Psychology and Aging 2014, Vol. 29, No. 4, 757-763
© 2014 American Psychological Association 0882-7974/14/512.00 DOI: 10.1037/a0036486
A Randomized Controlled Trial to Promote Volunteering in Older Adults Lisa M. Warner
Julia K. Wolff and Jochen P. Ziegelmann
German Centre of Gerontology, Berlin, Germany and Freie Universitat Berlin
German Centre of Gerontology, Berlin, Germany
Susanne Wurm German Centre of Gerontology, Berlin, Germany and Friedrich-Alexander University Erlangen-Nuremberg Volunteering is presumed to confer health benefits, but interventions to encourage older adults to volunteer are sparse. Therefore, a randomized controlled trial with 280 community-dwelling older German adults was conducted to test the effects of a theory-based social-cognitive intervention against a passive waiting-list control group and an active control intervention designed to motivate physical activity. Self-reports of weekly volunteering minutes were assessed at baseline (5 weeks before the intervention) as well as 2 and 6 weeks after the intervention. Participants in the treatment group increased their weekly volunteering minutes to a greater extent than participants in the control groups 6 weeks after the intervention. We conclude that a single, face-to-face group session can increase volunteering among older community-dwelling adults. However, the effects need some time to unfold because changes in volunteering were not apparent 2 weeks after the intervention. Keywords: volunteering, older adults, theory-based social-cognitive intervention, randomized controlled trial
Although almost half of the U.S. (Independent Sector, 2001) and almost one third of the German (Bundesministerium fur Familie, Senioren, Frauen und Jugend, 2010) populations aged 65 and older engage in volunteer work, Gottlieb and Gillespie (2008) predict a potential shortage of volunteers in the future, with consequences for various areas of society. Given the salutary benefits of volun teering for society and for volunteers themselves (Okun, Yeung, & Brown, 2013), this predicted shortage of volunteers has led to calls for developing interventions to increase volunteering (e.g., Okun et al., 2013). Therefore, the aim of this study was to test the effects of a short, face-to-face group intervention to encourage community-dwelling older adults to volunteer. Because socialcognitive frameworks have been demonstrated to be valid in the domain of volunteering (Greenslade & White, 2005; Warburton, Terry, Rosenman, & Shapiro, 2001), the intervention was based on social-cognitive behavior change techniques.
Longitudinal and Experimental Evidence for the Benefits of Volunteering The benefits of volunteering for society are manifold and range from economic benefits for institutions that employ volunteers to lower depressive symptoms and isolation for recipients of volun teer services (Wheeler, Gorey, & Greenblatt, 1998). Moreover, volunteers profit themselves in terms of higher quality of life, social integration, self-rated health, ability to cope with stressful life events, and number of years lived without morbidity as well as lower levels of depressive symptoms and physical limitations (Onyx & Warburton, 2003), and this is perhaps especially true for older and retired adults (Sneed & Cohen, 2013; Van Willigen, 2000). A recent meta-analysis further confirms a reliable associa tion between volunteering and mortality in older adults (Okun et al., 2013). Despite cross-sectional and prospective evidence, these findings are often confronted with the argument of reverse causal ity: Does volunteer work generate health benefits or are healthier individuals more likely to volunteer (Warburton & Peel, 2008)? Experimental studies suggest that volunteer work in fact con tributes to better health and health behavior. A recent randomized controlled trial on healthy 1Oth-grade students showed that being assigned to weekly volunteering sessions with elementary schoolchildren improved cardiovascular reactivity after 2 months as compared with a waiting-list control group (Schreier, SchonertReichl, & Chen, 2013). For older adults, the Experience Corps Program found that participants who were randomly assigned to volunteer in elementary schools experienced greater gains than wait-list controls in terms of higher physical activity, muscle strength, cognitive performance, and social integration at 4- to 8-month follow-up assessments (Carlson et al., 2009; Fried et al., 2004; Tan, Xue, Li, Carlson, & Fried, 2006). Although both
This article was published Online First August 18, 2014. Lisa M. Warner, German Centre of Gerontology, Berlin, Germany and Department of Health Psychology, Freie Universitat Berlin, Berlin, Germany; Julia K. Wolff and Jochen P. Ziegelmann, German Centre of Gerontology; Susanne Wurm, German Centre of Gerontology and Institute of Psychogeron tology, Friedrich-Alexander University Erlangen-Nuremberg, Berlin, Ger many. This work was supported by the German Federal Ministry of Education and Research. The authors thank the German Centre of Gerontology for logistic support and a team of highly motivated student assistants for helping to conduct the study. Correspondence concerning this article should be addressed to Lisa M. Warner. Health Psychology, Freie Universitat Berlin, Habelschwerdter Allee 45, 14195 Berlin, Germany. E-mail: [email protected]
