Evaluation of Social Norms Marketing Campaigns: Five Sites across 8 Years Adrienne Keller, Ph.D. National Social Norms Institute University of Virginia
Social Norms Interventions What they do NOT do What they DO Why they work Characteristics Evidence of effectiveness Limitations Conclusions
Social Norms Interventions What They Do NOT Do
Social Norms Interventions What They Do
Most (83%) Adults in Illinois Don’t Smoke* *Source: Centers for Disease Control, 2008 Tobacco Use in Illinois Adults
Teach
Promote Self-Efficacy
Why do social norms interventions change behavior? Think about “peer pressure” And how others’ choices influence our own…
BUT… Perceptions may not conform to reality
Source: NCHA, Spring 2008, n= 80,121 Q.: How many alcoholic drinks did you have the last time you “partied” or socialized? Q.: How many alcoholic drinks do you think the typical student at your school had the last time they “partied” or socialized?
AND… We don’t like contradictions.
a MINORITY of our youth Are addicted to tobacco Do use drugs Do have a serious problem with alcohol. MOST of our youth Don’t smoke cigarettes Don’t do drugs Don’t use alcohol or have an alcohol problem.
Harm Reduction: Decrease alcohol-related morbidity & mortality DESIRED IMPACT Decrease Alcohol Use ANTICIPATED OUTCOMES Misperceived norm Correct perception of norm INTERVENTION GOAL IDENTIFIED ISSUES Personal alcohol misuse Misperceived norm INTERVENTION STRATEGY Social norms marketing campaign
Characteristics of Social Norms Marketing
Positive Inclusive Empowering
Evidence of Effectiveness Good theory…great media… But does it work?
Participating Schools Five universities 4 public; 1 private Thru-out USA Between 14K to 46K students Funding: Anhauser-Busch Fndtn ACHA’s NCHA 2002 through 2009: some years
Social Norms Campaigns Each university developed its own campaign using the five step model
Five Step Model Choose the Audience Gather & Evaluate Data Develop & Test Message Implement Campaign Evaluate Effectiveness
Sample Characteristics Siten Mean Age % Female % White % Res Hall % Frat/ Soro Schl Schl Schl Schl Schl
But did the samples change over the years?
YES for 1 and 2 Schl 1: Race/Ethnicity only % White % Hispanic Schl 2: Everything! Age % Female % White % in Res Hall % Frat/Soro
YES for 3 and 4 Schl 3: Residence only % in Res Hall Schl 4: Sex, Race, Residence % Female % White % in Res Hall
and YES for 5 Schl 5: Everything! Age % Female % White % in Res Hall % Frat/Soro So, we controlled for everything in our analyses.
Linear Mixed Effects Models Did perceptions change across years? Perception: ACHA-NCHA question Controlled for demographics Within each school Did use change across years? Use: ACHA-NCHA question Controlled for demographics Within each school By perception
Results for Perception Changes Type III Tests of Fixed Effects Source Numerator df Denominator dfFSig. Year of survey school
Results for Actual Drinks Changes Type III Tests of Fixed Effects a Source Numerator df Denominator dfFSig. Percvd drinks Year of survey school
Duh…Beautiful Stats… But what do they mean?
Mean Perceived Drinks
Mean Actual Drinks
Mean Perceived-Actual Drinks 02-09
Limitations Internal Validity Exposure not included in analysis Other interventions not included Related events not included Moderate to low sample response rates External Validity Population characteristics Environmental characteristics Implementation fidelity
Conclusions Misperceptions in all 5 schools Real world settings Diverse schools Actual drinks decreased in 4 of 5 “Gap” decreased in 2 of 5 Importance of longitudinal data Single year results unclear Effects & relationship only become clear over the years
Thank You Questions, Comments ???