3805

Social Cognitive Correlates of Physical Activity Across 12 Months in Cohort Samples of Couples Without Children, Expecting Their First Child,

and Expecting Their Second Child

Ryan E. Rhodes University of Victoria

Chris M. Blanchard Dalhousie University

Cecilia Benoit University of Victoria

Ryna Levy-Milne BC Cancer Agency, Vancouver, Canada

Patti-Jean Naylor University of Victoria

Danielle Symons Downs The Pennsylvania State University

Darren E. R. Warburton University of British Columbia

Objective: The onset of parenthood has been identified as a critical risk period for physical inactivity, yet limited research has examined the correlates of physical activity (PA) using theoretical models in longitudinal designs with comparison groups of couples without children. The purpose of this study was to predict PA across 12 months among cohorts of couples with and without children using the theory of planned behavior (TPB). Method: Participants were 314 adults (102 not expecting a child, 136 expecting first child, and 76 expecting second child) who completed baseline demographics, measures of the TPB, and 7-day accelerometry, followed by assessments at six and 12 months. Results: Hierarchical linear modeling showed some TPB relationships were moderated by parental status and gender. Most notable, time-varying covariate analyses showed perceived behavioral control and intention decreased for new mothers compared to women without children across time. PA was predicted by intention for all cohorts, and intention was predicted by affective and instrumental attitudes and perceived behavioral control for husbands and wives, whereas subjective norm predicted intention only in husbands. For wives, the relationship between intention and instrumental attitudes and perception of behavioral control varied by parental status, and was larger for couples without children compared to second-time parents. Overall, there was considerable coordination in slopes and intercepts among couples, yet individual cognitions were better predictors than partner cognitions. Conclusion: The findings provide helpful information for targeting PA interventions among young adults, and suggest that interventions for new mothers may require greater effort to raise the absolute values of control when compared to women without children.

Keywords: physical activity, theory of planned behavior, parenthood

There is irrefutable evidence that regular physical activity (PA) contributes to the primary and secondary prevention of several chronic conditions, including cardiovascular disease, diabetes,

cancer, hypertension, obesity, and osteoporosis (Warburton, Charlesworth, Ivey, Nettlefold, & Bredin, 2010; Warburton, Nicol, & Bredin, 2006). In addition to these physical health benefits, the

This article was published Online First August 5, 2013. Ryan E. Rhodes, Behavioural Medicine Laboratory, University of

Victoria, Victoria, Canada; Chris M. Blanchard, Faculty of Medicine, Dalhousie University, Halifax, Canada; Cecilia Benoit, Department of Sociology, University of Victoria, Victoria, Canada; Ryna Levy-Milne, BC Cancer Agency, Vancouver, Canada; Patti-Jean Naylor, School of Exercise Science, Physical & Health Education, University of Victoria, Victoria, Canada; Danielle Symons Downs, Department of Kinesiology, The Pennsylvania State University, State College, Pennsylvania; and Darren E. R. Warburton, School of Kinesiology, University of British Columbia, Vancouver, Canada.

This research was funded through the Canadian Diabetes Association, and the Social Sciences and Humanities Research Council of Canada. Ryan E. Rhodes is supported by a Canadian Cancer Society Senior Scientist Award and Give to Live, Chris Blanchard is supported by the Canada Research Chairs Program. We thank Kai Riecken, Leila Pfaeffli, Rachel Mark, Cara Temmel, and Gabriella Nasuti for the hard work of data collection and data entry during this project.

Correspondence concerning this article should be addressed to Ryan E. Rhodes, Behavioural Medicine Laboratory, Faculty of Education, Univer- sity of Victoria, P.O. Box 3015 STN CSC, Victoria, BC V8W 3P1, Canada. E-mail: [email protected]

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

Health Psychology © 2013 American Psychological Association 2014, Vol. 33, No. 8, 792– 802 0278-6133/14/$12.00 http://dx.doi.org/10.1037/a0033755

792

documented mental health benefits include improved well-being, reduced depression and anxiety, and enhanced cognitive function- ing (Warburton, Katzmarzyk, Rhodes, & Shephard, 2007). Despite this overwhelming evidence, only 15% of Canadian adults aged 20 –79 years are meeting the revised PA recommendations of at least 150 min of moderate to vigorous PA a week (Colley et al., 2011); the majority of Canadian adults spend their waking hours in sedentary pursuits. Clearly, PA promotion is an essential target for preventive medicine.

Promotion efforts have tended to target both children and older adults, which is understandable given the obvious primary and tertiary prevention strategies available in these respective age groups (Kumanyika, Jeffrey, Morabian, Ritenbaugh, & Antipatis, 2002). In Canada, cross-sectional and cohort studies on over- weight/obesity prevalence, however, have suggested one of the most prominent decreases in PA is between ages 25–35 years (Jedwab, 2005; Statistics Canada, 2005). Further, the decrease tends to level out or plateau between ages 35– 45 years, and does not appear to resume thereafter (Jedwab, 2005; Statistics Canada, 2005).

One of the reasons for this decline appears to be the onset of parenthood (Allender, Hutchinson, & Foster, 2008; Bellows- Riecken & Rhodes, 2008). For example, a meta-analysis of 10 studies comparing parents and nonparents on moderate to vigorous PA showed an effect size (d) of .48, in favor of couples without children (Bellows-Riecken & Rhodes, 2008). Retrospective exam- ination of PA following parenthood has shown that over 50% of previously active women (at public health guideline levels) are no longer active even 5 years into motherhood (McIntyre & Rhodes, 2009), and longitudinal research has found new Australian mothers were 2.27 times more likely to be inactive compared to nonmoth- ers 4 years after the onset of motherhood (Brown & Trost, 2003). This same study also found that mothers who had more than one child were twice as likely to be inactive as those with only one child (Brown & Trost, 2003). Although most of this evidence has focused on motherhood, there has been some recent preliminary evidence that fathers experience similar (Berge, Larson, Bauer, & Neumark-Szainer, 2011) or even greater (Hull et al., 2010) de- clines in PA compared to mothers. Clearly, curbing the declining PA behavior of young adults, and new parents specifically, is an important task underpinned by a deeper knowledge of the corre- lates that contribute to it.

Strong theoretically based PA correlates are also important to ensure that the known mechanisms of PA behavior change are targeted (Rhodes & Nigg, 2011). There are limited theory-based studies on the PA behavior of parents. Within those, an assortment of studies using constructs from Bandura’s (2004) social– cognitive theory (Deflandre, Antonini, & Lorant, 2004; Hinton & Olson, 2001; Pereira et al., 2007) have shown that self-efficacy (i.e., confidence to overcome barriers) predicted PA frequency of mothers 1 year postpartum (� � .30; Hinton & Olson, 2001), while perceived lack of childcare predicted inactivity (odds ratio � 1.73) 6 months postpartum (Pereira et al., 2007). A more recent study applying Ajzen’s (1991) theory of planned behavior (TPB) showed that perceived behavioral control predicted PA continuation in a retrospective analysis of mothers (McIntyre & Rhodes, 2009).

Although these previous studies have provided promising initial evidence to guide future intervention efforts, they have several limitations. First, most of the studies omit fathers and thus may not

reflect gender differences in the PA transition. Fathers and mothers may have different motivational and control factors impacting PA. For instance, a recent cross-sectional TPB study on parents (Ham- ilton, Cox, & White, 2012), although showing relatively few gender differences, highlighted that social aspects (i.e., subjective norm) had a greater influence on the PA motivation of mothers. Longitudinal exploration seems warranted, as does including both fathers and mothers in the analyses to explore the influence of the couple.

