Escolar Documentos
Profissional Documentos
Cultura Documentos
A Thesis
MASTER OF ARTS
December 2007
Committee:
William H. O'Brien
Steven Jex
ii
ABSTRACT
improved participant weight loss (e.g., Black, Gleser, & Kooyers, 1990), however little is known
of the influence of naturally occurring (external to the intervention situation) support. Similarly
overlooked is the role of social support to the numerous behavior changes required for successful
weight loss. The current study evaluated the occurrence (i.e., frequency) and experience (i.e.,
outcomes. The primary goals were to examine the support – behavior change relationship for
evidence of specificity and to evaluate the utility of measuring support appraisals as a tool for
identifying resource-need match. Within these objectives, another aim of the study was to
returned largely null results. Small sample size and low power are important considerations in
explaining the null findings. However, attention is also called to other possible factors,
including stage of behavior change and the “obesogenicity” of modern environments, which may
have contributed to the current null findings and warrant further attention.
iii
ACKNOWLEDGMENTS
I am grateful for my academic mentors that have helped me to focus my curiosity and
encouraged me to ask questions that matter to me: Craig W. Berridge, Rob Carels, and Bill
O’Brien. I am grateful to the Clinical Nine, The Tree House Gang, and Listy McListerson
friendships and laughter for making graduate school the best time of this life so far. I am grateful
for Azure Pupper faithfully resting next to me during every single late night. I am grateful to my
spiritual teachers, including Dave Peters and Zen Master Dae Kwang, for “just this” teaching. I am
grateful to my Mom, Dad, Tanya and Eric for loving and believing in me through a lifetime of
challenges. And, everyday, I am grateful for the one hundred thousand ways that Dryw reminds me of
what matters.
v
TABLE OF CONTENTS
Page
INTRODUCTION ................................................................................................................. 1
METHOD …………………………………………………………………………………. 18
RESULTS ………………………………………………………………………………… 26
DISCUSSION……………………………………………………………………………… 35
REFERENCES ...................................................................................................................... 41
LIST OF TABLES
Table Page
Physiological Outcomes............................................................................................. 54
Obesity causes or exacerbates many chronic and life threatening physical health
cardiovascular disease, stroke, osteoarthritis, and sleep apnea (NIH, 1998, Pi-Sunyer, 1996). The
prevalence of overweight and obesity continue to rise, making it a leading public health concern.
In the United States, 66.3% of adults over the age of 20 are either overweight, (Body Mass Index
(BMI) of 25 – 29.9 kg/m2) or obese (BMI ± 30 kg/m2; Ogden, 2006; NIH, 1998).
Behavioral weight loss programs (BWLP) have emerged as the treatment of choice for
mild to moderate obesity (NIH, 1998). Such programs typically involve psycho-educational
classes that facilitate skill-development and problem solving to promote gradually increasing
energy expenditure through physical activity while progressively decreasing energy intake
through permanent lifestyle changes. Many of the health consequences of obesity can be
improved or reversed with modest weight loss. Studies examining the health benefits of weight
loss have determined that such benefits begin when approximately 10% of total body weight is
lost (Pi-Sunyer, 1996). The average weight loss of participants in BWLPs, 8 – 10% of total body
weight (NIH, 1998), just reaches the threshold of clinical significance. While BWLPs are
generally regarded as effective short-term treatments, rates of post-treatment weight regain are
high (Elfag & Rossner, 2005; Jeffery, Drewnowski, Epstein, Stunkard, Wilson & Wing, 2000). In
addition, some BWLP participants dropout of treatment or fail to lose enough weight to
overweight), psychological factors (e.g., self-efficacy, mood), and social factors (e.g., role
Experience of Social Support 2
responsibilities, social support) can affect treatment outcomes in weight loss interventions
(Carels, Cacciapaglia, Douglass, Rydin & O’Brien, 2003; Wing, Marcus, Epstein, & Jawad,
1991). One factor, social support, has long been identified as an important social variable in
health and health behavior change research (Hogan, Linden & Najarian, 2002), including
investigations of weight loss (Black, Gleser & Kooyers, 1990) and weight loss maintenance
efforts (Elfhag & Rossner, 2005). Of particular interest is the role of social support in facilitating
Social Support
Structural models. Several models have been developed to account for the social support
– health relationship. Early models correlated structural aspects of support with particular
health related outcomes. Structural measures emphasize the number and range of social
connections that comprise a social network, as well as the frequency of social contacts within the
network (House, 1987). Primary social relationships (spouse, family) are often highlighted. For
example, marital status, an important index of structural support is often related to health
outcomes (Cohen, 1988). The finding that married men live longer is one such example
(Kiecolt-Glaser & Newton, 2001). Not surprisingly, intimate relationships, particularly spouses
and co-habiting partners have consistently been identified as the single most influential source of
Structural models offer an important starting point for exploring the relationship of social
support to health outcomes. However, the simple presence of socially related others (spouse,
family, friend, etc) does not address the mechanisms for the apparent salutary effect.
networks with health outcomes, functional models of social support, attempt to understand
Experience of Social Support 3
something about how support is functionally related to particular outcomes (e.g. What resources
do married vs. unmarried men receive through their supportive relationships?). Functional
models assume that social support functions by helping an individual cope with challenges by
providing certain resources. More specifically, the matching hypothesis asserts that the
provision of support is most functional when it is matched to specific resource needs (Wills &
Fegan, 2001). In an effort to explore this, some measures of supportive behaviors are separated
into broad functional classes based on the apparent resources they provide: informational,
solving around stressful or challenging life events or changes. Emotional support is the
provision of reassurance, love, and acceptance. Instrumental support is the provision of material
resources or needed services (Cohen & Wills, 1985). These measures are used to test predictions
about the relative contribution of support functions to successful coping in the face of a given
challenge.
Similarly, specificity models argue the importance of ‘matching’ with respect to the
general content of enacted support behaviors. Specificity models, highlight the fact that support
needs may vary according to stressor and note the value of using measures that are specific with
respect to enacted support items that “match” the demands of the context (Vaux, 1992). For
engaging in a specific support behavior that matches the context (Walitzer & Derman, 2004).
Indeed, this approach is particularly well suited to behavior change interventions (Cohen &
Lichtenstein, 1990). In effect, this type of model calls for specificity in matching enacted
support behavior to relevant behavior change goals (Hutchinson, 1999). Weight loss requires
multiple behavior changes (e.g., eating, exercise), and, as such, examination of specificity and
Experience of Social Support 4
matching may be particularly relevant (McLean, Griffin, Toney & Hardeman, 2003). Of interest
in this area is the degree to which specificity and matching of support to behavior change
domains is related to behavioral outcomes within those domains. For instance, does support
behavior that targets dietary change tend to predict eating behavior changes; does support
targeting physical activity predict increased exercise? Understanding the specificity of these
support-change matches, may help guide the nature of support instruction in interventions that
emphasized the resource function of support without attention to the contribution of structural
features (e.g., source of support), some models have broadened to appreciate the interplay of
source and function of support (House, Umberson & Landis, 1988). For instance, a functional
account for the primacy of the spousal relationship in health outcomes recognizes that, in sharing
living space and resources, partners are uniquely situated to facilitate positive change or maintain
barriers that reinforce the status quo. Indeed, in adulthood, patterns of support from individuals
outside of this uniquely powerful dyad do not tend to vary substantially in the patterns of support
that they provide (Argyle, 1992; Sallis, Grossman, Pinski, Patterson & Nader, 1987). Thus,
while resources provided through broad network support (e.g., collapsing across friends and
family) are important and generally related to health outcomes, support in the context of an
structural and functional support are related to health outcomes, they do not tend to be highly
correlated with one another. That is, a large social network is not necessarily indicative of
greater access to resources through social support just as being married is not necessarily
Experience of Social Support 5
indicative of greater access to supportive behaviors from one’s spouse (Wills & Fegan, 2001).
Similarly, the simple occurrence of specific functional support behaviors does not necessarily
translate into useful/effective support experiences (Harris, 1992; Rook, 1984, 1992).
In health psychology, functional models propose that social support may affect health by
influencing health related behaviors (Cohen, 1988), for instance by increasing positive or
protective behaviors or decreasing harmful behaviors. Social support has been linked to a
number of health behavior changes, such as smoking cessation (Cohen & Lichtenstein, 1990;
Roski, Schmid & Lando, 1996), alcohol reduction (Walitzer & Dermen, 2004), and medication
adherence (DiMatteo, 2004; Simoni, Frick & Huang, 2006). However, attempting to make
predictions regarding the relative match of a particular functional support category to the
Informational (e.g., behavior change strategies), emotional (e.g., reassurance during discouraging
lapses), and instrumental (e.g., goods/services that facilitate greater access to positive health
behaviors) support resources may not always match the support needs of individuals engaged in
multiple ways, ranging from motivating to undermining, and may vary considerably from the
generally not functionally independent in the naturalistic contexts in which support is provided.
