Cognitive and Motivational Factors Associated with Sedentary Behavior: A Systematic Review

Excessive time spent in sedentary behavior (SB) is associated with numerous health risks. These associations remain even after controlling for moderate-to-vigorous physical activity (PA) and body mass index, indicating that efforts to promote leisure time physical activity alone are insufficient. Cognitive and motivation variables represent potentially modifiable factors and have the potential of furthering our understanding of sedentary behavior. Hence, a systematic review was conducted to synthesize and critique the literature on the relationship between cognitive and motivational factors and sedentary behaviors. In April 2016, four electronic databases (Psych info, Pub Med, SPORTDiscus, Web of Science) were searched and a total of 4866 titles and abstracts were reviewed. After meeting inclusion criteria, study characteristics were extracted and the methodological quality of each study was assessed according to the Downs and Black Checklist. PRISMA guidelines for reporting of systematic reviews were followed. Twenty-five studies (16 cross-sectional, 8 longitudinal and one examining two populations and employing both a cross-sectional and prospective design) assessed 23 different cognitive and motivational factors. Seventeen studies were theory-based and 8 did not employ a theoretical model. Results showed that among SB-related cognitions, risk factors for greater sedentary time included having a more positive attitude towards SB, perceiving greater social support/norms for SB, reporting greater SB habits, having greater intentions to be sedentary, and having higher intrinsic, introjected, and external motivation towards SB. Protective factors associated with lower sedentary time included having greater feelings of self-efficacy/control over SB and greater intentions to reduce SB. Among PA-related cognitions, protective factors for lower SB included a more positive attitude towards PA, having greater social support/norms for PA, greater self-efficacy/control for PA, higher PA intentions, and higher intrinsic and identified motivation towards PA. In addition, feeling more supported and empowered in general was related with lower levels of SB. The average methodological quality score for included studies was 69% (SD = 9.15%; range 35–80%). In conclusion, a number of cognitive and motivational factors were identified that were associated with sedentarism. These findings have come from reasonably high quality studies. To further extend our understanding of the relation between cognitive and motivational factors and SB, more longitudinal, theory-driven studies examining cognitions and motivation from a sedentary perspective are required.