Second, to the best of our knowledge, existing research has not yet explored the correlates of PA patterns using comparison co- horts of couples. Parenthood is only one of several changes during young adulthood that may affect PA (Allender et al., 2008). Adult couples without children and parents expecting their second child are helpful comparison groups to identify both unique and similar aspects in PA motivation and to explain variation in PA across time. Indeed, the psychological models used to explain PA are rarely tested with population subgroups leading to the possible overspecification of differences (Rhodes & Blanchard, 2007; Rho- des, Blanchard, & Blacklock, 2008). A comparison of subgroups of young adults could help to identify unique motivational chal- lenges during parenthood, while showing what is similar to all groups. This would seem helpful to intervention efforts, because the content in interventions may be different in some specific areas for parents but largely similar in other aspects.

Finally, much of the research focused on parenthood and PA has relied on retrospective self-report, which can introduce memory bias because parents may consider their preparenthood PA profiles through “rose-tinted glasses” (Prince et al., 2008). The use of direct measures of PA, specifically accelerometry, could improve the validity of research efforts. Research has shown that acceler- ometers are reliable and objective PA measurement devices, which provide robust quantitative data and overcome many of the meth- odological issues associated with self-report data (Esliger, Cope- land, Barnes, & Tremblay, 2005; Strath, Brage, & Ekelund, 2005; Welk, 2005).

With these limitations in mind, the purpose of this study was to examine the social cognitive correlates of PA across subgroups (cohorts) of young adult couples over 12 months. The novelty of this approach includes the inclusion of a couples without children comparison cohort, the analysis of predictors of PA changes across time, direct PA measurement using accelerometry and the multi- level modeling of couple dyads. In a prior article (Rhodes, et al., 2013), these data have shown that PA was higher for couples without children and first-time expectant parents at baseline com- pared to couples expecting their second child, yet new mothers significantly lowered their PA across the 12 months compared to couples without children. The cohorts were couples between the ages of 25 and 40 years who were: (a) without children, (b) first-time parents during the first year of parenthood, and (c) second-time parents during their first year of parenting after the birth of the second child. The TPB was chosen as the social– cognitive framework for exploring PA motivation, given its proven predictive utility in this domain (McEachan, Conner, Taylor, & Lawton, 2011; Symons Downs & Hausenblas, 2005). Briefly, this model proposes that behavior is the result of a person’s intention (i.e., summary motivation) and, to a lesser extent, perception of behavioral control (PBC; confidence and controllability to perform the behavior). Intention is subsequently determined by one’s af-

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

793PARENTHOOD AND PHYSICAL ACTIVITY

fective (enjoyment) and instrumental (utility) attitudes (evalua- tions of the behavior), subjective norm (perceived pressure to perform the behavior from others), and PBC. Our first objective was to assess whether attitudes, subjective norm, PBC, and inten- tion changed from baseline to 12 months, and whether any of the relationships were moderated by sex and/or parental status. We hypothesized that parents would demonstrate lower PBC profiles than nonparents due to the additional activities and workload that come with parenting. We also hypothesized that this lower PBC would subsequently manifest in lower intentions. Our second objective was to evaluate the longitudinal prediction utility of the overall TPB model. We hypothesized that our results would mimic general TPB research (McEachan et al., 2011; Symons Downs & Hausenblas, 2005) and past TPB studies that used cross-sectional designs with parents (Hamilton et al., 2012; McIntyre & Rhodes, 2009). Specifically, we hypothesized that intention would predict changes in PA, while affective attitude and PBC would predict intention. Finally, we conducted exploratory analysis using gender, couple status, and parenthood as moderators of these TPB predic- tor variables.

Method

Participants

Participants were 314 adults (157 common law or married couples), aged 25– 40 years, without children or with one child at the time of recruitment from the Victoria metropolitan area in British Columbia, Canada. One hundred thirty-six participants in this sample were expecting their first child at the time of recruit- ment, 76 participants had one child and were expecting their second child at the time of recruitment, and the remaining 102 participants were not expecting to have a child. The sample was delimited to couples (rather than single parents), between ages 25 and 40 years, who passed the Physical Activity Readiness Ques- tionnaire (Canadian Society for Exercise Physiology, 2012). Sim- ilarly, couples who experienced health complications due to preg- nancy or birth (e.g., gestational diabetes, preeclampsia, bed rest) were excluded (n � 2).

Procedures

We advertised our study at ultrasound clinics, coffee shops, newspapers, online parenthood lists, purchase lists (e.g., craigslist- victoria.com), physician and midwife offices, and outreach parent programs such as Best Babies, as well as prenatal classes and baby retail outlets. A stratified recruitment strategy of advertisement at all regional sections of the Victoria metropolitan area (i.e., Victo- ria, Oak Bay, Saanich, Esquimalt, View Royal, Colwood, Lang- ford, and the Highlands) was conducted to maximize representa- tion of the region. Rolling recruitment and subsequent data collection were ongoing from January 2007 to December, 2011. There were three measurement periods for parents (pregnancy, 6 months after child, and 1 year after child) and nonparents (base- line, 6 months, and 12 months) across the study. Short-term assessment of PA is often criticized for failing to convey patterns of the behavior (Rhodes & Nigg, 2011), and prediction tends to remain relatively invariant within a 6-month frame (Rhodes & Plotnikoff, 2005; Symons Downs & Hausenblas, 2005). Thus,

assessment of PA every 6 months seemed a reasonable time lag to observe potential changes in PA. From the perspective of parent- hood, assessment 6 months after the birth of the child allows for recovery from the physical experience of giving birth with mothers and the early establishment of behavioral patterns in the family. All participants signed ongoing informed consent for the study, and the study was approved by the University of Victoria ethics board. Eleven couples agreed to participate in the study by signing informed consent, but they subsequently dropped out of the study before completing baseline measures.

Measures

Basic demographic and health behavior measures. Basic demographic and health behavior measures were collected via self-report based on prior published protocols (Benoit, Jansson, Leadbeater, & McCarthy, 2002–2005; Rhodes, Courneya, Blanchard, & Plotnikoff, 2007). These questions ask participants their self-reported height and weight, date of birth, sex, education, family income, and minority status, along with brief checkbox items about smoking status (yes or no), and whether a physician or health professional has ever told them they have heart disease, diabetes, cancer, high blood pressure, or high cholesterol.

Physical activity. PA was measured objectively for 7 consec- utive days at baseline and at 6 and 12 months using the GT1M activity monitor (Actigraph). The GT1M device is designed to ascertain normal human movement without impeding activity; it has been shown to provide valid and reliable estimates of PA (Abel et al., 2008; Janz, 1994). The activity monitor is attached to an elastic belt and worn at the waist above the left hip. Participants wore the monitors for 5 weekdays and 2 weekend days from when they got up in the morning to when they went to bed. Participants were instructed to remove the monitors at night and while swim- ming, bathing, or showering. They also completed a daily PA log/diary that identified when the accelerometer was removed, unusual circumstances, and structured activities. Participant accel- erometry data was included if there was a minimum of 5 full days (1 weekend day and 4 weekdays) meeting the minimum wear time (600 min/day) as proposed by Eslinger and colleagues (Esliger et al., 2005; Esliger et al., 2010). If minimum wear time was not met, but detailed information showed that the unit was removed for PA (e.g., swimming for 45 min) and, if included, the day met the wear time requirement, the day was considered valid. Weekdays with missing data were modeled from the composite of the other 4 weekdays, and missing weekend days were modeled from the available existing weekend days (Esliger et al., 2005; Esliger et al., 2010).

