Negative Attitudes Toward Physical Activity: Measurement and Role in Predicting Physical Activity Levels Among Preadolescents

Abstract

Objectives To describe the development and validation of a measure of negative attitudes toward physical activity and examine the association between these attitudes and self-reported physical activity among preadolescents. Method A school-based sample of 382 fifth and sixth graders (mean age = 10.8) completed measures of attitudes toward physical activity and self-reported physical activity. Body mass index data for the participants was collected as a part of a standard school health assessment. Exploratory factor analysis, confirmatory factor analysis, and structural equation modeling were utilized to test the factor structure and predictive value of attitudes toward physical activity. Results Results supported the reliability and concurrent validity of the negative attitudes measure and found a significant association between negative attitudes and physical activity. Negative attitudes was found to be a stronger predictor of physical activity than positive attitudes, which have been the focus of previous research in this area. Conclusions The results suggest that negative attitudes toward physical activity can be reliably measured and may be an important target for intervention efforts to increase physical activity among children and adolescents.

Regular physical activity is recommended for all healthy children and adolescents as well as for many individuals with specific medical conditions (American Academy of Pediatrics Committee on Sports Medicine and Fitness and Committee on School Health, 2000). Daily exercise is central to a healthy lifestyle, and the U.S. Department of Health and Human Services and U.S. Department of Agriculture (2005) recommend that children and adolescents engage in at least one hour of moderate physical activity daily. However, despite the consensus regarding the necessity of regular physical activity in promoting pediatric health, many children and adolescents fall short of these standards. Recent estimates suggest that only 8% of adolescents in the United States engage in the recommended levels of physical activity, and activity levels tend to decline with age (Troiano et al., 2008). Preadolescence may be a particularly important time with regard to physical activity as previous research has shown that activity levels often begin to decline at this time (Broderson, Steptoe, Williamson, & Wardle, 2005; Kimm et al., 2000). The gap between recommendations and reality highlights a serious problem in our increasingly sedentary society, and efforts to better understand the factors that influence pediatric physical activity are needed to effectively address this growing public health concern.

The consequences of insufficient physical activity in childhood and adolescence are considerable and have the potential to last throughout the lifespan. Most obvious among these are the risk for pediatric obesity and the panoply of associated physical and psychological comorbidities (see Steele, Nelson, & Jelalian, 2008, for review). Beyond its role in obesity prevention and treatment, regular physical activity is often indicated for individuals with a variety of chronic medical conditions such as pediatric chronic pain (Long, Palermo, & Manees, 2008), pediatric migraine headaches (Powers & Hershey, 2005), and sleep disorders (e.g., Alessi, Yoon, Schnelle, Al-Samarrai, & Cruise, 1999; Fox, 1999). Increasing physical activity is also a key component of many behavioral treatments for depression (i.e., behavioral activation) and has been associated with a decrease in depression and anxiety in adolescents (Calfas & Taylor, 1994). Given the ubiquity of physical activity as a health recommendation, a thorough understanding of the factors that contribute to this important behavior in the pediatric population is needed.

As evidenced by consistently low rates of pediatric physical activity despite widely available information describing its benefits, knowledge alone is often not a sufficient impetus for behavior change (Glanz & Rimer, 2005). Therefore, in attempting to better inform prevention and intervention efforts, researchers have investigated numerous psychological and social factors believed to influence health behaviors. One such factor is attitudes toward health behaviors such as physical activity. Introduced in the Theory of Reasoned Action (TRA), and its later incarnation, the Theory of Planned Behavior (TPB), attitudes are conceptualized as stemming from a person's beliefs regarding a particular behavior and its consequences. In turn, attitudes go on to shape an individual's behavioral intentions and, ultimately, their consequent actions (Ajzen, 1991; Ajzen & Fishbein, 1980). Related constructs, such as outcome expectations in social cognitive theory (Bandura, 1997) and perceived benefits and barriers in the health beliefs model (Becker, 1974; Rosenstock, 1974) have also been proposed as important cognitive factors influencing health behaviors. Although the conceptualization of these factors may differ somewhat across theories, it is clear that attitudes and related constructs play a central role in the leading theories aimed at explaining health behaviors such as pediatric physical activity. In our conceptualization, attitudes toward physical activity are a broad construct encompassing both perceptions of the immediate experience of activity as well as the consequences of activity.

