What is relational mobility?

Relational mobility is a socio-ecological factor that represents how much freedom and opportunity a society or social context affords individuals to choose and dispose of interpersonal relationships based on personal preference (Thomson et al., 2018).

In low relational mobility environments, people tend to stay in long-standing relationships and groups, and it is hard to change them if they want to. Interpersonal relationships (friends and acquaintances etc.) are generally defined by existing social network structures (like hierarchies and histories of social groups, and work, school, and community groups in more recent times). Traditionally, human societies – such as small-scale tribal societies – tended to be low in relational mobility characterized by relatively ‘closed’ interpersonal networks and stable group memberships.

In high relational mobility environments, on the contrary, opportunity and freedom abounds to select friendships based on personal preference. In cities (vs. villages) and in some areas of the online world, for example, relationships are more likely to be formed through personal choice than external constraints (Adams & Plaut, 2003; Schug et al., 2009).

Where are high and low relational mobility environments typically found?

If we focus just on society-level relational mobility, it is high in North America, Western Europe, and Latin America. Relational mobility is low in East Asia, the Middle East, North Africa, and South Asia (Thomson et al., 2018). Relational mobility in the Thomson et al. (2018) study was highest in Mexico and Puerto Rico, and lowest in Japan and Malaysia. You can explore relational mobility around the world, visually, here: http://relationalmobility.org/#rmob-country-scores

The above findings from Thomson et al. (2018) replicate, in part, previous dual-country studies’ findings that North America has high relational mobility and East Asia has low relational mobility (e.g., Schug et al., 2009). Adams noted that certain societies in West Africa were also lower in relational mobility than North American society (Adams & Plaut, 2003).

Relational mobility is not limited to society-level comparisons. It is a contextual variable that also applied to within-nation regions (Yuki, Sato, Takemura, & Oishi, 2013), and even temporal categories (like first year college vs. second year college; Sato & Yuki, 2014).

What other theories are relevant to relational mobility?

Other theories that refer to the openness (or “closedness”) and fluidity of interpersonal relationships within a society span many areas of human and social science. The sociologist Tönnies in 1887 spoke about Gemeinschaft (community) and Gesellschaft (society), and how these different societal structures affected humans. Anthropologists (e.g., Henrich et al., 2005) show how community’s involvement in the market economy affects people’s sense of fairness. Evolutionary psychologists (e.g., Barclay, 2016) speak of the degree
of partner choice in biological markets. Social psychologists have argued that incentive structures–external to the self and determined by how easy it is to move between relationships–impact interpersonal trust (Yamagishi, 1998; 2011).

Relational mobility research employs an 'adaptionist' perspective. What does this mean?

Relational mobility acknowledges human being as a species that seeks to adapt to, or respond to incentives present in their immediate environment (see discussion about perceptions vs. reality below). This is nothing new in social science, of course. John Berry (1975; 2014) first proposed the idea in the 1970’s with his eco-cultural approach. Since then this idea has developed and evolved (see Georgas et al., 2004 and Oishi & Graham, 2010). Drawing on these concepts, we view relational mobility as being predominantly adaptive processes. With an adaptionist perspective like this, models of human behavior and psychology can be more parsimonious.

This idea has parallels with behavioral ecology in the natural sciences. In nature, plants and animals respond to their natural environment (their ecology) in order to flourish. “In a similar way, the the socio-ecological approach to variation in human behavior states that in order to flourish, humans adapt their behavior (both consciously and unconsciously) according to their surrounding physical and social environment” (Thomson & Yuki, 2015).

Conceptually, relational mobility is one characteristic of people’s social reality that exists outside of people’s minds. Behavior and psychological tendencies emerge as adaptations to this external social reality.

Looking to the future, more theoretical effort is needed to further refine relational mobility theory. Mathematical formulation, for example, should help us explicate the mutual influence between macro-level relational mobility and psychological tendencies of individuals who live in those social environments. Rigorous empirical tests will be required to test such theories, at the same time as current relational mobility theory is tested in a wider range of populations.

How do you measure relational mobility?

Relational mobility is measured using the relational mobility scale, a 12-item self-report Likert scale (Thomson et al., 2018). The scale asks respondents to evaluate the degree to which people around them have opportunities freedom to meet new people and choose their relationships based on personal preference. See the scale here: http://relationalmobility.org/the-relational-mobility-scale

Is relational mobility about perceptions or reality?

As a concept, relational mobility is concerned with objective social realities. Currently, however, to measure relational mobility we depend on respondents’ perceptions of the mobility of people around them, measured using the self-report relational mobility scale (see the scale here).

Just to reiterate, relational mobility is “a socioecological variable that represents how much freedom and opportunity a society affords individuals to choose and dispose of interpersonal relationships based on personal preference” (Thomson et al., 2018). In this sense, relational mobility is an objective social reality ‘out there’.

To fully flesh out the distinction between relational mobility as a reality versus perception, we need to talk about a) relational mobility as a concept and b) the measurement of relational mobility.

First, relational mobility as a concept does not assume or require people’s perception of it. There’s no need for someone to perceive or acknowledge how much relational mobility there is in their surrounding environment, in order for relational mobility to affect their behavior. Nor does relational mobility necessarily directly affect behavioral and psychological tendencies.

