The challenges of applied social psychology: Action hypotheses, Meta-Analysis and Intervention Studies.
An important but neglected topic in social psychology has been the process of formulating and testing policy recommendations on an empirical basis to solve practical problems like the increase of the innovation rate of individuals and organizations. The necessary preconditions and prerequisites are often not at all or only partially known and taught.
All social psychological theories(e.g. reasoned action theory, social identity theory or theory of values) cannot be transformed directly into policy recommendations. We have to distinguish between the theory itself as a set of postulates(structural model), the measurement theory representing the operationalization and its underlying assumptions(measurement model) and the action hypotheses representing the relationships between actions(policies) and latent constructs of the theory used(MIMIC-Model). Only the combination of all three components allows scientifically based policy interventions(Bamberg/Schmidt 2001). For testing all three components in one model one can use latent variable models which also allow to test for mediation, moderation and total and indirect effects of policy measures, which is not possible with traditional multivariate analyses.
In addition one needs meta-analyses instead of narrative reviews, which contain not only the structural and the measurement model, but also the action hypotheses to have a solid empirical basis for policy recommendations. However there are till now very few such metaanalyses dealing also with the action hypotheses( e.g. Albarazzin et al. 2005 and Lemmer/Wagner in print) which take into account relevant action hypotheses. Therefore it is often necessary to perform experiments and intervention studies with no prior explicit knowledge of empirically tested action hypotheses( see Shadish/ Cook/ Campbell 2002 and Aussems/Boomsma/ Snijders 2011.). Furthermore the use of experiments and quasiexperiments in field settings allows external validity and a more severe testing of causal effects(Rosenbaum/Rubin 1983).
To use the full advantage of different designs one needs a sequence of labaratory experiments, field studies and longitudinal intervention studies either experimental or quasiexperimental.
To demonstrate the procedure just discussed I firstly describe a large theory driven quasiexperiment to change the travel-mode choice of students in a region of Germany and in addition in Germany generally. It is shown how structural equation models can used to test the direct and indirect effects over time. In a second application I discuss, which possible experiments would be useful as a preparation for an intervention study to increase innovations in organizations.
Key words: Applied social Psychology, Action Hypothesis, Structural Model, measurement Model, Meta-Analysis, Latent-Variable Models, Multiple Indicator Multiple causes Model(MIMIC), Experiment, Quasi-Experiment,
Total effects, Mediation and Moderation, Values and innovation,