- Email: [email protected]
Less sitting and more moving in the office: Using descriptive norm messages to decrease sedentary behavior and increase light physical activity at work
Psychology of Sport and Exercise 19 (2015) 76e84
Contents lists available at ScienceDirect
Psychology of Sport and Exercise journal homepage: www.elsevier.com/locate/psychsport
Less sitting and more moving in the office: Using descriptive norm messages to decrease sedentary behavior and increase light physical activity at work Carly S. Priebe*, Kevin S. Spink University of Saskatchewan, Canada
a r t i c l e i n f o
a b s t r a c t
Article history: Received 27 November 2014 Received in revised form 26 February 2015 Accepted 27 February 2015 Available online 18 March 2015
Objectives: The sedentary nature of office workplaces is an increasing concern (Marshall & Ramirez, 2011). One avenue through which sedentary behavior may be decreased and beneficial activity increased is through social norms. Previous research (e.g., Priebe & Spink, 2012) has demonstrated that perceptions about others' behavior (i.e., descriptive norms; Cialdini, Reno, & Kallgren, 1990) can influence physical activity. However, it is unclear if descriptive norms can be used to impact sedentary time. This study examined whether descriptive norm messages would impact sedentary behavior and light activity in an office setting. Given the possible importance of personal and contextual characteristics of the norm reference group, a secondary purpose was to examine messages that varied in reference group characteristics. Design: This study utilized a pre-post experimental design. Method: Office workers were randomly assigned to receive one of four email messages containing descriptive norms about co-workers’ behavior. Sedentary behavior and activity were measured before and after message receipt. Results: A repeated measures MANOVA revealed a main effect for time, p < .001, hp 2 ¼ .316. Those who received descriptive norm messages about co-workers' lower sedentary behavior and greater stair use and walking decreased their own sitting time while increasing stair use and walking at the office, respectively, p's < .05. No differences emerged between participants receiving information about groups that varied in reference group characteristics, p's > .10. Conclusion: Results provide experimental evidence that descriptive norm messages may serve to decrease sedentary and increase light activity in an office setting. © 2015 Elsevier Ltd. All rights reserved.
Keywords: Descriptive norms Office workers Light activity Sedentary behavior Email
While it is generally accepted that most North Americans could improve their health by increasing physical activity levels (Cragg, Wolfe, Griffiths, & Cameron, 2007), less attention has been paid to the observation that individuals could benefit from decreasing their sedentary time (Marshall & Ramirez, 2011). This is surprising as studies have shown that sedentary behaviors, commonly defined as “behaviors that involve sitting and low levels of energy expenditure” (Marshall & Ramirez, 2011, p. 519), carry health risks independent of insufficient physical activity (Biddle, Gorely, Marshall, Murdey, & Cameron, 2004; Pate, O'Neill, & Lobelo, 2008).
* Corresponding author. College of Kinesiology, University of Saskatchewan, 87 Campus Drive, Saskatoon, Saskatchewan, S7N 5B2, Canada. Tel.: þ1 306 227 2406; fax: þ1 306 966 6464. E-mail address: [email protected] (C.S. Priebe). http://dx.doi.org/10.1016/j.psychsport.2015.02.008 1469-0292/© 2015 Elsevier Ltd. All rights reserved.
Despite these health risks, the typical adult in Canada is estimated to spend about 9.5 h of waking time being sedentary, with most of it sitting (Colley et al., 2011). Further, adults in office jobs are likely sedentary for more time when compared to those in occupations such as the trades (Hu, Li, Colditz, Willett, & Manson, 2003). As such, both the World Health Organization (2008) and researchers (e.g., Chau et al., 2010; Marshall & Ramirez, 2011) acknowledge the need for more research examining ways to decrease sedentary behavior (sitting) in office settings. It has been suggested that one practical way to reduce sitting time would be to break it up with light intensity activities (Chau et al., 2010; Marshall & Ramirez, 2011). Breaking up total sitting time appears to relate to health benefits as researchers have identified important health-related outcomes (e.g., improved metabolic profile) when individuals move from sitting to standing
C.S. Priebe, K.S. Spink / Psychology of Sport and Exercise 19 (2015) 76e84
(Hamilton, Hamilton, & Zderic, 2004). Further, in a large population-representative sample, Healy, Matthews, Dunstan, Winkler, and Owen (2011) found breaks in sedentary behavior (independent of sedentary behavior) were beneficially associated with a known marker of health, waist circumference. In addition, these associations between breaks in sedentary time and health have been found independent of overall physical activity (Gilson et al., 2009), emphasizing the need to consider sedentary behavior as its own entity (Owen, Healy, Matthews, & Dunstan, 2010).
77
by Terry and Hogg (1996), it is possible that descriptive norms are only effective when they refer to groups where the individual feels some type of similarity (e.g., more similar friends compared to public service announcements conveying that others are active, where similarity to others is less clear). In addition to having a conceptual underpinning, empirical support exists for a stronger relationship between descriptive norms and physical activity in those who highly identify with their reference group (e.g., Rimal, Lapinski, Cook, & Real, 2005). Personal and contextual similarity
Descriptive norms and activity Given the important links between physical activity, sedentary behavior, and health, there is a need for research understanding how we might change these behaviors, especially in an office setting. While there are a number of possible ways to target change in activity and sedentary behavior, one possibility is through social influence. One form that might hold particular potential in this area is descriptive norms. In focus theory of normative conduct, Cialdini, Reno, and Kallgren (1990) define descriptive norms as perceptions about the prevalent behavior of others. These norms have been examined with a number of behaviors (e.g., environmental conservation, Cialdini, 2003; alcohol consumption, Rimal, 2008; Rimal & Real, 2005), and recent research has emerged in the physical activity setting (e.g., Priebe & Spink, 2011, 2012). Descriptive norms and sedentary behavior As descriptive norm messages have shown some promise in the area of increasing light physical activity (stair use) in office workers (e.g., Priebe & Spink, 2012), it also might be important to extend this to examining norms for other types of light activity that might break up sedentary time (e.g., walking within the office and standing up during the work day). Also, as descriptive norms often are examined in relation to decreasing the prevalence of other negative behaviors (e.g., less alcohol consumption, Rimal & Real, 2005; less littering, Cialdini et al., 1990), using normative information to decrease sedentary behavior appears plausible (i.e., less sitting time). The main purpose of this experimental field study was to examine the effect of descriptive norm messages delivered via email on sedentary behavior and light physical activity in an office worker population. It was hypothesized that descriptive norm messages about co-workers’ lower sedentary behavior and greater light activity in the office would result in a decrease in sedentary behavior and an increase in light activity in individuals during work time. To be clear, by assessing both sedentary and light activity, we are taking the position of others that activity and sedentary behavior are not two sides of the same coin, but rather are independent behaviors that can co-exist in any given context (e.g., Biddle, 2007).
Goldstein, Cialdini, and Griskevicius (2008) argue that normative research has focused on the personal similarity of reference groups (i.e., how similar they are in values, morals, characteristics), but suggest that contextual similarity in terms of proximity and situation also might be important. They found that normative messages about hotel towel re-use were most effective when describing group behavior that occurred in the setting most closely aligned with the individuals' immediate context (e.g., “the majority of guests who also stayed in this room reused their towels”) when compared to normative messages about more distal groups (e.g., “other guests”, “fellow citizens”). Based on the theory, empirical evidence, and the suggestions of Goldstein et al. (2008), one would expect descriptive norms about physical activity to be more effective when they refer to groups high in personal (e.g., same motivations) or contextual (e.g., same location) similarity when compared to less similar groups. The secondary purpose of this study was to examine the effects of descriptive norm information about groups varying in personal and contextual similarity on sedentary behavior and light physical activity in an office worker population. Two secondary hypotheses were proposed: 1) There would be a main effect for similarity, such that descriptive norms about more contextually similar reference groups would result in lower sedentary behavior and greater light activity when compared to messages associated with less contextually similar groups. 2) There would be a main effect for personal similarity, such that descriptive norm messages about more personally similar reference groups would result in lower sedentary and higher light activity when compared to messages associated with less personally similar reference groups. Method This study employed a 2 (personal: low vs. high) 2 (context: low vs. high) pre-post-test independent groups design. Office workers were randomly assigned to one of four conditions (high personal/high contextual; high personal/low contextual; low personal/high contextual; low personal/low contextual) and reported sedentary behavior and light activity both before and after receiving descriptive norm email messages specific to their condition.
Characteristics of the norm reference group
Participants
While the existing correlational (Priebe & Spink, 2011) and experimental research (Priebe & Spink, 2012) supporting a relationship between descriptive norms and activity is promising, some important questions still remain unanswered. One relates to an examination of the characteristics of the norm reference group. Examining these characteristics might help explain results from past studies where the relationship between descriptive norms and behavior was only present when the reference group was framed as friends (Polonec, Major, & Atwood, 2006; Priebe & Spink, 2011). Consistent with the views of the social identity approach advocated
Participants (N ¼ 142) were office workers employed in the head office of one large private company. Employees had jobs that required them to spend the majority of their workday at a desk (e.g., human resources, corporate affairs, information technology). In terms of final numbers for the analyses, 38 individuals did not complete all surveys and 8 participants did not recall receiving the manipulation message, leaving 96 participants (see Fig. 1 for flow of participants through experiment). Mean age of the final sample was 40.30 years (SD ¼ 12.02), with 66% female and 34% male. Time with the company averaged 10.76 years (SD ¼ 10.68).
78
C.S. Priebe, K.S. Spink / Psychology of Sport and Exercise 19 (2015) 76e84
Fig. 1. Overview of study procedures.
Procedures Ethics approval was obtained from the University Ethics Review Board. After permission was received from the company, participants were recruited through an email sent by human resource personnel on the researchers' behalf to potential participants.