Introduction
Excessive time spent in sedentary behavior is associated with numerous health risks. An overview of 27 systematic reviews found that among adults, sedentary time is positively associated with all-cause mortality, fatal and non-fatal cardiovascular disease, type 2 diabetes, metabolic syndrome, and several types of cancers [1]. Among children and youth, the risks include obesity, increased blood pressure and total cholesterol, poorer self-esteem, social behavior problems, poorer physical fitness and lower academic achievement [1]. These associations remain even after controlling for moderate to vigorous physical activity and body mass index (BMI), indicating that efforts to promote leisure time physical activity alone are insufficient.
Sedentary behavior has been defined as "any waking behavior characterized by an energy expenditure ≤1.5 METs while in a sitting or reclining posture" [2]. Sedentary behaviors permeate all domains of life, including work, school, transportation, leisure/recreation, and spiritual/contemplative pursuits. The pervasiveness of sedentarism is evident through population-based studies, which indicate that Canadian and US adults spend an average of 9.7 and 7.7 hours per day, respectively, being sedentary [3,4]. The high prevalence of sedentarism and its adverse outcomes has added a whole new paradigm to the physical activity field focused on understanding and reducing sedentary time.
to include possible correlates or if they did not measure predictors and behavior within the same individual (e.g., studies examining the relationship between parental beliefs and children"s sedentary behavior were excluded). Studies examining mental health outcomes such as affect (e.g., depression, anxiety), quality of life, and physical self-perceptions were also excluded because these constructs are often viewed as consequences rather than antecedents of sedentary behavior. Finally studies that examined personality were excluded as they represent constructs that are considered stable and hence less modifiable.
All selected studies  were summarized in table format and data were extracted with regards to the author(s) and publication year, study population, sample size, sampling methods, study design, correlates/predictors examined, type and measurement of sedentary behavior or sedentary time, and the results pertaining to the relationship between behavior and significant correlates/predictors. In addition to summarizing the findings in table format and in text, we have visually represented the findings using what we have termed a pinwheel. The purpose of the pinwheel is to illustrate, at a glance, which constructs have been examined in the literature as well as whether a relationship emerged between the constructs. Within the health domain, sedentary behavior is considered a risk behavior. For this reason, the colour green was chosen to indicate a protective effect (i.e., lower sedentariness) due to its association with safety and the word "go-ahead" (e.g., its use in traffic lights). On the other hand, red is associated with a hazard and the word "stop". For this reason, we used the colour red to indicate an association between a factor and increased sedentary behavior. Yellow was chosen to indicate a null effect due to the fact that it is seen as in-between green and red (e.g., on a traffic light signal).
The methodological quality of individual studies was assessed using the Downs and Black checklist [60]. The Downs and Black instrument assessed study quality including strength of reporting, external validity, internal validity (bias), internal validity (confounding), and power. The checklist consists of 27 items with a maximum score of 32 points. A modified version of the checklist was employed with items that were not relevant to non-experimental studies removed (8, 13-15, 17, 19, and 21-24). The adapted checklist consisted of 20 items, including 14 items from the original list (1-3, 6-7, 9-12, 16, 18, 20, and 25-26); three items that were modified (4, 5, and 27); and three items created for purposes of this review. Reporting items 4 and 5 from the original list were reworded to align with non-intervention (i.e., cross-sectional and prospective) studies being examined in this review. Item 27, concerning power from the original list was modified to address the number of participants needed to detect a significant association between an exposure and sedentary behavior. Of the three items created, two were internal validity criteria and one was concerned with study power. We believe that changes made to the original checklist had merit and that modifications held value in assessing the methodological quality of studies included in this review. Each quality criterion was rated as positive (1), negative (0), or unknown/insufficiently described (0). A positive sign (+) was given if the publication provided a sufficient description of the item, per the predefined criteria, and met the quality criteria for the item. A negative sign (-) was allotted if the publication did not provide an adequate description or did not address and/or perform the quality criteria for the item. Finally, if an insufficient or unclear description of the item was provided, a question mark (?) was given. The maximum possible score for the modified checklist was 20 points (higher scores indicate higher quality). The methodological quality of individual studies was independently scored by SR and verified by HP; if disagreements between assessors occurred, consensus was achieved through discussion with a third reviewer (AG). For each study, an overall methodological quality score was calculated. In addition, the percentage of studies meeting each quality criterion was calculated.
Data were not pooled for a number of reasons. First, there was little consistency among studies with respect to exposures and even when the same exposures were examined by multiple studies, they often used different scales. Second, studies used varying methodologies and reported statistics inconsistently. Therefore, to synthesize the evidence and allow conclusions to be drawn regarding the relationship between cognitive and motivational factors and sedentary behavior, a best-evidence synthesis that has been used in previous reviews [61] was implemented. The findings for each cognitive and motivational variable were interpreted on the following basis: there was no evidence of an association if more than 50% of the cross-sectional and prospective studies reported no association; there was inconclusive evidence for an association if 50% of the studies reported no association and 50% reported a positive or negative association; there was some evidence of an association if more than 50% of the studies reported a positive or negative association; and there was consistent evidence of an association if all of the studies reported a positive or negative association.

