1
d
Co-Housing and loneliness
To what extent are adults
that live in cohousing in The Netherlands
less often socially lonely?
ergsdfg
Cohousing and Social Loneliness
A quantitative study on the relationship between
cohousing and social loneliness in the Netherlands
Bachelor thesis, June 17th, 2019
Author: Tijmen Kuyper
Supervisor: Beatriz Pineda Revilla
Thesis proposal, March 19th 2019
Student: Tijmen Kuyper
Supervisor: Beatriz Pineda Revilla
Co
2 The cover picture is from cohousing project Startblok Riekerhaven in Amsterdam.
(Startblok Riekerhaven, 2016)
“First we shape cities, then they shape us”
-
JAN GEHL (2010) Cities for People
Colofon
Course: Bachelors Thesis
University: University of Amsterdam
Bachelors: Human Geography & Urban Planning Teacher and first corrector: Beatriz Pineda Revilla
Second reader: Ori Rubin
Date: 17th of June 2019
Author: Tijmen Kuyper
UvA student ID: 11055073
3
Preface
You are reading the results of multiple months of research into the relationship between cohousing and social loneliness. This research has been conducted as thesis that concludes my bachelor Human Geography and Urban Planning at the University of Amsterdam. Human geography showed me the greatness of the impact of the way space is organised on the people that use it. Also, I learned that people consciously and unconsciously shape their space with social processes. The way we shape our neighbourhoods, shapes us. It is my passion to find ways to shape the limited space that we have in such a way that improves the experience of our lives. Currently many countries face a “loneliness epidemic”. Cohousing is a way to let people live in a community and is proclaimed by many as a medicine for the increasingly lonely society. In this thesis I investigate if there is indeed a relationship between cohousing and loneliness.
First, I would like to thank my supervisor Beatriz Pineda Revilla. Together with her colleagues she always swiftly provided me with useful feedback. Also, I want to thank my girlfriend Sanne for her mental support whenever I was pulling an all-nighter again. Furthermore, I thank her and my good friend Jesse for their helpful readthrough of the text. Exceptional praise goes to the hundreds of respondents who took the time to fill in my questionnaire with personal questions. Finally, this research really was made dozens of people who distributed the survey among residents, friends and neighbours. Thank you all!
Voorwoord
Voor u ligt het resultaat van een maandenlang onderzoek naar de relatie tussen Centraal Wonen (cohousing) en sociale eenzaamheid. Dit onderzoek is uitgevoerd als scriptieproject waarmee ik de bacheloropleiding Sociale Geografie en Planologie aan de Universiteit van Amsterdam afrond. Binnen de sociale geografie heb ik geleerd hoe veel invloed de invulling van de ruimte heeft op de mensen die die ruimte gebruiken. Maar ook hoe mensen zelf deze invulling vormgeven door bewuste en onbewuste sociale processen. De manier waarop wij onze buurten vormen, vormt ons. Het is mijn passie om op zoek te gaan naar manieren om de beperkte ruimte die we hebben, zo in te vullen dat we er als mensen in levensgenot op vooruit gaan. Momenteel kennen veel landen in de wereld een snelgroeiende “eenzaamheidepidemie”. Cohousing is een opkomende woonvorm waarvan velen beweren dat het mensen meer in een gemeenschap kan laten wonen en zodoende een medicijn is voor de steeds eenzamere maatschappij. In deze scriptie onderzoek ik of er inderdaad een relatie is tussen cohousing en eenzaamheid. Ik heb besloten om de scriptie in het Engels te schrijven om aan te sluiten op de internationale wetenschapsliteratuur, wel is er een korte Nederlandse samenvatting.
Ten eerste wil ik mijn begeleider Beatriz Pineda Revilla bedanken, die samen met haar collega’s, mij altijd snel en scherp van feedback voorzag. Ook wil ik mijn vriendin Sanne bedanken voor de mentale steun als ik weer eens een nacht aan het doorhalen was. Naast Sanne eveneens dank voor mijn goede vriend Jesse voor ook hun nauwkeurige blik op de tekst. Buitengewoon veel dankbaarheid heb ik voor de honderden respondenten die tijd hebben vrijgemaakt om de enquête met persoonlijke vragen in te vullen. Tot slot is dit onderzoek echt mogelijk gemaakt door de tientallen mensen die de enquête hebben verspreid onder huurders, vrienden en buren. Allen bedankt!
4
Abstract
Loneliness is a rising problem. Currently, 10% of the adults in the Netherlands are (severely) lonely. Cohousing is increasingly pointed at as part of the solution. There is no prior statistical research on the relationship between cohousing and loneliness. Hence, this study aims to determine to what extent adults that live in cohousing are less likely to be socially lonely. An online survey was used to gather data of people who (want to) live in cohousing (N=263). Correlation analyses showed that cohousing correlates with social loneliness. Frequency of neighbourhood contacts is the mediator in this relationship. People who live in cohousing are likely to have more neighbourhood contacts and are thus less socially lonely. However, those that live in cohousing and do not have frequent neighbourhood contacts are still likely to be socially lonely, despite living in cohousing. Logistic regression showed that while controlling for education, having an occupation and disability, cohousing residents are 80.5% less likely to be socially lonely than those who want to live in cohousing but not do so yet.
Samenvatting
Eenzaamheid is een toenemend probleem. 10% van de Nederlanders is nu (zeer) ernstig eenzaam. Centraal Wonen (cohousing) wordt steeds vaker aangewezen als onderdeel van de oplossing. Er is nog geen statistisch onderzoek gedaan naar de relatie tussen cohousing en eenzaamheid. Daarom probeert dit onderzoek aan te tonen in hoeverre de kans kleiner is dat mensen die in cohousing wonen niet sociaal eenzaam zijn. Een online vragenlijst is gebruikt om voor het eerst statistische data te verzamelen van mensen die (graag willen) wonen in cohousing (N=263). Correlatieanalyses tonen aan dat wonen in cohousing correleert met sociale eenzaamheid. Frequentie van burencontacten verklaart dit verband als mediator. Dit betekent dat de kans om sociaal eenzaam te zijn kleiner is voor mensen in cohousing, maar alleen voor mensen die vaak contact hebben met hun buren. Logistische regressieanalyse die controleert voor opleidingsniveau, dagelijkse bezigheid, en beperking door gezondheid, toont aan dat mensen die in cohousing wonen 80,5% minder waarschijnlijk sociaal eenzaam zijn dan mensen die in cohousing willen wonen maar dit (nog) niet doen.