Employees could follow a link to the consent form, and complete a series of online surveys on their own time. This study involved three online surveys (spaced over 10 days), each taking 5e10 min to complete (see Fig. 1). Of note, the dependent variables were assessed in surveys one and three. The first online survey assessed demographics, self-reported
C.S. Priebe, K.S. Spink / Psychology of Sport and Exercise 19 (2015) 76e84
activity, motivation for being active, self-reported sedentary behavior, walking, and stair use within the office, and initial descriptive norm perceptions. Three to four days after completing the first survey, participants were manually randomly assigned using random number tables to one of four conditions: high personal/high contextual (n ¼ 23), high personal/low contextual (n ¼ 24), low personal/high contextual (n ¼ 22), low personal/low contextual (n ¼ 27); participant numbers shown here reflect final numbers after attrition from survey one to three and recall of message receipt. An examination of the characteristics of those who dropped out or did not recall receiving the message versus those who remained revealed no significant differences in terms of overall activity levels, age, initial sedentary behavior, and initial light activity levels, p's > .14. Following assignment, participants received an email containing norm-based information (specific to their condition) outlining how fellow workers were being active at the office. This email was designed to appear as a follow up to the results from the first survey that participants had completed approximately three days earlier. In all cases, regardless of participants' actual initial survey responses, messages highlighted that other employees (norm reference group), were standing up from their desk, using the stairs, and walking around the office more than they were in a typical day (i.e., a descriptive norm for lower sedentary and higher light physical activity). Specifically, emails indicated that based on the previous survey results, “other employees stood up from their desk at least 3 more times than you, took the stairs at least 2 more times than you, and walked around the office at least 2 times more than you in a typical day.” In terms of the manipulation for the secondary hypotheses, the contextual similarity component was varied by the location of the norm reference group (“employees at YOUR [company name] head office” versus “employees at [company name] offices in OTHER PROVINCES”). Personally similarity was manipulated as the reason for being active (i.e., the message was either about employees who “are active for the SAME MOTIVATIONAL REASON (health vs. non-health) that you identified in the survey” or “are active for A DIFFERENT MOTIVATIONAL REASON (health vs. non-health) than you identified in the survey”). The procedure for collecting the data needed for this personal manipulation is detailed below in the measures section under “Motivations for activity.” A link was placed at the end of this email message that took participants to a second online survey that was used to confirm participant's condition assignment by asking for an email. The third survey was sent three work days after the email message/second survey and assessed self-reported (a) sedentary behavior (b) light activity in the office and (c) manipulation checks. Participants completed all surveys independently using a computer at a location of their choice. After completing the final survey, participants received an online debriefing letter. Measures To measure all sedentary or physical activity dependent variables, participants were asked to think about a typical day at the office in the last week (Pre-manipulation e Time 1) or last three days (Post-manipulation e Time 3), and record the frequency or length of behavior during the morning (i.e., time between 9am e noon) as well as the afternoon (1pme5pm) of that day. The day was broken up into these two time periods to aid participant recall as well as focus attention on time in the office, and exclude the lunch break. The morning and afternoon values were combined for analyses.
79
Sedentary behavior As there is no consensus on a “gold standard” measure of sedentary behavior, one was developed specifically for this study using recommendations gleaned from previous reviews of sedentary behavior research. While there is still ambiguity regarding a definition, sedentary behavior seems to be most commonly defined as a “distinct class of behaviors that involve sitting and low levels of energy expenditure, typically less than 1.5 metabolic equivalents (METs)” (Marshall & Ramirez, 2011, p. 519). As there are very few seated activities that are above 1.5 METs, Marshall and Ramirez (2011) suggest that sedentary behavior is best operationalized as sitting, and research should focus on domain-specific sitting time (i.e., in the workplace) or the breaking up of sitting time. This resulted in the creation of two measures for this study. Longest period of sitting time Length of sitting periods has been identified as an important consideration when it comes to sedentary behavior (Marshall & Ramirez, 2011). To measure the longest period of sitting during their workday, participants recalled a typical day at the office in the last week (Time 1 survey) or last three days (Time 3 survey), and recorded the longest period of continuous sitting that they did at one time. Participants recorded these times in hours and minutes. Standing up from desk Standing is one possible non-exercise thermogenic activity (i.e., an activity that does not necessarily qualify as physical activity) that might break up long periods of sitting (i.e., sedentary) time. To measure this behavior, participants recorded the number of times that they stood from their desk for a particular day. Participants were asked to think about times they stood for a stretch break, while on the phone, etc., and to report only those times when they intentionally stood up to be active. They were instructed to NOT include the times captured by other questions (i.e., walking or using stairs). Light activity in the office As light activity that breaks up sitting (i.e., sedentary) time has been associated with health benefits (Healy et al., 2011; Levine, Eberhardt, & Jensen, 1999), this study also examined measures of light activities that might decrease time spent sitting at a desk at one's office. These included the following: Walking in the office Participants record the number of times that they walked to be active during the day. To establish context, participants were provided with examples such as walking to talk to a colleague rather than calling (or sending an email) or walking during a break. As the goal of this study was to increase intentional light activity, participants were asked to exclude times when they walked because they had no other choice (e.g., to go the washroom, pick up supplies, use the photocopier) unless they walked further than they had to with the intention of walking to be active. Stair use in the office As another measure of a light activity that could be performed in an office setting (Priebe & Spink, 2012), participants were asked to record the number of times that they used the stairs rather than the elevator to go to another floor. The building housing the participants was multi-story and included both stairs and elevators between floors.
80
C.S. Priebe, K.S. Spink / Psychology of Sport and Exercise 19 (2015) 76e84
Motivations for activity The manipulation regarding personal similarity involved tailoring descriptive norm messages to be about others with “the same” or “different” motivations for being active. While the manipulation conditions were randomly assigned and were not related to participants' reported motivations, it was still necessary to collect information about motivations in the first survey so that the subsequent descriptive norm messages could be ostensibly linked to these motivations. Participants were asked in the first survey to select the best option between two possible categories of reasons for their being active: Health reasons (e.g., cardiovascular health, reduce risk of disease, control weight) or non-health reasons (e.g., socialize, appearance, challenged, competition, enjoyment). As a point of clarification, participants' answers to this question were not considered in the conditions, as message conditions were about others with “the same” or “different” reasons for being active.
about their usual weekly leisure-time physical activity in terms of strenuous, moderate, and light activities. Reported values for these intensities were multiplied by 9, 5, and 3, respectively, and the products were summed to obtain total weekly leisure activity. In addition to the above measures, demographics (e.g., age, years with company, sex, office city) were assessed. All measures were piloted with a small sample of office workers (N ¼ 8). In addition, researchers conferred with representatives of the company where participants were recruited. Company representatives (e.g., senior researcher from department research operations and human resource management) had the opportunity to preview procedures and questionnaires in the early stage of development and offer feedback. The consensus moving forward was that the procedures and measures were reasonable and relevant to the population. Data analyses
Norm perceptions As norms are thought to work by altering normative perceptions (Rimal, 2008), assessing pre-manipulation norm perceptions captures whether normative messages have the potential to influence perceptions and subsequent behavior (Schultz, Nolan, Cialdini, Goldstein, & Griskevicius, 2007). Three items were designed for this study to assess participants' initial perceptions of others' standing, walking, and stair use behavior within the office. For example, participants were asked, “In order to fit activity into their workday, on average, how many times do you think other employees in offices like yours stand up from their desk in a typical day (do not include times to walk or use the stairs)?” For this question and the walking question, participants were asked to circle the most appropriate answer on a scale with the following options: “Less than once an hour, once an hour, twice an hour, three times an hour, four times an hour, five or more times an hour”. The responses for the use of stairs instead of the escalator or elevator ranged from 0 times a day to 7 or more times a day. Norm perceptions were assessed in the initial survey before participants received norm information. Message quality check Four message quality check items were assessed following the norm manipulation (e.g., “The information about other's activity was … believable, relevant, easy to read, persuasive”). Responses were made on a 7-point Likert scale ranging from 1 (strongly disagree) to 7 (strongly agree). Message receipt and recall manipulation checks To check recall and receipt of messages, participants were asked, “Do you recall receiving and reading an email message with information about physical activity in the office within the last few days?” Participants with a “no” response were excluded from the study. In addition, three questions assessed participants' recall of the specific details in the messages (e.g., “Was the message about people who had a different or the same reason as you for being active?”). Answers to these questions were analyzed separately for each condition as correct answers depended on message received. Physical activity The Godin Leisure-Time Exercise Questionnaire was used to assess overall physical activity participation to ensure that participants' typical activity levels did not differ across the conditions. This questionnaire has demonstrated acceptable reliability and validity (Godin & Shephard, 1985; Jacobs, Ainsworth, Hartman, & Leon, 1993). In accordance with the questionnaire instructions (Godin & Shephard, 1985), participants were asked
Data were screened for outliers, and variables were checked for normality prior to the main analyses. ANOVA was used to test differences between the four conditions on demographic variables, potential confounding variables (e.g., age, physical activity levels, years with the company), and message fidelity checks (e.g., believability). A three-factor repeated measures MANOVA was used to test the main hypothesis regarding the effects of the descriptive norms on behavior over time as well as to test the two secondary hypotheses regarding contextual and personal similarity. Effect sizes for the hp 2 were interpreted as small (.0099), medium (.0588), or large (.1379) (Cohen, 1969). The within factor (time) compared pre-to post-manipulation behavior while the two between factors compared high vs. low contextual similarity and high vs. low personal similarity. The four dependent variables were sitting, standing, walking, and stair use behavior with each of these calculated as total minutes (sitting) or times (standing, walking, stair use) for the day (i.e., combination of morning and afternoon). In the case of a significant omnibus test, univariate main effects were examined for the hypotheses regarding time effects, context similarity effects, and personal similarity effects. Results Descriptive statistics Data were screened for outliers using histograms and standardized scores. Data were found to be normally distributed with the exception of pre-message manipulation standing, which was kurtotic. As transformation did not improve the distribution, raw scores were used. Means for dependent variables are reported in Table 1. Randomization and manipulation checks Results from an ANOVA testing for differences between the four conditions on demographic variables (e.g., age, physical activity levels, years with company) revealed no differences, p's > .20. Descriptive data for message quality checks suggested that the messages were moderately easy to read (M ¼ 4.86, SD ¼ 1.76), believable (M ¼ 4.76, SD ¼ 1.61), relevant (M ¼ 4.64, SD ¼ 1.75), and persuasive (M ¼ 3.82, SD ¼ 1.72) with all message quality items measured on 7-point scale, and there were no differences between conditions, p's > .10. Confirming that the messages could potentially change participants' perceptions about co-workers’ sedentary behavior and activity, pre-message norm perceptions indicated that participants
C.S. Priebe, K.S. Spink / Psychology of Sport and Exercise 19 (2015) 76e84
81
Table 1 Dependent variables by condition and overall. Variable
High personal/High contextual condition mean (SD)
Time 1 157.82 (88.43) Sittinga 123.70 (90.62) Time 2 Sittinga 7.37 (5.00) Time 1 Standingb 6.87 (4.86) Time 2 Standingb 5.26 (4.17) Time 1 Walkingb 6.74 (5.14) Time 2 Walkingb 3.41 (2.58) Time 1 Stair Useb 3.34 (2.98) Time 2 Stair Useb a b
High personal/Low contextual condition mean (SD)
Low personal/High contextual condition mean (SD)
Low personal/Low contextual Whole sample (Used to assess condition mean (SD) time effects) mean (SD)
187.39 (117.62)
196.59 (113.99)
179.11 (95.48)
180.00 (102.78)
141.49 (88.82)
127.09 (75.88)
131.14 (89.61)
130.68 (84.65)
7.11 (7.34)
6.48 (6.24)
7.66 (6.27)
7.04 (6.20)
6.87 (7.65)
7.96 (6.15)
10.55 (8.14)
8.18 (6.88)
6.93 (7.61)
3.75 (3.80)
5.77 (5.90)
5.42 (5.58)
7.98 (7.64)
6.21 (6.37)
7.36 (6.41)
7.11 (6.35)
3.76 (3.88)
2.44 (2.17)
3.31 (3.74)
3.22 (3.17)
3.83 (4.27)
4.85 (4.67)
4.26 (4.11)
4.12 (4.03)
Sitting was reported as longest sitting period during workday (excluding lunch hour), in minutes. Standing, walking, and stair use were reported as number of times during the workday.
