Social Isolation in America

by Kieran Healy on June 23, 2006

Here’s an important new paper by two former colleagues of mine (just departed for Duke) and one of our grad students here at Arizona. The paper compares the social network module of the 2004 General Social Survey (GSS) to the 1985 GSS, the last to include network questions. The key question of interest is this:

From time to time, most people discuss important matters with other people. Looking back over the last six months—who are the people with whom you discussed matters important to you? Just tell me their first names or initials.

The survey went on to probe the respondents about their relationship to the people they mentioned, and the relationship of these people to one another. The new findings are striking: since 1985, the number of people saying there is no-one with whom they discuss important matters nearly tripled, rising to about a quarter of the respondents. As McPherson et al say,

The modal respondent now reports having no confidant; the modal respondent in 1985 had three confidants. Both kin and non-kin confidants were lost in the past two decades, but the greater decrease of non-kin ties leads to more confidant networks centered on spouses and parents, with fewer contacts through voluntary associations and neighborhoods. … Educational heterogeneity of social ties has decrease, racial heterogeneity has increased.

The predicted probability of social isolation is much higher the fewer years of education one has. Also. “Young (ages 18—39), white, educated (high school degree or more) men seem to have lost more discussion partners than other groups.”

The observed differences are pretty big, as these things go. Are they real? It may be that the 2004 respondents differed from the 1985 respondents in their interpretation the words “discuss” and “important.” (People might interpret “discuss” as face-to-face discussion, when they may also be pouring out their hearts on a blog somewhere, for instance.) Because of these issues, the authors spend a lot of time investigating the validity of the measure. More interestingly, it may be that we really are observing a shift in patterns of network affiliation. Feel free to speculate in the comments, but also take a look at the paper—it discusses several of the most plausible interpretations of the shift, in addition to documenting the findings.

{ 1 trackback }

Feminist Law Professors » Blog Archive » Interesting Things To Read Elsewhere
06.26.06 at 6:12 am



gerry 06.23.06 at 6:13 pm

Maybe Putnam as on to something after all?


tina 06.23.06 at 7:12 pm

Such an interesting paper! I wonder, if these changes really do reflect changing network patterns, whether this is also related to the supposedly dramatic increase in people suffering from depression.


eweininger 06.23.06 at 10:02 pm

I’d speculate that their first possible explanation–change in what people consider “important”–plays a role in accounting for the decline in the educational effect, specifically. That is, highly educated respondents are less likely to count whatever it is that they discuss with their “confidants” as “important” now (2004) than they were in ’85.


lemuel pitkin 06.24.06 at 12:21 am

How about the theory that links atrophying social networks to increasing income inequality?


bad Jim 06.24.06 at 5:08 am

I talk to my mother, her sister, my sister, my brothers, my uncle, and his, her and their kids and their spouses and their kids.

It’s hard to have political conversations with the elders, resolute Roosevelt Democrats all, when the only permissible question is, “Which Democrat can win?” Discussions with my siblings and cousins and their offspring differ in detail but share their tenor.

The idea that someone has no one with whom to share an opinion is desolating.

But not quite convincing. The last time I enjoyed a day or two of jury duty, almost all of my fellow prospective jurors testified to having lived stupefyingly uneventful lives. A very few admitted to having purses snatched in Rome. Fewer had ever encountered the police (except for the one girl with the boyfriend who drove an Integra, and the other who’d served time for attemped murder).

Up to a point I can imagine that my fellow suburbanites live such sheltered experiences that they never talk to their neighbors and are never robbed. But then I have to sneeze.


ambi 06.24.06 at 8:37 am

Well, there’s a lot to discuss here.

But first I wanted to ask–are we supposed to discuss this paper, or discuss it?

I mean, that’s an important issue. Unless it’s merely important.

In good earnest, though–this is quite a striking finding. Thanks for bringing it to my attention.


harry b 06.24.06 at 9:52 am

eweiniger — your speculation implies a decline in the self-importance of educated people over the 2 decades, no? Highly unlikely, in my view.


etat 06.24.06 at 1:50 pm

This issue (which I will call insularity) is the subject of discussion at one of the Guardian (UK) blogs, and purports to reflect US social circumstances, i.e., Americans are social dweebs, good thing that isn’t the case in the UK.

Anecdotally, and OTOMH, people in the UK are just as insular, and I’m inclined to agree with Lemuel that it is significantly an economic issue (time pressured/starved middle classes) along with two social components a) that meaningful social dialogue is distanciated and alienated (happens via media, not down the pub/pitch), and b) pressure to behave professionally has meant that more people are less willing to discuss their work, thereby cutting off avenues of meaningful discussion.

If there’s any substance to this characterisation, I would expect insularity to increase among people establishing and maintianing their careers, but to be less pronounced among people in the earlier and later parts of their careers. In other words, I would expect that people in their mid-to-late 20s would not show symptoms of insularity, with women showing less insularity than men.

I’ve skimmmed the article with this in mind without seeing anything that refers to feelings of isolation by age group. But I did spot this telling line on p 373: ‘the increase has been the most dramatic among middle-aged, better-educated, higher-income families’.

One last thought: how does this relate to the perceptions of an increasing ideological divide?


eweininger 06.24.06 at 1:57 pm

Harry b—I guess there’s sort of an implicit self-flagellation amongst the educated in it.

But to be more precise, what I’m saying would be the following. Imagine a list of the k things people discuss with their “confidants”:

1) weather
2) ball games
3) curtains
4) whether particular in-laws dislike them
5) which second grade teacher Junior should be enrolled with next year
6) whether they should pursue a new job opportunity
k) the “big events of the day” (the cold war/the “clash of civilizations”—or whatever)

Now imagine that our two samples had exactly the same distribution across this (unmeasured) list—that is, people in ’85 were exactly as likely as their counterparts in ’04 to talk about ball games, curtains, second grade, etc. with their “confidants”.