5
Index
Preface ... 3 Abstract ... 4 Introduction ... 6 Theoretical framework ... 7Loneliness and neighbourhood contacts ... 7
Cohousing and neighbourhood contacts ... 8
Factors correlating with cohousing and loneliness ... 9
Personality... 10 Research question ... 11 Methodology ... 12 Research design ... 12 Data ... 13 Sampling ... 13 The survey ... 14
Process of data collection ... 15
Removing responses ... 19
Constructing the variables ... 19
Methods of analyses ... 23
Results ... 25
Overview of the respondents ... 25
Q1. Cohousing and Neighbourhood contacts ... 28
Q2. Neighbourhood contacts and Loneliness ... 29
Q3. Neighbourhood contacts as mediating variable ... 30
Q4 Testing for spurious relationships between Cohousing and Social Loneliness ... 31
Conclusion ... 33
Discussion ... 34
Literature ... 36
Appendix 1. Initial letter to cohousing projects ... 41
6
Introduction
Dutch citizens are now more connected to the world than ever (Katz, 2017). New technologies allow people to communicate in ways that were unimaginable for many people a few decades ago, yet more people become severely lonely every year (Goggin & McLelland, 2017; Beuningen & Witt, 2016; Movisie, 2018). Especially in the bigger cities the percentage of people that experience social and emotional loneliness increases rapidly (GGD, 2017). In Amsterdam the percentage of lonely people in 2008 was 40% (GGD, 2016). Eight years later already 48% of those living in the Dutch capital experience loneliness. More than a 100.000 of the 820.000 citizens are not just lonely but severely lonely. Pensioners are more likely to experience loneliness, yet 40% of the adults between 19 and 34 years old endure loneliness as well. Loneliness is often divided in emotional and social loneliness (Ministerie van Volksgezondheid, Welzijn en Sport, 2018). Emotional loneliness is the experienced lack of intimate relationships while social loneliness is the experienced lack of meaningful relationships with a wider group (Weiss, 1973).
Loneliness is not just an unpleasant feeling; it has personal health and societal consequences. Loneliness causes high levels of stress, reduces happiness in life, causes depressions and even shortens the lifespan (Luo, et al., 2012). Other western countries struggle with similar rising numbers of lonely people. In the US some researchers already claim that “the loneliness epidemic may present
a greater public health hazard than obesity.” (American Psychological Association, 2017). Lodder, et
al. (2016) point out that the increase of loneliness in society also brings along greater pressure on the welfare state as healthcare costs increase. Another problem is that lonely people are often caught in a negative spiral which leads them to increasingly avoid interactions (Heinrich & Gullone, 2006). This causes further societal problems. Dekker et al. (2002) show for example that lonely people are less likely to vote. In Bowling Alone (2000) sociologist Putnam warns that in western societies people are increasingly socially isolated, lonelier and less participatory in society.
The consequences of loneliness are thus diverse, and so are the causes. Loneliness is a highly complex problem to solve (Westelaken, 2012). Putnam (2000) puts most of the blame on the increased use of technology resulting in decreasing real live contact. This is however not the only cause. Other research found that one of the main factors that contribute to loneliness is living alone (Campen, et al., 2018; Klinenberg, 2012). CBS (2018) predicts that in the Netherlands in the next 11 years, 406.000 extra one-person households will be added to the already 3.1 million. In 2016, a Dutch nationwide health survey concluded that loneliness correlates with age, gender, education, (immigrant) origin, having an occupation, disability and the frequency of contact with neighbours (GGD, 2017). Not just societal trends and someone’s position in that society predicts loneliness, also personality and genetics are an important factor in predicting loneliness (Boomsma, et al., 2005). Evident is that loneliness is highly complex. While factors like genes and health are difficult, if not impossible to change, factors like neighbourhood contacts are relatively changeable. Furthermore, the frequency of neighbourhood contacts has a strong negative correlation with social loneliness (Masi, et al., 2011). This makes increasing the frequency of neighbourhood contacts a widely used prevention an intervention strategy (Ministerie van Volksgezondheid, Welzijn en Sport, 2018; Movisie, 2018).
Qualitative case studies show that people who live in cohousing have more frequent neighbourhood contacts (Bouma & Voorbij, 2009; Tummers, 2017; Williams, 2005, Fromm, 2000). Cohousing is defined as housing that features spaces and facilities for joint use by all residents who also maintain their own individual household (Franck and Ahrentzen, 1989). Currently there is a rapid rise in cohousing projects in the Netherlands (Tummers, 2017). Both cohousing researchers and loneliness researchers see the possibility that cohousing could be part of the solution for society’s growing loneliness problem. This potential has however never been researched. Holtzman (2012) stated that
7 more research is needed on the potential of cohousing to tackle social isolation and loneliness. In more recent research Hopwood & Mann (2018, p.3) also mention the potential of cohousing: “Cohousing may have the potential to promote socialization and neighbourliness and improve factors
affecting loneliness such as helping residents feel valued, useful and part of a community”. The Dutch
state research institute Movisie (2018, p.16) published a report on preventing loneliness and stated that modern ways of shared housing, that maintain sufficient privacy, could potentially help to prevent loneliness in the future. However, the report underscores that more research needs to be conducted about preventing loneliness among adults younger than 65 years old.
Since loneliness is complex and different for every person, there can never be one solution (Beuningen & Witt, 2016; Boomsma, et al., 2005; Lodder, et al., 2016; Westelaken, 2012). Still, cohousing could be part of the solution for this societal problem. The aim of this thesis is to research to what extent living in cohousing influences social loneliness for adults older than 18. With quantitative falsification this research tries to answer the question: to what extent are adults that live in cohousing in the Netherlands less likely to be socially lonely? Based on discussed literature the hypothesis is constructed that people who live in cohousing are less likely to be socially lonely due to more frequent neighbourhood contacts.
Theoretical framework
Loneliness and neighbourhood contacts
Humans are social by nature. They have a need to belong, and a need to have satisfying connections with other human beings (Heinrich & Gullone, 2006). Loneliness is not a lack of contacts but the subjective lack of contacts (Weiss, 1973). Loneliness is therefore not the same as social isolation. A person can be socially isolated without experiencing that as a negative situation. Loneliness is always perceived as unpleasant and involuntarily (De Jong Gierveld, 1984).
“Loneliness is a situation experienced by the individual as one where there is an unpleasant or inadmissible lack of (quality of) certain relationships. This includes situations, in which the number of existing relationships is smaller than is considered desirable or admissible, as well as situations where the intimacy one wishes for has not been realized.”