thought their co-workers were fairly inactive in the office. For example, most participants thought their co-workers either stood up from their desk less than once an hour (the lowest possible response; frequency 47%), walked less than once an hour (47%), and took the stairs 1e2 times per day (79%). Finally, message receipt manipulation checks revealed that there was some variability in the recall of message information. For the main purpose, 71% of participants correctly recalled message information that others were more active than them in the office. Results from the secondary purposes revealed less accurate recall. For the contextual similarity question asking where the people in the messages were from, 56% of participants answered correctly vis-a-vis their condition. For the personal similarity question, 64% of participants answered the question correctly about same or different reasons for being active. Main analyses Before running the MANOVA, assumptions specific to this technique (i.e., multivariate normality and homogeneity of covariance matrices) as well assumptions for ANOVA were checked and met (Tabachnick & Fidell, 2007). Results from the 2 (time: premanipulation vs. post-manipulation message behavior) 2 (contextual similarity: high vs. low) x 2 (personal similarity: high vs. low) repeated measures MANOVA revealed a significant main effect for time, Wilks' l ¼ .684, F(4, 89) ¼ 10.30, p < .001, hp 2 ¼ .316 supporting the first hypothesis. Given the significance of the overall test for the time effect, univariate main effects were examined for each of the four dependent variables (sitting, standing, walking, stair use). Longest period sitting in the office Results from the univariate test for longest period of sitting revealed a large effect size for time, hp 2 ¼ .264. Supporting hypothesis 1, post-message manipulation sitting (M ¼ 130.68 min, SD ¼ 84.65 min) was significantly less than pre-manipulation sitting (M ¼ 180.00 min, SD ¼ 102.78 min), F(1,92) ¼ 32.99, p < .001. Standing at the office No significant time main effect emerged for the measure assessing standing time while at the office.
Walking at the office In terms of differences in walking in the office, results revealed a medium effect size for time, hp 2 ¼ .061. In support of hypothesis 1, differences between pre-message manipulation (M ¼ 5.42 times/ day, SD ¼ 5.58 times/day) and post-message manipulation (M ¼ 7.11 times/day, SD ¼ 6.35 times/day) behavior emerged, with walking behavior in the office significantly increasing over time, F(1,92) ¼ 6.01, p ¼ .016. Stair use at the office Finally, consistent with hypothesis 1, results for stair use behavior revealed a medium effect size for time (hp 2 ¼ .074) with significant increases between pre- (M ¼ 3.22 times/day, SD ¼ 3.17 times/day) and post-message manipulation (M ¼ 4.12 times/day, SD ¼ 4.03 times/day), F(1,92) ¼ 7.33, p ¼ .008. Additional analyses based on manipulation checks Recall that 71% of participants correctly responded to the descriptive norm manipulation check question about the coworkers described in the messages having greater light activity in the office than they did. Of note, the distribution of correct responses was random across groups and there were no significant differences between groups. Interestingly, a follow-up MANOVA including only those participants who recalled this part of the message correctly (N ¼ 68) revealed a stronger main effect for time, Wilks' l ¼ .579, F(4, 61) ¼ 11.10, p < .001, hp 2 ¼ .421, when compared to the analyses including all participants (N ¼ 96). In terms of univariate main effect for time, results including only those who accurately recalled the descriptive norm component revealed stronger pre-and post-message norm message effect sizes for reducing sitting behavior, F(1,64) ¼ 37.10, p < .001, hp 2 ¼ .367; increasing walking behavior, F(1,64) ¼ 5.69, p ¼ .020, hp 2 ¼ .082; and increasing stair use, F(1,64) ¼ 11.29, p ¼ .001, hp 2 ¼ .150, when compared to the analyses that included all participants (hp 2 ¼ .264, .061, .074, respectively). Although the sample size was considerably smaller (N ¼ 28), implicating power issues, it is worth noting that there were no time effects in those who did not accurately recall the descriptive norm portion of the message manipulation for any of the dependent variables, (sitting time, p ¼ .544, hp 2 ¼ .016; walking, p ¼ .208, hp 2 ¼ .068; stair use, p ¼ .969, hp 2 ¼ .000; standing, p ¼ .961, hp 2 ¼ .000). Further, although there was less power to detect differences in those who did not recall compared to those
82
C.S. Priebe, K.S. Spink / Psychology of Sport and Exercise 19 (2015) 76e84
who did, means revealed greater differences in those who accurately recalled the descriptive norm message. Secondary hypothesis In terms of the secondary purpose, no support was found for the hypotheses about reference groups, as no significant betweengroup differences for personal (p ¼ .149, hp 2 ¼ .072) or contextual (p ¼ .500, hp 2 ¼ .037) similarity were found. Discussion The main purpose of this study was to examine the effect of descriptive norm messages on sedentary behavior and light activity in an office setting. Supporting the main hypothesis, descriptive norm messages resulted in a change in sedentary behavior and specific self-report light activity in an office setting. Specifically, differences emerged between pre-and post-manipulation for the behaviors of sitting time, walking, and stair use across all participants. The finding that light activity behaviors changed from pre-to post-descriptive norm manipulation is consistent with previous research reporting that descriptive norms influence office stair use (e.g., Priebe & Spink, 2012), as well as extends this to include a new activity - walking within the office. In addition, the finding that descriptive norms affected changes in sitting behavior extends past research to a new health-related behavior e sedentary behavior. Taken together, the findings that descriptive norms resulted in increased light activities that might break up sedentary time and decreased sitting behavior adds to the literature on sedentary behavior. Sedentary behavior has been linked to numerous negative health effects (e.g., Pate et al., 2008), and there is a specific need for research focused on decreasing this behavior in an office setting (WHO, 2008). As reported, change across time was seen in the behaviors of walking, stair-use, and sitting, but not in the measure of standing behavior. This latter finding deserves comment on two levels. First, standing was specifically targeted in the normative messages as it was viewed as a non-exercise thermogenic activity that might decrease sedentary behavior by breaking up long periods of sitting time. As standing did not change, but sedentary behavior decreased significantly across time, one is left to wonder what impacted the change in sitting time. Based on the current results, a displacement hypothesis (Ritchie, Price, & Roberts, 1987) might be a more appropriate explanation wherein the increase in stair use and walking in the office that was reported displaced sitting time. At another level, no change in the report of standing might be related to the fact that standing may be the most difficult to recall, as it might not be as easily tied to another task as the other behaviors (e.g., walking to talk to a colleague rather than emailing, using the stairs to go to a washroom on another floor, sitting for an entire meeting). Regardless of the reason, it appears that standing behavior was less prone to change in response to the descriptive norm messages. The secondary purpose of this study was to examine the effects of descriptive norm messages that varied in reference-group similarity (contextual and personal similarity). While there was a significant time effect in this study for three of four of the behaviors, an examination of reference group characteristics revealed no difference between conditions varying in high/low personal or high/ low contextual similarity, which was unexpected. Based on theory and current empirical findings in other areas, it was hypothesized that high contextual and personal similarity would produce greater behavior change than conditions reflecting lower contextual (e.g., Goldstein et al., 2008) and lower personal similarity (e.g., Rimal & Real, 2005). The failure to find differences between the similarity
conditions might reflect that these factors are not salient in the activity or office setting (i.e., descriptive norms are effective regardless of the similarity of the reference group). However, the fact that a strong relationship between descriptive norms and other types of activity behavior has only emerged in past research when norms were framed around specific reference groups (Priebe & Spink, 2011) makes this explanation less tenable. A more plausible explanation relates to the fidelity of the delivered manipulation in this field experiment. Manipulation checks revealed that participants who recalled receiving the messages varied in their recall of the reference group similarity information. Just over half of the participants were able to accurately recall the contextual (office location; 56%) or the personal similarity information (reason for being active; 64%) in their condition. Given this reported inaccuracy in recall, the present results do not provide a strong test of the proposed hypotheses regarding differences due to groups that varied in similarity. It is possible results would be different if participants processed the messages as intended. As an aside, examining differences among only those who recalled the contextual or personal manipulation questions correctly was not possible due to the small participant numbers in the sample that remained. Limitations There were limitations worth noting, and many of these relate to important future directions. First, a possible limitation regards the time effect reported in this study, and the fact that there was no control group. To maximize power through sample size, a control condition was excluded as there has already been an assessment of the effect of descriptive norms on activity compared to a nomessage control where differences were found (Priebe & Spink, 2012), and it was deemed acceptable to use participants as their own control by comparing post-message behavior to initial behavior. However, without a control group, it could be suggested that the increase in behavior from pre-to post-message manipulation was simply due to a Hawthorne effect (Roethlisberger & Dickson, 1939). Also, a control group could have been used to rule out the possibility that the differences found occurred because recall for activity and sedentary behavior at baseline were recorded for a “typical day” whereas post-message they were recorded as recall for the past three days. While these possibilities cannot be ruled out completely, there is support for the likelihood that the time effect was due to the manipulation. Specifically, follow-up analyses revealed pre-to postmessage manipulation differences in behavior among those participants who answered the descriptive norm manipulation question correctly (i.e., thought that the majority of co-workers engaged in activity in the office more than themselves, as indicated by the messages), whereas no differences emerged for those who did not (i.e., either thought the majority of co-workers were the same or less active in the office than themselves). While those who answered the descriptive norm question incorrectly are not a true control group, they do represent a group who did not report processing the message as delivered. Notwithstanding this point, a need exists to replicate the results with a control group. This is especially true of the findings regarding a relationship between descriptive norms and sedentary behavior, as this has yet to be examined using a true control. Another limitation refers to the fidelity of the manipulation delivery as recall results suggest that proposed differences between reference groups varying in contextual and personal similarity were not processed as delivered for many participants. Perhaps the contextual (i.e., different office) and personal (i.e., different reason for being active) factors targeted in the current study were not
C.S. Priebe, K.S. Spink / Psychology of Sport and Exercise 19 (2015) 76e84
strong enough to elicit substantial changes from group to group. In all conditions, the norm reference group was office workers who worked for the same company. It is possible that a stronger manipulation with greater contrast between reference groups would elicit a different behavior response. Another potential reason for the lack of differences across conditions could relate to possible contamination among conditions as all participants worked in the same office and could have discussed the email messages with one another. Further, self-report was used for the dependent variables. This study aimed to measure everyday activity behaviors that would benefit office workers and decrease the sedentary nature of their workday. Unfortunately, the nature of these behaviors (e.g., standing in the privacy of one's office, length of sitting time) meant that these measures were easiest to capture via self-report as opposed to observed or otherwise objectively recorded. Yet, there exists a possibility of inaccurate recall and social desirability influencing participants' self-reported behavior (Baranowski, 1988; Sallis & Saelens, 2000). However, it was assumed that these possible influences were randomly distributed across conditions. A final caveat, and suggestion for future research, relates to the time frame used. In this study, norms changed specific light activities and sedentary behavior over a 10-day period. The results cannot be generalized past this short time interval. To determine whether norms have any long-lasting effects, assessment of these behaviors over a longer time period is required. Strengths Despite these limitations, this study had a number of strengths. First, the norms presented to participants in the current study were relative to their past behavior (e.g., “co-workers took the stairs 2 more times than you”). As such, these messages helped to negate the potential for a boomerang effect, in which participants who might already be engaging at or above the level of the norm presented actually decrease their behavior to comply (Schultz et al., 2007). In addition, the collection of pre-manipulation information from the participants aided in making the norms more believable to participants and allowed the norms ostensibly to be about similar others (i.e., co-workers) as opposed to generic others. Though not immune to issues associated with self-report measures (Baranowski, 1988; Sallis & Saelens, 2000), the operationalization of sedentary behavior was another strength of this research. Most studies examining sedentary behavior have relied on measures of TV viewing or have measured sedentary behavior as the absence of physical activity (Marshall & Ramirez, 2011). The measure used in this study followed the recommendations of reviews of the area (e.g., Chau et al., 2010; Marshall & Ramirez, 2011), and focused on domain-specific sitting time (i.e., in the workplace) to operationalize sedentary behavior. In attempt to break up sedentary sitting time, and in line with recommendations (Healy et al., 2011), measures also targeted light activities such as standing, walking, and stair-use. Finally, participants were asked to only recall times that they engaged in standing, walking, or stair use behavior with the intent to be active. Focusing on specific behaviors performed with intention was done in attempt to minimize selfreport bias, as general measures of light activity tend to be among the least accurately recalled forms of activity (Sallis & Saelens, 2000). Measures of initial norm-perceptions and manipulation checks proved useful. For example, the poor recall of the personal and contextual similarity manipulation components of the messages revealed through the manipulation checks provided one possible explanation for why differences between high and low similarity were not found.