757
758
WARNER, WOLFF, ZIEGELMANN, AND WURM
studies were conducted in the field, they were highly structured, and participants were randomized to specific volunteer tasks (e.g., support in schools) to ensure comparable doses of volunteering (e.g., at least 15 hr/week). However, in everyday life, most older adults organize their volunteer work themselves. This provides them with the chance to select volunteer tasks that are meaningful to them and to self-determine the type and amount of work, but it requires high motivation as well as self-regulatory and social skills to find a suitable volunteer position and to maintain it. Therefore, interventions that aim to promote older adults’ motivation for volunteering should be more sustainable if they prompt selfregulatory strategies to help them find and maintain volunteer work by themselves rather than enrolling them in specific volun teer work for a defined study period.
Motivation to Volunteer Among Older Adults Earlier studies on people’s motivation to volunteer mainly fo cused on various functions of volunteering. They found that older adults are motivated to volunteer out of altruistic reasons (e.g., helping people in need or contributing to society), but that they also pursue self-oriented motives such as structuring and filling leisure time, being socially integrated, gaining social approval, and boosting their self-esteem (Okun, 1994; Warburton et al., 2001). However, simply asking current volunteers for their motives to volunteer neglects other cognitions, such as the costs of volunteer ing, whether people feel capable to volunteer, or whether or not they are supported by their environment (Greenslade & White, 2005). Therefore, more recent studies investigated the explanatory value of social-cognitive theories for volunteering. Their main findings are that the motivation to volunteer is based on positive attitudes toward volunteering, self-efficacy beliefs, and social sup port (Warburton et al., 2001). Hence, people who expect benefits of volunteering, feel capable of volunteering, and perceive that their significant others approve of and support volunteering are more likely to be motivated to volunteer than others (Grano, Lucidi, Zelli, & Violani, 2008; Greenslade & White, 2005; Morrow-Howell, Hong, & Tang, 2009). Although there is com prehensive research on the factors that predict volunteering, these studies typically use correlative designs and consider only people who volunteer (Fisher & Ackerman, 1998). Therefore, they give little insight into how organizations or interventionists could en courage older adults to initiate or maintain volunteering. One of the few randomized controlled trials to promote volun teering without assigning participants to specific volunteer work prompted volunteering among adolescents (Wilson, Allen, Strahan, & Ethier, 2008). In two 80-min interactive group sessions delivered over a 2-week period, interventionists provided informa tion about the benefits of volunteering and encouraged debates about strategies on how to overcome barriers to volunteer. Com pared with an active control group that debated body image in an identical setting, the intervention group reported higher intentions for various kinds of volunteer work 1 week later. Two further experimental studies in the laboratory (in samples of undergraduate students and parents of 4- to 17-year-old chil dren) varied the amount of expected social recognition and per ceived needs for volunteer services in specific organizations on flyers and posters (Fisher & Ackerman, 1998). The intention to
volunteer was higher among participants in the condition with high expected social approval and high organizational need, confirming the authors’ hypothesis of social influences on volunteering moti vation. Considering the gap between intentions and behavior (Sheeran, 2002), the evaluation of intervention effects in terms of short-term changes in intentions to volunteer rather than actual involvement in volunteer work provides little guidance on how to prompt volunteering behavior. Therefore, on the basis of broad correla tional evidence (but sparse experimental evidence), socialcognitive interventions to increase volunteering should contain behavior change techniques that focus on positive outcome expec tancies of volunteering, self-efficacy, and social support. A wellestablished social-cognitive framework that incorporates these constructs is the Health Action Process Approach (HAPA), which was found to be valid in various health behaviors (Schwarzer et al., 2007). Research on the effects of interventions to increase volun teering should further evaluate their success in terms of the amount of volunteer work rather than intentions and have a longitudinal design.
