1
The Lasting Lessons ofEarly Adolescent Friendships:
The Benefits of Autonomy and the Mixed Blessings of Early Intensity
Joseph P. Allen
Amanda Hare
University of Virginia
Paper presented at the Biennial Meeting of the Society for Research in Child Development. Boston, MA. March 31, 2007.
A.SLIDE#1) :
B.I went to my high school reunion a few weeks ago…I won’t say which one, but if you’ve ever been to a high school reunion, its striking how well it can reinforce a fundamental point in development that we often overlook.
1.The point is simply this: what seems to be so clearly either good or bad in terms of how kids are functioning socially inearly adolescence ultimately turns out to be a far more complex phenomenon as development progresses.
2.SLIDE#2) : I think those of us who study particular periods in development, and I include myself in this, have a bit of a tendency to equate whatever’s linked to positive outcomes at a specific point as being likely to always be linked to positive outcomes.
a.If being popular is good at 13, then it must predict good outcomes by age 20 aswell.
3.But of course, that’s not necessarily true:
4.Developmental needs change, different eras have different demands, and what’s good in the short run isn’t always so great in the long run.
5.I wanted to start with that as a preface, because I’m going to talk today about early adolescent friendships and about how qualities of those friendships (which we observed at age 13) predict outcomes as we get closer to the end of adolescence, at age 19.
6.Let me start with the take home message: As friendship researchers, its very tempting to see the strong, close, early adolescent friendship as one of the ultimate positive markers of functioning of this period.
a.I’m going to suggest today that simply having close friendships—while its associated with many positive outcomes in early adolescence—only sets the stage forlong term social outcomes, but doesn’t tell us how things will actually play out on that stage
i.Rather, the key is not going to be whether teens have close friendships, but what KINDS of close friendships they actually have.
7.To get at this point, and to make sure everyone’s awake and alert:
a.Let me start with an audience participation question: How many people have at some point or other had a close friend who really wasn’t functioning all that well for an extended period of time?
b.OK, right there, that should make clear that the simple presence of a close friendship does not shield a person from serious difficulties in functioning.
II.SLIDE#3) :So, to try to understand this idea a bit more thoroughly, we’re going to look at four basic questions today:
A.We’re going to start by examining the degree to which teens are highly engaged in friendships at age 13 and from this examine:
B.Concurrent markers of psychosocial functioning
C.LONG-term predictions of functioning in late adolescence (age 19?)
D.Whether we can use knowledge of autonomy processes in friendships to understand moderating processes in these long-term links?
E.And finally, whether these autonomy processes might have long-term predictions on their own?
III.Methods
A.SLIDE#4)Sample : We’re going to examine these questions in a community sample of 184 adolescents, drawn from the public school system of Charlottesville, Virginia, who were interviewed along with their best friends,
B. but more importantly who were also observed in interactions with those best friends.
1.The sample was evenly divided between males and females,
2.Adolescents were initially assessed at age 13, and we’ll be reporting re-assessments up through their 19th year.
3.The sample was a normal community sample, and was representative of the population of the surrounding community in both socioeconomic and racial/ethnic terms.
4.An Intensive Interview Study
5.We’ve managed to maintain extremely low attrition over time (we’ve only had 1 participant dropout and say don’t contact us further, and at any given data point we typically have better than 90% data coverage. Nonetheless, we’ll be using Full Information Maximum Likelihood techniques to handle any missing data that we do have today.
IV.So, let’s start with our first question: SLIDE#5) : What are the links of high engagement with a close friend to concurrent markers of functioning?
A. We assess engagement with an observational task, which we call the Supportive Behavior Task
1.SLIDE#6) :In this task, we have teens’ bring in the person they feel is their “closest” friend to our labs, and we then ask our teens to tell their friend about a question or problem about which they’d like some advice or help.
2.We then film the discussion that follows for the next 6-minutes.
3.Teens discuss anything from very practical issues (e.g., how to make money) to very emotion-laden issues (e.g., coping with a parent’s divorce).
4.We observe these discussions and code them for the degree to which the teen and their friend are highly engaged in the discussion.
5.We assess engagement in terms of how closely teens listen to and follow up on what each other is saying in the conversation and their verbal and non-verbal cues that they are closely attending to what each other is saying.
B.So to be clear: We’re assessing just one aspect of the “closeness” of a friendship here, which is based on our own observations of an adolescent’s degree of engagement with a close friend around an issue where they could use some help.