Kito, Yuki, & Thomson (2017) give the example of maintaining harmonious relationships in low relational mobility societies. They argue this is an important adaptive task in such societies, because relationships are long lasting and difficult to change – better not to damage relationships and “be trapped within uncomfortable, disharmonious relationships for a long time with few opportunities to seek alternative relationships, or (b) be rejected by a relationship partner in an environment where relatively few alternatives are available” (p. 119).

We argue that the reason why people do not need to perceive relational mobility in order for it to affect their behavior, is because high and low mobility societies have different incentive structures. In low mobility societies, behaviors that disrupt harmony, for example, are punished, and in high mobility societies behaviors that demonstrate one’s strengths and ‘social value’ are rewarded (Kito, Yuki, & Thomson, 2017).

Individuals who grow up under these respective incentive structures learn and internalize ‘default’ adaptive behaviors through reinforcement learning or trial and error, via interaction with parents, peers, and through education processes etc. They acquire heuristics (behavioral and psychological ‘shortcuts’ or rules) that allow them to quickly choose the most adaptive course of action in any given situation (Yamagishi, Hashimoto, & Schug, 2008; Yamagishi, 19982011; Yamagishi, 2013).

In this way, it is conceivable that relational mobility indirectly affects behavior via mindsets, values, and norms that are adaptive depending on how relationally mobile a society or social context is. This does not require a person’s perception of relational mobility. It only assumes that the relational mobility of their society or social context has caused people to intuitively react or ‘feel’ in particular ways, appropriate to the situation.

Secondly, let’s talk about the measurement of relational mobility. Measuring relational mobility is not easy, and the current tools we have are not perfect. The best that has been proposed so far is Yuki et al.’s (2007) relational mobility scale. At the measurement level, this scale taps into respondents’ perceptions of others around them.

While the relational mobility scale itself is measuring perceptions, however, Thomson et al. (2018) argue that those perceptions represent reality (i.e., the potential for movement between relationships) with some degree of accuracy. One reason to think that these perceptions are accurate is the fact that society-level relational mobility scores correlate with measures of actual movement (see Thomson et al., 2018, Table S6)

To measure relational mobility, why not use objective measures of actual relational movement, rather than perceptions of relational mobility?

We don’t use measures of actual relational movement (such as residential mobility; see Oishi & Talhelm, 2012), because for certain interpersonal behaviors and psychological tendencies, they don’t predict those tendencies as well as the potential for movement between relationships (i.e., relational mobility) (Thomson et al., 2018).

Think of a naval officer who could be ordered to transfer to another city (or country) at any time. If you were friends with that naval officer, the knowledge that your friend is about to be transferred to another city (quite possibly against their own preference) is not likely to increase your likelihood to engage in behaviors to try to retain that friend (e.g., by increasing intimacy, disclosing more). If your friend’s moving away is determined by an outside force, then any attempt to persuade the person to stay with you as a friend is for naught; it is maladaptive.

If, on the other hand, it is entirely up to your friend’s choice whether they move away or not, then why not try to convince them to stay, by engaging in relationship-retention behaviors? That would be the adaptive choice (Thomson et al., 2018).

Of course, this is a proposition that can be tested. Thomson et al. (2018) tested this by replacing relational mobility with variables that measure actual relational movement (such as residential mobility and the number of new acquaintances met in the last month), and found that relational mobility was a stronger and more reliable predictor of relational investment behaviors (see Thomson et al., 2018, Table S13).

Is the relational mobility scale reliable when used across different societies and samples?

Yes, the tool for measuring relational mobility–the relational mobility scale–is remarkably robust across cultures and societies. It has been used in a great number of cross-national studies, showing good reliability and relatively stable factor structures. In the largest multi-national study to date (Thomson et al., 2018), the scale proved to be mostly invariant in its measurement properties across 39 countries.

Furthermore, even though the scale measures people’s perceptions of the relational mobility of people around them, those perceptions seem to be accurate. Society-level relational mobility scores correlate well with other variables that indicate relative ease of movement between relationships such as the number of acquaintances met in the last month, justifiability of divorce, and residential mobility (Thomson et al., 2018).

Is relational mobility the same as individualism or independent self-construal?

No, we argue that relational mobility exists in a different domain to shared cultural mindsets and ways of viewing the self.

Thomson et al. (2018) argue that cultural self-construals–a characteristic of a human’s mind, albiet shared with others in one’s society–could, in fact, plausibly result from relational mobility–a characteristic of humans’ social ecology. In this sense, relational mobility can be seen as one possible mediator of the effect of distal historical or ecological factors (such as subsistence style and threat) on cultural syndromes such as interdependence and independence (see Talhelm et al., 2014). Here’s the theoretical causal chain: In response to threats and styles of subsistence, people form societies low or high in relational mobility, and this relational mobility is one factor that influences self-construal.