Despite the theoretical role of attitudes in influencing physical activity, reviews of the child and adolescent activity literature have found the role of attitudes to be inconsistent (e.g., Kohl & Hobbs, 1998; Sallis, Prochaska, & Taylor, 1999). This may be, in part, due to the oversimplified conceptualization and measurement of attitudes and the tendency to focus only on the role of positive attitudes (e.g., Baker, Little, & Brownell, 2003; Dishman et al., 2002). Studies focusing exclusively on positive attitudes are limited by at least one of three assumptions: (a) low positive attitude scores are indicative of high negative attitudes; (b) positive and negative attitudes cannot coexist; or (c) only positive attitudes influence behavior. The current study challenges these assumptions by suggesting that negative and positive attitudes toward physical activity are distinct constructs, and that negative attitudes are equally or even more important than positive attitudes in predicting physical activity.

Although attitudes, in general have received some attention in the pediatric literature, negative attitudes toward physical activity have been under-studied. Such attitudes may include beliefs that exercise is in some way unpleasant (e.g., it is painful) or has negative consequences (e.g., it is too time-consuming). Some researchers have recently begun to examine perceptions of peer criticism or teasing around exercise (e.g., Storch et al., 2007); however, more general investigation of the role of negative attitudes toward physical activity is lacking. Negative attitudes toward physical activity are more than simply the lack of positive attitudes and may exert a unique effect on pediatric activity levels. For example, an individual can hold high levels of positive attitudes (i.e., believe that physical activity provides a number of potential benefits) but also have high levels of negative attitudes (i.e., believe that physical activity is unpleasant or has important drawbacks or consequences). Understanding both positive and negative attitudes held by an individual may be important to gaining insight into that individual's decision-making process with regard to physical activity.

Based on the premise that positive and negative attitudes toward physical activity each have an effect on pediatric physical activity, individuals can be categorized along both dimensions, creating a typology of attitudes (i.e., High Positive/High Negative, Low Positive/High Negative, High Positive/Low Negative, Low Positive/Low Negative; Figure 1). Using these categorizations, predictions can be made for the four groups with regard to physical activity. Individuals who have high positives and low negatives might be considered likely exercisers because they perceive significant advantages associated with physical activity but few disadvantages. Conversely, those with low positives and high negatives may be considered unlikely exercisers because they perceive few benefits and many costs. Individuals reporting high levels of both positive and negative may be considered ambivalent because they recognize benefits to exercise but also acknowledge considerable drawbacks. Finally, the group of individuals who indicate low levels of both positive and negative attitudes might be considered to have low interest as they see neither substantial benefits nor costs associated with physical activity. This typology is proposed as a simple heuristic to conceptualize the interplay between positive and negative attitudes and help make clinically-relevant predictions.

Our model's focus on attitudes is consistent with cognitive-behavioral approaches to intervention, which highlight the central role of attitudes in health behaviors. Cognitive-behavioral therapy (CBT) often focuses on modifying thoughts—challenging negative perceptions and replacing them with more adaptive cognitions—which requires first identifying relevant attitudes as targets for change. In addition to predicting physical activity levels based on group, the proposed typology also suggests intervention strategies that might be appropriate for addressing these thoughts (Figure 1). For example, an individual with high positives but also high negatives might benefit from challenging these negative perceptions, while an individual with low positives and low negatives might require more attention to building positive perceptions about the advantages of physical activity.