C.And we find a number of cross-sectional correlates that I’ll just point to very briefly:
1.Cross-sectionallySLIDE#7) :
a.Being highly engaged with a close friend in these discussions is linked to:
i. Teens’ degree of engagement with their mothers –assessed using the same technique—thus showing that these kids are n’t engaged with their peers to the exclusion of their parents
ii.We also find links to teens’ qualities of relationship with their fathers.
iii. To their overall attachment security, assessed using the Adult Attachment Interview
iv. To them having fewer attitudes that support aggressive behavior
v. To their greater popularity
vi. And to one not so positive measure—a small relation to the degree of conflict in their friendships.
a)That’s another way of saying these 13-year olds with highly engaged friendships mostly appear to be doing very well, although it may be that the friendships are intense and engaged in ways that are also linked to some degree of conflict—which isn’t entirely surprising.
D.OK, So that’s what we see cross-sectionally…SLIDE#8) : Close friendships seem to be markers of mostly positive functioning across an array of indicators in early adolescence…
V. But what happens when we follow these kids over time?
A.Well, we can look at many different outcomes, but I’m going to focus on just a couple today starting with
B.SLIDE#9) : Future friendship quality:
C.Simply put, does having intense friendships early in adolescence actually predict anything about what kind of friendships one will ultimately have by the end of adolescence?
1.SLIDE#10) : We assess future friendship quality using the Total score from the Friendship Quality Questionnaire
2.In which we use PEERs’ reports to tell us how strong a close friendship our target teens are able to form in terms of the qualities you can see on the overhead.
a.IN essence, now we’re not observing one interaction but instead we’re getting an overall rating of the teen’s ability to form a strong positive close friendship, and we’re getting that rating from the person probably in the best position to know—our teen’s closest friend at age 19.
D.SLIDE#11) : So, looking first at the simple stability of this measure over time we find ,
1. moderate stability. (I should note that these are for the most part NOT the same peers as came in at age 13…so we’re looking not at how stable a given friendship is—only about 10% of people still had the same closest friend 6 years later, but rather at
a.The ability to form a strong, positive close friendship…which we see displays moderate stability over this period.
2. We also find that at Age 13, friendship quality is cross-sectionally linked to our observations of engaged interactions.
a.It’s a modest relationship though…intense friendships at 13 aren’t always good friendships as rated by friends.
3. But most interestingly, we find that if we look at how engaged a teen was with their best friend at 13 in our brief interaction, that that adds significantly to our prediction of relative changes in friendship quality OVER TIME.
a.In essence, this early intensity…or highly engaged interaction, is predicting a modest sized chunk of the change variance in close friendship quality over time. One possible (causal) interpretation is that closeness in early adolescent friendships begets (or at the very least foreshadows) further good friendships down the road, again, not too surprising, but interesting that we can capture it with a 6-minute interaction.
4.Previously, we’ve looked at other outcomes and found similar positive outcomes from these early qualities of interaction in this supportive behavior task with regards to other behaviors as well.
a.SLIDE#12) :For example, at SRA last year, my colleague Jill Antonishak reported finding that the quality of this supportive interaction was tightly linked to aggressive behavior over time.
b. We look at this relationship using a latent difference score model developed by Jack McArdle andAki Hamagami…, and I don’t want to frighten you with this next slide, because what we’re actually going to show you is quite simple really:
c.SLIDE#13) : This looks complicated but in essence all we did was begin with two manifest variables measured repeatedly SLIDE#14) :
d.SLIDE#15) : Then models intercept and slope parameters in classic growth curve fashion.
e.SLIDE#16) : Then considers dynamic autoregressive parameters (how where a variable starts affects how much it changes)
f.SLIDE#17) : and Finally, what we’re most interested in: Lagged predictors of Change that are CONSISTENT over time
g.i.e., where one variable consistently predicts future change in the other variable over a number of years.
h.SLIDE#18) : So what we’ve really got here is a repeated cross-lagged model assessed within a model that also accounts for some growth curve and dynamic change parameters.
5.And we find, in this simplified presentation, that we do in fact find these lagged predictions…Qualities of behavior in this brief supportive interaction task are consistently predicting lower levels of adolescent aggression over time.
a.Its as close to causality as we’re going to get with nonexperimental data.
E.So, this idea of intensity in early adolescent best friendships seems like a good thing, and in some ways it is, BUT…there’s another side to the picture.
F.If we look at kids’ levels of depression, we see some evidence of something different going on:
1.SLIDE#19) : We find first that Depression is also moderately stable over this fairly long span of time.
2.But more importantly for our purposes we find that being highly engaged with a close friendin a supportive behavior task at 13 predicts relative INCREASES (i.e. over and above this stability level) by age 19).
a.Kids who are highly engaged with their best friends at age 13 are becoming more depressed over time.
3.This is NOT at all a gender effect (we covary gender in these analyses). And gender has relatively modest effects in these data. There also aren’t any gender interactions in any of these findings (the effects apply similarly for males and for females).
G.And, lest you think this is just a fluke, I should note that we’ve previously found and reported similar results regarding anxiety.