That is, a supportive act may serve some combination of emotional, instrumental, and/or
informational resource needs. Indeed, many empirical studies attempting to separate support
functions demonstrate substantial intercorrelation between the scales (Sarason, Shearin, Peirce &
Experience of Social Support 6
Sarason, 1987; Schwarzer & Gutierrez-Dona, 2005; Simony, Frick & Huang, 2006). An
examination of such interrelations between support scales lead Sarason et al (1987) to conclude
that the trend to separate support into discrete functions does not improve the sensitivity of these
Both statistics, and common sense, seem to be returning our attention to a more holistic
view of support that recognizes that support behaviors can be multifunctional and that the
relative appraisal and impact of these functions are context specific (Martin, Davis, Baron,
Suls & Blanchard, 1994; Sarason et al, 1987). For instance, the provision of needed financial
help by a parent to an adult child for health care expenses certainly serves as instrumental
emotional support function. By contrast, the same financial support provided by an arrogant
younger sibling may retain its positive quality as an instrumental support, but be regarded as
humiliating, thus the emotional support function acquires a negative quality in this context.
These examples highlight the potential importance of assessing the appraisal of the supportive
support attempts may be especially important for understanding the relationship between enacted
support and health behavior change outcomes (Goldsmith, McDermott & Alexander, 2000;
The goal of the proposed investigation is to examine the specificity of the relationship of
support (e.g., from spouse and family and friends) as it relates to weight reducers’ behavior
change and health outcomes. Of particular interest is whether the reported experience (e.g.,
appraisal of helpfulness) of support predicts behavior change and health outcomes beyond the
Experience of Social Support 7
occurrence (e.g., frequency) of support. The following review begins with a broad discussion of
social support and BWLP outcomes research. Next, a rationale for examining appraisals of
resource – need match) is provided. Finally, the goals and hypotheses of the proposed
Functional models of the role of support in weight loss and physical fitness suggest that
support increases contact with resources that facilitate helpful/protective behaviors and block
understood in terms of its potential for increasing helpful/protective behavior and/or decreasing
harmful behavior. To illustrate, social support may function by reinforcing positive changes,
punishing lapses, problem-solving barriers to behavior change, and influencing contact with
In obesity research, social support for participants in weight loss interventions has
generally been investigated by involving supportive others, such as a spouse or family member,
in the intervention in some way. Though individual studies have been contradictory, taken
together, spouse and/or family involvement in weight loss interventions has generally been
related to better treatment outcomes (Black et al, 1990; McLean, Griffin, Toney & Hardeman,
participant-alone programs found that interventions that involve partners prove significantly
more effective in helping participants with initial weight loss than participant-alone
interventions. This effect held for up to three months follow-up. However, after this time point,
there was not a statistically significant advantage for the couples programs over the participant-
Experience of Social Support 8
alone programs, though a positive trend favoring couples remained. The couples-interventions
differed in the kinds of support that partners were encouraged to provide, and most studies did
not assess the range of support that partners actually provided. None of the studies assessed
In a unique departure from the usual couples’ weight loss interventions, Wing, Marcus,
Epstein and Jawad (1991) conducted a participant alone vs. participant - partner together
intervention in which the partner was also obese and a target of the weight loss intervention.
Each of the participants, though not necessarily their spouses, had Type II diabetes. Prior to
formal treatment, participants in both conditions were comparable in their reports of support.
Participants in the alone condition were married and taking part in the BWLP alone, with no
formal involvement of their partner. Participants in the together condition attended the standard
BWLP sessions with their spouses and sessions included attention to joint problem-solving and
support strategies relevant to weight loss. Interestingly, the groups both improved, and were
roughly equivalent with respect to weight loss, dietary change, and exercise from pretreatment to
posttreatment and one-year follow-up. Though not part of the investigators’ original hypotheses,
post-hoc analyses demonstrated that there was a significant condition by gender effect on
outcome, such that women did better when treated together and men did better when treated
alone. These effects held at the one-year follow-up (Wing et al, 1991). In addition to drawing
attention to the importance of considering gender effects, this study also serves to highlight the
questions that remain with respect to social support in this context; did the intervention change
the frequency or appraisal of support? Does the occurrence or appraisal of support tend to differ
The Wing et al (1991) study manipulated active spousal involvement in the intervention.
Such interventions have been conducted with the assumption that if significant others are taught
strategies for supporting their partner’s behavior change, the support person will be better able to
help facilitate the translation of intervention suggestions into the home environment and serve as
functional support from friends and family for exercise and diet was assessed, and found to be
equivalent, prior to the intervention. However, there were no follow-up assessments of spousal
support once active treatment had begun that might have helped determine whether or how the
social support intervention was translated into actual support behaviors or whether the support
involves an already naturally and appropriately supportive spouse is unlikely to be made more
effective by instructing the spouse in the support behaviors that are either already offered or are
Indeed, support for the notion that some interventions that target support strategies may
not actually broaden the support repertoire of participants or spouses was found in the Wing et al
(1991) study. Participants in both the treatment alone and the treatment together conditions
reported changes in the use of listening and support strategies from pre to post treatment despite
the fact that only the participants in the treatment together condition were taught such strategies.
This finding may suggest that a context that provides for personal goals to be articulated (i.e.,
desire to change behavior in order to lose weight), such as that provided by involvement in a
weight loss intervention, may be all that is necessary for support persons to become alert and
engaged in the support repertoire they already possess. Additionally, instruction in support
behaviors may have little bearing on the frequency of dietary and exercise support behaviors
Experience of Social Support 10
actually engaged by spouses outside of the intervention (McLean et al, 2003). The awareness of
support needs and the resources to engage in appropriate support repertoires will not necessarily
translate into effectively enacted supportive behavior. Finally, such support interventions have
failed to appreciate the potentially important role of participants’ experience of the support
In another study, Wing and Jeffrey (1999) attempted to examine the role of naturally
occurring social support, and compare it against experimental social support manipulations.
Participants in this study were recruited alone, or with three friends and family members. All
participants were administered a modified version of the Sallis Social Support for Eating and
Exercise survey (Sallis et al., 1987) at baseline and there were no significant differences among
the groups in reported frequency of diet and exercise support. Individuals (recruited alone) and
groups (recruited together) of participants were randomized to BWLP or BWLP plus social
BWLP plus a social support intervention (experimental condition) included intragroup activities
and intergroup weight loss and weight loss maintenance competitions. While there were
differences between the groups according to treatment, such that those receiving the support
intervention lost more weight, these differences were no longer significant after recruitment
strategy was taken into account. Additionally, like the Wing et al (1991) study discussed above
there were only pretreatment measures of the frequency of functional support for eating and
enacted support items was given. Since the measures ask participants about functional support
for eating and exercise prior to taking part in the intervention and attempting diet and exercise
changes, it is not surprising that these measures would not predict outcomes. That is, prior to
Experience of Social Support 11
initiating the change effort, such support is plausibly less likely to occur (Cohen & Wills, 1985)
and less likely to be experienced as helpful (Prochaska, DiClemente, Norcross, 1992). A better
understanding of the experience/appraisal of support received from spouses and others during the
behavior change effort, may help provide a clearer understanding of the cognitive factors that
For many obese individuals, successful weight reduction requires significant behavior
change in a number of areas (e.g., diet, physical activity), making it one of the more challenging
health behavior change efforts to engage and sustain. Social support has the potential to
facilitate or block such efforts in some or all of these areas. As noted earlier, traditional
functional models have failed to appreciate the complexity of the supportive interaction. It is
hoped that attention to participants’ appraisals of experienced helpfulness may foster a more
flexible and complete picture of the multiple, and sometimes contradictory, functions of social
Participants’ evaluations may provide a link between the occurrence of particular support
behaviors and behavior change outcomes (Cohen & Wills, 1985). Differential perceptions of
support have been linked to the relative fit between support resource offered and support need
(Jacobson, 1986), such that negative appraisals may indicate a poor fit between resources offered
and desired and positive appraisals may indicate a good fit between resources offered and
desired. Measuring the evaluation of enacted social support is one way to gauge the experience
of resource-need fit. Research in this area has demonstrated that support may be evaluated along
multiple dimensions: helpful, supportive, and sensitive. The helpful, supportive, and sensitive
relational assurance, and emotional awareness respectively (Goldsmith et al, 2000). As noted
previously, the experience of enacted support may be evaluated positively in some respects and
negatively in others. To illustrate, a BWLP participant may experience reminders from a support
person not to eat high fat, high calorie foods as condescending and unsatisfying (e.g., a negative
positive eating changes in another (e.g., a positive evaluation on the problem solving/helpfulness
experiences of resource-need fit. Given the challenges of negotiating multiple behavior changes,
gauging participants’ experience of the problem-solving utility of enacted support seemed most
appropriate for the task of evaluating perceptions of resource-need fit in this context. Thus,
helpfulness was chosen as the evaluative dimension for the current study.