(De Jong Gierveld, 1987, p. 120)
In most of the research and Dutch policy there is a distinction between two kinds of loneliness (Ministerie van Volksgezondheid, Welzijn en Sport, 2018). The first, emotional loneliness, is caused by an experienced severe lack of intimate relationships (Weiss, 1973; Tilburg, 2007). This can be the perceived absence of a partner or close friends. The other type is social loneliness. Someone is socially lonely if he or she misses meaningful relationships with a wider group of people like acquaintances, colleges, neighbours or people with the same interests (Weiss, 1973; Tilburg, 2007). Someone who has a loving relationship and a good friend but still experiences a lacking social network can still experience social loneliness. This thesis focusses on social loneliness because cohousing is more likely to influence someone’s wider social network like neighbourhood contacts than intimate relationships (Tummers, 2017).
Neighbourhood contacts is a factor that is repeatedly found to directly influence social loneliness. According to Vermeij (2008) people with close neighbourhood contacts are happier and less lonely. Movisie (2018) indicates that increasing the frequency of neighbourhood contacts is both a helpful intervention and prevention for loneliness.
8
Cohousing and neighbourhood contacts
Although cohousing has its origins in the 1960s, only recently “the third way of housing” has been getting wide attention (Tummers, 2017). Cohousing originated in Denmark as part of a feminist movement with the idea that sharing household activities would free up time for women (Lang, et al., 2018). Later, cohousing projects with more focus on environmental sustainability and governance were started (Tummers, 2017). The Netherlands is among a few Northern European countries where cohousing is relatively widespread (Brenton, 2008). Although there is no official count of cohousing communities in the Netherlands, there are over 800 woongemeenschappen (housing communities) registered at gemeenschappelijkwonen.nl (Gemeenschappelijk Wonen, 2019). A rough estimate is that each community houses approximately 30 residents. That would mean that around 24000 Dutch people (or 0,14% of the country) live in some form of cohousing. Cohousing has numerous advantages compared to traditional housing. Since in cohousing facilities are shared, it is generally more environmentally sustainable and compact (Tummers, 2017). Those factors alone make cohousing an attractive option for all the cities that struggle with sustainability goals and lack of space. Furthermore, research shows that cohousing improves social cohesion (Williams, 2005; Korpela, 2012).
There are numerous definitions for cohousing and there is not a single agreed academic definition. Some definitions, for example, focus more on environmental ambitions of communities, other more on the governance or community aspects (Vestbro, 2000). The definition that this thesis uses is a combination of two definitions. The definition that this thesis uses is: Cohousing is housing where
besides your own private household you share spaces and facilities for joint use by all residents designed to facilitate encounters and to create community. It is a combination of two other
definitions. One has a on community, the other on privacy. The first definition is: “Cohousing is a
form of collective housing which has four common characteristics” (Mccamant & Durrett, 1994).
These four characteristics are summarized below.
- Social contact design. This means the building was designed in a way that residents are able to contact each other and create a sense of community.
- Common facilities are extensively used. These are community areas which are designed for daily use where residents can meet, talk and engage with each other.
- Resident involvement in the recruitment, production and operational processes. This could be policy designing that can affect the entire community, solving problems or introducing new services that might be needed in the community.
- Collaborative lifestyles offering inter-independence, sociability, support networks, and sense of security.
This first definition with the four characteristics was chosen because it focuses on the community aspect of cohousing which is important for loneliness related research. This definition is, however, still too broad for this thesis as it would still include elderly care homes and communes that facilitate different social dynamics than what is generally perceived as cohousing (Vestbro, 2000). The second definition by Franck and Ahrentzen, (1989:3) is: “Housing that features spaces and facilities for joint
use by all residents who also maintain their own individual household”. This definition was added to
exclude housing with less privacy and to add focus to the research. Furthermore, the research gap indicated by Movisie (2018) explicitly stated that modern ways of shared housing, that maintain sufficient privacy, could potentially help to prevent loneliness in the future.
9 People that live in cohousing are likely to have more frequent contact with their neighbours. Fromm (2000) claims that cohousing residents have 370% - 400% more contact with neighbours in cohousing than in their previous housing. Since the frequency of neighbourhood contacts influences social loneliness, it is thus highly likely that cohousing influences social loneliness through the frequency of neighbourhood contacts. This has however yet to be researched (Hopwood & Mann, 2018). Research that has come close to analysing if cohousing predicts social loneliness, was done among Dutch elderly in 2007. The study by Thomése (2007) did not include cohousing in her research. She did, however, include multiple housing types and their influences on loneliness through neighbourhood contacts. In her research amongst elderly, those living in care homes were least often socially lonely. Those living in flats were most often lonely. This was explained by difference in neighbourhood contacts. She did not include cohousing as housing type in this research. This thesis fills that gap. On basis of the discussed literature the following hypothesis is constructed: Someone who lives in cohousing has more frequent contact with neighbours and is consequently less socially lonely. This hypothesis is shown schematically in figure 1. The relationship between cohousing and social loneliness has neighbourhood contacts as the mediating variable. A mediating variable or mediator is a variable that explains the relationship between the independent and dependent variable (Bryman, 2012). Again, in this model the frequency of neighbourhood contacts is the mediating variable that explains the relationship between cohousing and social loneliness.
Figure 1. The basic conceptual model with neighbourhood contacts (V2) as mediator of cohousing (V1) and social loneliness (V3).
Factors correlating with cohousing and loneliness
Many factors are associated with social loneliness. In 2016 a nationwide survey was conducted in the Netherlands among 457.000 adults. This survey found that besides contact with neighbours also age, gender, education, relationship status, (immigrant) origin, (un)paid occupation, living alone and disability have correlation with social loneliness (GGD, 2017; Beuningen & Witt, 2016). It is thus not easy to point out single factors directly influencing loneliness. Statistical research on loneliness should thus consider this list of possible spurious variables. These variables are shown in the extended conceptual model in figure 2.
Besides research on loneliness, should statistical research related to cohousing, also be wary of spuriousness. Most cohousing communities select who may live there. Cohousing projects in the Netherlands are often targeted at certain groups in society (Tummers, 2017). As in most other countries, is cohousing usually for the elderly or those with (mental) health issues, making this group overrepresented in the cohousing population (Brenton, 2008). Only recently cohousing projects are being built for (young) adults like projects from Socius Wonen in Utrecht and Amsterdam. These projects then again inhabit large amounts of refugees (Socius Wonen, 2019). For cohousing research, it is thus important to consider that cohousing residents are not representative of a wider population. In Denmark, for example, white, higher educated people with a high income are overrepresented in cohousing (Jakobsen & Larsen, 2018). A study in Sweden and Denmark showed that women are overrepresented in cohousing (Choi, 2004). The same variables that correlate with social loneliness, also correlate with the likelihood that someone lives in cohousing.
10 In the figure below, the conceptual model of this thesis is shown. It shows that if someone lives in cohousing, he or she is likely to have more frequent contact with neighbours. Also, it shows that the frequency of contact with neighbours influences social loneliness. Finally, it shows the hypothesis that the control variables (age, gender, education, living alone, origin, occupation and disability), correlate with both social loneliness and with if someone lives in cohousing or not.