83
The extension of normative research to sedentary behavior also is a strength. In light of the negative health consequences of sedentary behavior and the prevalence of this behavior, especially for those in “desk jobs”, there is a need for research in this area. Despite the need to examine this behavior given the sedentary nature of many adults' occupations, most research in the area has focused on TV viewing among children and adolescents, and there has been a lack of successful experimental manipulations influencing sedentary behavior in adults (Marshall & Ramirez, 2011). As such, there has been a call for more research aimed at understanding how to reduce sedentary behavior in adults, and specifically in those with office jobs (Chau et al., 2010). Finally, this study used an experimental design including multiple time points. Researchers in the area (e.g., Rimal, 2008) have highlighted the need for more experimental research when examining the effect of norms on behavior. Conclusion The findings of the current study provide a first step in terms of understanding if normative messages might contribute to decreased sedentary behavior and increased light activity in an office setting. While the effects did not change for individuals receiving messages about reference groups that varied in similarity, descriptive norm email messages about others' behavior did result in a decrease in extended sitting behavior and an increase in specific light activities in the office setting. Acknowledgements This work was supported by a Vanier Canada Graduate Scholarship (first author) from the Social Sciences and Humanities Research Council of Canada. References Baranowski, T. (1988). Validity and reliability of self-report of physical activity: an information processing perspective. Research Quarterly, 59, 314e327. Biddle, S. J. (2007). Sedentary behavior. American Journal of Preventive Medicine, 33, 502e504. Biddle, S. J., Gorely, T., Marshall, S. J., Murdey, I., & Cameron, N. (2004). Physical activity and sedentary behaviors in youth: issues and controversies. The Journal of the Royal Society for the Promotion of Health, 124, 29e33. Chau, J. Y., Van der Ploeg, H. P., van Uffelen, J. G. Z., Wong, J., Riphagen, I., Healy, G. N., et al. (2010). Are workplace interventions to reduce sitting effective? A systematic review. Preventive Medicine, 51, 352e356. http://dx.doi.org/ 10.1016/j.ypmed.2010.08.012. Cialdini, R. B. (2003). Crafting normative messages to protect the environment. Current Directions in Psychological Science, 12, 105e109. Cialdini, R. B., Reno, R. R., & Kallgren, C. A. (1990). A focus theory of normative conduct: recycling the concept of norms to reduce littering in public places. Journal of Personality and Social Psychology, 58, 1015e1026. Cohen, J. (1969). Statistical power analysis for the behavioral sciences. New York: Academic Press. Colley, R. C., Garriguet, D., Janssen, I., Craig, C. L., Clarke, J., & Tremblay, M. S. (2011). Physical activity of Canadian adults: accelerometer results from the 2007 to 2009 Canadian Health Measures Survey. Health Reports/Statistics Canada, Canadian Centre for Health Information, 22, 7e14. Retrieved from http://www.ncbi. nlm.nih.gov/pubmed/21510585. Cragg, C., Wolfe, R., Griffiths, J. M., & Cameron, C. (2007). Physical activity among Canadian workers: Trends 2001e2006. Ottawa, ON: Canadian Fitness and Lifestyle Research Institute. Gilson, N. D., Puig-Ribera, A., McKenna, J., Brown, W. J., Burton, N. W., & Cooke, C. B. (2009). Do walking strategies to increase physical activity reduce reported sitting in workplaces: a randomized control trial. International Journal of Behavioural Nutrition and Physical Activity, 6, 43. http://dx.doi.org/10.1186/14795868-6-43. Godin, G., & Shephard, R. J. (1985). A simple method to assess exercise behavior in the community. Canadian Journal of Applied Sports Science, 10, 141e146. Goldstein, N. J., Cialdini, R. B., & Griskevicius, V. (2008). A room with a viewpoint: using social norms to motivate environment conservation in hotels. Journal of Consumer Research, 35, 472e482. http://dx.doi.org/10.1086/586910.
84
C.S. Priebe, K.S. Spink / Psychology of Sport and Exercise 19 (2015) 76e84
Hamilton, M., Hamilton, D. G., & Zderic, T. W. (2004). Exercise physiology versus inactivity physiology: an essential concept for understanding lipoprotein lipase regulation. Exercise and Sport Sciences Reviews, 32, 161e166. Healy, G. N., Matthews, C. E., Dunstan, D. W., Winkler, E. A. H., & Owen, N. (2011). Sedentary time and cardio-metabolic biomarkers in US adults: NHANES 2003e06. European Heart Journal, 32, 590e597. Hu, F. B., Li, T. Y., Colditz, G. A., Willett, W. C., & Manson, J. E. (2003). Television watching and other sedentary behaviors in relation to risk of obesity and type 2 diabetes mellitus in women. The Journal of the American Medical Association, 289, 1785e1791. Jacobs, D. R., Ainsworth, B. E., Hartman, T. J., & Leon, A. S. (1993). A simultaneous evaluation of 10 commonly used physical activity questionnaires. Medicine and Science in Sport and Exercise, 25, 81e91. Levine, J. A., Eberhardt, N. L., & Jensen, M. D. (1999). Role of non-exercise activity thermogenesis in resistance to fat gain in humans. Science, 283, 212e214. Marshall, S. J., & Ramirez, E. (2011). Reducing sedentary behavior: a new paradigm in physical activity promotion. American Journal of Lifestyle Medicine, 5, 516e530. Owen, N., Healy, G. N., Matthews, C. E., & Dunstan, D. W. (2010). Too much sitting: the population-health science of sedentary behavior. Exercise and Sport Sciences Reviews, 38, 105e113. Pate, R., O'Neill, J. R., & Lobelo, F. (2008). The evolving definition of “sedentary.” Exercise and Sport Sciences Reviews, 36, 171e172. Polonec, L. D., Major, A. M., & Atwood, L. E. (2006). Evaluating the believability and effectiveness of the social norms message “most students drink 0 to 4 drinks when they party”. Health Communications, 20, 23e34. Priebe, C. S., & Spink, K. S. (2011). When in Rome: descriptive norms and physical activity. Psychology of Sport & Exercise, 12, 93e98. Priebe, C. S., & Spink, K. S. (2012). Using messages promoting descriptive norms to increase physical activity. Health Communication, 27, 284e291.
Rimal, R. N. (2008). Modeling the relationship between descriptive norms and behaviors: a test and extension of the theory of normative social behavior (TNSB). Health Communications, 23, 103e116. Rimal, R. N., Lapinski, M. K., Cook, R. J., & Real, K. (2005). Moving towards a theory of normative influences: how perceived benefits and similarity moderate the impact of descriptive norms on behaviors. Journal of Health Communication: International Perspectives, 10, 433e450. Rimal, R. N., & Real, K. (2005). How behaviors are influenced by perceived norms: a test of the theory of normative social behavior. Communications Research, 32, 389e414. Ritchie, D., Price, V., & Roberts, D. F. (1987). Reading and television: a longitudinal investigation of the displacement hypothesis. Communication Research, 14, 292e315. Roethlisberger, F. J., & Dickson, W. J. (1939). Management and the worker. Cambridge, MA: Harvard University Press. Sallis, J. F., & Saelens, B. E. (2000). Assessment of physical activity by self-report: status, limitations, and future directions. Research Quarterly for Exercise and Sport, 71, 1e14. Schultz, P. W., Nolan, J. M., Cialdini, R. B., Goldstein, N. J., & Griskevicius, V. (2007). The constructive, destructive, and reconstructive power of social norms. Psychological Science, 18, 429e434. Tabachnick, B. G., & Fidell, L. S. (2007). In Using multivariate statistics (5th ed.). Boston, MA: Allyn & Bacon. Terry, D. J., & Hogg, M. A. (1996). Group norms and the attitude-behavior relationship: a role for group identification. Personality and Social Psychology Bulletin, 22, 776e793. WHO/WEF. (2008). Preventing non-communicable diseases in the workplace through diet and physical activity: WHO/World economic forum report of a joint event. Geneva: World Health Organization/World Economic Forum.