Experience of Social Support 13
The current study emerges from a larger project examining the impact of a six-month
behavioral weight loss program and an experimental weight maintenance program on the health
outcomes of BWLP participants. This study examined BWLP participants’ reports of support
received during the initial weeks of the intervention, in relation to treatment outcomes. Both
behavioral outcomes (i.e., variables assessing eating and physical activity behavior change) and
physiological outcomes (i.e., variables assessing body composition and cardiorespiratory fitness)
are important for evaluating weight loss interventions. Thus, the following behavioral and
intake, energy expenditure through physical activity, submaximal oxygen consumption, percent
Broadly, the traditional functional model of social support predicts that, inasmuch as
support provides resources, it will generally be related to positive outcomes. Indeed, greater
overall functional support has generally been associated with positive BWLP treatment
outcomes. In the area of health psychology, researchers have argued the importance of
specificity in evaluating support behaviors with respect to the specific support needs of people
presented with particular health-related concerns (Simony et al, 2006). Thus, when health-
related concerns require behavior change (e.g., obesity), the examination of support behaviors
relevant to these change goals is of particular importance (Hogan, Linden & Najarian, 2002;
Vaux, 1992). However, while the involvement of supportive others in weight loss interventions
has been generally advantageous for weight reducers, specificity with respect to support
behaviors and changes in diet and exercise is not clear (Black et al. 1990). Thus, a goal of the
current investigation was to examine the relationship of support with behavioral treatment
Experience of Social Support 14
outcomes for evidence of specificity (e.g., healthy eating support and eating behavior outcomes
Moving beyond a traditional functional examination of support, the current study also
explored the relationship between participants’ appraisals of received support and BWLP
treatment outcomes. Guided by the matching hypothesis that suggests that support is most
effective when support resources match support needs (Wills & Fegan, 2001), the current study
explored the utility of helpfulness ratings as a method for detecting “resource – need fit”
(Goldsmith et al., 2000). Of particular interest, was whether participants’ ratings of the
Consistent with guidelines for social support research (Vaux, 1992), the current
Specifically, family/friend support and spousal support are examined separately. Furthermore,
given the primacy of the spousal relationship (Argyle, 1992), and the general tendency to favor
spousal involvement in interventions that include supportive others (Black et al., 1990), the
current study also includes measures and hypotheses that apply only to the subset of married
participants.
(appraisal) that may be related to support behaviors becoming functionally supportive (i.e.,
related to positive behavioral and physiological outcomes), it is hoped that this research will
provide a clearer direction for the active and productive engagement of social support in health
Hypotheses
1. Overall frequency. (a) For all participants, greater overall (i.e., eating and exercise
combined) support frequency from family and friends (not including spouse) will predict
positive behavioral (e.g., eating and exercise) and physiological (e.g. weight, body fat)
cardiorespiratory fitness, and greater percent weight and body fat loss. (b) For married
participants, the same relationships between overall spousal support frequency and
2. Overall appraisal. (a) For all participants, greater overall (i.e., eating and exercise
combined) ratings of support helpfulness from family and friends (not including spouse)
will predict positive behavioral (e.g., eating and exercise) and physiological (e.g., weight,
body fat) outcomes, including: greater caloric reduction, increased physical activity,
improved cardiorespiratory fitness, and greater percent weight and body fat loss. (b) For
married participants, the same relationships between overall ratings of spousal support
3. Frequency specificity. (a) For all participants, frequency factors for eating support and
exercise support, from friends and family (not including spouse), will demonstrate
prediction of eating outcomes (i.e., caloric intake), than will the exercise
support factor.
- The exercise support factor will contribute significantly more to the prediction
cardiorespiratory fitness), than will the eating support factors (b) For married
4. Appraisal specificity. (a) For all participants, appraisal factors for eating support and
exercise support, from friends and family (not including spouse), will demonstrate
prediction of eating outcomes (i.e., caloric intake), than will the exercise
support factor.
- The exercise support factor will contribute significantly more to the prediction
(b) For married participants, the same relationships between appraisal factors for spousal
5. Overall resource – need fit. (a) Positive appraisals of the overall (i.e., eating and exercise
combined) helpfulness of support received from family and friends (i.e., not including
spouse) will predict positive behavioral (e.g., eating and exercise) and physiological (e.g.,
weight, body fat) outcomes beyond the parallel (family and friend) frequency of support
measure. (b) Similarly, for married participants, overall positive appraisals of support,
frequency of support.
Experience of Social Support 17
6. Specificity in resource – need fit. (a) Positive appraisal factors for the helpfulness of
eating and exercise support, from family and friends (not including spouse), will predict
positive behavioral and physiological outcomes beyond the parallel frequency factors.
- The eating appraisal factor(s) will predict eating outcomes (i.e., caloric
- The exercise support factor will predict physical activity outcomes (i.e.,
(b) Similarly, for married participants, the addition of an appraisal factor, will
7. In the subset of married participants, spousal support will better predict behavioral (e.g.
eating and exercise) and physiological (e.g., weight, body fat) outcomes than will support
a. For married participants, overall greater frequency positive spousal support will
be a better predictor of positive outcomes than will the analogous family and
better predictor of positive outcomes than will the analogous family and friend
support score.
Experience of Social Support 18
Method
Participants
Fifty-four obese, sedentary adults who were recruited to participate in a BWLP through
local newspaper advertisements and campus e-mail announcements will be examined in this
survey investigation. Interested individuals were included if they were: a) obese (BMI ≥
consent and e) approved for participation by their primary care physician. Individuals were
excluded from participation if they had: a) past or current cardiovascular disease determined
program, of which $75 was returned upon completion of the posttest, with the remaining $25 to
be returned at the one-year follow-up assessment. This contingency contract was meant to
increase compliance with the assessments, and was waived for participants without the financial
All participants in dating or marital relationships were eligible to participate in the survey
assessing significant other support. Participants were recruited for participation through their
BWLP classes. No compensation was offered for participation in the survey study. All
procedures received human subjects review board approval. Of the 54 BWLP participants
family and friend support (n = 38 married and single BWLP participants) and significant other
support (n = 24 married/dating BWLP participants). All participants in the BWLP were eligible
The data collected for this study emerge from a larger investigation in which half of the
participants were randomly assigned to the six-month BWLP or to the BWLP plus an
weight, body fat, cardiorespiratory fitness, physical activity, and nutrition. Measures of social
support frequency and helpfulness were completed at three-weeks following initiation of the
program.
Intervention
75 minute, weekly sessions. The BWLP was based on the LEARN program (Brownell, 2004).
The LEARN program is an empirically supported approach to weight maintenance and physical
activity that includes five components: Lifestyle, Exercise, Attitudes, Relationships, and
Nutrition. The goals of the program are to achieve weight loss by gradually increasing energy
expenditure through physical activity while progressively decreasing energy intake through
permanent lifestyle changes. Additional information on the LEARN program can be found at the
graduate student, or two upper-level clinical psychology graduate students, administered the
Following the 16 week BWLP, half of the participants took part in an addition six-week
personal food and physical activity environments (Lowe, 2003) to limit exposure to obesegenic
cues and maximize exposure to cues associated with healthy eating and physical activity.
Experience of Social Support 20
Data analysis for social support will treat the participants as one group because outcome
measures used for this study were obtained at the close of the BWLP (i.e., that all participants
Body weight and composition. Body weight was assessed on a weekly basis using a
digital sacle (BF-350e; Tanita, Arlington Heights, IL) to the nearest 0.1 lb., and height was
measured in inches to the nearest 0.5 inch using a height rod on a standard spring scale. Height
and weight were used to calculate BMI (kg/m2). Although change in body weight is often
commonly used to measure weight loss success (Jeffery et. al., 2000) weight loss expressed as a
percentage of initial body weight reduces the likelihood that heavier participants will be
categorized as more successful while losing a smaller percentage of their total body weight than
lighter participants. In this study, percent change in body weight was defined as the percent
change in body weight from pretreatment baseline to post-treatment (i.e. baseline weight minus
posttreatment weight divided by baseline weight. Body fat was obtained using leg-to-leg
impedance analysis correlates highly with body composition estimates utilizing the underwater
weighing method in obese women (Utter, Nieman, Ward & Butterworth, 1999). Weight and
body fat assessments were obtained at pretreatment and at the end of the 16-week intervention.
participant completed a graded series of sub-maximal treadmill tasks in which expired air was
collected and analyzed. Heart rate via ECG was recorded at the end of each stage. Aerobic
capacity (VO2 max) was predicted from the regression equation for the relationship between
submaximal VO2 and heart rate at two or more submaximal treadmill tasks (ACSM, 2000).