Figure 2. The extended conceptual model including the potential spurious variables V4-V10.
Personality
Finally, in both cohousing and loneliness research there is one more important factor: personality. Boyer and Leland (2018) state that personality is an important factor on whether people (want to) live in cohousing. Since cohousing is still not a widespread housing option, they showed that mainly extravert people with an interest in cohousing actually live there. Many cohousing communities select new residents that want to actively contribute to the community. This way, people that are socially competent are usually the ones who get selected to live in cohousing communities. Other research shows that for those that do live in cohousing, personality is again an important factor (Bouma & Voorbij, 2009). It influences the frequency of neighbourhood contacts a resident has. For example, an introvert person that does live in cohousing may still rarely have any contact with neighbours. This makes personality a moderating variable between cohousing and neighbourhood contacts. A moderating variable is a variable that modifies the relationship between the independent and dependent variable (Bryman, 2012).
Moreover, personality traits also have a significant influence on loneliness (Luanaigh & Lawlor, 2008; Tilburg, 2007). Those with low self-esteem, social anxiety, inadequate social abilities, neuroticism and introversion are more likely to be lonely. In the conceptual model on the next page, the complexity of personality is integrated. First of all, as a variable influencing whether or not people live in cohousing in the first place. Secondly as a control variable that directly influences social loneliness. This makes personality a similar variable as V4-V10 in that it could cause a spurious relationship between cohousing and social loneliness. Finally, personality is also a moderating variable influencing the correlation between cohousing and neighbourhood contacts.
11 Figure 3. The conceptual model including personality.
So, in general, people who (want to) live in cohousing and people who are less likely to be socially lonely are extravert. It is thus important to control for personality. Ideally, this would be done by measuring personality traits like introvertism and extrovertism for each respondent. However, since even some of the shortest questionnaires to measure personality have at least 30 items and take 5 minutes to complete, personality is left out of this research due to this practical concern (Konstabel, et al., 2017). To still control for personality, only people that either want to live in cohousing or already live in cohousing are included in this study, since in general those have similar extravert personality traits.
Research question
Genetic twin research goes beyond societal predictors and personality and states that loneliness is mainly determined by genetics (Lodder, et al., 2016; Boomsma, et al., 2005). This research even argues that interventions that aim to facilitate social activities for lonely persons are probably not effective since loneliness is mostly predicted by genetics. This is, however, in contrast with more demographic studies that state that more contact with neighbours does lead to a reduction in loneliness (Masi, et al., 2011; GGD, 2017; Movisie, 2018)
It is likely that cohousing could be part of the solution for the loneliness problem that society faces. Researchers like Schifferes (2018) who argue that cohousing is the simple solution that society needs, are overshadowed by others who state that there is not one simple solution. Since every lonely person is different, loneliness prevention and interventions will never consist of a single solution (Beuningen & Witt, 2016; Boomsma, et al., 2005; Lodder, et al., 2016; Westelaken, 2012). Cohousing should thus not be considered as the magic fix that can easily bring down loneliness in society. However, it has potential to be one of the many preventions or interventions to tackle the “loneliness epidemic”. To test if cohousing is indeed a way to reduce the chance to be socially lonely, this thesis aims to research: to what extend are adults that live in cohousing in the Netherlands less likely to be socially
12
Methodology
Research design
In order to answer this question, a quantitative methodology is used because of multiple considerations. First, it does address a research gap. Most research on cohousing use qualitative case study research designs. “The lack of verifiable quantitative data does little to support the ‘believers’
who claim that cohousing is ‘the third way of housing’ of the (near) future” (Tummers, 2017, p55).
Also, an extensive literature review on cohousing concludes: “Overall, the results of this
output-oriented thematic area call for more quantitative research to explore statistically significance, as well as for more critical reflections on CH” (Lang, et al., 2018, p. 19). Secondly, it is an advantage that
quantitative research says something about an entire population instead of only single cases. Compared to qualitative research, the external validity is higher for quantitative research (Bryman, 2012). The final argument to use quantitative methodology is that most research on cohousing is usually in debt with a small N and with numerous variables (Tummers, 2017). This research focusses on only one aspect: social loneliness. The relevant variables (except personality and genetics) are measured in a 5-minute survey, making it possible to have a large N that allows for statistical analyses with a probability of significant correlations.
Because it is widely assumed that cohousing reduces loneliness while it has never been empirically researched, there is scientific relevance to try to falsify this claim. Falsifying is done by deducting a hypothesis from theory and then trying to reject this hypothesis on basis of empirical observations (Popper, 1995). This thesis attempts to falsify the following hypothesis:
Adults that live in cohousing in the Netherlands are less likely to be socially lonely.
If the hypothesis is not rejected it is the best possible proof that adults in cohousing are indeed less lonely. If the falsifying succeeds, it indicates that the theory behind the conceptual model or the model itself are likely not correct.
The ideal research design would be longitudinal. Unfortunately, the timeframe of this research did not allow for that as new cohousing residents should be followed for years. The second-best research design is cross-sectional when longitudinal is not possible (Bryman, 2012). A cross sectional comparison has therefore been made during an approximately single point in time between people living in cohousing and those with interest for cohousing. The choice has been made to only include people living in or interested in cohousing to minimalize the effect of personality on the other variables. Thus, by doing a cross-sectional comparison between people who want to live in cohousing and people who already live in cohousing, people with similar personality traits are compared. This partly solves the complexity of personality; it does however shrink the population of the study.
In order to answer the research question, multiple sub questions need to be answered first. In figure 4 the conceptual model is shown again but now with the hypotheses included that correspond with the questions. The analyses have been conducted with computer program IBM SPSS 26.
Q1: To what extent do cohousing residents have more frequent contact with neighbours? Q2: To what extent are people with frequent neighbourhood contacts likely to be less lonely Q3: To what extent is neighbourhood contacts a mediating variable between cohousing and social
___.loneliness
Q4: While controlling for age, gender, education, living alone, origin, occupation and disability, to
13 Figure 4. Analytical conceptual model
Data
The data from the 2016 GGD gezondheidsmonitor that includes over 457.000 respondents and includes V2-V10 could have been useful as a control group. However, getting access to this data takes multiple weeks and costs at least €2650,00 (Loket Gezondheidsmonitors, 2019). Due to these practical restraints, this data set could not be used for this research. No other relevant data set regarding loneliness was found that could be used for this research. Even if one was found, a similar extensive and expensive process is likely because of the high privacy concerns of data including health and loneliness information. Also, there are no Dutch datasets yet on people living in cohousing (Tummers, 2017).