1.SLIDE#20) : Being engaged at 13 predicts being more anxious 4 years later.
H.So what’s going on here?SLIDE#21) :
What we see is that adolescents who are highly engaged with their friends ina problem task at 13 are becoming better friends and less aggressive later in adolescence but also are more likely to be anxious and depressed?
I. So, now, we go back to our opening message: Simply having close friendships isn’t the whole story:Rather, we have to ask WHAT KIND of close friendships does an early teen have?
J.We have to start to focus on relationship QUALITY
VI.As we start to think about relationship quality, we go back to a notion that’s been central to our research with parent-teen interactions for a number of years: Which is that the ideal relationships are characterized not just by closeness, but by a combination of closeness AND autonomy.
A.So, the question we went back and asked was how do teens in these close friendships that we observed at age 13 handle autonomy issues—how do they handle disagreements.
B.SLIDE#22) : And specifically, whether these autonomy processes might moderate some of the slightly puzzling effects we’ve just described.
C.SLIDE#23) : To do this we gave our young teens a second discussion task, a disagreement task… which we called the “Mars task.”
1.In this task, young people are asked to make decisions about hypothetical characters who’ve been involved in an accident on a spaceship returning from Mars.
a.Each young person makes decisions independently, then they are brought together and asked to try to come to a consensus.
D.What we look at is the extent to which these 13 year old friends can handle the resulting interaction without it falling apart with them undermining the autonomy or the connection with their friend.
1.So, we look, for example, to see if they handle the disagreement by discussing their reasoning, or alternatively do they do so with behaviors that are pressuring, enmeshing or hostile.
E.And we look for interactions of these QUALITIES of friendship with our measures of INTENSITY of friendship in predicting our outcomes.
F.What we see is an interesting pair of interaction effects.
1.We see it with BOTH of our longitudinal predictions.
2.SLIDE#24) : For predictions of increasing overall quality of friendship, we see that when teens handle disagreement in ways that preserve one another’s autonomy,then the normal pattern we described earlier holds:
a.Teens who are highly engaged with their friends at 13 develop stronger friendships (not just with the same friend) over time.
b.This in essence is what we’ve already reported.
3.BUT, when at 13, these friendships are characterized by what we would call ‘enmeshing’ behaviors—pressuring, hostile and personalizing behaviors, SLIDE#25) : then we see that the beneficial ‘effect’ of being closely engaged with a peer disappears.
a. Engagement predicts increasing friendship quality ONLY for teens not in enmeshed peer relationships.
4.SLIDE#26) : For predictions of increasing levels of depressive symptoms, we see almost the exact converse of this picture:
a.when teens handle disagreement in ways that preserve one another’s autonomy, then being engaged with a friend doesn’t have any effect in predicting subsequent levels of depression.
5.BUT, when at 13, these friendships are characterized by what we would call ‘enmeshing’ behaviors—pressuring, hostile and personalizing behaviors in the face of a disagreement, SLIDE#27) : then we see that being closely engaged does indeed predict becoming more depressed over time.
a. Yes, engagement can foreshadow increasing depressive symptoms, but ONLY for teens in autonomy undermining, enmeshed friendships.
G.SLIDE#28) : It’s a simple effect, that’s really obvious in retrospect, but that frequently gets overlooked:
1.Its not just a matter of having close friends, it’s a matter of what kind of close friendship one has.
2.Said differently, Intensity increases the predictive value of a friendship, whether for good or for bad, depending on the friendship quality.
VII.And so for our final question, we take a look at this additional aspect of friendship quality—at autonomy processes in close friend interactions and we find that it has some noteworthy main effect predictions of its own
A.SLIDE#29) : We look at aggressive behavior (over a somewhat shorter age span here for this particular measure), and we find that the ability to demonstrate and maintain one’s autonomy in a disagreement with one’s close friend at 13 predicts lower levels of future overt aggression for example
1. So young people who can maintain their autonomy in disagreeing with their friends are less likely to need to be aggressive toward others 4 years later. Suggesting that this ability to negotiate disagreements may be important to understanding aggression in late adolescence and adulthood.
B.SLIDE#30) : We find that these autonomy processes in 13-year olds’ friendships also predict changes in how much peer pressure teens are subjected to (as reported by their friends)…with teens with more autonomous early adolescent relationships being subjected to less peer pressure as adolescence proceeds.
1. Again, we have this idea that these early friendships are setting the stage for handling autonomy-like negotiations well into the future.
VIII.SLIDE#31) : Now before I summarize these findings there are a few limitations in these data that bear note:
A.First, of course these are not experimental data, so even when we’re doing our most sophisticated cross-lagged modeling, we still aren’t able to establish causal relations within the data.