Experience of Social Support 21
Physical activity was assessed using the Paffenbarger Physical Activity Questionaire
(PPAQ). The PPAQ quantifies physical activity in terms of the number of calories spent per
week during sport, leisure, and recreational activities. Physical activity was calculated from
Dietary assessment. Participants recorded food and beverage intake over a period of four
days (two weekdays, two weekend days) at baseline and following the BWLP. Estimates for
total calories were derived using Nutribase 2001 Professional Nutrition software (Phoenix,
Arizona). Average daily caloric intake at pretest and posttest were calculated.
Modified versions of the twenty-three item Sallis Social Support for Eating & Exercise
Behaviors (Sallis et al., 1987) scale were used to assess participants’ receipt of support from
friends and family (Appendices A and B) and from significant other/spouse (Appendices C and
D). The Social Support for Diet and Exercise Behaviors Scales was originally developed and
validated for heart-healthy living (e.g., low fat, low sodium). For the current study, the surveys
were modified, slightly, to be appropriate for weight control (Wing & Jeffrey, 1999). A measure
of support appraisal (i.e., “helpfulness”) was also added to the existing scales.
Frequency and appraisals of helpfulness were assessed for each support item on the scale.
Participants rated “how often” their supportive other(s) provided the indicated support using a 5
point likert scale (e.g., 1 = never, 5 = very often; 8 = does not apply). The Sallis scales assess
support for two domains of behavior (eating and exercise). The following were scored 1 : 1)
positive support for eating (5 items), 2) negative support for eating (5 items), 3) exercise
participation (10 items). In the current study, the relationship of “support for eating” to
1
An optional three-item subscale, rewards and punishments for exercise, will not be scored or included in the
analyses. This scale had equivocal factor loadings in most validation samples.
Experience of Social Support 22
treatment-outcomes was evaluated using a “support for eating” factor score created by summing
the positive and negative items (reverse scored). Additionally, a composite “overall positive
support frequency” score was calculated by summing the eating support factor and the exercise
support factors
Participants also rated “how helpful/unhelpful” they felt each support item was (e.g., 1 =
very unhelpful; 5 = very helpful; 8 = does not apply). As noted in the introduction, helpfulness
ratings serve as an evaluation of the experience of the problem solving utility (Goldsmith et. al.,
2000) of a supportive gesture. Helpfulness factors analogous to those scored for support
frequency were calculated; however, the average (not the sum) was used. A composite “overall
positive support appraisal” score was calculated by averaging all items factors together. It is
important to note that the instructions for this scale asked participants to make their ratings of
Appendices A- D). Unfortunately, the presence of both a “neutral” helpfulness rating (i.e., three)
and a “does not apply” rating apparently confused participants; this seemed to be particularly
true when the frequency rating was “never.” It appears as though, instead of (correctly) rating
support items that were not received (frequency = never) as helpful or unhelpful in some way,
many participants chose the “does not apply rating.” Even more problematic is the fact that
several participants responded, “Does not apply” to each of the appraisal items – thus
significantly reducing the number of usable ratings. Helpfulness ratings of “does not apply”
Data Analysis
physiological outcomes and social support. T-tests and Chi-square analyses were used to
compare the demographics of those BWLP participants who also participated in the social
support survey study, with those who declined. ANOVA and correlational analyses examined
the association between demographic characteristics (i.e. age, gender, race, income, education,
and number of children living at home) and indices of social support. Significant relationships
Scores (i.e., both composite and factors) of frequency and appraisal of social support for
eating and exercise are the primary independent variables of interest. Five behavioral and
measures include: caloric intake (i.e., index of eating behavior), and the Paffenbarger Physical
Activity Questionaire (i.e., index of physical activity). Physiological outcome measures include:
submaximal oxygen consumption (i.e., index of cardiorespiratory fitness), body weight, and
Prior to hypothesis testing, correlation matrices including the intercorrelations among all
test the hypotheses that greater overall (i.e., composite score) frequency (hypothesis 1) and
appraisal (hypothesis 2) of social support would predict each of the behavioral and physiological
2
multiple regression were conducted, controlling for baseline values when possible (i.e., caloric intake, percent
body fat, cardiorespiratory fitness); bivariate analyses were used for dependent measures based on averages (i.e.,
percent weight loss and physical activity)
Experience of Social Support 24
outcomes. Analogous correlations were conducted for support from friends and family and
Hypotheses 3 & 4. Hypotheses three and four guide examination of the relationship of
support factor scores and behavioral outcomes for evidence of specificity. Hierarchical
regression analysis were used to assess eating support specificity; whether support for eating
predicts caloric intake (controlling for baseline caloric intake) beyond support for exercise.
Exercise support was entered at step one and eating support was entered at step two. Similarly,
hierarchical regression analyses were used to assess exercise support specificity; whether support
for exercise predicts physical activity outcomes (i.e., physical activity and cardiorespiratory
fitness; controlling for baseline cardiorespiratory fitness) beyond support for eating. Eating
support was entered at step one and exercise support was entered at step two. The above
analyses were conducted separately for support from friends and family and support from
spouse.
hypotheses that appraisals of support helpfulness would predict outcomes beyond support
frequency. Separate analyses were performed for the composite and factor-specific (e.g., eating,
exercise) support scores. For each equation, the support frequency score was entered in the first
step and the appraisal score entered in the second step. The above analyses were conducted
separately for support from friends and family and support from spouse.
Hypothesis 7. For married participants, comparisons between support sources were also
proposed. In the subset of married participants, separate hierarchical regression analyses were
used to evaluate the hypotheses that frequency and appraisal of spousal support (composite
scores) would predict each of the behavioral and physiological outcome variables. For each
Experience of Social Support 25
equation, the support from friends and family measure (i.e., appraisal or frequency) was entered
in the first step and the parallel spousal support measure was entered in the second step.
SPSS (Version 13.0) was used to conduct all statistical analyses. Alpha was set at .05,
Results
Descriptive Statistics
(BWLP) participants consented to participate in the adjunct survey study of social support. Nine
of the original 54 BWLP were lost to follow-up; six of these included social support survey
participants. Thus, complete data 3 available for thirty-two of the forty-five BWLP completers
(71.11 %). There was a non-significant trend such that those who completed the BWLP were
also somewhat more likely to complete the social support surveys, χ2(1, N = 51) = 3.06, p = .147.
All of the participants were eligible, and thirty-five participants completed surveys assessing
social support received from family or friends; the one individual who did not complete this
survey did complete the spousal support survey. Of these, 24 participants were either married
(23) or dating (1) and completed surveys assessing social support received from their significant
other. The average age of survey participants was 49.03 (SD = 9.76) years. 78.9 % of the
participants were female, and 87.9% of participants were Caucasian. In addition, 76.9% had a
college degree or higher and 36.7% had an income of greater than 45,000 dollars per year. The
mean baseline body mass index (BMI) of the participants was 41.89 (SD = 7.36; See Table 1).
Survey participants vs. non participants. T-tests and Chi Square analyses were used to
compare the demographics of those BWLP participants who also participated in the social
support survey study, with those who declined. The analyses indicated that there were no
significant differences between the two groups with respect to age, t(40) = .501, p = .614,
gender, χ2(1, N = 51) = 3.246, p = .096, race, χ2(3, N = 43) = 1.336, p = .720, income χ2(5, N =
3
Pretreatment, survey, posttreatment data is available for 32 of the original 38 consenting participants. Pretreatment
and survey data are available for 36 of the original 38; the two with missing data, dropped from the BWLP before
returning the survey.