The only option was to collect data specifically for this research. To reach as many respondents as possible the choice was made to distribute an online survey instead of an analogue survey. This is due to the measurement validity of loneliness as will be elaborated in the section The survey on the next page and due to the efficiency in reaching respondents as will be described in the section The process
of data selection.
Sampling
The population of this research, and the group that this thesis can generalize its results to, are adults (older than 18) in the Netherlands that either live in cohousing or would (someday) like to live in cohousing. The unit of analysis are individuals and not, for example, cohousing projects. The population of this research is delimited to only those with interest in cohousing to improve internal validity. Since roughly the same personalities show interest in cohousing as that are not likely to be socially lonely, the internal validity of the analyses is improved. Non-response is a problem in this research because lonely people are generally less inclined to participate in society, it is expected that they are underrepresented in the response (Westelaken, 2012). Only English and Dutch speaking people were able to respond since the survey was only available in these languages.
14
The survey
To determine if someone lives in cohousing, wants to live in cohousing or does not want to live in cohousing (V1), the following question was used: “Do you currently live in cohousing? Cohousing is
housing where besides your own private household you share spaces and facilities for joint use by all residents designed to facilitate encounters and to create community”. The answers were “Yes”, “No, but I would (someday) really like to” and “No, also I do not want to”. In the second answer
“(someday)” was added during the data gathering process and therefore not all respondents saw the exact same multiple-choice answers to the same question. This choice is further elaborated in de section process of data collection.
The frequency of neighbourhood contacts (V2) was measured with the following question: “Approximately, how often do you have contact with your neighbours?”. The possible answers were“daily”, “more than three times a week”, “two or three times a week”, “once per week”, “twice
per month”, “once a month”, “less than once a month” and “rarely or never”. This question, along
with the multiple-choice answers, are based on the national health survey Gezondheidsmonitor (GGD GHOR, 2016).
To measure social loneliness (V3) there are two widely used methods for statistical research: These are the UCLA loneliness scale and the De Jong Gierveld Loneliness scale (dJG-scale). Despite UCLA-scale being more widely used, mainly outside western Europe, Penning et al. (2014) show that the dJG-scale is better for cross-sectional and longitudal research. This is because the dJG-scale has strong measurement invariance across age groups and successive measurements. The second reason to choose the dJG-scale, is that it consists of 11 closed questions instead of 20. The third reason is that most statistical research in the Netherlands uses this exact scale. The final and most important reason is that the dJG-scale measures loneliness, severe loneliness, emotional loneliness and social loneliness (RIVM, 2016). The scale consists of 11 statements on which someone can answer “yes”, “more or
less” or “no”. The statements from the scale are on the bottom of this page. Five of these questions
measure social loneliness and six measure emotional loneliness. Someone is emotionally or socially lonely if he/she scores negatively on two of the dedicated questions. This scale also states if someone is moderately lonely (3 to 8 negative answers) or severally lonely (9 to 11 negative answers) (Gierveld & Tilburg, 2017). Research points out that this scale works best if the setting is anonymously because participants often feel ashamed for their loneliness (De Leeuw, 1992; Sudman and Bradburn, 1974). Hence the survey has been distributed online and was anonymous.
1. There is always someone I can talk to about my day-to-day problems 2. I miss having a really close friend
3. I experience a general sense of emptiness
4. There are plenty of people I can lean on when I have problems 5. I miss the pleasure of the company of others
6. I find my circle of friends and acquaintances too limited 7. There are many people I can trust completely
8. There are enough people I feel close to 9. I miss having people around me 10. I often feel rejected
11. I can call on my friends whenever I need them
The other questions in the survey address control variables (V4-V9). These questions and answers are also based on the survey from the nationwide Gezondheidsmonitor (GGD GHOR, 2016). The complete survey that has been used in this research can be seen in appendix 2.
15 Loneliness related questions can amplify feelings of loneliness (Gierveld & Tilburg, 2017). To lessen the negative consequences this survey could have on individual respondents, a message was displayed after the survey was submitted. This message thanked the respondent for participating and kindly asked if he/she would like to send the survey to others. Most importantly, the respondent was informed on the fact that loneliness is common in the Netherlands. If the respondent is lonely or knows someone who is lonely he or she could click on a link from the Dutch government for information about loneliness (Rijksoverheid, 2019). Also the telephone number of “De Luisterlijn” was displayed which people can always call for help or to talk to someone (De Luisterlijn, 2019).
Process of data collection
The process of data collection did not go as initially planned. In general, seven different strategies were used to gather responses. During this process the book YES! from Goldstein, et al. (2017) was used to increase the effectiveness of the communication towards respondents. In later emails, for example, the sentence “many people in cohousing already helped me by filling in the survey” was included. This proved to be effective in getting people to also be willing to fill in the survey. In this section the different strategies, the relevant choices and their effectiveness is discussed. At the end of this section the process is summarized in a figure that also shows when and which strategies were applied.
The initial strategy to reach respondents that live in cohousing, or want to live in cohousing, was to contact the central email address of cohousing communities. In an email they were invited to participate in this research. They were asked to distribute the link of the online survey among their residents and future residents who are on a waiting list.
The cohousing projects that were contacted were initially selected in two ways. First the projects were contacted that were already known by the researcher. The list of these 13 projects was made by searching on the internet and the personal network of the researcher. This was non-probability sampling as the projects were non-randomly selected. Out of these 13 initial projects, one responded positively which resulted in four respondents.
After these projects were contacted with limited results, more projects needed to be contacted. In the Netherlands there is no official organisation that tracks all cohousing projects in the country. However, Gemeenschappelijkwonen.nl does have a list of over 800 woongemeenschappen (Gemeenschappelijk Wonen, 2019). Not all cohousing projects in the Netherlands are on this list. Many, but not all on this list, fit into the cohousing definition of this thesis. As it was too time consuming to all 809, a selection was made. This was done in multiple steps. First the sampling frame was constructed. The list on the website was copied into an excel sheet. All the rows were sorted. By doing this, the column became a long list of all the municipality names, zip codes, project names and websites. Only the part of the list with websites were kept and thus only the projects with a website. The woongemeenschappen without website were difficult to contact and to determine if these projects fit within the cohousing definition. This did create a sampling bias as only 313 out of 809 projects link to a website. Presumably the larger projects have a website and would therefore remain in the sampling frame. The remaining 313 were than randomized. This was done by inserting the =RAND() formula in column B and then sorting column A and B on basis of the randomly generated numbers. A probability sample was created out of the woongemeenschappen on the website
gemeenschappelijkwonen.nl that have an own website. The next step was to look into the projects one
by one to check if they fitted into the cohousing definition of this thesis. If so, the project was contacted with an email address or a contact form found on the website. The first 10 on this list were
16 contacted, one responded positively which resulted in six respondents. The list that was used, is not included in the appendix for privacy reasons.