The key question would then be which of the k items gets tagged as “important” by the respondents. My surmise is that highly educated people in ’04 were less likely to apply the tag to items such as (5) and (6) than were highly educated people in ’85. Or again, things other than (k)—the “big world events”—were less likely to be considered important in ’04 than they were in ’85 by the highly educated. For this reason—or at least partly for this reason—the magnitude of the education effect is reduced in the later sample.

The assumption that has to be made is that at a normative level, “important” has been culturally re-defined in the intervening years to exclude things that are of significance only (or mainly) to the individual and her family.

Anyway, Kieran gave full license to speculate. If I was really going to take advantage of it, I’d make up a story about the more educated members of the ’04 sample having read lots and lots of articles in the middle-brow press about declining social capital and the like.


previously pre 06.24.06 at 3:31 pm

Up to a point I can imagine that my fellow suburbanites live such sheltered experiences that they never talk to their neighbors and are never robbed.

That’s a strange comment to see from my perspective. I can’t imagine maintaining eye contact with any of my neighbors, much less conversing with them for longer than five or six seconds.

That’s not anything against my neighbors, who actually seem like really nice people. I think. But the idea of engaging them in discussion, much less a discussion about a topic genuinely important to me, seems incomprehensible. Elevator rides (I live on the eighth story) feel icy when others ride too. Saying “hello” tends to elicit a very… uncomfortable response, as though the act bordered on personal-privacy invasion.


GJ Neely 06.24.06 at 8:04 pm

This has been a subject of heightened interest to me – occuring mainly in these last two years when I, a 58 year old female, left my last position of employment and find myself edging closer and closer to the subject fate. And so, I further ponder how I got here – and the largest change in my world is that I trust NO ONE. It is my belief that Americans (if you want to single us out) are becoming more and more self-absorbed and lazy. Oh yes, some might work their you-know-whats off but it is, I believe, so that we may continue to amass our material goods. People, in general, are becoming combative and vicious – want to find out – just go to the “beginner” tab in Spades/Pogo and make an errant bid/discard. If we are to save ourselves – we must break of these behaviors and become compassionate – however, personally, I think it’s too late – there is ABSOLUTELY no motivation. Oh, and yes, I have two confidants – my sister and my neighbor – neither of whom I completely trust – and therefore, there probably is no one who really knows me . . . And, about wealth, I have worked around persons of various wealth throughout several years and they experience the same problems us less wealthy people do – it’s just covered up and you’ll never find out if they have their way . . . .


Martin Bento 06.25.06 at 1:53 am

What interests me is: what do we think would be different in our society if this were not so?


I don’t see what the basis for your claim is. You seem to trying to refute the data. People’s definitions of “important” may have changed, but I think such an assertion would require independent evidence, especially as we’re not talking a huge span of time for a basic shift in ideas of what people consider important, and because it seems such a shift, if present, would leave other evidence in its wake. In any case, I think the report claims that people report less discussion of what they feel is important regardless of what those things are, in which case the constituents of “important” do not matter: whatever people consider important, they are not discussing it.


ingrid 06.25.06 at 3:52 am

I don’t have time to read the paper right now, though it looks very interesting and will do so later — thanks for drawing our attention to it. I wondered whether anyone looked at similar questions in the British Household Panel Survey — I know for sure that they have similar questions, and since the BHPS is meant for longitudinal analysis, I’d assume that the quesiton has remained exactly the same for the different waves. The questions are about social interactions (e.g. “how often do you meet your three best friends?”), but also social support (e.g “do you have anyone to talk to if you are distressed?”). Around 2002 I looked at these questions (for one wave only) when I was studying gender inequality within the capability framework, and these social support and social interaction questions were the only dimensions in which men were (slightly) worse off than women. But that leads to another discussion…


eweininger 06.25.06 at 8:34 am

I think the report claims that people report less discussion of what they feel is important regardless of what those things are, in which case the constituents of “important” do not matter: whatever people consider important, they are not discussing it.

Perhaps…but keep in mind that I was talking only about the reduction in the education effect, not the overall “increase in isolation” claim.

It does remain true that, as the authors explicitly acknowledge, all of the major conclusions of the paper–i.e. contraction of networks, reduced density, and thus greater isolation–hinge on assuming that the definitions of “discuss” and “important” remained constant across the 19 year interval. Otherwise, the ’04 people could have been behaving exactly like the ’85 people, but the data would still register a shift. Something changed, but it’s clear what it was.

Overall, I’m undecided on what to make of the increasing isolation claim. As the original post notes, the shift between the two time periods is quite large, and therefore writing it off tout court entails the assumption of a substantial cultural change (or of the emergence of major offsetting behaviors–the internet, presumably). On the other hand, the trope of the atomizing effects of modern society is an old one, and I’m reflexively a bit sceptical of it.


eweininger 06.25.06 at 11:17 am

Polical Theory Daily has this, which in turn begets that.


Noumenon 06.25.06 at 10:34 pm

Also. “Young (ages 18—39), white, educated (high school degree or more) men seem to have lost more discussion partners than other groups.”

Video games!


Kevin 06.28.06 at 9:06 pm

My theory is that newer information technologies are injecting efficiencies into personal relationships just like they are in other aspects of society (economic).

Instead of thinking that this is a negative, i.e. we are becoming more isolated, I think this just might be a positive. Reducing the number of people by a third who know your most inner-secrets would seem to be a good thing. We should not confuse “confidants’ with “friends”.

Comments on this entry are closed.