Experience of Social Support 27
40) = 2.716, p = .740, education χ2(2, N = 36) = .189, p = .910, or employment status, χ2(4, N =
Demographics & social support. ANOVA and correlation analyses were used to
examine the association between demographic characteristics and indices of overall frequency
and helpfulness of support from family/friends and from significant other. Ratings of the overall
frequency of family/friend support did not differ significantly according to gender F(1, 34) =
.036, p = .852, race, F(3,27) = 1.465, p = .246, marital status, F(3,30) = .668, p = .578, income
F(5, 24) = 1.041, p = .416, education F(3, 21) = .836, p = .489, employment status F(4,24) =
.926, p = .465, or baseline Body Mass Index (BMI) r(36) = -.194, p = .257. The overall
frequency of family/friend support did differ significantly by participant age r(30) = -.380, p =
.038, such that older participants tended to report lower overall frequency of support from
Ratings of the average overall helpfulness of family/friend support did not differ
significantly according to gender, F(1, 31) = .047, p = .830, race, F(3, 24) = .879, p = .466,
marital status, F(3, 27) = .971, p = .421, income, F(5, 21) = .938, p = .477, education, F(3,19) =
.514, p = .678, employment status F(4, 23) = .211, p = .930 or age r(27) = -.062, p = .758.
significantly by participant’s baseline BMI r(33) = -.361, p = .039, such that participant’s with
higher baseline BMI, had lower support helpfulness ratings. Baseline BMI was controlled for in
hypothesis testing.
Ratings of the overall (eating and exercise combined) frequency of support received from
significant other did not differ according to gender, F(1, 22) = 2.844, p = .106, race, F(2,18) =
1.023, p = .380, marital status, F(2, 21) = 1.092, p = .354, income, F(4, 14) = 3.558, p = .114
Experience of Social Support 28
education, F(3, 14) = .898, p = .466, employment status, F(4, 15) = 2.279, p = .109, BMI, r(24) =
-.394, p = .057 or age, r(20) = .348, p = .133. Similarly, appraisals of the average overall
helpfulness of support received from significant other did not differ according to gender, F(1,11)
= .739, p = .408, race, F(2, 8) = 2.116, p = .183, marital status, F(2, 10) = .058, p = .944, income,
F(3, 6) = .074, p = .972, education, F(2, 6) = 1.043, p = .409, employment status, F(3, 7) = .400,
including the intercorrelations among the pairs of predictors of family/friend support (see Table
2), and significant other support (see Table 3). As expected, factor scores tended to be
significantly correlated with composite scores. For example, total frequency support for eating
(Frequency Eat Support Factor), r(36) = .716, p = .000, was highly correlated with the overall
frequency of support for eating and exercise (Frequency Support Composite), r(36) = .691, p =
.000; the later scale included the score of the former scale. Importantly, the correlations between
the pairs of frequency and appraisal support indices do not yield dramatically high correlations.
The overall frequency of support from family/friends was moderately correlated with the overall
appraisals of family/friend support helpfulness from r(33) = .341, p = .052. The overall
frequency of support from significant other was minimally correlated with overall appraisals of
hypotheses (i.e., hypotheses 1a and 1b) that greater overall (i.e., eating and exercise combined)
support frequency would predict positive behavioral and physiological outcomes. Surprisingly,
4
For hypothesis 1a, controlled for age and baseline health indicators. For hypothesis 1b, controlled for baseline
health indicators.
Experience of Social Support 29
neither greater overall frequency of support from family and friends, nor from spouse, was a
cardiorespiratory fitness (VO2max), body fat, or percent weight loss (See Table 4) at post-
treatment. The only significant predictor of behavioral and physiological outcome values in these
equations was the corresponding baseline value of the behavioral or physiological indicator.
hypotheses (i.e., hypotheses 2a and 2b) that appraisals of greater overall (i.e., eating and exercise
combined) support helpfulness would predict positive behavioral and physiological outcomes.
Contrary to prediction, positive appraisals of family and friend (not including spouse) support
helpfulness, failed to predict positive changes in caloric intake, physical activity (PPAQ),
cardiorespiratory fitness (VO2max), body fat, or percent weight loss at post treatment (See Table
4). The only significant predictor of behavioral and physiological outcome values in these
equations was the corresponding baseline value of the behavioral or physiological indicator.
fitness (VO2max), and lower percent weight loss. Appraisals of greater overall spousal support
helpfulness were not significantly related to energy expenditure through physical activity
(PPAQ) or body fat. With respect to the hypotheses related to appraisals of spousal support, it is
important to note that the sample sizes for these analyses are especially low, in part due to the
response pattern (i.e., exclusive use of the “does not apply” option) described in the data analysis
5
For hypothesis 2a, controlled for baseline BMI and baseline health indicators. For hypothesis 2b, controlled for
baseline health indicators.
Experience of Social Support 30
section. The analyses reported here reflect sample sizes of less than fourteen participants and
hypotheses that the eating support and exercise support frequency factors, would demonstrate
evidence of specificity in predicting behavioral outcomes. The frequency factor being controlled
for was entered into step one; the factor specific to the behavioral outcome was entered into step
two. The hypothesis that eating specific support (e.g., eating factor) from family and friends
would significantly improve the prediction of caloric intake beyond exercise support (e.g.,
exercise factor) was not supported. Surprisingly, the frequency of family and friend support
directed at eating behavior did not predict caloric intake after controlling for support for exercise,
ΔF(1,20) = .293, p >.05. By contrast, eating specific spousal support did predict caloric intake
prediction. Specifically, greater frequency spousal support for eating change predicted poorer
caloric reduction at post-treatment, after removing the influence of spousal support targeting
exercise behavior change, ΔF(1,15) = 3.11, p < .05 (see Table 5). The adjusted R2 value
Separate analyses were conducted to test hypotheses that frequency ratings of the
exercise specific support received from family and friends, and from spouses, would
significantly improve the prediction of the exercise related outcomes beyond the frequency of
eating specific support. Neither set of predictions was supported for either of the support
6
Each of the analyses controlled for pretreatment levels of the outcome variable of interest. In cases were a
significant F was indicated, the only significant beta weight was that of the pretreatment control variable, unless
otherwise indicated. That is, except in the one instance noted, the frequency support factors were not significant
predictors in either step of the hierarchical model.
Experience of Social Support 31
sources. Specifically, ratings of higher frequency family/friend support for engaging in exercise,
did not significantly predict increases in either physical activity (PPAQ), ΔF(1,20) = .047, p >
.05, or cardiorespiratory fitness (VO2max), ΔF(1,23) = .001, p > .05, after controlling for
frequency of family and friend support directed at eating behaviors. Similarly, greater frequency
spousal support for engaging in exercise, failed to predict increases in either physical activity
(PPAQ), ΔF(1,13) = .136, p > .05, or cardiorespiratory fitness (VO2max), ΔF(1,17) = .901, p >
.05, after controlling for frequency of spousal support directed at eating behaviors (see Table 5).
hypotheses that the family and friend eating support and exercise support appraisal factors,
factor being controlled for was entered into step one; the factor specific to the behavioral
outcome was entered into step two. None of the hypotheses related to the specificity of appraisals
of support from family and friends were supported. That is, eating specific support appraisals
failed to improve prediction of caloric intake beyond appraisals of exercise support ΔF(1,17) =
1.961, p > .05. Similarly, appraisals of exercise specific support failed to improve prediction of
post treatment physical activity (PPAQ), ΔF(1,17) = .015, p > .05, or cardiorespiratory fitness
(VO2max), ΔF(1,20) = 2.688, p > .05, after controlling for appraisals of family and friend eating
There were insufficient 8 data to test the appraisal specificity hypothesis in the context of
the spousal relationship. As previously noted, the sample to test hypotheses related to appraisals
of spousal support was especially low. In the case of hypothesis tests specifying hierarchical
7
Each of the analyses controlled for pretreatment levels of the outcome variable of interest. In cases where a
significant F was detected, the only significant beta weight was that of the pretreatment control variable, unless
otherwise indicated.
8
n=9
Experience of Social Support 32
analyses, a ‘cut-off’ was identified such that samples of fewer than 15 participants were judged
Overall resource – need fit. Hierarchical regression analyses 9 were used to test the
hypotheses (i.e., hypothesis 5a) that positive appraisals of the overall (i.e., eating and exercise
composite) helpfulness of support received from family and friends (not including spouse) would
predict positive behavioral (see Table 6) and physiological (see Table7) outcomes beyond the
parallel (i.e., family and friend) ratings of overall frequency of support. The hypotheses were not
supported. Higher average appraisals of family and friends’ support helpfulness did not
.249, p > .05, physical activity (PPAQ), ΔF(1,16) = .355, p > .05, cardiorespiratory fitness
(VO2max), ΔF(1,17) = .679, p > .05, body fat, ΔF(1,18) = .075, p > .05, or percent weight loss,
ΔF(1,20) = .376, p > .05, beyond composite frequency ratings of family and friend support.