Aims of the Current Study The current randomized controlled trial was conducted to test a short, face-to-face group intervention to increase volunteering among community-dwelling older adults by means of prompting social cognitions based on the HAPA (Schwarzer, 2008). The intervention targeted participants’ self-regulatory skills to choose and organize volunteer work on their own initiative without im posing any specific volunteer services on them. It was hypothe sized that participants in the intervention group would show higher levels of volunteering at the 2- and 6-week follow-up than partic ipants who attended the active or passive control group.
Method Participants and Procedure A sample of community-dwelling adults aged 64 and older was recruited via newspaper articles and advertisements in a large German city. The study was named “Active Retirement,” and the communicated study purpose was that volunteering and physical activity in retirement and their relation to health and well-being would be discussed in randomized groups. Interested retirees called the research institute and left their contact details to be called back. In this telephone interview, potential participants were selected as eligible for participation if they were 64 years or older, not acutely physically impaired or disabled, not exercising on a regular basis, and not seriously cognitively impaired. In total, 647 older adults were assessed for eligibility. The CONSORT flow diagram in Figure 1 displays exclusion and dropout rates. Within the first telephone interview, participants were randomized into three groups using the software R (http:// cran.r-project.org/) via the function “sample” of the R package “base” with predefined group sizes for the volunteering interven tion group (VIG), the active control group (ACG), and the passive control group with no intervention (PCG). Three hundred and ten participants provided informed consent and completed the baseline paper-and-pencil questionnaire (Tl) at the research institute in
VOLUNTEERING INTERVENTION IN OLDER ADULTS
Figure 1.
759
CONSORT flow diagram.
Spring 2012. All participants of T1 were remunerated with €25. The intervention session and the ACG session took place 5 weeks after T l. Sessions for the ACG were randomized into sessions that had parallel content to the VIG and sessions that were nonparallel with the VIG. Participants in the nonparallel sessions (n = 30) were not considered in the analyses for this study. Therefore, the final sample for this study comprises 280 individuals. All three groups received follow-up questionnaires via mail with prepaid return envelopes 2 (T2) and 6 weeks (T3) after the intervention (the PCG at respective time intervals). The first follow-up (T2) was completed by 253 participants, and 244 participants completed T3. Participants were on average 70.29 years of age (SD = 4.95, range = 64-92 years) and reported 4.65 (SD = 2.78) illnesses at T l. Most were women (76%) and high-school graduates (59%); 44% lived with a partner. Ethical consent was granted from the Ethics Commission of the German Psychological Society (DGPs-SW 02_2012).
