Deliverable D 1.3 RESULTS OF META-ANALYSIS ... - cast-eu.org

More documents

Recommendations

Info

Campaigns and awareness raising strategies in traffic safety — Deliverable D-1.15. EFFECTS OF CAMPAIGNS: “STATE-OF-THE-ART” PER 2006The topic of estimating effects of campaigns has over the years beenaddressed in several literature reviews (Elliot, 1993; Delhomme et al, 1999;Elvik and Vaa, 2004 16 ; Vaa et al, 2004, Delaney et al, 2004). Thesecomprehensive studies differ in several ways. Of these five reviews all butDelaney et als’ study use meta-analysis as an instrument to assess anestimate of the effect of campaigns. Elliot’s study address awareness,knowledge, attitudes and behaviour, but not accidents. Elliot and Delaney etals’ studies are explicitly confined to mass media studies only. Finally, thethree studies specifically addressing effects on accidents are Delhomme et al,Elvik and Vaa (2004) and Vaa et al (2004). That strategy pursued here is theone of applying meta-analysis to assess a concrete estimate of the effect ofcampaigns on the number of accidents. While Delhomme et al, and Elvik andVaa, only use meta-analysis to assess a rather crude and un-sophisticatedestimate, the Vaa et al study (2004), has incorporated more sophistication ofmeta-analytic techniques in terms of correcting for publication bias, testingheterogeneity and applying meta-regression to assess partial effects of givenpredictors.It is in this sense, that the present description of effects is labelled state-ofthe-artregarding estimation of the effect of campaigns, ie. what we knew ofthe effects of campaigns when the CAST-project was kicked-off in February2006. The results presented here are from the INFOEFFEKT project whichTØI completed in 2004 on commission from the Swedish Public RoadsAdministration (“Vägverket”) (Vaa et al, 2004). It builds on the GADGETproject which was completed in 1999 (Delhomme et al, 1999). We alsoinclude a cost-benefit analysis and appraisals of cost-effectivenes of selectedSwedish campaigns as these topics are also addressed in the CAST project.The Swedish examples may, hence, serve as “good examples” on how toperform cost-benefit analysis of effects of campaigns on accidents.5.1 INFOEFFEKT 2004: RESULTS FROM META-ANALYSISThe report from the INFOEFFEKT project consists of three studies ofrelevance for the CAST project (Vaa et al, 2004):1) Effects of campaigns on road traffic accidents,2) Effects of campaigns on other areas of behaviour than road traffic,3) Swedish campaigns: Appraisal of cost effectiveness and benefit-costaspects16 Chapter 7.3: Road user information and campaigns, pp 949-954.40
Campaigns and awareness raising strategies in traffic safety — Deliverable D-1.15.1.1 Effects of campaigns on road traffic accidents,Study 1 in INFOEFFEKT considered estimation of effects of campaigns onroad traffic accidents. The basis for estimating the effects was theconstruction of a database, which comprised 86 results from a total of 30evaluation studies. Of these, 72 results were used for estimating the effects ofcampaigns during the campaign period and 14 results for estimating the effectin the after-period. Two types of analysis were presented, the first from metaanalysis,concerning the general, overall effect of campaigns, and the secondcomprising results from meta-regression, concerning factors – partial effects –helping to explain the effects of campaigns.The overall, general question, addressed, was whether campaigns can besaid to have effects on road traffic accidents. The premises for answering thisquestion were:1. Only evaluation designs using a control group or comparison groupwere included in the database2. The data were put through meta-analysis in order to calculate a bestestimate of a potential effect3. The data were tested for homogeneity and corrected for publicationbiasConsidering the studies that evaluated effects in the campaign period thenumber of campaigns against drink driving was 33. Statistical tests indicatedpublication bias (the tendency not to publish results from campaigns wherethe effect seems to have been zero or gone in the opposite direction of whatwould have been expected; see Chapter 6 or section 7.3 for furtherexplanation). A number of “campaigns” was therefore added tocounterbalance the effects of these biases. There was no publication biasamong the studies used to estimate effects in the after-periods.Under these premises the best estimates were calculated. (The results aresummarised here but results detailed in Appendix D of this report.) Theweighted average of the effect of road safety campaigns in the campaignperiod was found to be an 8.9 % reduction on the number of road accidents.In the after-period the effect of campaigns was estimated to a reduction of14.8%. Both estimates were statistically significant.Most campaigns were directed towards one single theme. The two largestsubgroups in the database were campaigns against drink driving and againstspeeding. Single-theme campaigns appeared to do better than multi-themecampaigns. Campaigns against drink driving and other single-themecampaigns showed statistically significant reductions on the number ofaccidents of 14.,4 and 10.4 % respectively. Multi-theme campaigns did nothave any effect at all, while campaigns against speeding showed a tendencyof reducing the number of accidents (by 8.5 %), but the effect was notstatistically significant.41
Page 1 and 2: Campaigns and awareness raising str
Page 39: Campaigns and awareness raising str
Page 91 and 92:
Campaigns and awareness raising str
Page 93 and 94:
Page 95 and 96:
Page 97 and 98:
Page 99 and 100:
Page 101 and 102:
Page 103 and 104:
Page 105 and 106:
Page 107 and 108:
Page 109 and 110:
Page 111 and 112:
Page 113 and 114:
Page 115 and 116:
Page 117 and 118:
Page 119 and 120:
Page 121 and 122:
Page 123 and 124:
Page 125 and 126:
Page 127 and 128:
Page 129 and 130:
Page 131 and 132:
Page 133 and 134:
Page 135 and 136:
Page 137 and 138:
Page 139 and 140:
Page 141 and 142:
Page 143 and 144:
Page 145 and 146:
Page 147 and 148:
Page 149 and 150:
Page 151:
show all

Deliverable D 1.3 RESULTS OF META-ANALYSIS ... - cast-eu.org

Create successful ePaper yourself

Delete template?

Save as template?