There were insufficient data 10 to evaluate the related hypothesis (i.e., hypothesis 5b) that
positive appraisals of the overall helpfulness of spousal support would predict behavioral and
Specificity in resource – need fit. It was hypothesized (hypothesis 6a) that positive
appraisals of family and friend eating specific support (i.e., appraisal eating factor) and exercise
specific support (i.e., appraisal exercise factor) would predict positive outcomes in eating and
physical activity respectively, beyond the parallel family and friend frequency factor (i.e,
frequency eating factor, frequency exercise factor) and baseline control variables (e.g, pre-
treatment caloric intake). In the hierarchical analyses conducted to examine this hypothesis, the
9
Each of the analyses controlled for pretreatment levels of the outcome variable of interest. In cases where a
significant F was detected, the only significant beta weight was that of the pretreatment control variable.
10
n < 13
Experience of Social Support 33
frequency factor being controlled for was entered into step one; the parallel appraisal factor was
Cardiorespiratory fitness (VO2 Max) was the only outcome variable for which specificity
in resource – need fit was supported. That is, positive helpfulness appraisals of the exercise
specific support received from family and friends improved the prediction of VO2 max beyond
the frequency ratings of such support, ΔF (1,22) = 3.504, p < .05. Positive helpfulness appraisals
of eating specific support from family and friends failed to significantly improve prediction of
reductions in post-treatment caloric intake beyond ratings of the frequency of family and friend
eating support, ΔF(1,18) = .258, p >.05. Similarly, positive helpfulness appraisals of the exercise
support received from family and friends failed to improve prediction of post treatment increases
in physical activity (PPAQ) beyond ratings of the frequency of family and friend exercise
There were insufficient data 11 to evaluate the related hypothesis (i.e., hypothesis 6b) that
more positive appraisals of spousal eating and exercise specific support would predict positive
eating and physical activity outcomes, beyond the parallel frequency of spousal support factors.
Hierarchical regression analyses 12 were used to test the hypotheses (hypothesis 7) that,
for the subset of married participants, spousal support (i.e., frequency and appraisal composites)
would better predict behavioral (e.g., eating and exercise) and physiological (e.g., weight, body
fat) outcomes than would support from family and friends. The family/friend support composite
to be controlled for was entered into step one; the analogous spousal support composite was
entered into step two. Overall ratings (i.e., eating + exercise composite) of spousal support
11
n <13
12
Each of the analyses controlled for pretreatment levels of the outcome variable of interest
Experience of Social Support 34
frequency did not significantly improve prediction of post-treatment caloric intake, ΔF(1,13) =
.266, p > .05, physical activity (PPAQ), ΔF(1,11) = .208, p > .05, cardiorespiratory fitness,
ΔF(1,15) = .035, p > .05, body fat ΔF(1,16) = .103, p >.05, or percent weight loss ΔF(1,18) =
.028, p > .05, beyond the family and friend support frequency composite (see Table 9). There
were insufficient data 13 to evaluate the related hypothesis (i.e., hypothesis 7b) that more positive
appraisals of overall spousal support helpfulness would predict positive eating and physical
activity outcomes, beyond the analogous appraisals of family and friend helpfulness.
13
n < 13
Experience of Social Support 35
Discussion
The current study evaluated the occurrence (i.e., frequency) and experience (i.e.,
BWLP participants’ behavioral and health outcomes. The primary goals were to examine the
support – behavior change relationship for evidence of specificity and to evaluate the utility of
measuring support appraisals as a tool for identifying resource-need match. Within these
objectives, another aim of the study was to identify potentially distinct contributions of different
sources of support. Hypotheses testing returned largely null results. This study’s null results
may be due to small sample and related lack of statistical power. For example, other studies
using social support as a predictor of dietary (e.g., to increase fruit and vegetable consumption)
or physical activity change have samples ranging from 200 – 1,300 participants (Eyler et. al.,
2002; Kelsey et. al., 1996, Sorensen, Hunt, Cohen, Stoddard, Stein, Phillips et. al., 1998; Steptoe,
Perkins-Porras, Rink, Hilton, & Cappucio,1998). Additional considerations are outlined in the
discussion below.
In the present study, higher support frequency failed to predict positive changes in any of
the behavioral or health indices. This was true regardless of the support source. Other studies
assessing BWLP participants’ reports of the frequency of support received prior to the initiation
of the intervention (Wing 1991; Wing & Jeffrey 1999) demonstrate a similar lack of predictive
utility. The notion that these earlier studies might have assessed for naturally occurring support
too early, before participants’ were attempting to make behavior changes, informed the design of
the current study. Thus, the current investigation assessed social support three-weeks following
initiation of the intervention, reasoning that this would be sufficient time for participants to
Experience of Social Support 36
become actively engaged in behavior change efforts. However, the behavior changes necessary
for weight reduction efforts may simply require more time for participants to learn and manifest.
It is plausible that the null results may be an indication that not enough time was allowed for
participants to fully prepare for and engage (Prochaska, et. al., 1992) eating and exercise
changes, and thus, the frequency of support for such change would be similarly nascent (Cohen
& Willis, 1995). With few change efforts underway, support persons external to the intervention
context, may not have had the cues necessary to acquire and/or engage the appropriate support
action, and maintaining change (Prochaska, et. al., 1992), support for such change is likely to be
dynamic as well – adjusting to these changing circumstances. Additionally, it may be the case
that the resources offered through social support, are particularly important to specific stages of
change. Support for this is suggested, at least in the case of exercise behavior change, by a
recent study that found utilization of supportive relationships to be a significant factor associated
with the progression from action into maintenance (Lowther, Mutrie, & Scott, 2007). Taken
together, assessing a single time point and asking participants to, in effect, mentally average their
experience of this dynamic process over the prior month may ultimately be an inappropriate
method. Another approach, such as ecological momentary assessment (Stone & Shiffman,
1994), that would allow for the real-time recording of instances and appraisals of support may
provide the flexibility needed to capture the influence of this important social exchange.
Another goal of this study was to examine the relationship between the behavioral
content of supportive behaviors (e.g., sharing in healthy food choices) with behavior change
Experience of Social Support 37
goals (e.g., caloric reduction) for evidence of specificity. These comparisons overwhelmingly
failed to support the notion that support behaviors targeting behavior change outcomes are
uniquely effective in producing such outcomes. The only significant finding was spousal support
specific to eating – and this was in the opposite direction of prediction. These hypotheses, if
supported, may provide potentially useful information for more focused and effective
involvement of supportive others in behavior change efforts. Alternatively, both eating and
exercise behavior changes are necessary to successful weight loss; a point regularly
communicated during the BWLP. This context, which reinforces simultaneous negotiation of
both change efforts, may render evaluations of support specificity insensitive and/or irrelevant.
Given the small sample of the current study, it would be premature to disregard this as an area
One aim of the current study was to evaluate the utility of helpfulness-ratings as a method
for detecting “resource-need fit,” and, whether this might be superior to pure frequency
measures. In this study, appraisals of family and friend helpfulness were no more informative
than frequency ratings in predicting health outcomes; neither were significant predictors of
health outcomes. It was suggested earlier that the current findings might reflect a situation in
which the behavior changes necessary for successful weight loss were still being learned and
developed. If true, such poorly timed support assessment would also plausibly affect appraisals
of the problem-solving utility (i.e., helpfulness) of support offered at this early stage of change.
For instance, support for behavior changes not yet underway may go unnoticed or be appraised
barrier to change (Jacobson, 1986). Given the null findings in the current study, it would seem
Experience of Social Support 38
that the former is most likely. That is, participants appear to have found the support that they
recognized receiving from family and friends during the first few weeks of the intervention to be
rather inconsequential to whatever behavior change efforts they were beginning to initiate.
Results related to spousal support helpfulness are a bit more difficult to understand.
Despite the early timing of support assessment, overall (i.e., composite) spousal support
appraisal was a significant predictor of some (but not all) outcomes in this study in which
spousal support frequency – outcome relationships were all non-significant. However, the
direction of this relationship was opposite of prediction. Higher average ratings of the
helpfulness of spousal support received during the first weeks of the BWLP were related to
poorer eating (caloric intake), fitness (VO2max), and weight loss outcomes. It is possible that
respect to the behavior changes being initiated. Such a relationship would be consistent with
stress-buffering models of support that identify the role of support in diffusing the experience of
stress (Cohen & Wills, 1985). For instance, it may be that the BWLP strugglers recognized their
need for support and thus viewed support provision positively; support was appreciated, helpful
those adapting more easily to behavior changes may have rated received support more negatively
would suggest that the potential for using helpfulness ratings as an indication of “resource-need
match” may be promising, but must be viewed differently than originally proposed.
outcome is guarded given the small sample used in these analyses. Further, given the small
sample, hierarchical regression to evaluate the utility of spousal support helpfulness appraisal in
Experience of Social Support 39
detecting married participant’s experience of ‘resource – need fit,’ beyond support frequency was
not warranted. Thus, the nature of the appraisal - outcome relationship after controlling for the
treatment outcome remains not well understood. In addition to clarifying whether spousal
support appraisal makes unique contribution to outcome prediction, future work should also
attend to other factors that may influence the perceptions of support frequency and helpfulness.
