Prosjekt: Social experiments in three sites

Finansiering: Norges forskningsråd

Prosjektansvarlig fra 2006: Ira Malmberg-Heimonen

Project Description

The social experiment is a multi-site study of the effects of activation programmes basedon randomized controlled trials (RCT) in three municipalities concurrently. We will focus on programmes based on different strategies i.e.Social Capital, Work First (i.e. moving people into ordinary work as quickly as possible), and Human Capital Development (i.e. providing some education and vocational training) (Hotz et al. 2000).

Research Questions

• Do these activation programmes work?
• What works for whom? Especially, do the effects of the different programmes vary according to social dimensions such as gender and ethnicity?
• Do social capital programmes work better than individually oriented activation programmes?
• How do participants experience taking part in these programmes and how do their children fare? Do the participants feel that their participation is useful, that they are moving closer to the labour market, and that they are expanding their social relationships, increasing social participation and improving their well being?
• Do the benefits from these activation programmes exceed the costs associated with running them?

Data and methods

Question 1 and 2 above, i.e. those related to differences within programmes, can be answered by analysing differences in outcomes between the programme and the control group. To answer question 3, i.e. focusing on differences between programmes, we will examine programme effects by means of multilevel, multiple linear and logistic regression analyses. This more complicated approach is necessitated by the need to control for possible differences between sites with respect to such factors as the composition of participants and con-textual circumstances such as labour market conditions (Hotz et al. 2000). Comprehensive individual and contextual information collected at each site will enable us to account for relevant factors. There are further pitfalls in the RCT design which must be dealt with when designing the experiment and planning the data collection. Potential bias includes randomization bias, replacement bias, substitution bias and attrition bias. Methods to deal with these potential sources of bias include linking annual official administrative data to control for changes in recruitment, participation in other programmes and for attrition bias. This will be supplemented by exploiting the local knowledge possessed by social workers about the community, and their clients and their whereabouts