Experimental Conditions The intervention development was based on previous correla tional and experimental evidence and the HAPA model (Schwarzer, 2008). This model postulates that behavioral intentions are built on risk perception (not assumed to be relevant for volunteer ing), outcome expectancies, and self-efficacy in the motivational phase, whereas intentions are translated into behavior via selfregulatory strategies such as planning and self-monitoring in the volitional phase. Because volunteers and nonvolunteers should be addressed likewise, motivational and volitional strategies were chosen. The following behavior change techniques were used in interactive group intervention sessions to prompt volunteering (Michie et al., 2011): information about the benefits of volunteer ing in old age (prompt for outcome expectancies), focus on past success (prompt for self-efficacy in biography worksheet), goal setting behavior and outcome (prompt for intention formation in worksheet), action planning and use of cues (prompt for if-then
760
WARNER, WOLFF, ZIEGELMANN, AND WURM
implementation intentions in worksheet), and modeling behavior (prompt for self-efficacy in a 5-min video clip with older person as role model). Furthermore, informational material on volunteering opportunities for older adults in their residential area was available for free, and self-monitoring of behavior was prompted between T2 and T3 (10-day volunteering diary). Because volunteering can have detrimental effects on health if performed excessively, par ticipants were also informed about keeping weekly volunteer min utes to an amount that would not burden them (Musick, Herzog, & House, 1999). The ACG received the same behavior change techniques (in cluding the diary) as the VIG in interactive group sessions with similar length, but all material was adapted to physical activity. The PCG received neither intervention session nor informational material nor a diary during that time. Two female psychologists with doctoral degrees developed the intervention content and were the main interventionists for all group sessions. Both learned the session protocols by heart and read the manual again before every session. Further, sessions were structured by standardized PowerPoint presentations to ensure comparability of session content. Most sessions were conjointly led by those two women (n = 25 sessions), but some were conducted with only one of them leading the session (e.g., because of illness) while a trained psychology student assisted with content-unrelated tasks such as writing benefits on the whiteboard or starting the video (n = 9 sessions).
Measures Weekly volunteering minutes were assessed with two items adapted from Ayalon (2008): (a) “During the past 4 weeks, on how many days per week did you do volunteer work?” and (b) “If you did volunteer work, how many minutes did one session last on average?” They were multiplied and divided by 4 (weeks) to create a score for the average weekly volunteering minutes. Outliers were truncated to 2 SD above the respective M within each measurement point in time (results do not change with raw data). Covariates were participants’ age, gender, education, partner status (1 = partner, 0 = without partner), and number of illnesses at T1 because these variables were shown to be associated with volunteering in previous studies (Thoits & Hewitt, 2001). Educa tion was assessed and classified according to the International Standard Classification of Education (ISCED; Unesco, 1997), with 1 indicating low education (< 9 years school education), 2 indi cating medium education (secondary school), and 3 indicating high education (qualifying for university admission). The number of illnesses was assessed with a list of 25 illnesses mentioned either in the Charlson Comorbidity Index (Charlson, Szatrowski, Peter son, & Gold, 1994) or the Functional Comorbidity Index (Groll, To, Bombardier, & Wright, 2005) and summed up.
Data Analyses The intervention effect was tested with latent change scores, modeling change in volunteering minutes from T1 to T2 and from T2 to T3, and level of volunteering at T2 in Mplus (Mun, von Eye, & White, 2009). Missing values on volunteering (T1 = 8.1%, T2 = 17.1%, T3 = 25.2%) were imputed via full information maximum likelihood estimation (FIML) in Mplus because FIML
makes use of all available data in model estimation (Arbuckle, 1996). Two dummy variables (ACG and PCG) using the VIG as the reference group were entered as predictors of the latent change scores. Significant associations of ACG versus VIG or PCG versus VIG with changes in volunteering can be interpreted as treatment effects. The model was statistically controlled for participants’ age, gender, education, partner status, and number of illnesses at T l.