Duration (in minutes) and frequency (in number of bouts) of moderate to vigorous PA was calculated using established accel- erometer cut points (� 1952 average acceleration counts/min; Freedson, Melanson, & Sirard, 1998; Trost, Loprinzi, Moore, & Pfeiffer, 2011). To calculate these variables, the monitor was programmed to store data at 15-s intervals and converted to aver- age counts per minute.

Theory of planned behavior. TPB constructs were phrased in accordance with the Public Health Agency of Canada’s recom- mended weekly activity guidelines at the time of data collection (Health Canada, 2002).” Specifically, participants were asked to consider regular PA as activities of at least moderate intensity,

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

794 RHODES ET AL.

performed at least 4 times a week, 30 min or more per time across the next 6 months when answering the items.

Attitude. Attitude was measured using two Likert scale items that reflected the affective (i.e., enjoyable and fun) component, and two items that reflected the instrumental (i.e., wise and beneficial) component on a 5-point Likert-type scale (Ajzen, 2006). Several studies have been conducted in the PA research domain, which support the distinction of these attitudinal components (Rhodes, Fiala, & Conner, 2009). Furthermore, 5-point Likert scaling has shown the same reliability and predictive utility as semantic dif- ferential styles of assessment or 7-point scaling with these con- structs (Rhodes, Matheson, & Mark, 2010). Cronbach’s alpha coefficient of internal consistency was adequate for Time 1 (af- fective � � .80 and instrumental � � .74), Time 2 (affective � � .79 and instrumental � � .82), and Time 3 (affective � � .75 and instrumental � � .84).

Subjective norm. Subjective norm was measured by combin- ing two items reflecting the injunctive component of subjective norm and one item reflecting the descriptive component. The combination of these injunctive and descriptive components was based on Ajzen’s (2006) recommendations and the results of Rhodes and colleagues (Rhodes, Blanchard, Matheson, & Coble, 2006, Rhodes & Courneya, 2003). The items were: (a) “Most people who are important to me think I should engage in regular physical activity”; (b) “ Most people whose opinions I value would want me to engage in regular physical activity”; and (c) “Most people who are important to me will engage in regular PA them- selves.” All three items were framed for prediction over the next 6 months and measured on 5-point scales, ranging from 1 (strongly disagree) to 5 (strongly agree). Internal consistency for the three aggregated subjective norm items was 0.66 for Time 1, 0.68 for Time 2, and 0.67 for Time 3.

Perceived behavioral control. PBC was measured by three previously recommended items (Rhodes & Courneya, 2003, 2004); the measure assessed people’s perceptions of their ability to perform PA over the next 6 months. On a 5-point Likert scale, ranging from 1 (strongly disagree) to 5 (strongly agree), the items asked participants to rate whether they were completely confident they could perform PA, whether PA was under their complete control, and whether PA was extremely easy for them, if they were really motivated. Internal consistency values for the aggregated items were alphas of .80 for Time 1, .81 for Time 2, and .83 for Time 3.

Intention. Intention was measured using the following two items: (a) “During the next six months, I am definitely motivated to participate in regular PA” and (b) “During the next six months, I am determined to engage in regular PA” (ranging from 1 � strongly disagree to 5 � strongly agree). The items are based on prior validation by Rhodes et al. (2006). The aggregated intention measures showed adequate reliability for Time 1 (� � .83), Time 2 (� � 84), and Time 3 (� � .86).

Analysis Plan

Preliminary analyses. Descriptive findings were generated for the demographic and clinical variables by cohort followed by the calculation of attrition rates. Next, the TPB variables were checked for normality, after which analyses were conducted in SPSS, Version 20, to examine the pattern of missing data across

the TPB and accelerometer data and to identify the appropriate data imputation procedure.

Main analyses. Once the data were imputed, descriptive sta- tistics across sex and parenthood status were generated. Given the nested structure of the data (i.e., repeated TPB/moderate to vigor- ous PA assessments nested within couples), we used hierarchical linear modeling to further analyze the data (Raudenbush, Brennan, & Barnett, 1995). The first set of analyses examined whether attitudes, subjective norm, PBC, and intention changed from base- line to 12 months, and whether any of the relationships were moderated by sex and/or parental status. To address this question, two-level regression models were conducted. For example, a Level-1 no-intercept model for PBC was specified such that a main effect was entered for husband (0 � wife, 1 � husband), wife (0 � husband, 1 � wife), a husband linear trend (0 � baseline, 1 � 6 months, 2 � 12 months), and a wife linear trend with all coeffi- cients set to random. The main effects for the husbands’ and wives’ intercepts represent their respective baseline PBC levels, whereas the linear trends represent the change in PBC (or not) over the 12-month interval. At Level 2 (i.e., the couple level), cross- level interactions were entered such parental status predicted the intercepts and slopes (i.e., to determine if husbands’ and wives’ baseline PBC scores and the change in their PBC scores were similar for the different parental groups). In this case, three dummy-coded variables were created: nonparents (0 � no, 1 � yes), new parents (0 � no, 1 � yes), and child 2nd (0 � no, 1 � yes); the nonparents’ variable was excluded from the analysis to make it the comparison group. Follow-up analyses were then conducted excluding child 2nd to compare new parents versus child 2nd. Next, the husbands’ versus wives’ parental coefficients were statistically compared using the multivariate hypothesis test- ing procedure (e.g., to determine if the magnitude of the nonparent vs. new parent coefficient predicting PBC was the same for hus- bands and wives). Finally, the correlations among the spouses’ intercepts and slopes were examined to determine, for example, whether the wives’ baseline PBC levels were significantly associ- ated with their own change in PBC and/or their husbands’ change in PBC (and vice versa). The same analyses were conducted for the remaining TPB variables.

To address Objective 2 (i.e., whether the TPB variables pre- dicted intention across time), attitudes, subjective norm, and PBC were treated as time-varying covariates. Specifically, models were created such that affective attitude, instrumental attitude, subjec- tive norm, and PBC at baseline predicted intention at baseline and 6-month assessment of PBC predicted 6-month intention. First, a Level-1 no-intercept model included the husband and wife inter- cepts, husband and wife linear trends, the time-varying TPB co- variates for the husbands, and the time-varying covariates for the wives. The time-varying covariates can be interpreted as a unit change in PBC, for example, corresponding to an average increase or decrease in moderate to vigorous PA for a husband. Second, the husbands’ versus wives’ coefficients were statistically compared using the multivariate hypothesis testing procedure (e.g., to deter- mine if the affective attitude and intention relationship was the same for husbands and wives). Third, the moderating role of parental status on the time-varying covariate and intention relationships were examined for each TPB construct separately. For each construct, the same Level-1 model was created; how- ever, cross-level interactions were entered such that parental

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

795PARENTHOOD AND PHYSICAL ACTIVITY

status predicted a given husband’s and wife’s TPB construct using the dummy-coding procedure for parental status outlined above. Finally, we explored potential partner effects (e.g., the potential association between the wives’ affective attitude with their husbands’ intentions) for each TPB construct in separate models due to the number of predictors required to run a model with all the TPB constructs (i.e., 20 at Level 1), which raised concern about power and collinearity. For example, all TPB time-varying covariates were entered at Level 1 similar to the previous analyses, however, two additional variables (e.g., the wives’ affective attitudes predicting their husbands’ intention and the husbands’ affective attitude predicting their wives’ intention) were also included. The same analysis was then conducted for the remaining TPB variables, after which cross- level interactions were entered to determine potential parental effects.