During the same time period, 74 personal emails were sent out to individual people who were actively looking for a cohousing project to live in. These emails had a more informal tone and were personalised based on the stories of the people on the forums. These people posted on www.woongroep.net (68) and on www.omslag.nl (16). The forums were found by doing desk research. Although highly time consuming, personally emailing people who were actively looking for cohousing proved to be a much more efficient strategy than emailing cohousing communities. Around a third replied with a message that they filled in the survey and presumably more than one third filled in the survey.
Another strategy to reach respondents was by snowball sampling. If respondents finished the survey, they were kindly thanked and asked if they could help this research by sending the survey to others they know whom live in cohousing or are interested to live in cohousing. Weeks after the last email was sent, a few respondents a week still filled in the online survey. Most of these respondents were from cohousing projects who were never contacted directly and likely reached due to the snowball effect.
Since emailing cohousing communities proved to be highly time consuming with almost no result, Facebook was used to directly get in contact with residents. Communities were found by searching for “centraal wonen”, “CW” and “gemeenschappelijk wonen”, the Dutch equivalents of cohousing. 11 Facebook groups were found and after being accepted to join these groups, a small informal message was posted to ask the (future) residents to fill in the survey. This resulted in approximately 70 respondents who live in cohousing. The exact number is hard to tell due to the snowball effect and since many of the strategies described in this section were executed simultaneously.
To gather more data from the group of people who are interested in living in cohousing, projects in development were contacted. Initially these were found again on the forums of www.woongroep.net and www.omslag.nl. Most projects responded positively and a few shared contact information of other cohousing projects in development. Again, a snowball response, but now on the scale of projects instead of individuals. In total 15 cohousing communities in development were contacted, two of them through a Facebook page, the others by email or their website. This resulted in approximately 20 more respondents.
When all the emails and Facebook pages were contacted as described above, around 110 respondents filled in the survey. 30 of them wanted to live in cohousing, 80 of them lived in cohousing. Finding people that were actively looking to live in cohousing, especially proved to be difficult. In all the strategies above, the survey only reached people if they lived in cohousing or were actively searching for a cohousing community to live in. As most options to contact people who are looking for cohousing were done and with the result of only 30 respondents, a new strategy was used. The answer to the question if someone lives in cohousing “No, but I would really like to” was changed to “No, but
I would (someday) really like to”. Usually it is highly problematic to change a survey during the
process of data gathering. However, since everyone that responded “No, but I would really like to” before the change was contacted directly because they were actively looking for or building a cohousing community, they would have also responded “No, but I would (someday) really like to”. At this point only one person had answered “No, also I do not want to”. This is the only respondent that possibly could have answered differently. Since people who answer “No, also I do not want to” are not used in the analyses, this change is not a problem.
17 This change was made because it makes the question if someone would really like to live in cohousing more hypothetical. This allowed for the survey to also be distributed in a less targeted way. Instead of sending the survey only directly to people living in or looking for cohousing that fit the definition of cohousing used in this thesis, the survey was brought under the attention of as many people as possible. It was now up to the people filling in the survey themselves to judge whether they lived in/really wanted to someday live in/or did not want to live in cohousing according to the definition of this thesis. This weakened the measurement validity as some respondents probably quickly decided if their housing would be cohousing instead of a thorough check done by the researcher to determine if someone’s housing is cohousing.
After this small change the survey was distributed to as many people as possible. This was done by sending it to friends and family, sharing it on the private Facebook page of the researcher and most importantly posting it in the following Facebook groups: “Amsterdam durft te vragen” (Amsterdam dare to ask a question), “Kamer gezocht in Amsterdam” (room wanted Amsterdam), “Kamer
gezocht/aangeboden Amsterdam” (room wanted/offered Amsterdam) and internal Facebook groups
for students of the BSc Social Geography and Planning at the UvA. Especially the post on the page “Amsterdam durft the vragen” reached many. The page has over 32.000 members and since 12 people commented on the post, the post stayed prominently visible on the page for a few hours. This strategy, along with the snowball effect, and late responses to earlier send emails, yielded an extra 260 respondents. 124 of those were people who answered “No, also I do not want to”, 110 of them answered “No, but I would (someday) really like to” and approximately 30 of them answered “yes”. Finally, an event about community living at Pakhuis de Zwijger in Amsterdam was visited. 3 more respondents anonymously filled in the survey at this event on a laptop.
In total 390 people filled in the survey. 120 of them lived in cohousing, 143 of them would (someday) really like to live in cohousing, and 126 had no interest to live in cohousing. The last group is not needed for the analyses, resulting in 263 useful respondents. In the figure below the different strategies are shown, when they were applied and the approximate number of respondents they yielded. Since there was no way to determine how the survey reached each individual respondent, and due to the overlapping of strategies, these numbers are imprecise estimates. For the last three strategies it was not possible to make a sensible guess. Hence these groups are estimated together. The line between the week of the 22nd of April and the 29th of April indicates the change to add “(someday)”.
19
Removing responses
Since there was no supervision on who filled in the survey, the data set was carefully reviewed. One seemingly fake response had been found. The respondent claimed to be 100 years old, be a refugee (yet filled in the survey in Dutch), have a different gender than male or female, have no education and works both 32 hours a week while being unable to work at the same time. The answers were so unlikely that this response was deleted. Another response was deleted because the respondent only filled in the first few questions and was somehow still able to submit the survey.
Constructing the variables
After cleaning up the data, the variables were constructed in a way that they could be used in the analyses. In table 1 all the variables that were used in the analyses are seen. The second column shows their corresponding survey question. See appendix 2 for the questions and answers. In the next column the scale of the variable is shown. The classification in dichotomous (D), ordinal (O) or ratio (R) is based on Bryman (2012, p.335-336). V3 and V4 have two variables because they needed a different scale depending on the analyses. “Frequency of neighbourhood contacts” will from now on be referred to as “Contact” for the sake of compactness. In the final column a summary about how the variables were constructed is shown.