The present study was designed around four primary objectives. First, we describe the development of a measure of negative attitudes toward physical activity and report on the initial evaluation of the measure's psychometric properties in a school sample. We expected that the new measure would demonstrate adequate reliability and preliminary evidence of construct validity in confirmatory factor analyses. Second, we establish negative attitudes toward physical activity as a meaningful and unique construct, distinct from positive attitudes. We hypothesized that negative attitudes would be moderately correlated with positive attitudes, but would represent a distinct (i.e., not redundant) construct. Third, we evaluate the concurrent validity of negative attitudes with regard to physical activity. Specifically, we expected to find a significant negative correlation between negative attitudes and physical activity. Furthermore, we hypothesized that this relationship would remain significant even after controlling for positive attitudes (i.e., a distinct construct that will represent a unique predictive value). Fourth, in a separate ANOVA analysis, we test the predictive value of the attitudes typology in understanding physical activity levels among preadolescents. We expected that individuals with low negative attitudes and high positive attitudes would report the highest levels of physical activity, while individuals with high negatives and low positives would report the lowest levels of activity. Individuals with high negatives/high positives or low negative/low positives were expected to report physical activity levels between the other two groups.

Method

Participants

A volunteer sample of 382 participants was recruited through a large Midwestern public school district. Eligibility criteria for participation in the investigation included: (a) the child was enrolled in either fifth or sixth grade (mean age = 10.8; SD = 0.65; 56.9% of participants were in the sixth grade); (b) the student spoke and read English; and (3) the child's parent or custodial caregiver provided informed consent for participation. All students meeting these criteria were deemed eligible regardless of weight status, sex, or ethnicity. Approximately 54% of the sample was male. Ethnic composition of the sample was as follows: 6.6% African American, 5.8% Asian, 59.2% European American, 9.7% Hispanic, 6.3% Native American, 10% Other, and 2.1% Biracial. Individual information regarding socio-economic status was not available because only child report measures were completed; however, the school district reported that 43.3% of children attending the six schools sampled qualified for free and reduced lunch. The agregate school district percentage of children eligible for free and reduced lunch was 32.1%. All participants in this study had three required physical education classes per week.

Procedure

Recruitment and Data Collection

Information about the study and consent forms were distributed to children in the fifth and sixth grades of six selected elementary schools with instructions to deliver these documents to their parents. As an incentive for children to return the consent forms, classes with a consent form return rate of 80% or higher, regardless of whether parental consent or non-consent was indicated, received a 15-min visit from the school mascot of the authors’ academic institution. Of the 602 consent forms sent home to parents, 474 (79%) were returned. Of the returned consent forms, 401 (84%) indicated consent for participation. Of the 401 forms indicating consent, 382 (95%) were present on the day of the study and completed study measures. Listwise deletion of the 27 participants who provided incomplete data on the variables used in this study (∼7% of those completing measures) resulted in a final sample of 355. Participating children completed the study measures in their regular classrooms. Research assistants read each measure aloud to the students to eliminate reading comprehension as a confounding factor in study procedures. Additional research assistants were available to ensure participant understanding of directions and compliance with instructions. These procedures were approved by the Human Subjects Committee at the authors’ institution.

Development of Negative Attitudes Toward Physical Activity Scale

A new measure assessing negative attitudes toward physical activity was created by the authors of this article. The measure is modeled after the attitudes toward physical activity scale (Motl et al., 2000), but focuses on negative beliefs about physical activity (see Table I for items, descriptive statistics and factor loadings). Consistent with our conceptualization of attitudes toward physical activity, items assessing both perceptions of the immediate experience and consequences of physical activity were included. The same stem (i.e., “If I were to be physically active on most days …”) and five-point Likert scale (1 = disagree a lot; 5 = agree a lot) used in the Motl et al. measure were used in creating the negative attitudes scale. Based on clinical experience in health promotion with children and adolescents, the authors generated a preliminary list of 12 items representing potential negative attitudes toward physical activity, which were subsequently reviewed by the larger research team. The items were discussed and the scale was reduced to the nine items believed to best represent the negative attitudes construct based on consensus within the team. All nine items were included in the questionnaire packet administered to the participants. Eight of the nine items were retained based on their performance in exploratory factor analysis (EFA) procedures (see Results section). The final scale demonstrated good internal consistency in the present sample (α =.82).

