narrow age distribution of swingers, the alignment between predicted and actual overlaps supports the robustness of the prediction model and the quality of the dataset. In contrast, the calculated frequency of solo-couple duplications is considerable. Population size and density appear to drive the corresponding user behavior. Nevertheless, after adjustment by the respective duplication estimates, the resulting correlations to demographic data remain substantially the same. This means that the profile duplications do not introduce meaningful bias into the study’s primary finding, which confirms the robustness of its results also in this regard. The results reported here therefore remain unaffected by both types of duplications.

Study 2: Guests at Swinger Events

Study 2, in contrast to Study 1, focuses on observable behavior in the context of physical encounters by analyzing admissions to swinger events advertised on Joyclub. Swinger events represent a central format through which participants initiate contact and engage in shared experiences. The resulting data offer an independent empirical basis to assess the plausibility of patterns identified in Study 1 and to gain additional insights into preferences, travel behavior, and financial commitment as reflected in entrance fees and pricing structures.