Page 224 - The Handbook of Persuasion and Social Marketing

212 The Handbook of Persuasion and Social Marketing

Unfortunately, almost without exception, the effects are rather small.
The average across all studies, with all beneficial effects coded to have a
positive sign, is r = .092. The largest effect is d = .45 (r = .22), obtained for
stress management behaviors (Johnson, Scott-Sheldon, & Carey, 2010)
and for dimensions of recycling and conservation efforts (Osbaldiston
& Schott, 2012). The next largest is OR = 2.0 (r = .19), for programs
to increase condom use in developing countries. However, the authors
report that the quality of the studies analyzed is low due to a lack of rand-
omized controlled trials, demographic differences between test and con-
trol groups, and so on (Sweat, Denison, Kennedy, Tedrow, & O’Reilly,
2012). Also, other interventions related to sexual activities have an average
r of just .05.
Two sets of meta-analyses evaluated by Johnson, Scott-Sheldon, and
Carey (2010) show the effects of interventions on participation in health
services (r = .17) and screening or treatment behaviors for women
(r = .10). The interventions excluded mass media programs, which Snyder
and colleagues (2004) found to have an effect of r = .09 in health commu-
nication campaigns. Similarly, Cugelman, Thelwall, and Dawes (2011)
found d = .19 (r = .095) for online interventions aimed at health behavior
changes.
Johnson and colleagues (2010) examined 13 addiction-related meta-
analyses and reported an average finding of d = .21 (r = .10). This average
is at the high end of results for meta-analyses of substance abuse and
smoking interventions, with one study of mass-communications cam-
paigns reporting a result of r = .02 (Derzon & Lipsey, 2002).
The best result effect for youth weight loss from obesity interventions
is d = .29 (r = .14; Katz, O’Connell, Njike, Yeh, & Nawaz, 2008). Johnson
and colleagues (2010) combined meta-analyses on eating and physical
activity and found d = .22 (r = .11). Several other studies showed
smaller effects in terms of r values or specific reductions in body mass
index (BMI).
Interventions on the remaining topics in Table 8.1 show r values of
.09 or lower. Given the topics studied, small effects may nevertheless
justify social marketing programs. For example, any traffic accident
may have important financial or health consequences, potentially justify-
ing driver improvement programs with r = .03 (Masten & Peck, 2004).
Every additional organ donor may save multiple lives, so r = .05 for
donation campaigns is not trivial (Feeley & Moon, 2009). The small effect
sizes in Table 8.1 may therefore offer realistic expectations without neces-
sarily leading to pessimism about the potential for social marketing
programs.