The same analytical approach outlined for intention was used to determine whether the intention and PBC time-varying covariates (i.e., measured at baseline and 6 months) had significant relation- ships with moderate to vigorous PA (i.e., measured at 6 and 12 months) and whether these relationships were similar for husbands and wives and the different parental groups controlling for baseline PA. Finally, potential partner effects were also examined for intention and PBC.

Results

Preliminary Analyses

Baseline descriptive statistics are presented in Table 1. Couples without children were younger, had lower household incomes, and were more likely to be unemployed than couples expecting their first or second child. Body mass index was lower for women without children compared to expectant mothers, yet first-time expectant mothers were also lower than second-time expectant

mothers. There were no differences across groups on educational achievement, visible minority status, or health condition profiles, with the exception of smoking behavior. Women without children were more likely to be smokers than expectant mothers. Expectant mothers were often in their second trimester of pregnancy during the time of recruitment.

Fifteen couples did not return for our second wave of data collection, representing a 10% attrition rate. This included eight couples without children, five couples with their first child, and two couples expecting their second child. The reasons for this drop-out were that couples: (a) moved away (n � 2), (b) were too busy (n � 3), (c) dissolved their relationship (n � 2), or (d) had a health complication (n � 1). Seven couples did not disclose a reason. An additional 12 couples did not return for our third wave of data collection; this was distributed across the couples without children (n � 5), couples who had their first child (n � 5), and couples expecting their second child (n � 2). This represented an 8% attrition rate. The reasons for this drop-out were that couples: moved away (n � 2), were too busy (n � 1), dissolved their relationship (n � 4), or had a health complication (n � 1). Four reasons were undisclosed.

In terms of normality, skewness and kurtosis values exceeded 2.0 for the 6- and 12-month accelerometer data; therefore, the outliers were transformed using the next highest value �1 crite- rion. In terms of the missing data analyses, results showed that missingness at 6 and 12 months was significantly related to edu- cation, and the data were considered missing at random (MAR; i.e., the probability of missing a particular data point such as a moderate to vigorous PA score at 6 months is not related to its particular value, but can be dependent on other variables in the model; Allison, 2002). Unfortunately, using listwise deletion when one has data MAR may lead to biased estimates. Therefore, missing values were imputed in SPSS, Version 20, using the expectation maximization algorithm (Allison, 2002).

Table 1 Demographic and Health Profile of Couples at Baseline

Without children Onset of first child Onset of second child

(n � 102) (n � 136) (n � 76)

Characteristic Women Men Women Men Women Men

Demographic profile Mean agea 27.52� (5.12) 29.35� (5.33) 31.14 (4.78) 33.10 (4.71) 32.27 (3.87) 34.34 (4.91) Visible minorityb 6.0 12.0 9.8 9.0 7.8 10.5 Completed universityb 68.7 62.5 83.1 70.6 81.5 55.6 Household income � $75,000b 45.2� 45.5� 66.7 65.2 52.8 58.8 Currently employedb 54.2� 71.5� 88.7 89.7 73.7 83.4

Health profile Mean months pregnanta NA 5.75 (1.96) 5.70 (2.08) Mean BMI (kg/m2)a 22.65� (2.89) 25.48 (3.68) 24.90� (3.14) 26.18 (3.51) 26.49� (4.26) 26.53 (3.52) Smokerb 5.9� 8.2 0.0 2.9 0.0 7.9 Heart diseaseb 0.0 0.0 0.0 0.0 0.0 0.0 Diabetesb 2.1 0.0 4.2 0.0 2.7 5.3 Cancerb 2.1 0.0 1.4 1.5 0.0 0.0 High blood pressureb 0.0 10.2 1.4 7.5 5.4 0.0 High cholesterolb 0.0 3.9 1.4 4.5 0.0 2.6

Note. NA � not applicable; BMI � body mass index. a Values are M (SD). b Values are percent. � Significantly (p � .05) different from the other cohorts by sex.

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

796 RHODES ET AL.

Means of TPB Constructs Across Time

Descriptives in the form of mean values of TPB constructs can be found in Table 2 and analysis of these values over time can be found in Table 3. In terms of the baseline values, results from Table 3 show that for husbands, men without children had signif- icantly higher instrumental attitudes and PBC compared to fathers expecting a second child, whereas new fathers had significantly higher instrumental attitudes and intentions compared to parents with a second child. For wives, women without children had significantly higher PBC and intentions compared to mothers expecting a second child, whereas new parents had significantly higher PBC and intentions compared to those with a second child.

With respect to the change in the TPB constructs for husbands, results showed that interactions were present for affective attitudes such that they decreased for men without children and new fathers, whereas they increased for fathers with a second child. It was also shown that instrumental attitude declined for men without chil- dren, whereas it increased for fathers with a second child.

With respect to wives, results showed that the PBC levels of women without children and mothers with a second child were relatively stable, whereas new mothers’ PBC levels declined.

Finally, the correlations in Table 4 show that the husbands’ and wives’ intercepts (i.e., baseline values for a given TPB construct) were highly correlated as were their slopes (i.e., husbands’ and wives’ change in the TPB constructs were significantly related).

Prediction of Intention

As can be seen from Table 5, intention was significantly pre- dicted by affective attitude, instrumental attitude, and PBC for husbands and wives, whereas subjective norm only predicted in- tention for husbands. Further, the follow-up parental comparison analyses in wives showed that the intention/instrumental attitude and the intention/PBC relationships were significantly stronger for couples without children compared to parents with a second child.

Finally, Table 6 shows that the wives’ subjective norm signifi- cantly predicted their husbands’ intentions (� � .22), however, this was the only partner effect found. This relationship was not moderated by parental status.

Prediction of Moderate to Vigorous PA

As can be seen from Table 7, results showed that, on average, intention had a significant and positive association with PA for husbands (� � 8.34) and wives (� � 10.69). However, the intention and PA relationship was similar for all three parental groups. The relationship of PBC with PA was also examined, however, its inclusion in the model led to significant collinearity issues for the husband coefficients. Further, when examined inde- pendently from intention, it was found to be a nonsignificant predictor for husbands, but significant for wives. Therefore, we only examined the potential partner effect for intention. Results showed that the husbands’ intentions were not related to their wives’ PA (� � 1.37, p � .74), nor were the wives’ intentions related to their husbands’ PA (� � 2.31, p � .55).

