Introduction

Attachment styles refer to the ways in which people think, feel, and behave in some of their most important relationships, such as those with parents and romantic partners. (See my previous blog post for a general overview of attachment styles.)

One of the enduring questions about attachment concerns the way attachment styles change over time–especially as people transition from early adulthood to later adulthood. This post summarizes what we have learned thus far about age-based differences in attachment using data from the yourPersonality Attachment Project. Specifically, I examine the ways in which people’s scores on the two fundamental attachment dimensions–anxiety and avoidance–vary across people of different ages, ranging from 18 to 60.

Because the overall sample size in this project is relatively large (i.e., 70417 people have completed the survey at least once), I illustrate the average/mean attachment score for people at each age (e.g., age 18, 19, . . ., 60; represented by the colored squares. The colors, ranging from deep blue to yellow, represent the differnet ages, which are also plotted on the x-axis. The coloring isn’t necessary; I just like colors. I also summarize the age-based linear trends (black lines) as a statistical form of connecting the dots.

Finally, I also examine the question of whether age-based trends are similar to developmental ones using longitudinal data from 2813 people who have completed the attachment survey multiple times.

Global Attachment Across Age

The graphs below illustrate the ways in which attachment-related anxiety and avoidance vary across people of different age groups. I focus first on global attachment: The ways in which people generally think about close relationships.

Older adults tended to be more secure in the way they generally think about relationships compared to younger adults. Older adults tended to have lower attachment anxiety scores, indicating that they were less worried that others may reject or abandon them. Older adults also had lower attachment avoidance scores, indicating that they were more comfortable than younger adults opening up to others and depending on them. The age differences for avoidance, however, are much less dramatic than those for anxiety. In fact, most of the age differences seem to be concentrated in early adulthood.

In summary, when considering the way people relate to others in general, older people in our sample tended to be more secure (i.e., less anxious, less avoidant) than younger people.

Attachment to Parents Across Age

The graphs below show the average level of attachment-anxiety and attachment-avoidance people report in their relationships with their parents. Each data point represents the average score for people at that specific age. Please note that the graphs for attachment to mothers are based only on people who reported that their mothers were alive at the time of assessment. Similarly, the graphs for attachment to fathers are based on data from people who reported that their fathers were alive at the time of assessment.

These graphs demonstrate that older people were more insecure than younger people in their relationships with their parents. Why might this be the case? It is possible that, as people’s parents age, people begin to worry about their parents’ health and, ultimately, their availability. This could lead people to experience some attachment-related anxiety concerning their parents. People also are more likely to live further from their parents, potentially contributing to greater insecurity with their parents.

Ultimately, these kinds of explanations are developmental. That is, they assume that, as people grow older, their relationships begin to change. However, it is difficult to unpack the meaning of these patterns using cross-sectional age data alone: Data in which age differences can reflect both differences due to development/aging and differences in the cohort or generation in which people were raised. I’ll attempt to separate cohort-based differences from developmental ones in the second-half of this post.

Attachment with Romantic Partners

The next set of graphs show the average level of attachment-anxiety and attachment-avoidance that people report in their relationships with their romantic partners (i.e., boyfriends, girlfriends, spouses). Each data point represents the average score for people at a specific age. These graphs are based on the 42189 people who reported being in a romantic relationship at the time of the assessment.

Notice that these patterns are both similar to and different from the ones we saw previously with parental attachment. Compared to younger people, older people tended to be less anxious in their romantic relationships. That is, they were less worried that their partners may abandon them or be unavailable to them. However, older people were more avoidant in their romantic relationships compared to younger people. They were less comfortable opening up to and depending on their partners.

It is possible that older people were less anxious in their romantic relationships than younger people because their relationships have lasted longer. That is, someone in their 50s has likely been with their spouse longer than someone in their 30s. Thus, older people may have more evidence that their partner will not abandon them and, as a result, are less concerned about such issues.

One way to examine this possibility is to graph attachment-related anxiety as a function of relationship length (i.e., how long people have been with their partners) while holding age constant. There are fun ways to do this with statistics, but, for simplicity, I’ve plotted below attachment anxiety as a function of relationship length among people who are all of the same age: 30 years old. Thus, in this subset of people, adults do not vary in age, but they do vary in how long their relationships have lasted.

Notice that those who have been in their relationships longer (e.g., 10 years vs. 2 years) are less anxious and avoidant in their relationships.

In summary, it seems possible that one reason older adults are less anxious in their romantic relationships than younger adults is that their relationships have lasted longer. When I keep age constant (age = 30) and examine the association between relationship length and attachment insecurity, people who have been in relationships longer are more secure in their attachments than those who have been in relationships for shorter periods of time.

Separating Age and Time

The previous graphs were designed to illustrate the way attachment styles vary across people of different ages. In psychological research, such age effects can be the result of at least two broad classes of factors.

First, they can result from cohort effects. Cohort effects refer to differences between generations. People who were born in the 1960s, for example, have had different experiences than those born in the 2000s. If those differences are relevant for understanding insecure attachment, then we could find that older people are more insecure in their relationships with their parents than younger people are.

Second, age effects can result from developmental effects. Developmental or aging effects refer to differences due to time: The way life unfolds across people in the same cohort.