For instance whether, as just proposed, the experience of the behavior change process (e.g.,
to better understand the influence of marital quality on the supportive process (e.g., both
frequency and appraisals) as it relates to behavior change efforts. Marital researchers, highlight
the influence of global marital satisfaction on spousal support attributions (Bradbury & Fincham,
1992). It is possible that helpfulness appraisals simply reflect the overall quality or satisfaction
with the marriage rather than a more fine-grained analysis of the actual utility of the support
behaviors (Gurung, Sarason, & Sarason, 1997). There has been limited study of this relationship
While a number of study-specific limitations have been introduced, there are additional
challenges that should be considered in the design of future investigations of support for dietary
and exercise behavior change. Despite the behavioral focus of most items on the Sallis support
measure used in the current study, it was not developed or tested for use in predicting behavior
change in those attempting weight loss. Thus, it is possible that the support measures simply did
not assess the support behaviors most important to facilitating or blocking dietary and exercise
behavior change in this population. Such a situation would return little evidence of a support –
Experience of Social Support 40
behavior change relationship, as was the case in this study. Even with a thoroughly appropriate
set of support measures, it may be the case that support is just not particularly influential in
making these behavior changes. It is also possible that more salient factors dwarf any influence
of social support that does exist. As the obesity epidemic continues to grow, weight loss
interventions continue to achieve only marginal and short-term success. Researchers have been
calling more attention to environmental factors, such as easy access and regular exposure to
nutrient dense food and labor-saving devices, as especially potent contributors to the obesity
epidemic (French S., Story M., & Jeffrey R., 2001; Swinburn, Egger, Raza, 1999). Until public
policy catches up with public health needs, it is possible that factors such as social support will
continue to matter little in the overall success of individuals attempting to make these health
References
American College of Sports Medicine. (2000). ACSM's guidelines for exercise testing and prescription.
Baumann (Eds.), The meaning and measurement of social support (pp. 13-31). New
York: Hemisphere.
Black, D., Gleser, L., & Kooyers, K. (1990). A meta-analytic evaluation of couples weight-loss
Bradbury T. & Fincham. F. (1992). Attributions and behavior in marital interactions. Journal of
Brownell K. (2004). The LEARN Program for weight management 10th Edition. Dallas, Texas:
Carels,R., Cacciapaglia, H., Douglass, O., Rydin, S., & O’Brien. (2003). The early identification
of poor treatment outcome in a women’s weight loss program. Eating Behaviors, 4(3),
265-82.
Cohen, S. (1988). Psychosocial models of the role of social support in the etiology of physical
Cohen, J., Cohen, P., West, S., & Aiken, L. (2003). Applied Multiple Regression/Correlation
Analysis for the Behavioral Sciences (3rd ed.). Mahwah, New Jersey: Lawrence Erlbaum
Associates.
Cohen, S., & Wills, T. (1985). Stress, social support, and the buffering hypothesis.
Cohen, S., & Lichtenstein, E. (1990). Partner behaviors that support quitting smoking. Journal
Experience of Social Support 42
DiMatteo, M. (2004). Social support and patient adherence to medical treatment: a meta-
Elfhag, K., & Rossner, S. (2005). Who succeeds in maintaining weight loss? A conceptual
review of factors associated with weight loss maintenance and weight regain. Obesity
reviews : an official journal of the International Association for the Study of Obesity,
6(1), 67-85.
Eyler, A., Wilcox, S., Matson-Koffman, D., Evenson, K., Sanderson, B., Thompson, J. et al.
(2002). Correlates of physical activity among women from diverse racial/ethnic groups.
French S., Story M., & Jeffrey R. (2001). Environmental influences on eating and physical activity.
Goldsmith, D., McDermott, V., & Alexander, S. (2000). Helpful, supportive and sensitive:
Baumann (Eds.), The meaning and measurement of social support (pp. 293 – 307). New
York: Hemisphere.
Gurung, R., Sarason, B., & Sarason, I. (1997). Personal characteristics, relationship quality, and
social support perceptions and behavior in young adult romantic relationships. Personal
Harris, T. (1992). Some reflections on the process of social support and nature of unsupportive
Experience of Social Support 43
behaviors. In H.F.Veiel & U. Baumann (Eds.), The meaning and measurement of social
Hogan, B., Linden, W., & Najarian, B. (2002). Social support interventions: Do they work.
House. (1987). Social support and social structure. Sociological Forum, 2(1), 135.
House, J., Umberson, D., Landis, K. (1988). Structures and Processes of Social Support.
Hutchison, C. (1999). Social support: factors to consider when designing studies that measure
Jacobson, D. (1986). Types and timing of social support. Journal of Health and Social
Jeffery R., Drewnowski A., Epstein L., Stunkard A., Wilson T. & Wing R. (2000) Long-term
maintenance of weight loss: Current status. Health Psychology 19(1 Suppl), 5-16.
Kelsey, K., Kirkley, B., De Vellis, R., Earp, J., Ammerman, A., Keyserling, T., Shannon, J., &
Kiecolt-Glaser, J., & Newton, T. (2001). Marriage and health: his and hers. Psychological
Lowe M. (2003). Self-regulation of energy intake in the prevention and treatment of obesity: Is it
Lowther, M., Mutrie, N., & Scott, M. (2007). Identifying Key Processes of Exercise Behavior
Change Associated with Movement through the Stages of Exercise Behavior Change.
Martin, R., Davis, G., Baron, R., Suls, J., & Blanchard, E. (1994). Specificity in social
support: perceptions of helpful and unhelpful provider behaviors among irritable bowel
Martire, L., Stephens, M., Druley, J., & Wojno, W. (2002). Negative reactions to received
21(2), 167-76.
McLean, N., Griffin, S., Toney, K., & Hardeman, W. (2003). Family involvement in weight
NHLBI Obesity Education Initiative Task Force Members. (1998). Clinical guidelines on the
identification, evaluation, and treatment of overweight and obesity in adults: The evidence
Ogden C., Carroll M., Curtin L., McDowell M., Tabak C. & Flegal K. (2006). Prevalence of
Pi-Sunyer F. (2003). A review of long-term studies evaluating the efficacy of weight loss in
Prochaska, J., DiClemente, C. & Norcross, J. (1992) In search of how people change:
Reynolds, J., & Perrin, N. (2004). Mismatches in social support and psychosocial adjustment
Rook, K. (1984). The negative side of social interaction: impact on psychological well-being.
literature. In H.F.Veiel & U. Baumann (Eds.), The meaning and measurement of social
Roski, J., Schmid, L., & Lando, H. (1996). Long-term associations of helpful and harmful
Sallis, J., Grossman, R., Pinski, R., Patterson, T. & Nader, P. (1987). The development of
scales to measure social support for diet and exercise behaviors. Preventative Medicine,
Sarason, B., Shearin,E., Pierce, G. & Sarason, I. (1987). Interrelations of social support
Schwarzer & Gutierrez-Dona. (2005). More Spousal Support for Men Than for Women: A
Simoni, J., Frick, P., & Huang, B. (2006). A longitudinal evaluation of a social support model of
Sorensen, G., Hunt, M., Cohen, N., Stoddard, A., Stein, E., Phillips, J. et al. (1998).
Worksite and family education for dietary change: the Treatwell 5-a-Day program.
Steptoe, A., Perkins-Porras, L., Rink, E., Hilton, S., & Cappuccio, F. (2004). Psychological and
social predictors of changes in fruit and vegetable consumption over 12 months following
Stone, A., & Shiffman, S. (1994). Ecological momentary assessment (EMA) in behavorial
Experience of Social Support 46
Swinburn, B., Egger, G., Raza, F. (1999). Dissecting obesogenic environments: the development
Utter A., Nieman D., Ward A., & Butterworth D. (1999). Use of the leg-to-leg bioelectrical impedance
method in assessing body composition change in obese women. American Journal of Clinical
Vaux., A. (1992). Assessment of social support. In H.F.Veiel & U. Baumann (Eds.), The
meaning and measurement of social support (pp. 193 - 214). New York: Hemisphere.