Discussion

The purpose of this study was to examine the social– cognitive correlates of PA across cohorts of young adult couples over 12 months. The cohorts were couples between the ages of 25 and 40 years who were: (a) without children, (b) first-time parents during the first year of parenthood, and (c) second-time parents during their first year of parenting after the birth of the second child. This study advances the parenting and PA literature by: comparing cohorts of couples; using a longitudinal design; focusing on the transitional first year after having a baby; direct assessment of PA, and the use of hierarchical linear modeling to account for the nested relationship of couples. We uncovered some differences and many similarities in PA motivation across the subgroups of

Table 2 Means and Standard Deviations of the Theory of Planned Behavior Constructs by Time, Sex, and Parenthood Status

Baseline Six months Twelve months

Constructs Women Men Women Men Women Men

Couples without children Intention 4.27 (0.77) 3.87 (0.64) 4.30 (0.77) 4.00 (0.79) 4.38 (0.80) 4.01 (0.75) PBC 3.91 (0.78) 3.87 (0.77) 3.91 (0.70) 3.91 (0.85) 4.01 (0.74) 3.96 (0.76) Affective attitude 4.32 (0.57) 4.23 (0.57) 4.26 (0.60) 3.96 (0.84) 4.21 (0.69) 4.22 (0.65) Instrumental attitude 4.89 (0.30) 4.69 (0.39) 4.80 (0.47) 4.57 (0.62) 4.83 (0.35) 4.59 (0.51) Subjective norm 3.74 (0.57) 3.75 (0.52) 3.81 (0.61) 3.89 (0.57) 3.85 (0.58) 3.78 (0.62)

Couples expecting their first child Intention 4.15 (0.75) 4.11 (0.69) 4.10 (0.64) 3.91 (0.77) 4.10 (0.75) 4.15 (0.76) PBC 4.08 (0.74) 3.90 (0.86) 3.49 (1.06) 3.47 (0.79) 3.36 (1.01) 3.55 (0.88) Affective attitude 4.24 (0.80) 4.15 (0.79) 4.37 (0.75) 4.22 (0.72) 4.18 (0.67) 4.16 (0.72) Instrumental attitude 4.90 (0.26) 4.74 (0.46) 4.85 (0.29) 4.74 (0.42) 4.90 (0.30) 4.72 (0.52) Subjective norm 3.82 (0.72) 3.88 (0.53) 3.93 (0.63) 3.92 (0.56) 4.00 (0.66) 3.85 (0.59)

Couples expecting their second child Intention 3.67 (0.71) 3.60 (0.89) 4.05 (0.63) 3.90 (0.76) 4.03 (0.71) 3.90 (0.71) PBC 3.65 (0.78) 3.58 (0.92) 3.09 (0.82) 3.31 (0.96) 3.38 (0.98) 3.46 (0.95) Affective attitude 4.03 (0.66) 3.95 (0.89) 4.23 (0.78) 4.31 (0.78) 4.15 (0.66) 4.24 (0.69) Instrumental attitude 4.80 (0.35) 4.48 (0.60) 4.74 (0.49) 4.61 (0.60) 4.70 (0.47) 4.65 (0.58) Subjective norm 3.74 (0.64) 3.88 (0.53) 4.01 (0.58) 4.00 (0.61) 3.87 (0.70) 3.96 (0.52)

Note. PBC � perception of behavioral control.

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

797PARENTHOOD AND PHYSICAL ACTIVITY

couples. This knowledge will contribute to intervention efforts targeting the declining PA levels among younger adults by iden- tifying these key correlates and mean differences.

First, the profiles of TPB constructs across time helped to identify how motives for PA may change across parenthood. There has been considerable research into changes in behavioral profiles of parenthood across time that has shown PA declines (Allender et al., 2008; Bellows-Riecken & Rhodes, 2008), but relatively little research has explored changes in proposed antecedent constructs. It was hypothesized that parents would demonstrate lower PBC and subsequent intention profiles than nonparents. The rationale for this hypothesis was based on prior evidence from retrospective analyses (McIntyre & Rhodes, 2009) that parents had low control over PA in the face of additional activities and workload that may come with children. This hypothesis had some support. Indeed, PBC and intention were lower for mothers who were expecting their second child at baseline, compared to mothers expecting their first child and women without children. Second-time expectant fathers had significantly lower PBC than men without children and lower intentions than first-time expectant fathers. Further, the PBC of mothers expecting their first child at baseline changed across time (after delivery) to resemble the PBC of second-time mothers, with both groups of mothers showing significantly lower PBC at 12 months compared to women without children. The results support the lowering of control over PA at the onset of parenthood, particularly for women; at the same time, our findings support the

premise that it is the first child and not multiple children who has the greatest impact on PBC. The results are interesting because they may highlight the shift of control by first-time mothers, presumably as they attend to the daily workload of infant care. In our sample, 85% of mothers were on 12-month maternity leave or equivalent. It may be that new fathers did not change in PBC because their basic routines did not shift. An examination of PBC over the subsequent years would be helpful as mothers reenter the workforce after their 12-month maternity leave. The baseline re- sults of parents already with children suggested limited gender difference in PBC, so it may be that both parents negotiate com- parable child rearing or household chores following this year of leave. It may also be that parents are able to converge their parenting duties of multiple children so that time and workload responsibilities remain relatively constant (O’Reilly, 2004).

The main analyses of this study addressed the prediction of intention and PA over time. The prediction of intention with TPB variables generally supported our hypotheses. Specifically, affec- tive and instrumental attitudes and PBC showed independent medium-sized effects on intention. The results advance the knowl- edge gleaned from previous TPB literature, which used cross- sectional designs to predict intention, by following intention lon- gitudinally and allowing intention to vary across time (Symons Downs & Hausenblas, 2005), but these findings also replicate prior cross-sectional findings with parents (McIntyre & Rhodes, 2009). There was also an unexpected significant moderation by parent-

Table 3 Coefficients for Growth Curve Parameters and the Potential Moderating Effect of Parental Status for Husbands and Wives

Affective attitude Cognitive attitude

Subjective norm PBC Intention

Parameter � �a � �a � �a � �a � �a

Baseline level Husbands 4.23�� 4.75�� 3.64�� 3.96�� 3.97��

1. NonP vs. NewP �.04 �.03 .04 .04 .02 .02 �.11 �.06 .14 .08 2. NonP vs. P2nd �.16 �.09 �.21� �.19� .11 .07 �.39� �.19� �.24 �.13 3. P2nd vs. NewP .12 .08 .24� .25� �.08 �.06 .29 .16 .39� .25�

Wives 4.25�� 4.87�� 3.66�� 3.91�� 4.23��

4. NonP vs. NewP �.03 �.02 .02 .03 �.04 �.03 .12 .06 �.12 �.07 5. NonP vs. P2nd �.13 �.08 �.08 �.09 �.05 .03 �.38� �.18� �.53�� �.29��

6. P2nd vs. NewP .09 .07 .10 .14 .002 .01 .50�� .28�� .40�� .26��

Rate of change Husbands �.14� �.09�� .09 �.05 �.03

7. NonP vs. NewP .07 .05 .03 .03 .02 .02 �.14 �.08 .01 .01 8. NonP vs. P2nd .25�� .14�� .13� .12� .01 .01 �.03 �.01 .17 .09 9. P2nd vs. NewP �.18� �.12� �.10 �.10 .01 .01 �.11 �.06 �.16 �.10

Wives �.06 �.05 .09 .08 �.01 10. NonP vs. NewP .06 .04 .01 .01 .06 .05 �.50�� �.28�� .00 .01 11. NonP vs. P2nd .08 .05 .001 .01 .05 .03 �.17 �.08 .19 .11 12. P2nd vs. NewP �.03 �.02 �.001 .01 .01 .01 �.33�� �.18�� �.18 �.11

Multivariate hypothesis tests (spousal comparisons) 1 vs. 4 2(1) � 0.002 2(1) � 0.05 2(1) � 0.28 2(1) � 1.77 2(1) � 2.09 2 vs. 5 2(1) � 0.03 2(1) � 1.72 2(1) � 1.19 2(1) � .005 2(1) � 2.07 3 vs. 6 2(1) � 0.01 2(1) � 2.40 2(1) � 0.39 2(1) � 1.04 2(1) � 0.01 7 vs. 10 2(1) � 0.01 2(1) � 0.37 2(1) � 0.22 2(1) � 4.82� 2(1) � 0.002 8 vs. 11 2(1) � 2.27 2(1) � 3.36 2(1) � 0.17 2(1) � 0.76 2(1) � 0.04 9 vs. 12 2(1) � 2.05 2(1) � 2.14 2(1) � .001 2(1) � 3.15 2(1) � 0.07

Note. Comparisons 1, 2, 4, 5, 7, 8, 10, and 11 were made in the first regression using nonparents as the comparison group. Comparisons 3, 6, 9, and 12 were made in the second regression using parents with a 2nd child as the comparison group. PBC � perception of behavioral control; �a � standardized beta; NonP � nonparent; NewP � new parent; P2nd � parent second time. � p � .05. �� p � .01.