Variable Survey
Question Code Scale Groups
Cohousing 3 V1 D “Lives in cohousing” and “Wants to live in cohousing”
Contact 16
V2a O The 8 answers ranging from “Daily” to “Rarely” V2b D “Frequent” (more than once a week) and
“Not frequent” (less than once a week)
Social
Loneliness 17
V3a O A 0 – 5 score according to the manual of the loneliness scale (LonSocScr in figure 6)
V3b D “Socially lonely” and “Not socially lonely” if the score is 3 or higher (LonSocDi in figure 6)
Age 1 V4 R The raw data
Gender 2 V5 D “Male” and “Female” excluding “Other”
Education 9 V6 D
“High” (HAVO/VWO, HBO and WO) “Not high” (the other education levels) Living alone 8 V7 D “Living alone” and “Not living alone”
Origin 18 V8 D ”Dutch or western immigrant” and “Not-wester immigrant” (CBS, 2016)
Occupation 10 and 12 V9 D “Occupation, student or volunteer” and “No occupation”
Disability 15 V10 D “Limited by health” and “Not limited by health” Table 1. Variables
21 The first variable, “Cohousing” (V1), was constructed by recoding the answers from the survey into a new variable. The answer “No, but I would (someday) like to” was constructed into the shorter label “wants to live in cohousing”. “Yes” was constructed into “lives in cohousing”. The respondents that had no interest to live in cohousing were not included into the new variable since they were not needed for the analyses. Another variable “Sample” was also constructed with one group with all those living in or interested in cohousing and another group with those with no interest. This variable was used to turn off the irrelevant respondents while keeping them in the dataset.
The survey question about the frequency of neighbourhood contacts had 8 multiple choice answers. Besides being used as the ordinal variable V3a, this variable was also recoded into dichotomous variable V3b. Eight answers were recoded into two values, so this variable could be used in logistic regression. Answers were grouped in such a way that approximately half of the respondents were in each group. More importantly the “more than once a week” divide between “frequent” and “not frequent” was chosen because it is common in the Netherlands to have weekly contact with neighbours, but uncommon to have contact with neighbours more than once a week (Kloosterman & Houwen, 2013). In the table below it is shown which answers belonged to the categories: “Frequent” or “Not frequent”.
Constructing dichotomous contact variable V3b
Approximately, how often do you have contact with your neighbors?
Answer Daily More than three times a week Two or three times a week Once per week Twice per month Once a month Less than once a month Rarely or never Count 70 39 37 37 27 11 11 31 % 26.60% 14.80% 14.10% 14.10% 10.30% 4.20% 4.20% 11.80%
Group “Frequent” “Not Frequent”
% 55.5% 46.5%
Table 2. Constructing the variable “Contact”
The ordinal (V3a) and dichotomous (V3b) variables for Social Loneliness were constructed in multiple steps. In addition to these two variables, variables for emotional loneliness and loneliness in general were also constructed, so they can possibly be used in future research. Social loneliness, emotional loneliness and loneliness in general all have an ordinal variable and a categorical variable. For emotional and social loneliness that the dichotomous variables have the groups “Not lonely” or “Lonely”. For loneliness in general the categories are “Not lonely”, “Moderately lonely” and “Severely lonely”. These variables were constructed based on the manual for the dJG-scale (Gierveld & Tilburg, 2017). The process is summarized in figure 6. The data from the survey was initially divided over 11 variables that each had 4 values: 1 “yes”, 2 “more or less” and 3 “no” and -1 for the missing values. These eleven variables had to be constructed into 6 useful variables. In the analyses in this thesis the dichotomous (LonSocDi) and ordinal (LonSocScr) are used. In figure 6 these are highlighted by the dotted line along the box.
22 Figure 6. Constructing the loneliness variables.
The variables “Age” (V4), “Gender” (V5), “Education” (V6), “Living alone” (V7), “Origin” (V8), “Occupation” (V9) and “Disability” (V10) were all constructed as stated in table 1. Besides recoding there were no SPSS processes that require more elaborate explanation. However, in order to make the variables V5-V10 dichotomous, some choices were made. For “Gender”, those who do not identify as “male” or “female” but as “other”, were excluded since this group cannot be assigned to either the male or the female group. To make “Education” dichotomous, it was not possible to use the usual categorization (standaardonderwijsindeling) from the Dutch government, since this contains three groups instead of two (CBS, 2018). Hence the choice was made to include HAVO and VWO students in the category “high”, since it is likely that these persons to end up in high education. The variable “Occupation” divides the respondents in two groups. The first group contains all those who either have a job, follow education and/or is a volunteer. The other group is none of the above. Finally, “Disability” has one group that is “limited by health” and one that is “not limited by health”. Limited by health means that for the last 6 months or longer the respondent was limited by health problems in activities that other people normally do.
23
Methods of analyses
This section discusses the methods that were used for each sub question. For all analyses an α of 0.05 was used to determine if the null hypothesis was rejected and the alternative hypothesis could be accepted. Before the sub questions were answered, the response was analysed. The sample (N = 263) was compared to the people that filled in the survey but had no interest in cohousing (N = 126) and with the Dutch average. This comparison was made because it gives some insight whether common characteristics of respondents in the sample can be explained due to the fact that these correlate with them (wanting to) live in cohousing or due the sampling. In this response analyses characteristics were used that are comparable but mostly different from the variables that were used in the actual analyses. This is because information of the Dutch average was not available that matches the variables which are used for the analyses. After the response analyses the sub questions were answered.
The first sub question (Q1): “To what extent do cohousing residents have more frequent contact with
neighbours?” has been answered by using Gamma and a bivariate contingency table. Gamma was
chosen because of the ordinal scale and because there are many tied observations. The outcome of the test is a positive or negative score between 0 and 1 or -1. (Bryman, 2012, p.344). The further the outcome is from 0, the stronger the relationship. Whether the outcome is positive or negative determines the direction of the relationship. The test was done by using the Crosstabs function in SPSS (Bryman & Cramer, 2011). In this analysis the dichotomous “Cohousing” (V1) variable was used and the ordinal “Contact” (V2a) variable with 8 categories. If the outcome of the test was not significant, it showed that with the certainty of 5%, it could be stated that in this population cohousing does not correlate with neighbourhood contacts. (Pallant, 2010, p. 171-180).
The second sub question (Q2): “To what extent are people with frequent neighbourhood contacts
likely to be less lonely?” was analysed in a similar way. A bivariate contingency table and a Gamma
test were used to determine if there is indeed a correlation. For this, the ordinal “Contact” variable with 8 categories was used and the ordinal variable “Social Loneliness” with 5 categories (LonSocScr).
Then the third sub question (Q3): “To what extent is neighbourhood contacts a mediating variable
between cohousing and social loneliness?” was analysed by using a three-dimensional contingency
table. This was done as described in Quantitative Data Analysis with IBM SPSS by Bryman & Cramer (2011, p.283). The Cramer’s V test was used to determine if there is a correlation between “Cohousing” and “Social Loneliness” while controlling for “Contact”. Cramer’s V is used when both variables are nominal or both are dichotomous (Bryman, 2012, p. 340). The outcome of the test is a value between 0 and positive 1. The closer to one, the stronger the outcome. If there was a correlation in the total table but not within the “Contact” groups, “Contact” was either a mediator or spurious variable. Since it was not statistically possible to tell the difference, it comes down to logic and previous research to determine the difference (Bryman & Cramer, 2011).