Attrition Analysis Those 36 participants who dropped out were examined for significant differences compared with participants with complete data at T l. There were no statistically significant differences between participants who dropped out and those who stayed in the study in terms of weekly volunteering minutes f(255) = 0.49, p = •26 (^dropouts = 120.30 min, = 164.27 min; ^completers = 141-85 min, SDcompleters = 228.49 min), age r(275) = 0.26, p = .64 (Mdropouts = 70.09 years, SDdropouts = 5.11 years, A^compjeters 73.32 years, 5Dcornpjeters — 4.93 years), gen der x2(l) = 0-22, p = .64 (dropouts 75.8% women, completers 77.8% women), education r(276) = -0 .4 4 , p = .57 (Mdropouts = 2.51, SD dTapoutx = 0.70; Mcompleters = 2.46, SDcompleters = 0.72), partner status x2(l) = 0-25, p = .62 (dropouts 44.3% with partner, completers 38.9% with partner), or number of illnesses
© 2014 American Psychological Association 0882-7974/14/512.00 DOI: 10.1037/a0036486
A Randomized Controlled Trial to Promote Volunteering in Older Adults Lisa M. Warner
Julia K. Wolff and Jochen P. Ziegelmann
German Centre of Gerontology, Berlin, Germany and Freie Universitat Berlin
German Centre of Gerontology, Berlin, Germany
Susanne Wurm German Centre of Gerontology, Berlin, Germany and Friedrich-Alexander University Erlangen-Nuremberg Volunteering is presumed to confer health benefits, but interventions to encourage older adults to volunteer are sparse. Therefore, a randomized controlled trial with 280 community-dwelling older German adults was conducted to test the effects of a theory-based social-cognitive intervention against a passive waiting-list control group and an active control intervention designed to motivate physical activity. Self-reports of weekly volunteering minutes were assessed at baseline (5 weeks before the intervention) as well as 2 and 6 weeks after the intervention. Participants in the treatment group increased their weekly volunteering minutes to a greater extent than participants in the control groups 6 weeks after the intervention. We conclude that a single, face-to-face group session can increase volunteering among older community-dwelling adults. However, the effects need some time to unfold because changes in volunteering were not apparent 2 weeks after the intervention. Keywords: volunteering, older adults, theory-based social-cognitive intervention, randomized controlled trial
Although almost half of the U.S. (Independent Sector, 2001) and almost one third of the German (Bundesministerium fur Familie, Senioren, Frauen und Jugend, 2010) populations aged 65 and older engage in volunteer work, Gottlieb and Gillespie (2008) predict a potential shortage of volunteers in the future, with consequences for various areas of society. Given the salutary benefits of volun teering for society and for volunteers themselves (Okun, Yeung, & Brown, 2013), this predicted shortage of volunteers has led to calls for developing interventions to increase volunteering (e.g., Okun et al., 2013). Therefore, the aim of this study was to test the effects of a short, face-to-face group intervention to encourage community-dwelling older adults to volunteer. Because socialcognitive frameworks have been demonstrated to be valid in the domain of volunteering (Greenslade & White, 2005; Warburton, Terry, Rosenman, & Shapiro, 2001), the intervention was based on social-cognitive behavior change techniques.
Longitudinal and Experimental Evidence for the Benefits of Volunteering The benefits of volunteering for society are manifold and range from economic benefits for institutions that employ volunteers to lower depressive symptoms and isolation for recipients of volun teer services (Wheeler, Gorey, & Greenblatt, 1998). Moreover, volunteers profit themselves in terms of higher quality of life, social integration, self-rated health, ability to cope with stressful life events, and number of years lived without morbidity as well as lower levels of depressive symptoms and physical limitations (Onyx & Warburton, 2003), and this is perhaps especially true for older and retired adults (Sneed & Cohen, 2013; Van Willigen, 2000). A recent meta-analysis further confirms a reliable associa tion between volunteering and mortality in older adults (Okun et al., 2013). Despite cross-sectional and prospective evidence, these findings are often confronted with the argument of reverse causal ity: Does volunteer work generate health benefits or are healthier individuals more likely to volunteer (Warburton & Peel, 2008)? Experimental studies suggest that volunteer work in fact con tributes to better health and health behavior. A recent randomized controlled trial on healthy 1Oth-grade students showed that being assigned to weekly volunteering sessions with elementary schoolchildren improved cardiovascular reactivity after 2 months as compared with a waiting-list control group (Schreier, SchonertReichl, & Chen, 2013). For older adults, the Experience Corps Program found that participants who were randomly assigned to volunteer in elementary schools experienced greater gains than wait-list controls in terms of higher physical activity, muscle strength, cognitive performance, and social integration at 4- to 8-month follow-up assessments (Carlson et al., 2009; Fried et al., 2004; Tan, Xue, Li, Carlson, & Fried, 2006). Although both