Contents lists available at ScienceDirect
Psychology of Sport and Exercise journal homepage: www.elsevier.com/locate/psychsport
Less sitting and more moving in the office: Using descriptive norm messages to decrease sedentary behavior and increase light physical activity at work Carly S. Priebe*, Kevin S. Spink University of Saskatchewan, Canada
a r t i c l e i n f o
a b s t r a c t
Article history: Received 27 November 2014 Received in revised form 26 February 2015 Accepted 27 February 2015 Available online 18 March 2015
Objectives: The sedentary nature of office workplaces is an increasing concern (Marshall & Ramirez, 2011). One avenue through which sedentary behavior may be decreased and beneficial activity increased is through social norms. Previous research (e.g., Priebe & Spink, 2012) has demonstrated that perceptions about others' behavior (i.e., descriptive norms; Cialdini, Reno, & Kallgren, 1990) can influence physical activity. However, it is unclear if descriptive norms can be used to impact sedentary time. This study examined whether descriptive norm messages would impact sedentary behavior and light activity in an office setting. Given the possible importance of personal and contextual characteristics of the norm reference group, a secondary purpose was to examine messages that varied in reference group characteristics. Design: This study utilized a pre-post experimental design. Method: Office workers were randomly assigned to receive one of four email messages containing descriptive norms about co-workers’ behavior. Sedentary behavior and activity were measured before and after message receipt. Results: A repeated measures MANOVA revealed a main effect for time, p < .001, hp 2 ¼ .316. Those who received descriptive norm messages about co-workers' lower sedentary behavior and greater stair use and walking decreased their own sitting time while increasing stair use and walking at the office, respectively, p's < .05. No differences emerged between participants receiving information about groups that varied in reference group characteristics, p's > .10. Conclusion: Results provide experimental evidence that descriptive norm messages may serve to decrease sedentary and increase light activity in an office setting. © 2015 Elsevier Ltd. All rights reserved.
Keywords: Descriptive norms Office workers Light activity Sedentary behavior Email
While it is generally accepted that most North Americans could improve their health by increasing physical activity levels (Cragg, Wolfe, Griffiths, & Cameron, 2007), less attention has been paid to the observation that individuals could benefit from decreasing their sedentary time (Marshall & Ramirez, 2011). This is surprising as studies have shown that sedentary behaviors, commonly defined as “behaviors that involve sitting and low levels of energy expenditure” (Marshall & Ramirez, 2011, p. 519), carry health risks independent of insufficient physical activity (Biddle, Gorely, Marshall, Murdey, & Cameron, 2004; Pate, O'Neill, & Lobelo, 2008).
* Corresponding author. College of Kinesiology, University of Saskatchewan, 87 Campus Drive, Saskatoon, Saskatchewan, S7N 5B2, Canada. Tel.: þ1 306 227 2406; fax: þ1 306 966 6464. E-mail address: [email protected] (C.S. Priebe). http://dx.doi.org/10.1016/j.psychsport.2015.02.008 1469-0292/© 2015 Elsevier Ltd. All rights reserved.
Despite these health risks, the typical adult in Canada is estimated to spend about 9.5 h of waking time being sedentary, with most of it sitting (Colley et al., 2011). Further, adults in office jobs are likely sedentary for more time when compared to those in occupations such as the trades (Hu, Li, Colditz, Willett, & Manson, 2003). As such, both the World Health Organization (2008) and researchers (e.g., Chau et al., 2010; Marshall & Ramirez, 2011) acknowledge the need for more research examining ways to decrease sedentary behavior (sitting) in office settings. It has been suggested that one practical way to reduce sitting time would be to break it up with light intensity activities (Chau et al., 2010; Marshall & Ramirez, 2011). Breaking up total sitting time appears to relate to health benefits as researchers have identified important health-related outcomes (e.g., improved metabolic profile) when individuals move from sitting to standing
C.S. Priebe, K.S. Spink / Psychology of Sport and Exercise 19 (2015) 76e84
(Hamilton, Hamilton, & Zderic, 2004). Further, in a large population-representative sample, Healy, Matthews, Dunstan, Winkler, and Owen (2011) found breaks in sedentary behavior (independent of sedentary behavior) were beneficially associated with a known marker of health, waist circumference. In addition, these associations between breaks in sedentary time and health have been found independent of overall physical activity (Gilson et al., 2009), emphasizing the need to consider sedentary behavior as its own entity (Owen, Healy, Matthews, & Dunstan, 2010).
77
by Terry and Hogg (1996), it is possible that descriptive norms are only effective when they refer to groups where the individual feels some type of similarity (e.g., more similar friends compared to public service announcements conveying that others are active, where similarity to others is less clear). In addition to having a conceptual underpinning, empirical support exists for a stronger relationship between descriptive norms and physical activity in those who highly identify with their reference group (e.g., Rimal, Lapinski, Cook, & Real, 2005). Personal and contextual similarity
Descriptive norms and activity Given the important links between physical activity, sedentary behavior, and health, there is a need for research understanding how we might change these behaviors, especially in an office setting. While there are a number of possible ways to target change in activity and sedentary behavior, one possibility is through social influence. One form that might hold particular potential in this area is descriptive norms. In focus theory of normative conduct, Cialdini, Reno, and Kallgren (1990) define descriptive norms as perceptions about the prevalent behavior of others. These norms have been examined with a number of behaviors (e.g., environmental conservation, Cialdini, 2003; alcohol consumption, Rimal, 2008; Rimal & Real, 2005), and recent research has emerged in the physical activity setting (e.g., Priebe & Spink, 2011, 2012). Descriptive norms and sedentary behavior As descriptive norm messages have shown some promise in the area of increasing light physical activity (stair use) in office workers (e.g., Priebe & Spink, 2012), it also might be important to extend this to examining norms for other types of light activity that might break up sedentary time (e.g., walking within the office and standing up during the work day). Also, as descriptive norms often are examined in relation to decreasing the prevalence of other negative behaviors (e.g., less alcohol consumption, Rimal & Real, 2005; less littering, Cialdini et al., 1990), using normative information to decrease sedentary behavior appears plausible (i.e., less sitting time). The main purpose of this experimental field study was to examine the effect of descriptive norm messages delivered via email on sedentary behavior and light physical activity in an office worker population. It was hypothesized that descriptive norm messages about co-workers’ lower sedentary behavior and greater light activity in the office would result in a decrease in sedentary behavior and an increase in light activity in individuals during work time. To be clear, by assessing both sedentary and light activity, we are taking the position of others that activity and sedentary behavior are not two sides of the same coin, but rather are independent behaviors that can co-exist in any given context (e.g., Biddle, 2007).
Goldstein, Cialdini, and Griskevicius (2008) argue that normative research has focused on the personal similarity of reference groups (i.e., how similar they are in values, morals, characteristics), but suggest that contextual similarity in terms of proximity and situation also might be important. They found that normative messages about hotel towel re-use were most effective when describing group behavior that occurred in the setting most closely aligned with the individuals' immediate context (e.g., “the majority of guests who also stayed in this room reused their towels”) when compared to normative messages about more distal groups (e.g., “other guests”, “fellow citizens”). Based on the theory, empirical evidence, and the suggestions of Goldstein et al. (2008), one would expect descriptive norms about physical activity to be more effective when they refer to groups high in personal (e.g., same motivations) or contextual (e.g., same location) similarity when compared to less similar groups. The secondary purpose of this study was to examine the effects of descriptive norm information about groups varying in personal and contextual similarity on sedentary behavior and light physical activity in an office worker population. Two secondary hypotheses were proposed: 1) There would be a main effect for similarity, such that descriptive norms about more contextually similar reference groups would result in lower sedentary behavior and greater light activity when compared to messages associated with less contextually similar groups. 2) There would be a main effect for personal similarity, such that descriptive norm messages about more personally similar reference groups would result in lower sedentary and higher light activity when compared to messages associated with less personally similar reference groups. Method This study employed a 2 (personal: low vs. high) 2 (context: low vs. high) pre-post-test independent groups design. Office workers were randomly assigned to one of four conditions (high personal/high contextual; high personal/low contextual; low personal/high contextual; low personal/low contextual) and reported sedentary behavior and light activity both before and after receiving descriptive norm email messages specific to their condition.
Characteristics of the norm reference group
Participants
While the existing correlational (Priebe & Spink, 2011) and experimental research (Priebe & Spink, 2012) supporting a relationship between descriptive norms and activity is promising, some important questions still remain unanswered. One relates to an examination of the characteristics of the norm reference group. Examining these characteristics might help explain results from past studies where the relationship between descriptive norms and behavior was only present when the reference group was framed as friends (Polonec, Major, & Atwood, 2006; Priebe & Spink, 2011). Consistent with the views of the social identity approach advocated
Participants (N ¼ 142) were office workers employed in the head office of one large private company. Employees had jobs that required them to spend the majority of their workday at a desk (e.g., human resources, corporate affairs, information technology). In terms of final numbers for the analyses, 38 individuals did not complete all surveys and 8 participants did not recall receiving the manipulation message, leaving 96 participants (see Fig. 1 for flow of participants through experiment). Mean age of the final sample was 40.30 years (SD ¼ 12.02), with 66% female and 34% male. Time with the company averaged 10.76 years (SD ¼ 10.68).
78
C.S. Priebe, K.S. Spink / Psychology of Sport and Exercise 19 (2015) 76e84
Fig. 1. Overview of study procedures.
Procedures Ethics approval was obtained from the University Ethics Review Board. After permission was received from the company, participants were recruited through an email sent by human resource personnel on the researchers' behalf to potential participants.