The final sub question (Q4): “While controlling for age, gender, education, living alone, origin,
occupation and disability, to what extent are cohousing residents less likely to be socially lonely?”
was answered in multiple steps. Logistic regression works better if there are less variables in the regression (Field, 2009). Thus, only the variables that matter are included. The first was to determine if the conceptual model is right by checking for correlation of V4-V10 with V1 and V3. For this, the appropriate tests were used which was Cramer’s V in most cases, since the variables are mostly dichotomous (Bryman, 2012). Only V4 (Age) used the Spearman’s Rho test since the ratio scale in combination with a dichotomous dependent variable, allowed for it (Field, 2009). All the variables
24 that had no correlation with social loneliness (V3) were excluded from the regression. If a variable had a correlation with cohousing but not with social loneliness, this variable was not included in the regression. This is because this variable has no relationship with the dependent variable (social loneliness), but with an independent variable (cohousing). To improve the model by reducing multicollinearity, this variable will be excluded in the regression in favour of cohousing. Multicollinearity is when independent variables have a correlation with each other (Bryman & Cramer, 2011). The next step was to check for sample size, to make sure some groups were not too small. For this the crosstabs function was used. If cells were filled below 5, the variable was not included in the logistic regression analysis. Finally, a full check for multicollinearity was done. A regression model works less precise if included predictors have a strong correlation with each other. If this is the case, variables need to be taken out of the model until the multicollinearity is at acceptable levels. To check this the procedure as described by Pallant (2010, chapter 13) was followed. Multicollinearity is also the reason V2 was left out of the regression, since neighbourhood contacts and cohousing have a strong correlation. Finally, with the variables that are left, the logistic regression was performed that determined if cohousing residents are less likely to be socially lonely while controlling for the other variables. The Forced Entry Method was used because other techniques are disputed and because the previous steps already filtered the useless variables (Pallant, 2010; Field, 2018). If the outcome of the predictor “Cohousing” was not significant it showed with the certainty of 5% that it can be stated that in this population cohousing does not influence social loneliness and that the relationship is caused by at least one spurious variable.
If all null hypotheses were rejected it is more likely that cohousing influences social loneliness through neighbourhood contacts. However, a causal relationship could not be proven. What could be answered is the main question: to what extent are adults that live in cohousing in the Netherlands less often socially lonely?
25
Results
Overview of the respondents
In table 3a, 3b and 3c the characteristics of the respondents are shown along with those that filled in the survey without an interest in cohousing. The cells that are bold show a large difference from the other columns. On the right the Dutch average is shown for each characteristic. Possible explanations are given for the differences, these are however suggestions, and are by no means statistically tested. Since this is the first dataset on Dutch cohousing residents, it is hard to tell if certain traits are caused by the type of person that is interested in cohousing or by the methods of reaching respondents. The characteristics in this table are similar to the groups of the variables used in the analyses. However, they differ most of the time in classification and/or content to match available data of the Dutch average. For the variable “Occupation” it was not possible to find National data that could be combined into two groups “working/being a student/volunteering” and “not working/being a student/volunteering”. Hence, the characteristic if someone follows an education is used in (table 3b) since that data was possible to find and combine. The same accounts for “Disability” that has been replaced by “Subjective health”. The characteristics are grouped per table and shown in an order that the data is most easily discussed.
In table 3a, gender and origin alone are shown. A surprisingly large percentage (74,9%) of the sample is female. This could be explained by the fact that simply more women live in cohousing and thus more female respondents end up in the sample. A study done in Sweden and Denmark showed that women are overrepresented in cohousing in those countries (Choi, 2004). It is thus possible that more women than men (want to) live in Dutch cohousing communities. However, the other 126 people that filled in the survey are 76,2% women. An even higher percentage than those who (want to) live in cohousing. It is thus also possible that women were more willingly to participate in this research and more women than men who received the link to the online survey, actually filled it in.
People with a western or Dutch origin are, like women, overrepresented among the respondents. The difference between the sample and the Dutch average can likely partly be explained by the way of sampling but probably also in part by the characteristics of those living in cohousing. The same study that showed women are more likely to live in cohousing also showed that people with a local origin were more likely to live in cohousing (Choi, 2004).
Overview of the respondents, table a Characteristics of the respondents Sample N No interest
in cohousing N NL average Source Gender Man 23,6% 62 23,8% 30 49,6% (CBS StatLine, 2018) Woman 74,9% 197 76,2% 96 50,4% Other 1,5% 4 0% 0 - Total 100,0% 263 100,0% 126 100% Origin
1st and 2nd gen.
non-western 3,9% 10 6,0% 7 12%
(CBS, 2016) Western or Dutch 96,1% 246 94,0% 110 88%
Total 100% 256 100% 117 100%
26 The difference in age between the three columns can probably be explained for the most part by the strategy to reach respondents. Since the survey was distributed online by email and social media it did not reach anyone who does not use these communication methods. These people are mainly elderly (Vee, et al., 2018). Also, the survey was filled in by people in the personal network of the researcher. These are mainly students. The fact that the survey reached mostly young people becomes clear when looking at the middle column, that shows 67,2% of that group were between 18 and 27 years old. It is thus complicated to say anything about common age groups in Dutch cohousing projects with this data.
Like the age characteristic, the large percentage of people who follow an education in the sample probably can be explained by the way the survey was distributed. This is indicated by the even larger percentage of students among those who filled in the survey without interest in cohousing. Again, these are all presumptions.
As for the level of education, among the respondents both interested in cohousing and not, low educated are underrepresented in favour of middle educated respondents if compared to the Dutch average. This could be explained by the number of friends and acquaintances of the researcher that filled in the survey who are middle (mostly not yet highly) educated. Also, it is possible that low educated people that saw the survey on Facebook were less motivated to participate in this study. Low educated people tend to have a low response rate for survey research (Riele, 2002). Because there is almost no difference between the sample and the second column it is hard to say anything about the characteristics of those who (want to) live in cohousing.
Overview of the respondents, table b Characteristics of the
respondents Sample N
No interest
in cohousing N NL average Source
Age Young (18-27) 39,5% 96 67,2% 80 15.70% (CBS, 2018) Middle (28-49) 26,7% 65 16,0% 19 34.60% Old (>50) 33,7% 82 16,8% 20 49.70% Total 100,0% 243 100,0% 119 100% Following an education Yes 26,6% 70 46,8% 59 8,5% (CBS Statline, 2018) No 73,4% 193 53,2% 67 91,5% Total 100% 263 100% 126 100% Education level Low 4,9% 12 4,8% 6 31% (CBS, 2013) Middle 63,9% 168 69,0% 87 40% High 31,2% 82 26,2% 33 29% Total 100,0% 262 100,0% 126 100%