This article was published Online First August 18, 2014. Lisa M. Warner, German Centre of Gerontology, Berlin, Germany and Department of Health Psychology, Freie Universitat Berlin, Berlin, Germany; Julia K. Wolff and Jochen P. Ziegelmann, German Centre of Gerontology; Susanne Wurm, German Centre of Gerontology and Institute of Psychogeron tology, Friedrich-Alexander University Erlangen-Nuremberg, Berlin, Ger many. This work was supported by the German Federal Ministry of Education and Research. The authors thank the German Centre of Gerontology for logistic support and a team of highly motivated student assistants for helping to conduct the study. Correspondence concerning this article should be addressed to Lisa M. Warner. Health Psychology, Freie Universitat Berlin, Habelschwerdter Allee 45, 14195 Berlin, Germany. E-mail: [email protected]
757
758
WARNER, WOLFF, ZIEGELMANN, AND WURM
studies were conducted in the field, they were highly structured, and participants were randomized to specific volunteer tasks (e.g., support in schools) to ensure comparable doses of volunteering (e.g., at least 15 hr/week). However, in everyday life, most older adults organize their volunteer work themselves. This provides them with the chance to select volunteer tasks that are meaningful to them and to self-determine the type and amount of work, but it requires high motivation as well as self-regulatory and social skills to find a suitable volunteer position and to maintain it. Therefore, interventions that aim to promote older adults’ motivation for volunteering should be more sustainable if they prompt selfregulatory strategies to help them find and maintain volunteer work by themselves rather than enrolling them in specific volun teer work for a defined study period.
Motivation to Volunteer Among Older Adults Earlier studies on people’s motivation to volunteer mainly fo cused on various functions of volunteering. They found that older adults are motivated to volunteer out of altruistic reasons (e.g., helping people in need or contributing to society), but that they also pursue self-oriented motives such as structuring and filling leisure time, being socially integrated, gaining social approval, and boosting their self-esteem (Okun, 1994; Warburton et al., 2001). However, simply asking current volunteers for their motives to volunteer neglects other cognitions, such as the costs of volunteer ing, whether people feel capable to volunteer, or whether or not they are supported by their environment (Greenslade & White, 2005). Therefore, more recent studies investigated the explanatory value of social-cognitive theories for volunteering. Their main findings are that the motivation to volunteer is based on positive attitudes toward volunteering, self-efficacy beliefs, and social sup port (Warburton et al., 2001). Hence, people who expect benefits of volunteering, feel capable of volunteering, and perceive that their significant others approve of and support volunteering are more likely to be motivated to volunteer than others (Grano, Lucidi, Zelli, & Violani, 2008; Greenslade & White, 2005; Morrow-Howell, Hong, & Tang, 2009). Although there is com prehensive research on the factors that predict volunteering, these studies typically use correlative designs and consider only people who volunteer (Fisher & Ackerman, 1998). Therefore, they give little insight into how organizations or interventionists could en courage older adults to initiate or maintain volunteering. One of the few randomized controlled trials to promote volun teering without assigning participants to specific volunteer work prompted volunteering among adolescents (Wilson, Allen, Strahan, & Ethier, 2008). In two 80-min interactive group sessions delivered over a 2-week period, interventionists provided informa tion about the benefits of volunteering and encouraged debates about strategies on how to overcome barriers to volunteer. Com pared with an active control group that debated body image in an identical setting, the intervention group reported higher intentions for various kinds of volunteer work 1 week later. Two further experimental studies in the laboratory (in samples of undergraduate students and parents of 4- to 17-year-old chil dren) varied the amount of expected social recognition and per ceived needs for volunteer services in specific organizations on flyers and posters (Fisher & Ackerman, 1998). The intention to
volunteer was higher among participants in the condition with high expected social approval and high organizational need, confirming the authors’ hypothesis of social influences on volunteering moti vation. Considering the gap between intentions and behavior (Sheeran, 2002), the evaluation of intervention effects in terms of short-term changes in intentions to volunteer rather than actual involvement in volunteer work provides little guidance on how to prompt volunteering behavior. Therefore, on the basis of broad correla tional evidence (but sparse experimental evidence), socialcognitive interventions to increase volunteering should contain behavior change techniques that focus on positive outcome expec tancies of volunteering, self-efficacy, and social support. A wellestablished social-cognitive framework that incorporates these constructs is the Health Action Process Approach (HAPA), which was found to be valid in various health behaviors (Schwarzer et al., 2007). Research on the effects of interventions to increase volun teering should further evaluate their success in terms of the amount of volunteer work rather than intentions and have a longitudinal design.
