Positive, Negative, and Ambivalent Interactions With Family and Friends: Associations With Well‐being

• Author: Hyo Jung Lee,Maximiliane E. Szinovacz
• Published date: 01 June 2016
• DOI: http://doi.org/10.1111/jomf.12302
• Date: 01 June 2016

H J L  M E. S University of Massachusetts Boston

Positive, Negative, and Ambivalent Interactions

With Family and Friends: Associations

With Well-being

Although the relationship between social rela-

tionships and mental health is well established,

debate continues about the relative importance

of specic sources (spouses, children, relatives,

friends) as well as of positive and negative

interactions. The authors examined the asso-

ciations of positive, negative, and ambivalent

interactions with life satisfaction and depres-

sive symptoms for spouses, children, relatives,

and friends, using data from the 2008 Health

and Retirement Study (N=6,418). The nd-

ings generally showed positive associations

between positive interactions and mental health

and negative associations between negative

or ambivalent interactions and mental health.

These associations were most pronounced for

relationships with spouses and children. Gender

differences were found in life satisfaction but

not in depressive symptoms. These results imply

that future research on older adults needs to

consider both positive and negative relationship

features from diverse sources separately and in

combination to disentangle their relative effects

and their additive or compensatory potential.

Social relationships have been consistently

linked to mental health (Antonucci, Fiori,

Department of Gerontology, McCormack Graduate School

of Policy and Global Studies, University of Massachusetts,

100 Morrissey Blvd., Boston, MA 02125

(hyojung.lee001@umb.edu).

This article was edited by Linda Waite.

Key Words: aging, ambivalence, families in middle or later

life, mental health, social strains, social supports.

Birditt, & Jackey, 2010; Bertera, 2005; Rook,

1984, 1990, 1998); however, debate continues

on several contingencies of this relationship.

First, it remains unclear whether positive or

negative interactions are most important for

mental health. The relative effect of positive

and negative relationships on mental health may

depend on which indicators of mental health are

used (Ingersoll-Dayton, Morgan, & Antonucci,

1997) and on whether positive and negative

interactions are examined separately or included

simultaneously in analysis models (Stafford,

McMunn, Zaninotto, & Nazroo, 2011).

Second, some research suggests that relation-

ships characterized by both high positive and

negative interactions (ambivalence) are partic-

ularly likely to undermine mental health (Fin-

german, Pitzer, Lefkowitz, Birditt, & Mroczek,

2008). Although the ambivalence perspective

has been widely adopted, questions remain

about how to best conceptualize and measure

the combined effects of positive and negative

relationship features (Lendon, Silverstein, &

Giarrusso, 2014; Lettke & Klein, 2004).

Third, the strength of associations between

social interactions and mental health depends

on the source. Relationships with very close

network members (e.g., a spouse) tend to have

stronger associations with mental health than

interactions with more peripheral network

members (Chen & Feeley, 2014; Stafford et al.,

2011). Furthermore, the relative effects of

positive and negative interactions or of ambiva-

lence may be source specic. For example,

negative interactions with a spouse may be

more important for mental health than positive

660 Journal of Marriage and Family 78 (June 2016): 660–679

DOI:10.1111/jomf.12302

Social Interactions and Well-being 661

interactions with a spouse, whereas the opposite

may hold for interactions with friends.

To address these issues, we examined the

associations between positive and negativeinter-

actions as well as ambivalence on two indica-

tors of mental health, namely, life satisfaction

and depressive symptoms. This examination was

guided by three research questions:

1. Which aspects of social relations (positive

interactions, negative interactions, ambiva-

lence) are most strongly related to mental

health?

2. Does the relative strength of these associa-

tions depend on which indicator of mental

health is used (life satisfaction vs. depressive

symptoms) and on whether or not positive

and negative interactions are included sepa-

rately or simultaneously in analyses models?

3. Which relationship sources (spouse, children,

other relatives, friends) are most important

for mental health, and do the associations

between positive and negative interactions

or ambivalence and mental health vary by

source?

Because gender has been linked both to the

prevalence of positiveand negative relationships

as well as to their association with mental health,

we further assessed gender differences across

all three research questions. Our analyses relied

on the 2008 wave of the Health and Retirement

Study (HRS), a representative survey of individ-

uals age over 50 years and their spouses (see

http://hrsonline.isr.umich.edu/).

T B  L

R

Our hypotheses were informed by four different

perspectives: (a) research focusing on the rela-

tionship between social interactions and mental

health, (b) the interplay between positive and

negative interactions (ambivalence), (c) hier-

archies of social relationships in individuals’

social networks, and (d) gender roles.

Positive and Negative Social Ties and Mental

Health

Much of the literature on the effects of social

relationships on mental health has emphasized

the supportive functions of social interactions.

For example, the original intergenerational

solidarity model focused exclusively on positive

relationship features such as association, affec-

tion, or consensus (Bengtson & Roberts, 1991).

In general, research supports the assumption

that positive interactions contribute to mental

health. Similarly, past studies provide relatively

consistent evidence that negative social inter-

actions such as conict, irritation, or criticism

undermine mental health (Newsom, Nishishiba,

& Morgan, 2003; Newsom, Rook, Nishishiba,

Sorkin, & Mahan, 2005; Rook, 2001). It is

less clear, however, whether positive and neg-

ative interactions are equally important for

mental health, whether they have independent

or interrelated effects on mental health, and

whether they have similar or divergent inu-

ences on distinct dimensions of mental health

(Ingersoll-Dayton et al., 1997; Newsom et al.,

2003, 2005; Rook, 1990, 1998).

To capture the potentially differential associ-

ations between positive and negative exchanges

and selected dimensions of mental health,

Ingersoll and colleagues (1997) distinguished

among four models. The positivity effects model

suggests that solely positive exchanges affect

both positive and negative indicators of mental

health, whereas negative exchanges have no

effect. In contrast, the negativity effects model

posits that only negative exchanges inuence

mental health, regardless of domain (positive

vs. negative indicators of mental health). The

third model (the domain specic effects model)

proposes that the effects of social exchanges

are domain specic; that is, positive exchanges

inuence only positive indicators of mental

health but not negative ones, whereas negative

exchanges are solely associated with negative

indicators of mental health. The fourth, com-

bined model proposes both domain-specic

and crossover effects, so that both positive and

negative exchanges are related to both positive

and negative indicators of mental health.

The few studies that have simultaneously

included measures of positive and negative

exchanges (investigations addressing positive

and negative exchanges with specic sources

are discussed below) provide a mixed pic-

ture. Some indicate domain-specic effects for

positive exchanges but both domain-specic

and crossover effects for negative exchanges

(Ingersoll-Dayton et al., 1997; Newsom et al.,

2005); others support crossover effects for

both negative and positive exchanges (Uchino,

Holt-Lunstad, Smith, & Bloor, 2004); and yet

others suggest that support for specic models