Walitzer, K., & Dermen, K. (2004). Alcohol-focused spouse involvement and behavioral couples
Wills, T., & Fegan, M. (2001). Social networks and social support. In A. Baum, T., Revenson &
Wing, R., & Jeffery, R. (1999). Benefits of recruiting participants with friends and increasing
social support for weight loss and maintenance. Journal of Consulting and Clinical
Wing, R., Marcus, M., Epstein, L., & Jawad, A. (1991). A"family-based"approach to the
Appendix A
Below is a list of things family and friends (other than spouse/significant other) might do or say to someone who is trying to
improve their eating habits. We are interested in high fat and high calorie foods. If you are not trying to make any of these dietary
changes, then some of the questions may not apply to you, but please read and give an answer to every question.
Please give two ratings for each question in the columns provided next to the questions using the relevant rating scales given
below. In the first column, use the “how often” scale provided to rate how often your family and friends (NOT including
spouse/significant other) have said or done what is described during the last month. In the second column, use the
“helpful/unhelpful” scale to rate how helpful you felt this support was to you.
Please write one number from the appropriate rating scales provided below in each space.
SAMPLE:
A. If my family and friends rarely make fun of the foods I eat,
and this is very helpful, I would answer like this: How often Helpful/Unhelpful
How Often
a few very does not
never rarely times often often apply
_________________________________________________________________________
1 2 3 4 5 8
Helpful/Unhelpful
very somewhat somewhat very does not
unhelpful unhelpful neutral helpful helpful apply
__________________________________________________________________________
1 2 3 4 5 8
During the past month, my family & friends (NOT including spouse): How Often Helpful/Unhelpful
1. Encouraged me not to eat “unhealthy” foods (cake, chips) when I’m tempted to do
so ______ ______
2. Discussed my eating habit changes with me (asked how me how I’m doing with my
eating changes) ______ ______
3. Reminded me not to eat high fat, high calorie foods
______ ______
4. Complimented me on changing my eating habits (“keep it up”, “I’m proud of you”)
______ ______
5. Commented if I went back to my old eating habits ______ ______
______ ______
8. Brought home foods I’m trying not to eat
______ ______
9. Got angry when I encouraged them to eat low calorie, low fat foods
______ ______
10. Offered me food I’m trying not to eat ______ ______
Note:
Positive Items 1 – 5
Negative Items (reverse score) 6 – 10
Experience of Social Support 48
Appendix B
Below is a list of things family and friends (other than spouse/significant other) might do or say to someone who is trying to
exercise regularly. If you are not trying to exercise, then some of the questions may not apply to you, but please read and give an
answer to every question.
Please give two ratings for each question in the columns provided next to the questions using the relevant rating scales given below.
In the first column, use the “how often” scale provided to rate how often your family or friends (NOT including spouse/significant
other) have said or done what is described during the last month. In the second column, use the “helpful/unhelpful” scale to rate
how helpful you felt this support was to you.
Please write one number from the appropriate rating scales provided below in each space.
How Often
a few very does not
never rarely times often often apply
_________________________________________________________________________
1 2 3 4 5 8
Helpful/Unhelpful
very somewhat somewhat very does not
unhelpful unhelpful helpful helpful apply
__________________________________________________________________________
1 2 3 4 5 8
During the past month, family & friends (NOT including spouse) : How Often Helpful/Unhelpful
3. Gave me reminders to exercise (“Are you going to exercise tonight”) ______ ______
12. Asked me for ideas on how they can get more exercise ______ ______
13. Talked about how much they like to exercise ______ ______
Experience of Social Support 49
Appendix C
Below is a list of things a spouse or significant other might do or say to their partner who is trying to improve their eating habits.
We are interested in high fat and high calorie foods. If you are not trying to make any of these dietary changes, then some of the
questions may not apply to you, but please read and give an answer to every question.
Please give two ratings for each question in the columns provided next to the questions using the relevant rating scales given
below. In the first column, use the “how often” scale provided to rate how often your spouse or significant other has said or done
what is described during the last month. In the second column, use the “helpful/unhelpful” scale to rate how helpful you felt this
support was to you.
Please write one number from the appropriate rating scales provided below in each space.
SAMPLE:
A. If my spouse/significant other rarely makes fun of the foods I eat,
and this is very helpful, I would answer like this: How often Helpful/Unhelpul
How Often
a few very does not
never rarely times often often apply
_________________________________________________________________________
1 2 3 4 5 8
Helpful/Unhelpful
very somewhat somewhat very does not
unhelpful unhelpful neutral helpful helpful apply
__________________________________________________________________________
1 2 3 4 5 8
During the past month, my spouse or significant other: How Often Helpful/Unhelpful
1. Encouraged me not to eat “unhealthy” foods (cake, chips) when I’m tempted to do so
______ ______
2. Discussed my eating habit changes with me (asked how me how I’m doing with my
eating changes) ______ ______
9. Got angry when I encouraged him/her to eat low calorie, low fat foods ______ ______
Note:
Positive Items 1 – 5
Negative Items (reverse score) 6 – 10
Experience of Social Support 50
Appendix D
Below is a list of things a spouse or significant other might do or say to their partner who is trying to exercise regularly. If you are
not trying to exercise, then some of the questions may not apply to you, but please read and give an answer to every question.
Please give two ratings for each question in the columns provided next to the questions using the relevant rating scales given
below. In the first column, use the “how often” scale provided to rate how often your spouse or significant other has said or done
what is described during the last month. In the second column, use the “helpful/unhelpful” scale to rate how helpful you felt this
support was to you.
Please write one number from the appropriate rating scales provided below in each space.
How Often
a few very does not
never rarely times often often apply
_________________________________________________________________________
1 2 3 4 5 8
Helpful/Unhelpful
very somewhat somewhat very does not
unhelpful unhelpful neutral helpful helpful apply
__________________________________________________________________________
1 2 3 4 5 8
During the past month, my spouse or significant other: How Often Helpful/Unhelpful
3. Gave me reminders to exercise (“Are you going to exercise tonight”) ______ ______
12. Asked me for ideas on how he/she can get more exercise ______ ______
13. Talked about how much he/she likes to exercise ______ ______
Experience of Social Support 51
Table 1.
Demographic Characteristics of Support Survey Participants vs Non-Participants
M (SD) M ( SD)
Age 49.03 (9.76) 50.90 (11.42)
Baseline BMI 41.89 (7.36) 37.62 (6.36)
Gender
Male 21.1% 0.0%
Female 78.9% 100.0%
Race
Caucasian 87.9% 100.0%
Marital Status
Single 19.4% 10.0%
Dating 0.00% 20.0%
Married 61.1% 70.0%
Separated 0.02% 0.0%
Divorced 6.7 % 0.0%
Education
No College Degree 23.1% 30.0%
College Degree 53.8 % 50.0%
Post Grad Degree 23.1% 20.00%
Income
< $15,000 9.7% 0.0%
> $75,000 16.1% 22.2%
Employment Status
Full-Time 61.3% 66.1%
Retired 10.0% 11.19%
Experience of Social Support 52
Table 2.
Frequency Appraisal
Note. an = 32, bn = 36, cn = 31, dn = 29, en = 33. *p < .05, **p < .01
Experience of Social Support 53
Table 3.
Frequency Appraisal
Table 4.
Note. a = n > 19, b = n < 14. c = analyses controlled for participant age. d = analyses controlled for baseline BMI. *p < .05; **p < .01
Experience of Social Support 55
Table 5.
Appraisal Support
Eat Factor -.324 .090 - - - -
Exercise Factor - - -.029 .452 -.250 .058
Significant Other
Frequency Support
Eat Factor .402 .049* - - - -
Exercise Factor - - .059 .359 .161 .178
Appraisal Supportc
Eat Factor - - - - - -
Exercise Factor - - - - - -
Note. a = values are for the second step of the hierarchical model; that is, the exercise factor
controlled for in step 1 and the eat factor entered in step two. b. values are for the second step of
the hierarchical model; the eat factor is controlled for in first step and the exercise factor is
entered in the step two. c = insufficient data to report. *p < .05
Experience of Social Support 56
Table 6.
Note. Analyses controlled for participant age and BMI at baseline. a = The values in the final
equation, with all predictors entered. No significant outcomes at p < .05.
Experience of Social Support 57
Table 7.
Note. Analyses controlled for participant age and BMI at baseline. a = The values in the final
equation, with all predictors entered. b = the control variable, baseline bodyfat, was a highly
significant predictor of post treatment body fat. No significant Beta weights at p < .05.
Frequency and Appraisal 58
Table 8.
Note. a = The values in the final equation, with all predictors entered. *p < .05
Frequency and Appraisal 59
Table 9.
Summary of Hierarchical Analyses for Frequency Support Variables Predicting Behavioral and Physiological Outcomes of Married
Participants.
Note. a = The values in the final equation, with all predictors entered. No significant Beta weights at p < .05