Measures

Positive Attitudes toward Physical Activity

Positive attitudes were assessed using an eight-item self-report measure designed to examine children's beliefs about the benefits of being physically active (Motl et al., 2000). All items followed the stem “If I were to be physically active on most days …” Children rated various positive (e.g., “… it would help me cope with stress”) completions of this stem on a five-point scale ranging from 1 (disagree a lot) to 5 (agree a lot). Motl and colleagues (2000) demonstrated that the items included in this measure conformed to a unilateral model that was invariant across one year with eighth and ninth grade females and also demonstrated evidence of factorial validity. Previous research using this measure found these attitudes to be a significant predictor of behavioral intentions related to physical activity as well as a small but significant bivariate correlation with measures of physical activity (Motl et al., 2002). In the current study, two items were excluded from the analyses based on their performance in EFA procedures (see Results section). The resulting six-item scale, including means, standard deviations, and factor loadings is presented in Table I. The scale demonstrated good internal consistency in the current sample (α =.74).

Physical Activity

Physical activity was measured using the Self-Administered Physical Activity Checklist (SAPAC; Sallis et al., 1996). The SAPAC is a self-report measure developed for use with preadolescent children that consists of a list of 21 different physical activities, and allows children to report their engagement in physical activities before, during, and after school. Additional spaces are provided for reporting activities not appearing on the questionnaire. Sallis and colleagues (1996) assessed the reliability of this measure by comparing self-report and interview format reports. This comparison yielded correlations of r =.64 to r =.79. Additionally, Sallis and colleagues (1996) demonstrated that the self-report SAPAC was moderately and significantly correlated with objective measures of physical activity (e.g., heart rate monitor, r =.59, accelerometer, r =.32).

In its original development, this measure required children to report one-day recall of minutes engaged in each activity as well as subjective levels of intensity. In order to develop a broader profile of activity and to avoid reported recall issues related to duration and intensity (Dishman et al., 2004), this measure was modified in the following ways: (a) children were not required to report minutes engaged in activity; (b) children were not asked to give subjective reports of intensity; and (c) children were asked to report activity engagement over a three day period. McMurray and colleagues (2004) demonstrated the validity of this activity-based approach, reporting significant correlations with data from accelerometers. This study also found that the activity-based approach correlated more highly with accelerometer data for vigorous physical activity than a time-based approach. All children completed study measures on a Wednesday, ensuring that each child's reporting period included one weekend day and two week days.

Consistent with Sallis et al. (1996), a total weighted physical activity score was computed for each participant using metabolic equivalent task (MET) values obtained from the Compendium of Physical Activities (Ainsworth et al., 1993). Each physical activity was multiplied by its MET value and weighted activity scores were subsequently summed to obtain a total weighted activity value for each participant. The mean weighted physical activity score for the present sample was 216.6 (SD = 145.8) and this variable was positively skewed.

Body Mass Index

Participants’ height (in.) and weight (lbs) were collected by school nurses as part of a district-mandated health assessment conducted during the first quarter of the academic year. This information was provided to study personnel by the school district for all consenting participants. Using height and weight values, Body Mass Index (BMI) was calculated for each individual according the following formula: weight (lbs)/[height (in.)]² × 703 (Centers for Disease Control and Prevention, 2007). BMI percentile was then calculated for each individual using EZ BMI Software (Vosbury, 2007). For the present study, 65.9% of the participants were in the healthy weight range, 18.3% were at-risk for overweight, 13.8% were overweight, and 2.0 were underweight, based on CDC criteria (Centers for Disease Control and Prevention, 2007).