Results
The electronic search produced 4,866 articles (1298 from PsycINFO, 2595 from PubMed, 699 from SPORTDiscus, and 274 from Web of Science; Figure 1). After removing duplicates (n = 1121), a total of 3745 publications remained. After titles and abstracts were examined, 86 full-text articles were read and assessed further for eligibility. Of those, 21 articles were identified as suitable.  (Table S1). Eight [21,23,26,28,29,32,34,44] of the 25 reviewed studies did not specify a theoretical orientation in their study design and/or in the cognitive and motivational factors examined. Of these, only two [23,28] were longitudinal or prospective in nature while the remaining six [21,26,29,32,34,44] employed an observational, cross-sectional design. Researchers have emphasized the need for more longitudinal, prospective studies to be completed to fully understand temporal changes in sedentary time and corresponding psychological predictors [5,17]. Five studies [21,28,29,32,34] [23,26,28,29,32,44] of the eight studies investigated correlates across multiple levels of influence (i.e., socio-demographic, physical environmental, social environmental, social-cognitive, psychosocial, health-related, work-related, behavioral) and two [21,34] examined only cognitive variables. Furthermore, only four [23,26,34,44] of the eight studies assessed cognitive factors from a sedentary perspective or in a sedentary-specific manner. One study [21] examined cognitive factors from a general point of view, while three studies [28,29,32] assessed the associations between physical activity and/or exercise-specific cognitive factors and sedentary behavior.  Table S1 and illustrated in Figure 2. Overall, the associations reported in Table S1 were small to medium in size. Five studies [23,26,29,34,44] investigated the relationship between attitudes and sedentary behavior. Of these, one study [29] found more positive attitudes towards exercise to be associated with lower sedentary behavior. Four studies [23,26,34,44] found more positive attitudes towards sedentary behavior to be associated with higher sedentary behavior. Contrary to expectations, one study [26] found more positive attitudes towards sedentary behavior to be associated with lower sedentary behavior. Five studies [21,23,26,28,32] examined the relationship between social support and/or norms and sedentary behavior. One study [21] found greater support in life to be associated with lower sedentary behavior, while one [32] study found greater support for physical activity to be associated with lower sedentary behavior. Three studies [26,28,32]   Of the 25 studies included in this review, 17 were theoretically driven in their approach (see Table S1). Of these, 10 [11], Protection Motivation Theory (PMT) [12], and Self-Determination Theory (SDT) [15]. Furthermore, 11  Cognitive and motivational factors grounded in a theory-based framework and their respective associations to sedentary behavior are summarized in Table S1 and illustrated in Figure 2. Overall, the associations reported in Table S1  . The average score of the included studies for the quality sub-scales of reporting, external validity, internal validity, and power were 88%, 31%, 71%, and 12%, respectively. Also highlighted through the assessment was the percentage of studies meeting each item on the checklist ( Table 1). The majority of studies satisfied the reporting criteria (items 1-9) with >80% of studies meeting each of the items 1-8.
However, only 42% of studies reported actual probability values for the main outcomes except where the probability value is less than 0.001 (item 9). In terms of the external validity criteria, items 10 and 11 attempt to address the representativeness of the findings of the study and whether they may be generalized to the population from which the study subjects were derived. Only 35% and 27% of studies met these items, respectively. The proportion of studies meeting the quality items with respect to internal validity (items 12-18) varied considerably per item, with only 35% of studies measuring the cognitive and/or motivation variables at a time prior to the assessment of sedentary behavior (item 13). Further, only 12% of studies scored positive on item 16 and included an objective assessment or some corroboration of the objective and subjective assessment in the measurement of sedentary behavior. For the power criteria (items [19][20], 88% of studies did not report a formal power calculation for determining the association between an exposure and sedentary behaviors (item 19).