Aims of the Current Study The current randomized controlled trial was conducted to test a short, face-to-face group intervention to increase volunteering among community-dwelling older adults by means of prompting social cognitions based on the HAPA (Schwarzer, 2008). The intervention targeted participants’ self-regulatory skills to choose and organize volunteer work on their own initiative without im posing any specific volunteer services on them. It was hypothe sized that participants in the intervention group would show higher levels of volunteering at the 2- and 6-week follow-up than partic ipants who attended the active or passive control group.
Method Participants and Procedure A sample of community-dwelling adults aged 64 and older was recruited via newspaper articles and advertisements in a large German city. The study was named “Active Retirement,” and the communicated study purpose was that volunteering and physical activity in retirement and their relation to health and well-being would be discussed in randomized groups. Interested retirees called the research institute and left their contact details to be called back. In this telephone interview, potential participants were selected as eligible for participation if they were 64 years or older, not acutely physically impaired or disabled, not exercising on a regular basis, and not seriously cognitively impaired. In total, 647 older adults were assessed for eligibility. The CONSORT flow diagram in Figure 1 displays exclusion and dropout rates. Within the first telephone interview, participants were randomized into three groups using the software R (http:// cran.r-project.org/) via the function “sample” of the R package “base” with predefined group sizes for the volunteering interven tion group (VIG), the active control group (ACG), and the passive control group with no intervention (PCG). Three hundred and ten participants provided informed consent and completed the baseline paper-and-pencil questionnaire (Tl) at the research institute in
VOLUNTEERING INTERVENTION IN OLDER ADULTS
Figure 1.
759
CONSORT flow diagram.
Spring 2012. All participants of T1 were remunerated with €25. The intervention session and the ACG session took place 5 weeks after T l. Sessions for the ACG were randomized into sessions that had parallel content to the VIG and sessions that were nonparallel with the VIG. Participants in the nonparallel sessions (n = 30) were not considered in the analyses for this study. Therefore, the final sample for this study comprises 280 individuals. All three groups received follow-up questionnaires via mail with prepaid return envelopes 2 (T2) and 6 weeks (T3) after the intervention (the PCG at respective time intervals). The first follow-up (T2) was completed by 253 participants, and 244 participants completed T3. Participants were on average 70.29 years of age (SD = 4.95, range = 64-92 years) and reported 4.65 (SD = 2.78) illnesses at T l. Most were women (76%) and high-school graduates (59%); 44% lived with a partner. Ethical consent was granted from the Ethics Commission of the German Psychological Society (DGPs-SW 02_2012).
