Research Proposal

Declines in efficacy of anti-bullying programs among older adolescents: Theory and a three-level meta-analysis☆

David Scott Yeager a,⁎, Carlton J. Fong a, Hae Yeon Lee a, Dorothy L. Espelage b

a University of Texas at Austin, United States b University of Illinois at Urbana–Champaign, United States

a b s t r a c ta r t i c l e i n f o

Available online 21 January 2015

Keywords: Bullying Meta-analysis Adolescence Interventions Victimization

Highly visible tragedies in high schools thought to involve bullying have directly contributed to public support for state-mandated K-12 anti-bullying programming. But are existing programs actually effective for these older adolescents? This paper first outlines theoretical considerations, including developmental changes in (a) the manifestation of bullying, (b) the underlying causes of bullying, and (c) the efficacy of domain-general behav- ior-change tactics. This review leads to the prediction of a discontinuity in program efficacy among older adoles- cents. The paper then reports a novel meta-analysis of studies that administered the same program to multiple age groups and measured levels of bullying (k = 19, with 72 effect sizes). By conducting a hierarchical meta- analysis of the within-study moderation of efficacy by age, more precise estimates of age-related trends were possible. Results were consistent with theory in that whereas bullying appears to be effectively prevented in 7th grade and below, in 8th grade and beyond there is a sharp drop to an average of zero. This finding contradicts past meta-analyses that used between-study tests of moderation. This paper provides a basis for a theory of age- related moderation of program effects that may generalize to other domains. The findings also suggest the more general need for caution when interpreting between-study meta-analytic moderation results.

© 2014 Elsevier Inc. All rights reserved.

Contents

Part 1: Theoretical expectations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 Overview. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 Changes in the form of bullying. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39 Changes in the underlying causes of bullying. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

Social competence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39 Motives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40 Causes related to sex, sexuality, and race/ethnicity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

Changes in responsiveness to domain-general behavior-change strategies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41 Summary. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 A new meta-analysis is needed . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42

Part 2: A meta-analysis of within-study age moderation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43

Data. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 Measures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43

Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44 Do anti-bullying programs decline in efficacy with age? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44 Comparison to standard between-study techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46

Journal of Applied Developmental Psychology 37 (2015) 36–51

☆ The authors would like to thank the following people for responding to queries to provide additional data: Alexander Vazsonyi, Arthur Horne, Pamela Orpinas, Caroline Hunt, Christopher Cheng, Dennis Wong, Tanya Beran, Maila Koivisto, Donna Cross, Scott Menard, Karin Frey, Antti Karna, Anne Williford, Aaron Boulton, Mario Gollwitzer, Sally Kuykendall, Katherine Raczynski, Chris Bell, Carol Metzler, Ilse De Bourdeaudhuji, Susan Limber, Anna Baldry, David Farrington, Allison Fahsl, Robin Heydenberk, Leila Rahey, Wendy Hoglund, Martha Bleeker, Miia Sainio, Bonnie Leadbeater, Eric Brown, and Annis Fung. We also thank Carol Dweck for her comments on the manuscript and Mike Cheung for answering questions about three-level meta-analysis. ⁎ Corresponding author.

E-mail address: [email protected] (D.S. Yeager).

http://dx.doi.org/10.1016/j.appdev.2014.11.005 0193-3973/© 2014 Elsevier Inc. All rights reserved.

Contents lists available at ScienceDirect

Journal of Applied Developmental Psychology

Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 Potential limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47

Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 Appendix A. Supplementary data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48

Bullying is an aggressive act in which one or more individuals with relatively higher social power systematically and intentionally cause harm to an individual with relatively lower-power (Olweus, 1993). By now, the data are quite clear that victims of bullying suffer in terms of their social, emotional, academic, and physical development (Cook, Williams, Guerra, Kim, & Sadek, 2010; Copeland et al., 2014; Reijntjes, Kamphuis, Prinzie, & Telch, 2010; Reijntjes et al., 2011; Ttofi, Farrington, Losёl, & Loeber, 2011). The harm of victimization alone is reason for public action against bullying. But in the past decade and a half, there have also been a number of high-profile shootings and suicides carried out by older adolescents in high schools. Popular media interpretations have emphasized the role of bullying in these events (Gibbs, 2010; Grossman, 2009). It is unclear to what extent bullying actually contributes to such rare, extreme tragedies, but it is clear that such events have galvanized public support for laws requiring school-wide K-12 anti-bullying programs (Bierman, 2010; School Bullying Prohibited: Bullying Prevention Plan Act, 2010). However, do existing programs work among older adolescents, the age when many of the most visible tragedies have occurred?

Recent meta-analyses of past anti-bullying interventions have suggested that, although there is some notable variability across nations

and types of programs, “on average, anti-bullying campaigns have had some modest success” (Smith, 2011, p. 419; for null effects, see Merrell, Gueldner, Ross, & Isava, 2008; for evidence that programs are only effective in European countries, see Evans, Fraser, & Cotter, 2014; for modest positive effects, see Smith, Schneider, Smith, & Ananiadou, 2004; Ttofi & Farrington, 2011). Since past meta-analyses aggregated many studies conducted with many different grade levels, it has further been possible to conduct “meta-regression” analyses to test whether existing programs are more or less effective for older adolescents. In one of the most recent and comprehensive meta-analyses, a meta- regression produced a significant positive effect of grade level (Ttofi & Farrington, 2011), leading the authors to conclude that “programs should be targeted on children aged 11 years or older rather than on younger children” (p. 46). Thus, based on the published record, policymakers may have been justified in requiring older adolescents to receive anti-bullying programs.

However, evaluations of best-practices anti-bullying programs in- volving tens of thousands of adolescents sometimes show the opposite pattern: modest effects for younger children, and null effects for older adolescents (Kärnä, Voeten, Little, Alanen, et al., 2011). Indeed, theory and data in developmental psychology might lead one to predict this

(A) Within-study moderation

0

0.1

0.2

0.3

0.4

0.5

2nd 3rd 4th 5th 6th 7th 8th 9th 10th

C oh

en 's

d

Grade Level

Ertesvag & Vaaland (2007)

0

0.1

0.2

0.3

0.4

6 8

C oh

en 's

d

Average Grade Level

Ertesvag & Vaaland (2007)

(B) Between-study moderation

Fig. 1. Very simplified illustration that (A) within-study and (B) between-study moderation tests in meta-analyses can produce age-related trends in the opposite direction. Panel (A) clearly shows a decline to zero in efficacy among high school students, whereas Panel (B) shows an increase with age. Studies were cherry-picked to more clearly illustrate the potential for the two techniques to produce opposite developmental trends; see Figs. 4 and 5 for all effects from all studies. Grade levels were converted to United States grade levels for comparability.

48

37D.S. Yeager et al. / Journal of Applied Developmental Psychology 37 (2015) 36–51

ID Title Pages

359659 Declines in efficacy of anti-bullying programs among older adolescents: Theory and a three-level meta-

analysis ☆ 16

http://fulltext.study/article/359659

http://FullText.Study