Discussion
The purpose of this paper was to systematically review and critique the current literature on the role that cognitive and motivational processes play in understanding sedentary behavior. While other reviews have been conducted on socio-demographic and behavioral correlates of sedentary behavior, to our knowledge this is the first to focus exclusively on cognitive and motivational factors. self-efficacy or social support to sit less in the next month at work and occupational sitting time.
Contrary to expectations, De Cocker and colleagues found that adults who perceived higher social norms towards sitting less at work, reported greater benefits of sitting less, and had greater intentions to sit less at work reported higher occupational sitting time compared to respective comparison counterparts. They also found that employment status and occupational classification had a moderating effect on the association between control to sit less at work and occupational sitting time such that lack of control to sit less at work was positively associated with occupational sitting time among full-and part-time workers and white-collar and professional workers only. These findings suggest that those who are full-time, white-collar and/or professional workers may have positive attitudes towards sitting less and intentions to sit less; however, these individuals are also more likely to be employed in jobs that require prolonged sitting. Thus, in the absence of control, even attitudes and intentions are insufficient to lead to reduced sedentary behavior.
In a longitudinal study, Busschaert and colleagues [23] examined the relationship between changes in social-cognitive variables from baseline to one-year follow-up with changes in context-specific sitting times. They found that positive attitudes towards watching TV and computer use was associated with more sitting while watching TV and more sitting while using a computer, respectively. Higher perceived modeling of sedentary behavior (i.e., time partner spends watching TV) was associated with more sitting while watching TV and higher norms associated with computer use and motorized transport was associated with more sitting in those contexts. Self-efficacy to reduce computer use was associated with less sitting time while using a computer, whereas self-efficacy to use active transportation was associated with less sitting during motorized transport. Given that the associations between cognitive factors, motivational factors and sedentary behavior or sedentary time were small to medium in size, researchers interested in targeting these modifiable variables will need to take this into consideration when using these as agents of change for sedentary behavior interventions. Furthermore, these findings suggest that both physical activity related and sedentary-specific cognitive and motivational factors will play a role in understanding sedentarism. With respect to movement-related factors, research has shown a strong, inverse correlation between sedentary behavior and light-intensity physical activity [62], as well as a small to medium inverse correlation between sedentary behavior and leisure time physical activity [17,63]. If these behaviors are associated with one another, then it is highly likely that physical activity related cognitions could be associated with time spent sedentary. The findings, herein, serve to confirm this rationale and demonstrate that physical activity related cognitive and motivational factors are correlates of sedentary behavior. In order to maximize the contribution of studies examining physical activity related factors to our understanding of sedentary behavior determinants; researchers might need to measure these cognitions as they pertain to specific types of physical activity (i.e., total physical activity, light-intensity physical activity).
Based on the Downs and Black checklist [60] for assessment of the methodological quality, the findings from the included studies in this systematic review come from reasonably high quality studies (see Tables 1 and 2). For instance, 22 of the 26 reported studies had overall quality scores ≥65% and 11 of the 26 studies had overall quality scores ≥75%. We found no difference between the average quality Owen and colleagues [5] suggested that the "primary strategic goal for research on sedentary behavior determinants and interventions is to integrate evidence to identify effective or promising strategies to reduce sitting time." Further, Rhodes et al. [17] proposed that cognitive, social, and environmental correlates seem better suited for intervention efforts to reduce sedentary behavior.

Conflict of Interest
The authors declare no conflict of interest.

Association of family functioning and friendship quality with self-reported sedentary behaviors:
-Boys from better functioning families were less likely to report playing video games at the weekend (OR; 95% confidence interval: 0.73; 0.57,0.93) or reading for pleasure (weekday: 0.73; 0.56, 0.96 weekend: 0.75; 0.58,0.96).
-Boys who attained higher scores on the good friendship qualities scale were less likely to play video games at the weekend (0.61; 0.44, 0.86) or report high homework on weekdays (0.54; 0.31, 0.94).
-A higher score for good friendship qualities was associated with lower odds of girls playing video games during the week weekend days  -Perceived Competence significantly predicted SB (β = -0.28; 95% CI: -0.22, -0.14). Attraction to PA statistically significantly predicted SB in all age groups (β = -0.49; 95% CI: -0.22, -0.14). Thus, the students who felt more competent in PA and attracted to PA were more likely to be active and less sedentary. The effect of Perceived competence on SB was reduced but remained statistically significant after controlling for the effects of attraction to PA. Bootstrapping mediation analysis confirmed that perceived competence had a statistically significant indirect effect on SB (IE = 0.13, p < 0.05). -A one-unit higher score for 'I enjoy watching TV for many hours' (attitude 3) and 'I find TV a way to relax' (attitude 4) was associated with respectively 19 and 12 % more sitting while watching TV. Also, a one-unit higher score for 'time partner spend watching TV' (modelling 1) was associated with 5 % more sitting while watching TV.
-A one-unit higher score for 'I think using a computer is pleasant' (attitude 1), 'I enjoy using a computer for many hours' (attitude 3) and 'I think that I spend too much time on the computer' (norm) was associated with respectively 34, 17 and 24 % more sitting while using a computer. A one-unit higher score for 'I consider it possible that I do not use a computer for some days in the week' (self-efficacy 1) was associated with 13 % less sitting while using a computer.
-A one-unit higher score for 'I think that I -occupational desks at work or not spend too much time using motorized transport' (norm) was associated with 14 % more sitting during motorized transport. A one-unit higher score for 'I consider it possible to take the bicycle or to go by foot spontaneously even if it is possible to use a car' (self-efficacy 3) was associated with 19 % less sitting during motorized transport.