Employees could follow a link to the consent form, and complete a series of online surveys on their own time. This study involved three online surveys (spaced over 10 days), each taking 5e10 min to complete (see Fig. 1). Of note, the dependent variables were assessed in surveys one and three. The first online survey assessed demographics, self-reported
C.S. Priebe, K.S. Spink / Psychology of Sport and Exercise 19 (2015) 76e84
activity, motivation for being active, self-reported sedentary behavior, walking, and stair use within the office, and initial descriptive norm perceptions. Three to four days after completing the first survey, participants were manually randomly assigned using random number tables to one of four conditions: high personal/high contextual (n ¼ 23), high personal/low contextual (n ¼ 24), low personal/high contextual (n ¼ 22), low personal/low contextual (n ¼ 27); participant numbers shown here reflect final numbers after attrition from survey one to three and recall of message receipt. An examination of the characteristics of those who dropped out or did not recall receiving the message versus those who remained revealed no significant differences in terms of overall activity levels, age, initial sedentary behavior, and initial light activity levels, p's > .14. Following assignment, participants received an email containing norm-based information (specific to their condition) outlining how fellow workers were being active at the office. This email was designed to appear as a follow up to the results from the first survey that participants had completed approximately three days earlier. In all cases, regardless of participants' actual initial survey responses, messages highlighted that other employees (norm reference group), were standing up from their desk, using the stairs, and walking around the office more than they were in a typical day (i.e., a descriptive norm for lower sedentary and higher light physical activity). Specifically, emails indicated that based on the previous survey results, “other employees stood up from their desk at least 3 more times than you, took the stairs at least 2 more times than you, and walked around the office at least 2 times more than you in a typical day.” In terms of the manipulation for the secondary hypotheses, the contextual similarity component was varied by the location of the norm reference group (“employees at YOUR [company name] head office” versus “employees at [company name] offices in OTHER PROVINCES”). Personally similarity was manipulated as the reason for being active (i.e., the message was either about employees who “are active for the SAME MOTIVATIONAL REASON (health vs. non-health) that you identified in the survey” or “are active for A DIFFERENT MOTIVATIONAL REASON (health vs. non-health) than you identified in the survey”). The procedure for collecting the data needed for this personal manipulation is detailed below in the measures section under “Motivations for activity.” A link was placed at the end of this email message that took participants to a second online survey that was used to confirm participant's condition assignment by asking for an email. The third survey was sent three work days after the email message/second survey and assessed self-reported (a) sedentary behavior (b) light activity in the office and (c) manipulation checks. Participants completed all surveys independently using a computer at a location of their choice. After completing the final survey, participants received an online debriefing letter. Measures To measure all sedentary or physical activity dependent variables, participants were asked to think about a typical day at the office in the last week (Pre-manipulation e Time 1) or last three days (Post-manipulation e Time 3), and record the frequency or length of behavior during the morning (i.e., time between 9am e noon) as well as the afternoon (1pme5pm) of that day. The day was broken up into these two time periods to aid participant recall as well as focus attention on time in the office, and exclude the lunch break. The morning and afternoon values were combined for analyses.
79
Sedentary behavior As there is no consensus on a “gold standard” measure of sedentary behavior, one was developed specifically for this study using recommendations gleaned from previous reviews of sedentary behavior research. While there is still ambiguity regarding a definition, sedentary behavior seems to be most commonly defined as a “distinct class of behaviors that involve sitting and low levels of energy expenditure, typically less than 1.5 metabolic equivalents (METs)” (Marshall & Ramirez, 2011, p. 519). As there are very few seated activities that are above 1.5 METs, Marshall and Ramirez (2011) suggest that sedentary behavior is best operationalized as sitting, and research should focus on domain-specific sitting time (i.e., in the workplace) or the breaking up of sitting time. This resulted in the creation of two measures for this study. Longest period of sitting time Length of sitting periods has been identified as an important consideration when it comes to sedentary behavior (Marshall & Ramirez, 2011). To measure the longest period of sitting during their workday, participants recalled a typical day at the office in the last week (Time 1 survey) or last three days (Time 3 survey), and recorded the longest period of continuous sitting that they did at one time. Participants recorded these times in hours and minutes. Standing up from desk Standing is one possible non-exercise thermogenic activity (i.e., an activity that does not necessarily qualify as physical activity) that might break up long periods of sitting (i.e., sedentary) time. To measure this behavior, participants recorded the number of times that they stood from their desk for a particular day. Participants were asked to think about times they stood for a stretch break, while on the phone, etc., and to report only those times when they intentionally stood up to be active. They were instructed to NOT include the times captured by other questions (i.e., walking or using stairs). Light activity in the office As light activity that breaks up sitting (i.e., sedentary) time has been associated with health benefits (Healy et al., 2011; Levine, Eberhardt, & Jensen, 1999), this study also examined measures of light activities that might decrease time spent sitting at a desk at one's office. These included the following: Walking in the office Participants record the number of times that they walked to be active during the day. To establish context, participants were provided with examples such as walking to talk to a colleague rather than calling (or sending an email) or walking during a break. As the goal of this study was to increase intentional light activity, participants were asked to exclude times when they walked because they had no other choice (e.g., to go the washroom, pick up supplies, use the photocopier) unless they walked further than they had to with the intention of walking to be active. Stair use in the office As another measure of a light activity that could be performed in an office setting (Priebe & Spink, 2012), participants were asked to record the number of times that they used the stairs rather than the elevator to go to another floor. The building housing the participants was multi-story and included both stairs and elevators between floors.
80
C.S. Priebe, K.S. Spink / Psychology of Sport and Exercise 19 (2015) 76e84
Motivations for activity The manipulation regarding personal similarity involved tailoring descriptive norm messages to be about others with “the same” or “different” motivations for being active. While the manipulation conditions were randomly assigned and were not related to participants' reported motivations, it was still necessary to collect information about motivations in the first survey so that the subsequent descriptive norm messages could be ostensibly linked to these motivations. Participants were asked in the first survey to select the best option between two possible categories of reasons for their being active: Health reasons (e.g., cardiovascular health, reduce risk of disease, control weight) or non-health reasons (e.g., socialize, appearance, challenged, competition, enjoyment). As a point of clarification, participants' answers to this question were not considered in the conditions, as message conditions were about others with “the same” or “different” reasons for being active.
about their usual weekly leisure-time physical activity in terms of strenuous, moderate, and light activities. Reported values for these intensities were multiplied by 9, 5, and 3, respectively, and the products were summed to obtain total weekly leisure activity. In addition to the above measures, demographics (e.g., age, years with company, sex, office city) were assessed. All measures were piloted with a small sample of office workers (N ¼ 8). In addition, researchers conferred with representatives of the company where participants were recruited. Company representatives (e.g., senior researcher from department research operations and human resource management) had the opportunity to preview procedures and questionnaires in the early stage of development and offer feedback. The consensus moving forward was that the procedures and measures were reasonable and relevant to the population. Data analyses
Norm perceptions As norms are thought to work by altering normative perceptions (Rimal, 2008), assessing pre-manipulation norm perceptions captures whether normative messages have the potential to influence perceptions and subsequent behavior (Schultz, Nolan, Cialdini, Goldstein, & Griskevicius, 2007). Three items were designed for this study to assess participants' initial perceptions of others' standing, walking, and stair use behavior within the office. For example, participants were asked, “In order to fit activity into their workday, on average, how many times do you think other employees in offices like yours stand up from their desk in a typical day (do not include times to walk or use the stairs)?” For this question and the walking question, participants were asked to circle the most appropriate answer on a scale with the following options: “Less than once an hour, once an hour, twice an hour, three times an hour, four times an hour, five or more times an hour”. The responses for the use of stairs instead of the escalator or elevator ranged from 0 times a day to 7 or more times a day. Norm perceptions were assessed in the initial survey before participants received norm information. Message quality check Four message quality check items were assessed following the norm manipulation (e.g., “The information about other's activity was … believable, relevant, easy to read, persuasive”). Responses were made on a 7-point Likert scale ranging from 1 (strongly disagree) to 7 (strongly agree). Message receipt and recall manipulation checks To check recall and receipt of messages, participants were asked, “Do you recall receiving and reading an email message with information about physical activity in the office within the last few days?” Participants with a “no” response were excluded from the study. In addition, three questions assessed participants' recall of the specific details in the messages (e.g., “Was the message about people who had a different or the same reason as you for being active?”). Answers to these questions were analyzed separately for each condition as correct answers depended on message received. Physical activity The Godin Leisure-Time Exercise Questionnaire was used to assess overall physical activity participation to ensure that participants' typical activity levels did not differ across the conditions. This questionnaire has demonstrated acceptable reliability and validity (Godin & Shephard, 1985; Jacobs, Ainsworth, Hartman, & Leon, 1993). In accordance with the questionnaire instructions (Godin & Shephard, 1985), participants were asked
Data were screened for outliers, and variables were checked for normality prior to the main analyses. ANOVA was used to test differences between the four conditions on demographic variables, potential confounding variables (e.g., age, physical activity levels, years with the company), and message fidelity checks (e.g., believability). A three-factor repeated measures MANOVA was used to test the main hypothesis regarding the effects of the descriptive norms on behavior over time as well as to test the two secondary hypotheses regarding contextual and personal similarity. Effect sizes for the hp 2 were interpreted as small (.0099), medium (.0588), or large (.1379) (Cohen, 1969). The within factor (time) compared pre-to post-manipulation behavior while the two between factors compared high vs. low contextual similarity and high vs. low personal similarity. The four dependent variables were sitting, standing, walking, and stair use behavior with each of these calculated as total minutes (sitting) or times (standing, walking, stair use) for the day (i.e., combination of morning and afternoon). In the case of a significant omnibus test, univariate main effects were examined for the hypotheses regarding time effects, context similarity effects, and personal similarity effects. Results Descriptive statistics Data were screened for outliers using histograms and standardized scores. Data were found to be normally distributed with the exception of pre-message manipulation standing, which was kurtotic. As transformation did not improve the distribution, raw scores were used. Means for dependent variables are reported in Table 1. Randomization and manipulation checks Results from an ANOVA testing for differences between the four conditions on demographic variables (e.g., age, physical activity levels, years with company) revealed no differences, p's > .20. Descriptive data for message quality checks suggested that the messages were moderately easy to read (M ¼ 4.86, SD ¼ 1.76), believable (M ¼ 4.76, SD ¼ 1.61), relevant (M ¼ 4.64, SD ¼ 1.75), and persuasive (M ¼ 3.82, SD ¼ 1.72) with all message quality items measured on 7-point scale, and there were no differences between conditions, p's > .10. Confirming that the messages could potentially change participants' perceptions about co-workers’ sedentary behavior and activity, pre-message norm perceptions indicated that participants
C.S. Priebe, K.S. Spink / Psychology of Sport and Exercise 19 (2015) 76e84
81
Table 1 Dependent variables by condition and overall. Variable
High personal/High contextual condition mean (SD)
Time 1 157.82 (88.43) Sittinga 123.70 (90.62) Time 2 Sittinga 7.37 (5.00) Time 1 Standingb 6.87 (4.86) Time 2 Standingb 5.26 (4.17) Time 1 Walkingb 6.74 (5.14) Time 2 Walkingb 3.41 (2.58) Time 1 Stair Useb 3.34 (2.98) Time 2 Stair Useb a b
High personal/Low contextual condition mean (SD)
Low personal/High contextual condition mean (SD)
Low personal/Low contextual Whole sample (Used to assess condition mean (SD) time effects) mean (SD)
187.39 (117.62)
196.59 (113.99)
179.11 (95.48)
180.00 (102.78)
141.49 (88.82)
127.09 (75.88)
131.14 (89.61)
130.68 (84.65)
7.11 (7.34)
6.48 (6.24)
7.66 (6.27)
7.04 (6.20)
6.87 (7.65)
7.96 (6.15)
10.55 (8.14)
8.18 (6.88)
6.93 (7.61)
3.75 (3.80)
5.77 (5.90)
5.42 (5.58)
7.98 (7.64)
6.21 (6.37)
7.36 (6.41)
7.11 (6.35)
3.76 (3.88)
2.44 (2.17)
3.31 (3.74)
3.22 (3.17)
3.83 (4.27)
4.85 (4.67)
4.26 (4.11)
4.12 (4.03)
Sitting was reported as longest sitting period during workday (excluding lunch hour), in minutes. Standing, walking, and stair use were reported as number of times during the workday.
