Can we make environmental citizens? A randomised control trial of the effects of a school-based intervention on the attitudes and knowledge of young people

Abstract

The aim of this study was to determine whether school-based intervention can help in making environmental citizens. A randomised control trial of 448 primary school students and their families in 27 primary schools located in Vale Royal, North West England was carried out between January and July 2008. The interventions were two types of class-based instruction on environmental issues, one long and the other short, which were designed to increase environmental awareness. Environmental attitudes and behaviours were measured by surveys completed by the students in their classes and in their homes before and after the interventions. The analysis reports school averages of the questionnaire responses, followed by regression analysis using robust clustered standard errors. The results show no statistically significant differences between schools in the intervention groups compared to the control group schools. The rising environmental awareness of the control group during the intervention may partly explain the positive results of existing non-experimental studies.

Keywords:

• environmental
• attitudes
• behaviour
• schools
• randomised controlled trial

Notes

 1. Evans et al. (2007) rely on a sample of children who were primarily from well-educated and affluent families residing in small towns and rural areas. Parents self-selected into the research project, resulting in a likely selection bias toward more pro-environmental families.

 2. For an alternative view see Reid and Sa'di (1997, p. 478).

 3. One example is Bradley et al. (1999), who evaluated a 10-week environmental science course on 475 high school students across 18 schools. Though findings suggested that participation exerted positive effects on environmental knowledge and attitudes the research was not based on comparison with a control group.

 4. Bogner's (1999) study of the effects of a conservation education programme in secondary schools in Switzerland similarly found positive effects.

 5. In terms of design, and as the authors note, the study focused on low and middle SES schools and the control group came from one of the two treatment schools and not both.

 6. Leeming et al. (1997) evaluate a programme designed to support teachers to undertake a series of in-class activities during a school year, such as recycling items and planting trees. Participating classes engaged in a minimum of eight environmentally relevant activities. Improved environmental attitudes were not matched in respect to environmental knowledge which did not significantly alter. The parents of the participants also displayed significantly more pro-environmental behaviors after their children's activities than did parents of the control group.

 7. One possible reason suggested by Evans et al. (2007) concerns the young ages of the sample (6–8 years). As children mature it might be that environmental attitudes and behaviours begin to converge.

 8. The effectiveness of school-based interventions has also been demonstrated in other areas such as the reduction of prejudice (e.g. Aboud and Fenwick 1999). On the utility of randomised clinical trial methodology see T.D. Cook and D.T. Campbell (1979) Quasi-Experimentation: Design and Analysis for Field Settings (Boston: Houghton Mifflin Company); D.J. Torgerson and C.J. Torgerson (2008) Designing Randomised Trials in Health, Education and the Social Sciences (Basingstoke: Palgrave Macmillan).

 9. This data has been collected from the Department of Health (2008) Health Profile for Vale Royal 2006. Department of Health. Available online: http://www.apho.org.uk/resource/item.aspx?RID=52243 (Accessed 20 September 2008).

10. In May 2006 Vale Royal Borough Council launched the ‘SMYLE’ campaign (‘Supporting My Local Environment’) which aimed to improve residents' environmental awareness. As part of the SMYLE campaign, in December 2006 (prior to our intervention) all local residents were sent a SMYLE newsletter which introduced the campaign and encouraged residents to make five environmental pledges: (i) I will reduce the impact my actions have on the environment by ensuring I do not commit environmental crimes, (ii) I will reduce the impact my waste has on the environment by reducing, reusing and recycling as much as possible, (iii) I will help improve the quality of my local environment by supporting community action and reporting environmental crime, (iv) I will reduce the amount of energy I use in the home, (v) I will make better travel choices to reduce my impact on the environment. Through a second ‘Recycle for Cheshire’ campaign (also introduced in 2006) the council focused on improving recycling participation rates amongst local residents and delivered surveys to all households. The survey results were provided to the research team by Vale Royal Borough Council. The first survey (May 2005) had a response rate of 3 per cent whilst the second survey (June 2006) had a response rate of 4 per cent. In addition to those statistics cited in-text (and though not systematic) it is worth noting in addition: 97 per cent of respondents considered having a ‘clean, safe and green local environment’ to be important; 74 per cent of respondents said they were satisfied with local environmental quality; 33 per cent said they had seen a reduction in litter over the past 6 months and 28 per cent had seen a reduction in graffiti since the commencement of the campaign.

11. Details of the randomisation procedure are provided in a later section.

12. Further information about the intervention, including a description by the trained officer, is available in a policy report submitted to the North West Improvement Network (NWIN) and which is available from the lead author.

13. Alternative methods were investigated including multi-level modelling (Green and Vavreck, 2008). We also investigated different ways of specifying the dependent variable for example, by dichotomising and using a logistic regression, using an ordered probit, and an interval regression to account for potential censoring. The general null findings were reproduced by different methods. However, there is little evidence of a consistent pattern of association and we believe this finding is adequately captured by the results reported here.