[What, why, and which experiments?
],
[Moving beyond simple experiments], [Making it happen],
[Zero variable cost data and MusicLab], [3 Rs]
Matthew J. Salganik
Department of Sociology
Princeton University
1) Introduction
2) Observing behavior
3) Asking questions
4) Running experiments
5) Mass collaboration
6) Ethics
7) The future
Optimization experiments vs Understanding experiments
Optimization experiments + Understanding experiments
PS YC HOLOGICA L SC IENCE
Research Article
The Constructive, Destructive,
and Reconstructive Power of
Social Norms
P. Wesley Schultz,1 Jessica M. Nolan,2 Robert B. Cialdini,3 Noah J. Goldstein,3 and Vladas Griskevicius3
1
California State University, San Marcos; 2University of Arkansas; and 3Arizona State University
ABSTRACT—Despite a long tradition of effectiveness in Tabanico, & Rendón, in press). Such social-norms marketing
laboratory tests, normative messages have had mixed campaigns have emerged as an alternative to more traditional
success in changing behavior in field contexts, with some
[Link] approaches (e.g., information campaigns, moral exhortation, fear-
studies showing boomerang effects. To test a theoretical inducing messages) designed to reduce undesirable conduct
account of this inconsistency, we conducted a field exper- (Donaldson, Graham, Piccinin, & Hansen, 1995).
iment in which normative messages were used to promote The rationale for the social-norms marketing approach is
household energy conservation. As predicted, a descriptive based on two consistent findings: (a) The majority of individuals
normative message detailing average neighborhood usage overestimate the prevalence of many undesirable behaviors,
produced either desirable energy savings or the undesir- such as alcohol use among peers (e.g., Borsari & Carey, 2003;
Figures from Allcott (2011)
Figures from Allcott (2011)
Figures from Allcott (2011)
If you want to move beyond simple experiments:
I Validity
I Heterogeneity of treatment effects
I Mechanisms
Validity
Validity
I statistical conclusion validity
I internal validity
I construct validity
I external validity
Heterogeneous treatment effects
Heterogeneous treatment effects
I lots of people, lots of pre-treatment information: stop treating people like widgets
Heterogeneous treatment effects
I lots of people, lots of pre-treatment information: stop treating people like widgets
I beware of fishing (preregistration, data splits)
Heterogeneous treatment effects
I lots of people, lots of pre-treatment information: stop treating people like widgets
I beware of fishing (preregistration, data splits)
I Lots of new methods in this area
Mechanisms
Mechanisms
I Eating citrus prevents scurvy
Mechanisms
I Eating citrus prevents scurvy
I The mechanism through which citrus prevents scurvy vitamin C
I Not as easy as scurry and limes, “Enough already about black box experiments”
Green et al (2010)
Mechanisms
I Eating citrus prevents scurvy
I The mechanism through which citrus prevents scurvy vitamin C
I Not as easy as scurry and limes, “Enough already about black box experiments”
Green et al (2010)
I Experiments specifically designed to test mechanisms: Ludwig et al (2011), Imai
et al (2012), Pirlott and MacKinnon (2016)
Optimization experiments + Understanding experiments
[What, why, and which experiments?],
[Moving beyond simple experiments], [Making it happen],
[Zero variable cost data and MusicLab], [3 Rs]
Matthew J. Salganik
Department of Sociology
Princeton University