thought their co-workers were fairly inactive in the office. For example, most participants thought their co-workers either stood up from their desk less than once an hour (the lowest possible response; frequency 47%), walked less than once an hour (47%), and took the stairs 1e2 times per day (79%). Finally, message receipt manipulation checks revealed that there was some variability in the recall of message information. For the main purpose, 71% of participants correctly recalled message information that others were more active than them in the office. Results from the secondary purposes revealed less accurate recall. For the contextual similarity question asking where the people in the messages were from, 56% of participants answered correctly vis-a-vis their condition. For the personal similarity question, 64% of participants answered the question correctly about same or different reasons for being active. Main analyses Before running the MANOVA, assumptions specific to this technique (i.e., multivariate normality and homogeneity of covariance matrices) as well assumptions for ANOVA were checked and met (Tabachnick & Fidell, 2007). Results from the 2 (time: premanipulation vs. post-manipulation message behavior) 2 (contextual similarity: high vs. low) x 2 (personal similarity: high vs. low) repeated measures MANOVA revealed a significant main effect for time, Wilks' l ¼ .684, F(4, 89) ¼ 10.30, p < .001, hp 2 ¼ .316 supporting the first hypothesis. Given the significance of the overall test for the time effect, univariate main effects were examined for each of the four dependent variables (sitting, standing, walking, stair use). Longest period sitting in the office Results from the univariate test for longest period of sitting revealed a large effect size for time, hp 2 ¼ .264. Supporting hypothesis 1, post-message manipulation sitting (M ¼ 130.68 min, SD ¼ 84.65 min) was significantly less than pre-manipulation sitting (M ¼ 180.00 min, SD ¼ 102.78 min), F(1,92) ¼ 32.99, p < .001. Standing at the office No significant time main effect emerged for the measure assessing standing time while at the office.
Walking at the office In terms of differences in walking in the office, results revealed a medium effect size for time, hp 2 ¼ .061. In support of hypothesis 1, differences between pre-message manipulation (M ¼ 5.42 times/ day, SD ¼ 5.58 times/day) and post-message manipulation (M ¼ 7.11 times/day, SD ¼ 6.35 times/day) behavior emerged, with walking behavior in the office significantly increasing over time, F(1,92) ¼ 6.01, p ¼ .016. Stair use at the office Finally, consistent with hypothesis 1, results for stair use behavior revealed a medium effect size for time (hp 2 ¼ .074) with significant increases between pre- (M ¼ 3.22 times/day, SD ¼ 3.17 times/day) and post-message manipulation (M ¼ 4.12 times/day, SD ¼ 4.03 times/day), F(1,92) ¼ 7.33, p ¼ .008. Additional analyses based on manipulation checks Recall that 71% of participants correctly responded to the descriptive norm manipulation check question about the coworkers described in the messages having greater light activity in the office than they did. Of note, the distribution of correct responses was random across groups and there were no significant differences between groups. Interestingly, a follow-up MANOVA including only those participants who recalled this part of the message correctly (N ¼ 68) revealed a stronger main effect for time, Wilks' l ¼ .579, F(4, 61) ¼ 11.10, p < .001, hp 2 ¼ .421, when compared to the analyses including all participants (N ¼ 96). In terms of univariate main effect for time, results including only those who accurately recalled the descriptive norm component revealed stronger pre-and post-message norm message effect sizes for reducing sitting behavior, F(1,64) ¼ 37.10, p < .001, hp 2 ¼ .367; increasing walking behavior, F(1,64) ¼ 5.69, p ¼ .020, hp 2 ¼ .082; and increasing stair use, F(1,64) ¼ 11.29, p ¼ .001, hp 2 ¼ .150, when compared to the analyses that included all participants (hp 2 ¼ .264, .061, .074, respectively). Although the sample size was considerably smaller (N ¼ 28), implicating power issues, it is worth noting that there were no time effects in those who did not accurately recall the descriptive norm portion of the message manipulation for any of the dependent variables, (sitting time, p ¼ .544, hp 2 ¼ .016; walking, p ¼ .208, hp 2 ¼ .068; stair use, p ¼ .969, hp 2 ¼ .000; standing, p ¼ .961, hp 2 ¼ .000). Further, although there was less power to detect differences in those who did not recall compared to those
82
C.S. Priebe, K.S. Spink / Psychology of Sport and Exercise 19 (2015) 76e84
who did, means revealed greater differences in those who accurately recalled the descriptive norm message. Secondary hypothesis In terms of the secondary purpose, no support was found for the hypotheses about reference groups, as no significant betweengroup differences for personal (p ¼ .149, hp 2 ¼ .072) or contextual (p ¼ .500, hp 2 ¼ .037) similarity were found. Discussion The main purpose of this study was to examine the effect of descriptive norm messages on sedentary behavior and light activity in an office setting. Supporting the main hypothesis, descriptive norm messages resulted in a change in sedentary behavior and specific self-report light activity in an office setting. Specifically, differences emerged between pre-and post-manipulation for the behaviors of sitting time, walking, and stair use across all participants. The finding that light activity behaviors changed from pre-to post-descriptive norm manipulation is consistent with previous research reporting that descriptive norms influence office stair use (e.g., Priebe & Spink, 2012), as well as extends this to include a new activity - walking within the office. In addition, the finding that descriptive norms affected changes in sitting behavior extends past research to a new health-related behavior e sedentary behavior. Taken together, the findings that descriptive norms resulted in increased light activities that might break up sedentary time and decreased sitting behavior adds to the literature on sedentary behavior. Sedentary behavior has been linked to numerous negative health effects (e.g., Pate et al., 2008), and there is a specific need for research focused on decreasing this behavior in an office setting (WHO, 2008). As reported, change across time was seen in the behaviors of walking, stair-use, and sitting, but not in the measure of standing behavior. This latter finding deserves comment on two levels. First, standing was specifically targeted in the normative messages as it was viewed as a non-exercise thermogenic activity that might decrease sedentary behavior by breaking up long periods of sitting time. As standing did not change, but sedentary behavior decreased significantly across time, one is left to wonder what impacted the change in sitting time. Based on the current results, a displacement hypothesis (Ritchie, Price, & Roberts, 1987) might be a more appropriate explanation wherein the increase in stair use and walking in the office that was reported displaced sitting time. At another level, no change in the report of standing might be related to the fact that standing may be the most difficult to recall, as it might not be as easily tied to another task as the other behaviors (e.g., walking to talk to a colleague rather than emailing, using the stairs to go to a washroom on another floor, sitting for an entire meeting). Regardless of the reason, it appears that standing behavior was less prone to change in response to the descriptive norm messages. The secondary purpose of this study was to examine the effects of descriptive norm messages that varied in reference-group similarity (contextual and personal similarity). While there was a significant time effect in this study for three of four of the behaviors, an examination of reference group characteristics revealed no difference between conditions varying in high/low personal or high/ low contextual similarity, which was unexpected. Based on theory and current empirical findings in other areas, it was hypothesized that high contextual and personal similarity would produce greater behavior change than conditions reflecting lower contextual (e.g., Goldstein et al., 2008) and lower personal similarity (e.g., Rimal & Real, 2005). The failure to find differences between the similarity
conditions might reflect that these factors are not salient in the activity or office setting (i.e., descriptive norms are effective regardless of the similarity of the reference group). However, the fact that a strong relationship between descriptive norms and other types of activity behavior has only emerged in past research when norms were framed around specific reference groups (Priebe & Spink, 2011) makes this explanation less tenable. A more plausible explanation relates to the fidelity of the delivered manipulation in this field experiment. Manipulation checks revealed that participants who recalled receiving the messages varied in their recall of the reference group similarity information. Just over half of the participants were able to accurately recall the contextual (office location; 56%) or the personal similarity information (reason for being active; 64%) in their condition. Given this reported inaccuracy in recall, the present results do not provide a strong test of the proposed hypotheses regarding differences due to groups that varied in similarity. It is possible results would be different if participants processed the messages as intended. As an aside, examining differences among only those who recalled the contextual or personal manipulation questions correctly was not possible due to the small participant numbers in the sample that remained. Limitations There were limitations worth noting, and many of these relate to important future directions. First, a possible limitation regards the time effect reported in this study, and the fact that there was no control group. To maximize power through sample size, a control condition was excluded as there has already been an assessment of the effect of descriptive norms on activity compared to a nomessage control where differences were found (Priebe & Spink, 2012), and it was deemed acceptable to use participants as their own control by comparing post-message behavior to initial behavior. However, without a control group, it could be suggested that the increase in behavior from pre-to post-message manipulation was simply due to a Hawthorne effect (Roethlisberger & Dickson, 1939). Also, a control group could have been used to rule out the possibility that the differences found occurred because recall for activity and sedentary behavior at baseline were recorded for a “typical day” whereas post-message they were recorded as recall for the past three days. While these possibilities cannot be ruled out completely, there is support for the likelihood that the time effect was due to the manipulation. Specifically, follow-up analyses revealed pre-to postmessage manipulation differences in behavior among those participants who answered the descriptive norm manipulation question correctly (i.e., thought that the majority of co-workers engaged in activity in the office more than themselves, as indicated by the messages), whereas no differences emerged for those who did not (i.e., either thought the majority of co-workers were the same or less active in the office than themselves). While those who answered the descriptive norm question incorrectly are not a true control group, they do represent a group who did not report processing the message as delivered. Notwithstanding this point, a need exists to replicate the results with a control group. This is especially true of the findings regarding a relationship between descriptive norms and sedentary behavior, as this has yet to be examined using a true control. Another limitation refers to the fidelity of the manipulation delivery as recall results suggest that proposed differences between reference groups varying in contextual and personal similarity were not processed as delivered for many participants. Perhaps the contextual (i.e., different office) and personal (i.e., different reason for being active) factors targeted in the current study were not
C.S. Priebe, K.S. Spink / Psychology of Sport and Exercise 19 (2015) 76e84
strong enough to elicit substantial changes from group to group. In all conditions, the norm reference group was office workers who worked for the same company. It is possible that a stronger manipulation with greater contrast between reference groups would elicit a different behavior response. Another potential reason for the lack of differences across conditions could relate to possible contamination among conditions as all participants worked in the same office and could have discussed the email messages with one another. Further, self-report was used for the dependent variables. This study aimed to measure everyday activity behaviors that would benefit office workers and decrease the sedentary nature of their workday. Unfortunately, the nature of these behaviors (e.g., standing in the privacy of one's office, length of sitting time) meant that these measures were easiest to capture via self-report as opposed to observed or otherwise objectively recorded. Yet, there exists a possibility of inaccurate recall and social desirability influencing participants' self-reported behavior (Baranowski, 1988; Sallis & Saelens, 2000). However, it was assumed that these possible influences were randomly distributed across conditions. A final caveat, and suggestion for future research, relates to the time frame used. In this study, norms changed specific light activities and sedentary behavior over a 10-day period. The results cannot be generalized past this short time interval. To determine whether norms have any long-lasting effects, assessment of these behaviors over a longer time period is required. Strengths Despite these limitations, this study had a number of strengths. First, the norms presented to participants in the current study were relative to their past behavior (e.g., “co-workers took the stairs 2 more times than you”). As such, these messages helped to negate the potential for a boomerang effect, in which participants who might already be engaging at or above the level of the norm presented actually decrease their behavior to comply (Schultz et al., 2007). In addition, the collection of pre-manipulation information from the participants aided in making the norms more believable to participants and allowed the norms ostensibly to be about similar others (i.e., co-workers) as opposed to generic others. Though not immune to issues associated with self-report measures (Baranowski, 1988; Sallis & Saelens, 2000), the operationalization of sedentary behavior was another strength of this research. Most studies examining sedentary behavior have relied on measures of TV viewing or have measured sedentary behavior as the absence of physical activity (Marshall & Ramirez, 2011). The measure used in this study followed the recommendations of reviews of the area (e.g., Chau et al., 2010; Marshall & Ramirez, 2011), and focused on domain-specific sitting time (i.e., in the workplace) to operationalize sedentary behavior. In attempt to break up sedentary sitting time, and in line with recommendations (Healy et al., 2011), measures also targeted light activities such as standing, walking, and stair-use. Finally, participants were asked to only recall times that they engaged in standing, walking, or stair use behavior with the intent to be active. Focusing on specific behaviors performed with intention was done in attempt to minimize selfreport bias, as general measures of light activity tend to be among the least accurately recalled forms of activity (Sallis & Saelens, 2000). Measures of initial norm-perceptions and manipulation checks proved useful. For example, the poor recall of the personal and contextual similarity manipulation components of the messages revealed through the manipulation checks provided one possible explanation for why differences between high and low similarity were not found.