Results

Analytic Plan

To allow for cross-validation of the factor structure, the sample was split in half by randomly assigning participants to one of two sub-samples. The first sub-sample was used for EFA procedures to identify the items to be included in the final factors and the second sub-sample was used for the CFA procedures to confirm the factor structure. The full sample was used for concurrent validity analyses using SEM to examine the relationship between attitudes and self-reported physical activity. In these analyses, positive and negative attitudes were represented as latent constructs, with each retained item loading on its respective factor, and BMI, gender, and physical activity were represented as observed variables (Figure 2). Because the χ² statistic is highly sensitive to sample size (Kline, 2005), alternative fit statistics such as RMSEA, CFI, and NNFI were used to evaluate model fit for all CFA and SEM analyses. For each analysis, the sample size exceeded the number of parameters estimated and was generally considered sufficient to produce reliable estimates based on guidelines outlined by Kline (2005). Finally, to examine the typology of attitudes, median splits for positive and negative attitudes were created, and a 2 × 2 Analysis of Variance (ANOVA) was conducted. Median splits were used to divide the sample into “high” and “low” on each type of attitudes to correspond to the heuristic developed in the attitudes typology. CFA and SEM analyses were conducted using LISREL 8.80 (Jöreskog & Sörbom, 2007).

Exploratory Factor Analysis

All items from the positive and negative attitudes scales were included in an EFA using principal axis factoring and a varimax rotation. Given the hypothesized two-factor structure, the solution was constrained to produce two factors. Items that performed poorly in the EFA (i.e., loadings <.40 or cross-loading >.30) were removed one at a time until an acceptable solution was produced. A total of three items were dropped due to insufficient loadings and/or high cross-loadings, resulting in eight negative items and six positive items that were retained. The final two-factor solution accounted for 47.8% of the variance. Alternative models constrained to one- and four-factors were also tested. The one-factor model produced several items with insufficient loadings and the four-factor model produced an uninterpretable solution with numerous insufficient loadings and high cross-loadings. Consistent with expectations, the two-factor model produced an acceptable solution with items loading on either positive or negative attitudes to be used in the CFA cross-validation.

Confirmatory Factor Analysis

The factor structure found in the EFA was cross-validated using a two-factor CFA in the second sub-sample. The latent constructs were allowed to correlate but no cross-loadings were allowed. This model demonstrated acceptable fit, χ² (76, N = 178) = 159.54, p <.001; RMSEA =.079 (CI₉₀ =.062−.096), CFI =.93; NNFI =.92, and the latent correlation between positive and negative attitudes was.50. The two-factor model was then compared to an alternative one-factor model using a nested model comparison technique. The latent correlation between positive attitudes and negative attitudes was constrained to −1.0 to test the model assuming that positive and negative attitudes represent a unitary (but perfectly negatively correlated) construct. The one-factor model fit poorly, χ² (77, N = 178) = 266.82, p <.001; RMSEA =.137 (CI₉₀ =.122−.153); CFI =.79; NNFI =.75 and represented a significant degradation in fit compared to the two-factor model using the χ² difference test (Steiger, Shapiro, & Browne, 1985), χ²Δ(1) = 107.28, p <.01. These analyses suggest that the two-factor model is empirically superior to the one-factor model; that is, positive and negative attitudes toward physical activity are correlated but distinct constructs.

Concurrent Validity Analyses

To establish the concurrent validity of the negative attitudes measure, a model with negative attitudes as the independent variable and self-reported physical activity as the dependent variable was examined, controlling for gender and BMI percentile. The measurement model, which was conducted prior to the path model and included free latent correlations but no beta paths (i.e., directional regression paths), demonstrated acceptable fit, χ² (41, N = 355) = 110.42, p <.001; RMSEA =.073 (CI₉₀ =.058–.089); CFI =.95; NNFI =.93. When beta paths were added in the path model, negative attitudes significantly predicted physical activity (β = −.24, p <.01), controlling for gender and BMI percentile. Neither gender (β =.03, p >.05) nor BMI percentile (β =.02, p >.05) significantly correlated with physical activity.