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

798 RHODES ET AL.

hood status and instrumental attitude/PBC for women. These data suggested that the PA intention of women without children was predicted by appraisals of the benefit and utility of PA and control more than that of second-time mothers. This effect may represent a ceiling effect at which once a threshold level of utility or control

is reached and parenthood demands increase, other factors are more influential. Still, the moderation effect was relatively small. Overall, the findings suggest that young adults have relatively similar motivational antecedents. The results also mimic meta- analyses that have shown that the affective judgments are larger predictors than the instrumental judgments of PA (Nasuti & Rho- des, 2013; Rhodes et al., 2009). Overall, the majority of variance in PA intention can be accounted for by appraisal of the enjoy- ability of PA and PBC, regardless of parenthood status or gender. This contrasts the work of Hamilton et al. (2012), who found subjective norm predicted intention in parents; the problem is that this study did not separate instrumental and affective attitudes, so the difference may derive from simple measurement discrepancies.

The hypothesis that intention, treated as a time-varying covari- ate, would predict PA was supported. The findings illustrated that intention significantly predicted moderate to vigorous PA. This is commensurate with the tenets of the TPB model (Ajzen, 1991). Still, the findings are the first to demonstrate the prediction using directly measured PA among young adult couples and parents, and one of the few to demonstrate TPB with a longitudinal design that simultaneously accounted for changes in accelerometry and the TPB constructs (McEachan et al., 2011). Interestingly, the results also show that the robustness of the intention– behavior relation- ship to gender and parenthood.

Although couples were coordinated in their PA motives, as noted by significant correlations in the slopes and intercepts of TPB constructs, it was interesting to show that almost no cross- partner effects were associated with PA or intention after control- ling for individual effects. Specifically, only wives’ subjective norm predicted husbands’ intention in these models. This suggests that couples’ PA behavior and motivation, while relatively sym-

Table 4 Correlations Between Intercepts and Slopes for Husbands and Wives for the TPB Constructs

Constructs Husband intercept

Husband slope

Wife intercept

Affective attitude Husband slope �.46�

Wife intercept .57� �.50�

Wife slope �.08 .63� �.41�

Instrumental attitude Husband slope �.13 Wife intercept .64� �.02 Wife slope �.36� .20� .48�

Subjective norm Husband slope �.49�

Wife intercept .63� �.50�

Wife slope �.32� .61� �.42�

PBC Husband slope �.40�

Wife intercept .63� �.65�

Wife slope �.19� .63� .08 Intention

Husband slope �.55�

Wife intercept .33� �.13 Wife slope �.26� .58� �.57�

Note. PBC � perception of behavioral control. � p � .05.

Table 5 Coefficients for Intention Growth Curve Parameters and the Potential Moderating Effect of Parental Status for Husbands and Wives

Parameter � �a Follow-up parental comparisons

Husbands 1. Intercept 3.92 2. Linear trend .05 .01 3. Affective attitude .39�� .38�� NonP � NewP � P2nd 4. Instrumental attitude .24�� .15�� NonP � NewP � P2nd 5. Subjective norm .13� .10� NonP � NewP � P2nd 6. PBC .33�� .37�� NonP � NewP � P2nd

Wives 7. Intercept 4.01 8. Linear trend .21� .11�

9. Affective attitude .42�� .19� NonP � NewP � P2nd 10. Instrumental attitude .33�� .15�� NonP � NewP, NewP � P2nd, NonP � P2nd 11. Subjective norm .03 .02 NonP � NewP � P2nd 12. PBC .27�� .32�� NonP � NewP, NewP � P2nd, NonP � P2nd

Multivariate hypothesis tests (spousal comparisons) 1 vs. 7 2(1) � 0.03 2 vs. 8 2(1) � 4.28�

3 vs. 9 2(1) � 0.13 4 vs. 10 2(1) � 0.39 5 vs. 11 2(1) � 1.72 6 vs. 12 2(1) � 1.21

Note. PBC � perception of behavioral control; �a � standardized beta; NonP � nonparent; NewP � new parent; P2nd � parent second time. � p � .05. �� p � .01

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

799PARENTHOOD AND PHYSICAL ACTIVITY

metrical, is not based on the other’s motivation after accounting for their own.

Taken together, these findings provide helpful practical infor- mation for targeting PA interventions among young adults. First,

the relatively similar findings about the constructs of the TPB model suggests that parents and couples without children have the same basic motivational base for engaging in PA, so much of the promotional efforts could be similar. In this case, changing inten- tion through the promotion of (a) enjoyable and pleasant PA while (b) fostering a strong perception of control and (c) some promotion of the utility and benefit of PA seems warranted. Affective judg- ments and positive experiences are the strongest postintervention predictors of successful PA and weight-change interventions among middle-aged women (Silva et al., 2011), so our findings reflect the importance of this intervention target. Nevertheless, parents, particularly new mothers, reported significantly lower PBC over PA, so an intervention for parents may require a more intensive PA control strategy than couples without children. The clinical implication of this finding is that parents may have lower absolute values of control over PA and thus greater room for improvements in interventions. A targeted approach to couples with children may be warranted. Our results suggest that normative aspects, similar to the general TPB literature (Hagger, Chatzisa- rantis, & Biddle, 2002; McEachan et al., 2011), have little utility for intervention. Interestingly, our results also suggest that inter- vention at the couple level, including both men and women, may be the most efficacious unit of intervention because couples’ motives for PA are coordinated but not dependent on each other.

Although this study had several methodological strengths, in- cluding direct assessment of PA and homogenous assessment of three cohorts across time, there are study limitations that warrant consideration. First, our sample was clearly more educated, less overweight (Statistics Canada, 2007), and more physically active (Colley et al., 2011) than the average for the Canadian population, suggesting there may have been selection bias. Our nonrandom

Table 6 Results From the Hierarchical Linear Models Exploring Partner Relationships With Intention for Each Theory of Planned Behavior Variable

Models � �a

AA Husband AA ¡ Husband intention .41� .38�

Husband AA ¡ Wife intention .04 .04 Wife AA ¡ Wife intention .41� .37�

Wife AA ¡ Husband intention �.06 �.05 IA

Husband IA ¡ Husband intention .25� .15�

Husband IA ¡ Wife intention .01 .01 Wife IA ¡ Wife intention .42� .15�

Wife IA ¡ Husband intention �.05 �.02 SN

Husband SN ¡ Husband intention .12� .10�

Husband SN ¡ Wife intention .03 .02 Wife SN ¡ Wife intention .04 .03 Wife SN ¡ Husband intention .22� .18�

PBC Husband PBC ¡ Husband intention .32� .35�

Husband PBC ¡ Wife intention .01 .01 Wife PBC ¡ Wife intention .26� .30�

Wife PBC ¡ Husband intention .02 .02

Note. �a � standardized beta; AA � affective attitude; IA � instrumen- tal attitude; SN � subjective norm; PBC � perceived behavioral control. � p � .01.