Experimental Conditions The intervention development was based on previous correla tional and experimental evidence and the HAPA model (Schwarzer, 2008). This model postulates that behavioral intentions are built on risk perception (not assumed to be relevant for volunteer ing), outcome expectancies, and self-efficacy in the motivational phase, whereas intentions are translated into behavior via selfregulatory strategies such as planning and self-monitoring in the volitional phase. Because volunteers and nonvolunteers should be addressed likewise, motivational and volitional strategies were chosen. The following behavior change techniques were used in interactive group intervention sessions to prompt volunteering (Michie et al., 2011): information about the benefits of volunteer ing in old age (prompt for outcome expectancies), focus on past success (prompt for self-efficacy in biography worksheet), goal setting behavior and outcome (prompt for intention formation in worksheet), action planning and use of cues (prompt for if-then
760
WARNER, WOLFF, ZIEGELMANN, AND WURM
implementation intentions in worksheet), and modeling behavior (prompt for self-efficacy in a 5-min video clip with older person as role model). Furthermore, informational material on volunteering opportunities for older adults in their residential area was available for free, and self-monitoring of behavior was prompted between T2 and T3 (10-day volunteering diary). Because volunteering can have detrimental effects on health if performed excessively, par ticipants were also informed about keeping weekly volunteer min utes to an amount that would not burden them (Musick, Herzog, & House, 1999). The ACG received the same behavior change techniques (in cluding the diary) as the VIG in interactive group sessions with similar length, but all material was adapted to physical activity. The PCG received neither intervention session nor informational material nor a diary during that time. Two female psychologists with doctoral degrees developed the intervention content and were the main interventionists for all group sessions. Both learned the session protocols by heart and read the manual again before every session. Further, sessions were structured by standardized PowerPoint presentations to ensure comparability of session content. Most sessions were conjointly led by those two women (n = 25 sessions), but some were conducted with only one of them leading the session (e.g., because of illness) while a trained psychology student assisted with content-unrelated tasks such as writing benefits on the whiteboard or starting the video (n = 9 sessions).
Measures Weekly volunteering minutes were assessed with two items adapted from Ayalon (2008): (a) “During the past 4 weeks, on how many days per week did you do volunteer work?” and (b) “If you did volunteer work, how many minutes did one session last on average?” They were multiplied and divided by 4 (weeks) to create a score for the average weekly volunteering minutes. Outliers were truncated to 2 SD above the respective M within each measurement point in time (results do not change with raw data). Covariates were participants’ age, gender, education, partner status (1 = partner, 0 = without partner), and number of illnesses at T1 because these variables were shown to be associated with volunteering in previous studies (Thoits & Hewitt, 2001). Educa tion was assessed and classified according to the International Standard Classification of Education (ISCED; Unesco, 1997), with 1 indicating low education (< 9 years school education), 2 indi cating medium education (secondary school), and 3 indicating high education (qualifying for university admission). The number of illnesses was assessed with a list of 25 illnesses mentioned either in the Charlson Comorbidity Index (Charlson, Szatrowski, Peter son, & Gold, 1994) or the Functional Comorbidity Index (Groll, To, Bombardier, & Wright, 2005) and summed up.
Data Analyses The intervention effect was tested with latent change scores, modeling change in volunteering minutes from T1 to T2 and from T2 to T3, and level of volunteering at T2 in Mplus (Mun, von Eye, & White, 2009). Missing values on volunteering (T1 = 8.1%, T2 = 17.1%, T3 = 25.2%) were imputed via full information maximum likelihood estimation (FIML) in Mplus because FIML
makes use of all available data in model estimation (Arbuckle, 1996). Two dummy variables (ACG and PCG) using the VIG as the reference group were entered as predictors of the latent change scores. Significant associations of ACG versus VIG or PCG versus VIG with changes in volunteering can be interpreted as treatment effects. The model was statistically controlled for participants’ age, gender, education, partner status, and number of illnesses at T l.
Attrition Analysis Those 36 participants who dropped out were examined for significant differences compared with participants with complete data at T l. There were no statistically significant differences between participants who dropped out and those who stayed in the study in terms of weekly volunteering minutes f(255) = 0.49, p = •26 (^dropouts = 120.30 min, = 164.27 min; ^completers = 141-85 min, SDcompleters = 228.49 min), age r(275) = 0.26, p = .64 (Mdropouts = 70.09 years, SDdropouts = 5.11 years, A^compjeters 73.32 years, 5Dcornpjeters — 4.93 years), gen der x2(l) = 0-22, p = .64 (dropouts 75.8% women, completers 77.8% women), education r(276) = -0 .4 4 , p = .57 (Mdropouts = 2.51, SD dTapoutx = 0.70; Mcompleters = 2.46, SDcompleters = 0.72), partner status x2(l) = 0-25, p = .62 (dropouts 44.3% with partner, completers 38.9% with partner), or number of illnesses