Although cohort and developmental effects can sometimes parallel one another, that is not always the case. For example, although people become more wise over time (i.e., a developmental/aging effect), it is not necessarily the case that people born in older cohorts are wiser than those born in younger cohorts. In this case, older adults will be wiser than younger adults, but that is a result of developmental and not cohort/generational differences. Another example: Younger adults are more tech savvy than older adults. This is probably not due to developmental processes; it seems unlikely that, as people get older, they become technologically inept. The difference is probably due to cohort effects: People born more recently are more immersed in technology than people born 50 years ago. In this case, age-based differences reflect cohort effects and not developmental ones.

It is possible to separate cohort and developmental effects in the yourPersonality Project because, not only do participants vary in their ages (i.e., some people are older than others), but some participants have completed the survey multiple times. Within that set of people, it is possible to examine whether they tend to become more or less secure over time above and beyond cohort differences.

The next set of graphs illustrate time-based differences in a subsample of 2813 people who completed the survey 3 or more times. The developmental trends are represented by the short, colored lines. Specifically, I plot people’s average attachment score at a specific age (e.g., 20) and their average trajectory over a 6-month period of time. This process is repeated for people of different ages, creating plots that look a bit like a comb. If the lines are going down, that implies that people are becoming more secure over time. If the lines are going up, that implies that people are becoming more insecure over time. The cohort trends are, as before, represented by the solid, gray line. That line shows the age- or cohort-based differences in attachment. These lines are similar, but idential to, the ones illustrated before because these analyses come from a subsample of the originally plotted data.

Global Attachment

As shown before, when it comes to global attachment, older people tended to be more secure (i.e., less anxious and avoidant with respect to attachment concerns) than younger people. In the graphs below we can see this pattern again in the longitudinal subsample, but we can also see in the colored lines that, for people of any given age, people tend to become less insecure over time. In fact, these changes are fairly large for attachment-anxiety: The typical person reports being much less anxious after having completed the survey at least once.

In summary, with global attachment, people tend to become more secure over time (as shown by the short, colored lines). Moreover, older people, compared to younger people, tend to be more secure (as shown by the gray lines). The cohort-based and developmental-based trends are in the same direction.

Attachment to Parents

As we saw before, older people, compared to younger people, are more insecure in their relationships with their parents (gray lines). If this age difference reflects a developmental effect, then we should see that, over time, people become more insecure in their parental relationships. However, we see the opposite (colored lines): Over repeated assessments, the same people become less insecure in their parental relationships.

This kind of finding is sometimes referred to as Simpson’s Paradox–a reference to how statistical patterns can shift directions when examined at different levels of analysis (e.g., across cohorts vs. across time). It suggests that we need different kinds of explanations to understand the way cohort differences and developmental processes may impact attachment.

What could explain the developmental/longitudinal pattern? One possibility is that, on average, people become more secure in their parental relationships for the traditional reasons given by attachment theorists. Namely, the continued and uncompromised nature of the bond leads people to feel safe and secure in those relationships.

What could explain the cohort differences? One possibility is that, compared to younger people, older people in our sample experienced more formal or less emotionally warm relationships with their parents. Physical punishment (e.g., spanking), for example, was much more common among children growing up in the 1970s than it is among children growing up now. Thus, one hypothesis–one that needs to be evaluated more carefully in future work–is that older people, compared to younger people, experienced less warm and supportive relationships with their parents. And, as a result, older people are more insecure in their parental relationships than younger people are.

Attachment to Partners

What about romantic relationships? The graphs below illustrate the cohort-based and developmental-based trends for attachment to romantic partners in a subsample of people who were involved in romantic relationships.

Recall that we saw previously that older people are less anxious in their romantic relationships than younger people. The time-based trends are consistent with this too: The people who completed multiple attachment assessments over time became less anxious over time regardless of their ages. (In fact, these drops in attachment anxiety are fairly sizable.)

We do, however, see a manifestation of Simpson’s Paradox with respect to attachment avoidance. Namely, although older people are more avoidant than younger people in their romantic relationships, both younger and older people tend to become less avoidant over time. This suggests that the developmental- and cohort-based effects oppose each other. Even though people tend to become less avoidant over time in their romantic relationships, there may be cohort-based or generational reasons why younger people tend to be less avoidant in their relationships than older people overall.

Summary

The extent to which people feel secure or insecure in their close relationships varies across age. But, the yourPersonality Attachment Project reveals that there may be different processes that explain these age differences.

Specifically, the finding that older people tend to be more insecure than younger people in their parental relationships appears to be due to cohort rather than development or aging. That is, people do not become more insecure over time. But people born several decades ago, despite becoming more secure over time, tend to start off more insecure than people born in more recent decades.

OVerall, these data suggest that there is a trend towards increasing security, both in people’s global attachment styles and in their relationship-specific attachment styles. Not only do people tend to become more secure over time, younger generations appear to be more secure than older generations.

Technical notes:

This post was created by R. Chris Fraley at the Department of Psychology at the University of Illinois at Urbana-Champaign. To learn more about attachment theory and research, please visit one of our overviews.

This is an R Markdown document. Markdown is a simple formatting syntax for authoring HTML, PDF, and MS Word documents. For more details on using R Markdown see http://rmarkdown.rstudio.com.

The data used for this post come from the Voluntary Sample from the yourPersonality Attachment Project.

Cite: Fraley, R. C. (2018, June 18). How do Attachment Styles Vary Across Age? Retrieved from https://yourpersonality.net/attachment/blog2.html.