Table 7 Coefficients for Physical Activity Growth Curve Parameters and the Potential Moderating Effect of Parental Status for Husbands and Wives

Parameter � �a Follow-up parental comparisons

Husbands Intercept (�10) 56.21 Baseline PA (�11) 0.22

�� 0.23��

Linear trend (�30) �6.59 �0.05 Intention (�50) 8.34

� 0.28� NonP � NewP � P2nd Wives

Intercept (�20) 62.71 Baseline PA (�21) 0.26

�� 0.25��

Linear trend (�30) �20.25 �� �0.14��

Intention (�60) 10.69 �� 0.37�� NonP � NewP � P2nd

Multivariate hypothesis tests (spousal comparisons) �50 vs. �60

2(1) � 0.22

Note. PA � physical activity; �a � standardized beta; NonP � nonparent; NewP � new parent; P2nd � parent second time. � p � .05. �� p � .01. Level-1 Model PA � 1 (husband intercept) � 2 (wife intercept) � 3 (husband linear trend) � 4 (wife linear trend) � 5 (husband intention) � 6 (wife intention) � e Level-2 Model 1 � �10 � �11 (husband’s baseline PA) � r1 2 � �20 � �21 (wife’s baseline PA) � r2 3 � �30 � r3 4 � �40 � r4 5 � �40 � r5 6 � �40 � r6

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

800 RHODES ET AL.

sample appears to have possessed high initial motivation for health. Given these data limitation, it is likely that changes in PA across time may be even larger for a population with lower baseline PA and motivation status. There is also the possibility that our cohorts had differences between groups that were not con- trolled for in the analyses. The impact of these factors is a possible limitation to the results. We are also unable to provide estimates of the reach of recruitment due to our passive (e.g., flyers, advertise- ments) recruitment strategy. The longitudinal nature of the study is a strength, but a longer time frame may show even larger hetero- geneity in PA and TPB variables as time elapses. The extant literature on parenthood and PA has suggested that the first 5 years represents the largest decline in activity (Rhodes, Symons Downs, & Bellows Riecken, 2008), so a longitudinal study over 5 years may be ideal.

References

Abel, M. G., Hannon, J. C., Sell, K., Lillie, T., Conlin, G., & Anderson, D. (2008). GT1M accelerometers for walking and running in adults. Ap- plied Physiology, Nutrition, and Metabolism, 33, 1155–1164.

Ajzen, I. (1991). The theory of planned behavior. Organizational Behavior and Human Decision Processes, 50, 179 –211.

Ajzen, I. (2006). Constructing a theory of planned behavior questionnaire. Retrieved from http://people.umass.edu/aizen/pdf/tpb.measurement.pdf

Allender, S., Hutchinson, L., & Foster, C. (2008). Life-change events and participation in physical activity: A systematic review. Health Promo- tion International, 23, 160 –172.

Allison, P. D. (2002). Missing data. Thousand Oaks, CA: Sage. Bandura, A. (2004). Health promotion by social cognitive means. Health

Education and Behavior, 31, 143–164. Bellows-Riecken, K. H., & Rhodes, R. E. (2008). The birth of inactivity?

A review of physical activity and parenthood. Preventive Medicine: An International Journal Devoted to Practice and Theory, 46, 99 –110.

Benoit, C., Jansson, M., Leadbeater, B., & McCarthy, B. (2002–2005). The impact of stigma on marginalized populations’ work, health and access to services. Canadian Institutes of Health Research.

Berge, J. M., Larson, N., Bauer, K. W., & Neumark-Szainer, D. (2011). Are parents of young children practicing healthy nutrition and physical activity behaviors? Pediatrics, 127, 881– 887.

Brown, W. J., & Trost, S. G. (2003). Life transitions and changing physical activity patterns in young women. American Journal of Preventive Medicine, 25(2), 140 –143.

Canadian Society for Exercise Physiology. (2012). Physical activity read- iness questionnaire. Retrieved from http://www.csep.ca/english/view .asp?x�698

Colley, R. C., Garriguet, D., Janssen, I., Craig, C. L., Clarke, J., & Tremblay, M. S. (2011). Physical activity of Canadian adults: Acceler- ometer results from the 2007 to 2009 Canadian Health Measures Survey. Health Reports/Statistics Canada, 22, 7–14.

Deflandre, A., Antonini, P. R., & Lorant, J. (2004). Perceived benefits and barriers to physical activity among children, adolescents and adults. International Journal of Sport Psychology, 35, 23–36.

Esliger, D. W., Copeland, J. L., Barnes, J. D., & Tremblay, M. S. (2005). Standardizing and optimizing the use of accelerometer data for free- living physical activity monitoring. Journal of Physical Activity & Health, 3, 366 –383.

Esliger, D. W., Tremblay, M. S., Copeland, J. L., Barnes, J. D., Huntington, G. E., & Bassett, D. R., Jr. (2010). Physical activity profile of Old Order Amish, Mennonite, and contemporary children. Medicine & Science in Sports & Exercise, 42, 296 –303. doi:10.1249/MSS.0b013e3181b3afd2

Freedson, P. S., Melanson, E., & Sirard, J. (1998). Calibration of the Computer Science And Applications, Inc. accelerometer. Medicine & Science in Sports & Exercise, 50, 777–781.

Hagger, M., Chatzisarantis, N. L. D., & Biddle, S. J. H. (2002). A meta-analytic review of the theories of reasoned action and planned behavior in physical activity: Predictive validity and the contribution of additional variables. Journal of Sport & Exercise Psychology, 24, 1–12.

Hamilton, K., Cox, S., & White, K. M. (2012). Testing a model of physical activity among mothers and fathers of young children: Integrating self- determined motivation, planning, and the theory of planned behavior. Journal of Sport & Exercise Psychology, 34, 124 –145.

Health Canada. (2002). Physical activity. Retrieved from http://www.phac- aspc.gc.ca/hp-ps/hl-mvs/pa-ap/index-eng.php

Hinton, P., & Olson, C. (2001). Postpartum exercise and food intake: The importance of behavior-specific self-efficacy. Journal of the American Dietetic Association, 101, 1430 –1437.

Hull, E. E., Rofey, D. L., Robertson, R. J., Nagle, E. F., Otto, A. D., & Aaron, A. J. (2010). Influence of marriage and parenthood on physical activity: A two-year prospective analysis. Journal of Physical Activity & Health, 7, 577–583.

Janz, K. F. (1994). Validation of the CSA accelerometer for assessing children’s physical activity. Medicine & Science in Sports & Exercise, 26, 369 –375.

Jedwab, J. (2005). Actively Canadian: Who’s the most active of us all? Montreal, Canada: Association for Canadian Studies.

Kumanyika, S., Jeffrey, R. W., Morabian, A., Ritenbaugh, C., & Antipatis, V. J. (2002). Obesity prevention: The case for action. International Journal of Obesity, 26, 425– 436.

McEachan, R. R. C., Conner, M., Taylor, N. J., & Lawton, R. J. (2011). Prospective prediction of health-related behaviors with the theory of planned behavior: A meta-analysis. Health Psychology Review, 5, 97– 144.

McIntyre, C. A., & Rhodes, R. E. (2009). Correlates of physical activity during the transition to motherhood. Women and Health, 49, 66 – 83.

Nasuti, G., & Rhodes, R. E. (2013). Affective judgment and physical activity in youth: Review and meta-analysis. Annals of Behavioral Medicine, 45, 357–376. doi:10.1007/s12160-012-9462-6

O’Reilly, M. (2004). Achieving a new balance: Women’s transition to second-time parenthood. Journal of Obstetric, Gynecologic, & Neonatal Nursing, 33, 455– 462.