Relationship between changes in socialcognitive predictors from baseline to follow-up and changes in TV-viewing, computer use, motorized transport and occupational sitting:
-An increase from baseline to follow-up with one unit on the five-point Likert scale for 'I enjoy watching TV for many hours at a time' (attitude 3) was associated with 7.96 min/day more sitting while watching TV at follow-up. An increase from baseline to follow-up with one unit on the eight-point Likert scale for 'time partner spend watching TV' (modelling 1) was associated with 9.91 min/day more sitting while watching TV at follow-up.
-An increase from baseline to follow-up with one unit on the five-point Likert scale for 'I consider it possible to park the car somewhat further spontaneously and to walk the remaining distance' (self-efficacy 2) was associated with 8.48 min/day more sitting during motorized transport at follow-up. More active transport to go to work/school (modelling 1) from baseline to follow-up of the partner was associated with 16.47 min/day more sitting during motorized transport at follow-up of the respondent. -A higher number of minutes of sedentary behavior were associated with lower levels of empowerment (r = -0.498, p < 0.001) and self-efficacy for PA (r = -0.297, p < 0.001)

Predictors of sedentary behavior:
-Empowerment was found to be the strongest predictor of a high level of sedentary behavior (β = -0.394, p < 0.001).    -Employment status and occupational classification moderated the association between control to sit less and  [4], five hours or more [5]. -Among males, self-efficacy related to barriers to PA (B = -2.16 (-3.60, -0.73)) was inversely related to an increase in TST, indicating a decrease of around 2.2 hours per week per unit increase in selfefficacy score.

Predictors of tracking of high TST:
-Results of the multinomial regression analysis show that, among girls, children with low self-efficacy related to barriers to PA were more likely to track high TST (OR = 2.30, C.I. = 1.13-4.69, p < 0.05) compared to children with high selfefficacy.
-Among males, boys with low selfefficacy related to barriers to PA were also more likely to track high TST (OR = 6.83, CI = 3.22-14.45, p < 0.001) than the group with high self-efficacy.

Differences in variables btw low-and high-screen users:
-A significantly smaller proportion of high-screen users held negative attitudes about screen use (P < 0.01) -Intentions: More than two thirds of children indicated that they would elect to spend more time engaged in physical activities if they were ""given the choice""; however, fewer high-screen users than low-screen users (P < 0.01) chose to do so.
-Habit strength was a significant, positive predictor of sedentary behavior (γ 03 ), so that people with stronger habits for sedentary behavior engaged in more sedentary behavior.
-The interaction between daily planning and sedentary behavior habit strength was not a significant predictor of daily sedentary behavior (γ 05 ).
-Participants who had stronger usual intentions to limit or interrupt sedentary behavior had lower usual levels of physical activity (γ 06 ).
-On days when participants intended to limit or interrupt sitting time more than was typical for them, they reported lower levels of sedentary behavior (γ 10 ).
-Intentions also had weak-to-moderate positive correlations with sedentary behavior risk perceptions and lightintensity physical activity outcome expectations (rs = 0.20, 0.06, respectively) at the between-person level.
-Intraclass correlation coefficients (ICC) were calculated to describe the proportion of variance in each variable attributable to between-person differences. ICCs indicated that approximately half of the behavior -ActivPAL3 activity monitors used physical symptoms) and wore the activity monitor on their thigh during all sleeping and waking hours. variance in self-reported and objectively measured sedentary behavior and two thirds of the variance in task self-efficacy, intentions, and planning was the betweenperson variance, with the remainder driven by within-person factors and measurement error.