To examine the predictive value of the negative attitudes construct in relation to positive attitudes, positive attitudes was added to the model as a predictor of physical activity (Figure 2). The measurement model demonstrated acceptable fit, χ² (112, N = 355) = 293.01, p <.001; RMSEA =.072 (CI₉₀ =.063−.082); CFI =.92; NNFI =.90 (see Table II for latent correlations among constructs). In the structural model, positive and negative attitudes were allowed to correlate with each other and predictive paths to physical activity were tested to evaluate the unique contribution of each construct “controlling” for the other. Negative attitudes remained a significant predictor of self-reported physical activity in the model controlling for positive attitudes (β = −.21, p <.01) and accounted for 4.4% of the variance in predicting physical activity. In contrast, positive attitudes was not a significant predictor of physical activity with negative attitudes in the model (β =.05, p >.05). A summary of fit statistics for all structural equation models is presented in Table III.

Test of the Attitudes Typology

To examine the usefulness of the attitudes typology proposed in this article, a 2 × 2 ANOVA was conducted using self-reported physical activity as the dependent variable and median splits of positive and negative attitudes as the independent variables. A main effect was found for negative attitudes, F(1,352) = 11.57, p =.001, partial η² =.032. The main effect for positive attitudes was only marginally significant, F(1,352) = 2.93, p =.09. The positive × negative interaction was not significant. Inspection of the means supports the hypotheses proposed regarding the physical activity levels of each group (Figure 3).

Discussion

The present study examined the psychometric properties and predictive value of a new measure of negative attitudes toward physical activity among a preadolescent school sample. Consistent with our hypotheses, the results indicated that the measure demonstrated good internal consistency and provided preliminary evidence of construct validity. Furthermore, negative attitudes toward physical activity were shown to represent a unique (i.e., not redundant) construct, distinct from positive attitudes. The results also supported the concurrent validity of the negative attitudes construct based on a significant correlation with self-reported physical activity. The relationship remained significant even after controlling for positive attitudes, demonstrating a unique predictive value associated with negative attitudes. Furthermore, the results for physical activity were consistent with predictions based on our proposed attitudes typology, giving the model preliminary support as a potentially useful heuristic in conceptualizing the role of attitudes in pediatric physical activity.

The results from the current study highlight the role of negative attitudes in predicting pediatric physical activity and are consistent with theories that focus on attitudes and related constructs. As discussed earlier, several major theories, such as the Theory of Reasoned Action (and Theory of Planned Behavior), social cognitive theory, and the health beliefs model focus on the importance of similar cognitive factors in directly or indirectly influencing health behaviors. Although this study did not explicitly test these theories, it may be useful in guiding future examinations by highlighting the importance of negative attitudes in understanding cognitive factors affecting pediatric physical activity.

Consistent with our hypotheses, the present study found negative and positive attitudes toward physical activity were related, but distinct constructs. The latent correlation between the two constructs was moderate and in the expected direction. More importantly, the variance that was unique to negative attitudes showed a unique value in predicting physical activity, although the amount of variance explained was relatively small. These results provide initial evidence to counter the assumption that negative attitudes are merely the absence of positive attitudes and, instead, highlight the potentially meaningful role of negative attitudes in relation to physical activity. Negative attitudes, encompassing negative perceptions of the experience of physical activity as well as its potential consequences, may represent a cognitive obstacle to meeting recommended activity standards, even in the presence of significant positive attitudes. The attitudes typology presented in this study further supported the role of negative attitudes and group means were consistent with expectations.