Pereira, M. A., Rifas-Shiman, S. L., Kleinman, K. P., Rich-Edwards, J. W., Peterson, K. E., & Gillman, M. W. (2007). Predictors of change in physical activity during and after pregnancy: Project Viva. American Journal of Preventive Medicine, 32, 312–319.

Prince, S. A., Adamo, K. B., Hamel, M. E., Hardt, J., Connor Gorber, S., & Tremblay, M. (2008). A comparison of direct versus self-report measures for assessing physical activity in adults: A systematic review. The International Journal of Behavioral Nutrition and Physical Activity, 5, 56. doi:10.1186/1479-5868-1185-1156

Raudenbush, S., Brennan, R., & Barnett, R. (1995). A multivariate hier- archical model for studying psychological change within married cou- ples. Journal of Family Psychology, 9, 161–174.

Rhodes, R. E., & Blanchard, C. M. (2007). Just how special are the physical activity cognitions in diseased populations? Preliminary evi- dence for integrated content in chronic disease prevention and rehabil- itation. Annals of Behavioral Medicine, 33, 302–312.

Rhodes, R. E., Blanchard, C. M., Benoit, C., Levy-Milne, R., Naylor, P. J., Symons Downs, D., & Warburton, D. E. (2013). Physical activity and sedentary behavior across 12 months in cohort samples of couples without children, expecting their first child, and expecting their second child. Journal of Behavioral Medicine. Advance online publication. doi:10.1007/s10865-013-9508-7

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

801PARENTHOOD AND PHYSICAL ACTIVITY

Rhodes, R. E., Blanchard, C. M., & Blacklock, R. E. (2008). Do physical activity beliefs differ by age and gender? Journal of Sport & Exercise Psychology, 30, 412– 423.

Rhodes, R. E., Blanchard, C. M., Matheson, D. H., & Coble, J. (2006). Disentangling motivation, intention, and planning in the physical activ- ity domain. Psychology of Sport and Exercise, 7, 15–27.

Rhodes, R. E., & Courneya, K. S. (2003). Self-efficacy, controllability, and intention in the theory of planned behavior: Measurement redundancy or causal independence? Psychology & Health, 18, 79 –91.

Rhodes, R. E., & Courneya, K. S. (2004). Differentiating motivation and control in the theory of planned behavior. Psychology, Health & Med- icine, 9, 205–215.

Rhodes, R. E., Courneya, K. S., Blanchard, C. M., & Plotnikoff, R. C. (2007). Prediction of leisure-time walking: An integration of social cognitive, perceived environmental, and personality factors. The Inter- national Journal of Behavioral Nutrition and Physical Activity, 4, 51. doi:10.1186/1479-5868-4-51

Rhodes, R. E., Fiala, B., & Conner, M. (2009). Affective judgments and physical activity: A review and meta-analysis. Annals of Behavioral Medicine, 38, 180 –204.

Rhodes, R. E., Matheson, D. H., & Mark, R. (2010). Evaluation of social cognitive scaling response options in the physical activity domain. Measurement in Physical Education and Exercise Science, 14, 137–150.

Rhodes, R. E., & Nigg, C. R. (2011). Advancing physical activity theory: A review and future directions. Exercise and Sports Sciences Reviews, 39, 113–119.

Rhodes, R. E., & Plotnikoff, R. C. (2005). Can current physical activity act as a reasonable proxy measure of future physical activity? Evaluating cross-sectional and passive prospective designs with the use of social cognition models. Preventive Medicine: An International Journal De- voted to Practice and Theory, 40, 547–555.

Rhodes, R. E., Symons Downs, D., & Bellows Riecken, K. H. (2008). Delivering inactivity: A review of physical activity and the transition to motherhood. In L. T. Allerton & G. P. Rutherfode (Eds.), Exercise and women’s health: New research (pp. 105–127). Hauppauge, NY: Earth- link Science Press.

Silva, M. N., Markland, D., Carraça, E. V., Vieira, P. N., Coutinho, S. R., Minderico, C. S., . . . Teixeira, P. J. (2011). Exercise autonomous motivation predicts 3-yr weight loss in women. Medicine & Science in Sports & Exercise, 43, 728 –737. doi:10.1249/MSS.0b013e3181f3818f

Statistics Canada. (2001). 2001 community profile-Victoria. Retrieved from http://www12.statcan.gc.ca/census-recensement/2011/dp-pd/prof/ details/page.cfm?Lang�E&Geo1�CSD&Code1�5917034&Geo2� PR&Code2�59&Data�Count&SearchText�Victoria&SearchType� Begins&SearchPR�01&B1�All&GeoLevel�PR&GeoCode�5917034

Statistics Canada. (2004). Health of Canadians living in census metropol- itan areas. Retrieved from http://www5.statcan.gc.ca/bsolc/olc-cel/olc- cel?catno�89-613-MIE2004002&lang�eng

Statistics Canada. (2005). Canadian Community Health Survey 1994 – 2003. Retrieved from http://www23.statcan.gc.ca/imdb/p2SV .pl?Function�getSurvey&SDDS�3226&Item_Id�50653&lang�en

Statistics Canada. (2007). 2006 Census. Retrieved from http://www12 .statcan.gc.ca/census-recensement/2006/dp-pd/prof/92-591/

Strath, S. J., Brage, S., & Ekelund, U. (2005). Integration of physiological and accelerometer data to improve physical activity assessment. Medi- cine & Science in Sports & Exercise, 37(11 Suppl.), S563–S571.

Symons Downs, D., & Hausenblas, H. A. (2005). Exercise behavior and the theories of reasoned action and planned behavior: A meta-analytic update. Journal of Physical Activity & Health, 2, 76 –97.

Trost, S. G., Loprinzi, P. D., Moore, R., & Pfeiffer, K. A. (2011). Com- parison of accelerometer cut points for predicting activity intensity in youth. Medicine & Science in Sports & Exercise, 43, 1360 –1368.

Warburton, D. E. R., Charlesworth, S., Ivey, A., Nettlefold, L., & Bredin, S. S. D. (2010). A systematic review of the evidence for Canada’s physical activity guidelines for adults. The International Journal of Behavioral Nutrition and Physical Activity, 7, 39. doi:10.1186/1479- 5868-7-39

Warburton, D. E. R., Katzmarzyk, P. T., Rhodes, R. E., & Shephard, R. J. (2007). Evidence-informed physical activity guidelines for Canadian adults. Applied Physiology, Nutrition, and Metabolism, 32(Suppl. 2E), S16 –S68.

Warburton, D. E. R., Nicol, C. W., & Bredin, S. S. (2006). Health benefits of physical activity: The evidence. CMAJ: Canadian Medical Associa- tion Journal, 174(6), 801– 809.

Welk, G. J. (2005). Principles of design and analyses for the calibration of accelerometry-based activity monitors. Medicine & Science in Sports & Exercise, 37(11 Suppl.), S501–S511.

Received November 22, 2012 Revision received May 17, 2013

Accepted May 21, 2013 �

T hi

s do

cu m

en t

is co

py ri

gh te

d by

th e

A m

er ic

an P

sy ch

ol og

ic al

A ss

oc ia

ti on

or on

e of

it s

al li

ed pu

bl is

he rs

. T

hi s

ar ti

cl e

is in

te nd

ed so

le ly

fo r

th e

pe rs

on al

us e

of th

e in

di vi

du al

us er

an d

is no

t to

be di

ss em

in at

ed br

oa dl

y.

802 RHODES ET AL.