Multilevel model of daily sedentary behavior:
-Multilevel models predicting behavior revealed that sedentary behavior was (a) negatively associated with planning to reduce sedentary behavior at the withinperson, and (b) positively associated with sedentary behavior habit strength (monitored behavior: γ 02 = 19.97, p = 0.04).
-There were no differences in objectively monitored sedentary behavior between participants who tended to form stronger or weaker plans (γ 01 = -0.41, p = 0.24) but, as hypothesized, participants were less sedentary on days when they formed stronger-than-usual plans to limit sedentary behavior (γ 10 = -0.51, p = 0.005). 2 -As indicated by the pseudo-R , this model accounted for 14% of the variance in objectively measured sedentary behavior, with habit strength accounting for 9% and daily planning accounting for 5% of the explained variance.
Multilevel model of daily plans to limit SB: -Plans to limit sedentary behavior were (a) positively associated with task selfefficacy at the within-person level (γ 10 = 0.14, p = 0.001), but (b) negatively associated at the between-person level (γ 01 = -0.59, p = 0.04), and (c) positively associated with intentions at the between-(γ 02 = 1.17, p = 0.001) and within-person level (γ 20 = 0.20, p = 0.004). 2 -As indicated by the pseudo-R , this model accounted for approximately 20% of the variance in daily plans to limit sedentary behavior. Daily intentions accounted for 23%, daily task selfefficacy accounted for 10%, and usual intentions and task self-efficacy each accounted for 2% of the explained variance.

Multilevel model of intentions to limit SB:
-Intentions to limit sedentary behavior were (a) positively associated with task self-efficacy at the between (γ 01 = 0.96, p = 0.001) and within-person level (γ 10 = 0.61, p = 0.001), but (b) not associated with light-intensity physical activity outcome expectations, sedentary behavior risk perceptions, or sedentary behavior habit strength. 2 -As indicated by the pseudo-R , this model accounted for approximately 44% of the variance in daily intentions to limit sedentary behavior, with daily task selfefficacy accounting for 80% and usual task self-efficacy accounting for 4% of the explained variance.

Multivariate model for girls:
-Included all of the variables that were associated with the outcome from the unadjusted bivariate analyses.
-The R 2 for the main-effects model was 0.25, and the inclusion of the interaction term increased the R 2 to 0.28. The Hosmer-Lemeshow test indicated that the fit of the model was good (P = 0.25). -Intention was correlated with attitude (0-4) in only one model, but was related to attitude (half) and attitude (12)(13)(14)(15)(16) in three models. Subjective norms were associated with intention in four of the five models and PBC showed an association only in one model.
-For behavior, intention emerged as a significant correlate in all five models. Behavior was related with attitude (0-4) in one model, attitude (half) in three models, and attitude (12)(13)(14)(15)(16) in two models, SN in three models and PBC in a single model.
Variables predicting sitting and socializing more than 8 hr/week: measure -Frequency and duration of participation in walking, moderateintensity activity, vigorous activity, and total leisuretime activity.

Associations of psychosocial variables with leisure-time internet use:
-Concerning the psychosocial factors, perception of higher social norm from family towards Internet use (β = 0.161, p = 0.011) and more cons (β = 0.187, p = 0.002) were related to more leisure-time Internet use. Moreover, more pros (β = -0.116, p = 0.009) and higher selfefficacy about reducing leisure-time Internet use were associated with less Internet use (β = -0.285, p < 0.001). Global me" (β = 0.10) expressing a more positive attitude towards sitting with increasing sitting durations.
Variables predicting work-related sitting time: -In model 4, for men, the belief "Sitting for long periods does not matter to me" (recoded) (β = 0.10) was positively correlated with work-related sitting time, reflecting more positive attitudes towards sitting with increasing sitting durations.
-For women, for the cognitive variables, no associations were found. -In the general model, scheduling SE productive/focused (r = -0.13, p < 0.05) and scheduling SE studying in library/Wi-Fi area (r = -0.14, p < 0.05) were significantly related to sedentary behavior.

Variables predicting sedentary behavior:
-For goal intention, 5% and 1% of the variance was explained in the general and leisure model, respectively. RE and scheduling SE studying at home were significant contributors for the general model only.
-Intention: goal intention (GI), implementation intention (II) activities Exercise behavior: Leisure Score Index (LSI) of the Leisure Time Exercise Questionnaire -Four-item assessment that measures intensity and frequency of physical activity -For implementation intention, 10% and 16% of the variance was explained in the general and leisure model, respectively. In the general model, PV, RE, and scheduling SE productive/focused were significant contributors. For the leisure model, PV, RE, and scheduling SE studying at home were significant contributors.
-For sedentary behavior, 3% and 1% of the variance was explained in the general and leisure model, respectively. Goal intention was a significant contributor in the leisure model only.