Limitations

A number of limitations to the present study should be noted. First, the negative attitudes scale was developed based on the clinical experience of the authors, and other methods such as expert consultation or focus groups that might have further contributed to the content validity of the measure were not employed. Second, the present study consisted of only one data collection, so there was not an opportunity to refine the negative attitudes measure and test it within a second sample. Instead, a split-sample strategy was required to conduct an EFA and CFA on different random subsets of the sample. The single data collection in this study limits generalizability, and future research should further validate the factor structure of the negative attitudes measure found in this investigation. Third, the use of retrospective self-report for physical activity (the SAPAC; Sallis et al., 1996) could have affected the accuracy of the activity estimates and the non-significant relationship between positive attitudes and physical activity. However, previous research with the SAPAC found evidence of convergent validity with objective measures, suggesting it was appropriate for this preliminary investigation. To overcome this limitation, future research should attempt to replicate the findings of the present study using objective measures of activity (e.g., pedometers, accelerometers) or instruments with shorter recall periods (e.g., daily diaries).

Fourth, it should be noted that, although the negative attitudes construct was a significant predictor of physical activity, it accounted for a relatively small portion of the total variance. Clearly, preadolescent attitudes are only one of numerous factors contributing to pediatric activity levels. Future research should examine the role of attitudes within the context of other important influences, such as the individual's physical environment, parental attitudes, and peer activity patterns, to develop a more complete understanding of preadolescent physical activity. Fifth, the current study did not include data on the socio-economic status (SES) of individual participants. Given disparities in health and many health-related behaviors, future studies may incorporate examinations of SES to determine if the role of attitudes in predicting health behaviors differs by SES. Finally, the current study did not examine potential ethnic differences in responses to the measures in this study, which could be a useful area of future investigation.

Clinical Implications

Despite the limitations of the current investigation, the results suggest important clinical implications regarding the role of negative attitudes toward physical activity in cases where increased physical activity is an identified goal. This study suggests that negative attitudes are a contributor to exercise behavior and may actually play a larger role than the positive attitudes that have been studied previously. Based on these results, clinicians working with preadolescents on improving physical activity should consider explicitly assessing negative attitudes early in an intervention and throughout treatment. Merely asking youths about positive attitudes (e.g., “Do you think physical activity is fun?,” “Do you see benefits to physical activity?”) may neglect important negative attitudes that can represent obstacles to activity. For individuals with ambivalent attitudes (i.e., high positives and high negatives), identifying the negative attitudes that may offset or even overshadow more positive perceptions of activity, can be an important step in isolating targets for intervention. Challenging these negative attitudes and working with the individual to develop creative strategies for reducing the unpleasant experiences or consequences of activity is consistent with CBT approaches focused on challenging negative cognitions that affect behavior (e.g., Brownell, Kelman, & Stunkard, 1983; Coates & Thoresen, 1981; Herrera, Johnston, & Steele, 2005). Given the hypothesized value of challenging attitudes toward physical activity, changes in attitudes could be an important mediator of intervention outcomes and future research should explore the malleability of these attitudes as well as the potential mediating role of such attitude change in bringing about changes in activity levels.

Conclusions and Future Directions

Consistent with existing theories of health behavior, the current study suggests that attitudes toward physical activity have a significant, although modest, impact on activity levels among preadolescents. Furthermore, this study highlights the role of negative attitudes, specifically, as potentially an even stronger influence than positive attitudes in predicting physical activity. Future work should expand upon these findings by investigating potential mediator models involving attitudes, self-efficacy, physical activity, and weight status in longitudinal studies. We recommend incorporating measurement and targeting of negative attitudes into treatments for increasing physical activity, which would allow for investigation of the modifiability of maladaptive attitudes and the effect of attitude change on treatment outcomes. Finally, the role of negative attitudes should be examined among different populations beyond the current non-clinical, preadolescent sample. Negative attitudes among adolescents may be an important area for study as attitudes become more solidified and responsibility for health behaviors increases. Also, studying negative attitudes among certain chronically ill populations, such as individuals who are extremely obese (e.g., >99th percentile for BMI) or those with chronic pain, might inform intervention efforts within a wider range of individuals.

Acknowledgment

The authors express their gratitude to Dr Ric Steele for his assistance with this manuscript.

Conflict of interest: None declared.

References

Author notes