83
The extension of normative research to sedentary behavior also is a strength. In light of the negative health consequences of sedentary behavior and the prevalence of this behavior, especially for those in “desk jobs”, there is a need for research in this area. Despite the need to examine this behavior given the sedentary nature of many adults' occupations, most research in the area has focused on TV viewing among children and adolescents, and there has been a lack of successful experimental manipulations influencing sedentary behavior in adults (Marshall & Ramirez, 2011). As such, there has been a call for more research aimed at understanding how to reduce sedentary behavior in adults, and specifically in those with office jobs (Chau et al., 2010). Finally, this study used an experimental design including multiple time points. Researchers in the area (e.g., Rimal, 2008) have highlighted the need for more experimental research when examining the effect of norms on behavior. Conclusion The findings of the current study provide a first step in terms of understanding if normative messages might contribute to decreased sedentary behavior and increased light activity in an office setting. While the effects did not change for individuals receiving messages about reference groups that varied in similarity, descriptive norm email messages about others' behavior did result in a decrease in extended sitting behavior and an increase in specific light activities in the office setting. Acknowledgements This work was supported by a Vanier Canada Graduate Scholarship (first author) from the Social Sciences and Humanities Research Council of Canada. References Baranowski, T. (1988). Validity and reliability of self-report of physical activity: an information processing perspective. Research Quarterly, 59, 314e327. Biddle, S. J. (2007). Sedentary behavior. American Journal of Preventive Medicine, 33, 502e504. Biddle, S. J., Gorely, T., Marshall, S. J., Murdey, I., & Cameron, N. (2004). Physical activity and sedentary behaviors in youth: issues and controversies. The Journal of the Royal Society for the Promotion of Health, 124, 29e33. Chau, J. Y., Van der Ploeg, H. P., van Uffelen, J. G. Z., Wong, J., Riphagen, I., Healy, G. N., et al. (2010). Are workplace interventions to reduce sitting effective? A systematic review. Preventive Medicine, 51, 352e356. http://dx.doi.org/ 10.1016/j.ypmed.2010.08.012. Cialdini, R. B. (2003). Crafting normative messages to protect the environment. Current Directions in Psychological Science, 12, 105e109. Cialdini, R. B., Reno, R. R., & Kallgren, C. A. (1990). A focus theory of normative conduct: recycling the concept of norms to reduce littering in public places. Journal of Personality and Social Psychology, 58, 1015e1026. Cohen, J. (1969). Statistical power analysis for the behavioral sciences. New York: Academic Press. Colley, R. C., Garriguet, D., Janssen, I., Craig, C. L., Clarke, J., & Tremblay, M. S. (2011). Physical activity of Canadian adults: accelerometer results from the 2007 to 2009 Canadian Health Measures Survey. Health Reports/Statistics Canada, Canadian Centre for Health Information, 22, 7e14. Retrieved from http://www.ncbi. nlm.nih.gov/pubmed/21510585. Cragg, C., Wolfe, R., Griffiths, J. M., & Cameron, C. (2007). Physical activity among Canadian workers: Trends 2001e2006. Ottawa, ON: Canadian Fitness and Lifestyle Research Institute. Gilson, N. D., Puig-Ribera, A., McKenna, J., Brown, W. J., Burton, N. W., & Cooke, C. B. (2009). Do walking strategies to increase physical activity reduce reported sitting in workplaces: a randomized control trial. International Journal of Behavioural Nutrition and Physical Activity, 6, 43. http://dx.doi.org/10.1186/14795868-6-43. Godin, G., & Shephard, R. J. (1985). A simple method to assess exercise behavior in the community. Canadian Journal of Applied Sports Science, 10, 141e146. Goldstein, N. J., Cialdini, R. B., & Griskevicius, V. (2008). A room with a viewpoint: using social norms to motivate environment conservation in hotels. Journal of Consumer Research, 35, 472e482. http://dx.doi.org/10.1086/586910.
84
C.S. Priebe, K.S. Spink / Psychology of Sport and Exercise 19 (2015) 76e84
Hamilton, M., Hamilton, D. G., & Zderic, T. W. (2004). Exercise physiology versus inactivity physiology: an essential concept for understanding lipoprotein lipase regulation. Exercise and Sport Sciences Reviews, 32, 161e166. Healy, G. N., Matthews, C. E., Dunstan, D. W., Winkler, E. A. H., & Owen, N. (2011). Sedentary time and cardio-metabolic biomarkers in US adults: NHANES 2003e06. European Heart Journal, 32, 590e597. Hu, F. B., Li, T. Y., Colditz, G. A., Willett, W. C., & Manson, J. E. (2003). Television watching and other sedentary behaviors in relation to risk of obesity and type 2 diabetes mellitus in women. The Journal of the American Medical Association, 289, 1785e1791. Jacobs, D. R., Ainsworth, B. E., Hartman, T. J., & Leon, A. S. (1993). A simultaneous evaluation of 10 commonly used physical activity questionnaires. Medicine and Science in Sport and Exercise, 25, 81e91. Levine, J. A., Eberhardt, N. L., & Jensen, M. D. (1999). Role of non-exercise activity thermogenesis in resistance to fat gain in humans. Science, 283, 212e214. Marshall, S. J., & Ramirez, E. (2011). Reducing sedentary behavior: a new paradigm in physical activity promotion. American Journal of Lifestyle Medicine, 5, 516e530. Owen, N., Healy, G. N., Matthews, C. E., & Dunstan, D. W. (2010). Too much sitting: the population-health science of sedentary behavior. Exercise and Sport Sciences Reviews, 38, 105e113. Pate, R., O'Neill, J. R., & Lobelo, F. (2008). The evolving definition of “sedentary.” Exercise and Sport Sciences Reviews, 36, 171e172. Polonec, L. D., Major, A. M., & Atwood, L. E. (2006). Evaluating the believability and effectiveness of the social norms message “most students drink 0 to 4 drinks when they party”. Health Communications, 20, 23e34. Priebe, C. S., & Spink, K. S. (2011). When in Rome: descriptive norms and physical activity. Psychology of Sport & Exercise, 12, 93e98. Priebe, C. S., & Spink, K. S. (2012). Using messages promoting descriptive norms to increase physical activity. Health Communication, 27, 284e291.
Rimal, R. N. (2008). Modeling the relationship between descriptive norms and behaviors: a test and extension of the theory of normative social behavior (TNSB). Health Communications, 23, 103e116. Rimal, R. N., Lapinski, M. K., Cook, R. J., & Real, K. (2005). Moving towards a theory of normative influences: how perceived benefits and similarity moderate the impact of descriptive norms on behaviors. Journal of Health Communication: International Perspectives, 10, 433e450. Rimal, R. N., & Real, K. (2005). How behaviors are influenced by perceived norms: a test of the theory of normative social behavior. Communications Research, 32, 389e414. Ritchie, D., Price, V., & Roberts, D. F. (1987). Reading and television: a longitudinal investigation of the displacement hypothesis. Communication Research, 14, 292e315. Roethlisberger, F. J., & Dickson, W. J. (1939). Management and the worker. Cambridge, MA: Harvard University Press. Sallis, J. F., & Saelens, B. E. (2000). Assessment of physical activity by self-report: status, limitations, and future directions. Research Quarterly for Exercise and Sport, 71, 1e14. Schultz, P. W., Nolan, J. M., Cialdini, R. B., Goldstein, N. J., & Griskevicius, V. (2007). The constructive, destructive, and reconstructive power of social norms. Psychological Science, 18, 429e434. Tabachnick, B. G., & Fidell, L. S. (2007). In Using multivariate statistics (5th ed.). Boston, MA: Allyn & Bacon. Terry, D. J., & Hogg, M. A. (1996). Group norms and the attitude-behavior relationship: a role for group identification. Personality and Social Psychology Bulletin, 22, 776e793. WHO/WEF. (2008). Preventing non-communicable diseases in the workplace through diet and physical activity: WHO/World economic forum report of a joint event. Geneva: World Health Organization/World Economic Forum.