Of course, this is an empirical question. Thus, Thomson et al. (2018) ran analyses to test this model in their 39-society dataset (threats/subsistence style→relational mobility→independence/interdependence). As a measure of interdependent self-construal, they used Vignoles et al.’s (2016) self-construal variables that are conceptually related to independence/interdependence. In short, relational mobility does seem to act as a mediator for distal social ecologies’ effect on self-construals (see Thomson et al., 2018, Table S15).

Thomson et al. (2018) Figure S8 - A model representing the indirect effect of subsistence style and threat on dimensions of interdependent and independent cultural self-construals via relational mobility.

Granted, there is very likely a feedback loop at play. For example, where relational mobility is low, interdependent self-construals are prevalent. At the same time, to the extent that an individual develops an interdependent self-construal, the more unlikely they will be to venture outside of existing relationships (because they see the self as intrinsically connected with others), thus perpetuating a state of low relational mobility.

In the PNAS study, did you use Facebook to recruit participants?

Yes, in the Thomson et al. (2018) study, we paid Facebook to display ads to Facebook users in 46 countries, encouraging Facebook users to take part in our World Relationships Survey. You can see a sample of the ads on the World Relationships Survey Behind the Scenes page. Clicking on an ad in Facebook directed participants to a survey hosted on an online survey software Qualtrics. Participants did not provide any access to their Facebook data (nor were they required to).

As an incentive to take part, participants were shown a ‘report page’ at the end of the survey. This report page compared their responses to the aggregated responses of others who had taken the survey in their language: average number of past romantic partners, average level of self-disclosure to a romantic partner or friend, average level of intimacy and similarity with one’s friend etc. There are some report page examples on the Behind the Scenes page.

There are a couple of advantages to doing recruitment this way. The first advantage is access to broader sample characteristics. So long as a person had a smartphone (or a computer) and a Facebook account, they could choose to take part in the survey. Many cross-cultural studies use university students as their participants. However, in some countries university students are markedly more wealthy than the general population, so using Facebook ads allowed us exposure to a much more diverse set of respondents.

A second advantage is that this is a very cost-effective method of recruiting. It is much cheaper than using commercial survey companies’ services.

The distinct disadvantage of this recruiting method is selection bias. Even though the Facebook ad stats told us the ads were shown to a roughly 50/50 split of males and females, over 80% of our final sample was female. The content of the survey was clearly more interesting to females, which means our sample is quite biased. As far as we could tell, there wasn’t any difference between men and women in relational mobility (see Thomson et al., 2018, Supplementary Information Section 1.1.1), but this is something that needs to be addressed in future studies.

In the PNAS study, how effective were the Facebook ads for recruitment?

Effectiveness is difficult to define, but here’s some stats that should give some context.

When recruiting participants for the study published in the PNAS paper, our Facebook ads (see examples here) had on average a 2.8% (SD = 1.4%) click through rate. That is, of all the people who were shown an ad, 2.8% of people chose to click on the ad. Of those 2.8% that clicked on the ad, on average about 16% of them completed the survey (valid responses; i.e., cases where we could use the data for analyses).

Overall, we paid about 1,500,000yen in total in Facebook advertising fees. Taking into account that we ended up with just over 16,000 valid responses, that works out to be around 95yen per valid response.

Of course, there was a very big difference between countries when it came to cost per click, click through rates, and response rates. If you’re looking at these numbers and trying to estimate costs for your own study in one particular country, you’re very unlikely to get an accurate estimate. That’s because you can have a country whose cost-per-click is very low (e.g., Egypt in our case), but due to a very low response rate (e.g., Egypt in our case), the cost per valid response ends up higher than a country with a high cost-per-click (e.g., Japan in our case) but high response rate (e.g., Japan in our case).

Here’s an Excel sheet that breaks down all these things by country for our study. NOTE: This sheet was prepared in 2016 based on preliminary data. Country N’s may differ from the PNAS paper. It also doesn’t include South Korea data. Almost 6 years has passed since we did the first pilot using Facebook ads – since then costs may have increased. Also really important to note is that we used a fairly strong intrinsic incentive to encourage people to take part – an informal ‘analysis’ of their friendships and romantic relationships based on responses they provided in the survey (see the Behind the Scenes for more details). While we’re not experts on online survey motivation, we feel that graphic design, incentives, survey length, survey content, as well a myriad of other things will affect how well recruitment will work using Facebook ads. Take our experiences with a grain of salt 🙂

Can I download your PNAS analysis syntax from somewhere?

Yes. We have uploaded all the data and syntax related to those analyses on the project’s Open Science Framework page here: https://osf.io/qfbjc/.

You may want to view that data and syntax along with Section 1.2.3 of the Thomson et al. (2018) Supplementary Information.

In the Thomson et al. (2018) study, we used Mplus software for the relational mobility scale measurement invariance testing, as well as the multilevel structural equation modelling.

Can I download your PNAS analysis syntax from somewhere?

Yes. We have uploaded all the data and syntax related to those analyses on the project’s Open Science Framework page here: https://osf.io/qfbjc/.

You may want to view that data and syntax along with Section 1.2.3 of the Thomson et al. (2018) Supplementary Information.

In the Thomson et al. (2018) study, we used Mplus software for the relational mobility scale measurement invariance testing, as well as the multilevel structural equation modelling.