Drug education in victorian schools (DEVS): the study protocol for a harm reduction focused school drug education trial
© Midford et al; licensee BioMed Central Ltd. 2012
Received: 18 January 2012
Accepted: 10 February 2012
Published: 10 February 2012
This study seeks to extend earlier Australian school drug education research by developing and measuring the effectiveness of a comprehensive, evidence-based, harm reduction focused school drug education program for junior secondary students aged 13 to 15 years. The intervention draws on the recent literature as to the common elements in effective school curriculum. It seeks to incorporate the social influence of parents through home activities. It also emphasises the use of appropriate pedagogy in the delivery of classroom lessons.
A cluster randomised school drug education trial will be conducted with 1746 junior high school students in 21 Victorian secondary schools over a period of three years. Both the schools and students have actively consented to participate in the study. The education program comprises ten lessons in year eight (13-14 year olds) and eight in year nine (14-15 year olds) that address issues around the use of alcohol, tobacco, cannabis and other illicit drugs. Control students will receive the drug education normally provided in their schools. Students will be tested at baseline, at the end of each intervention year and also at the end of year ten. A self completion questionnaire will be used to collect information on knowledge, patterns and context of use, attitudes and harms experienced in relation to alcohol, tobacco, cannabis and other illicit drug use. Multi-level modelling will be the method of analysis because it can best accommodate hierarchically structured data. All analyses will be conducted on an Intent-to-Treat basis. In addition, focus groups will be conducted with teachers and students in five of the 14 intervention schools, subsequent to delivery of the year eight and nine programs. This will provide qualitative data about the effectiveness of the lessons and the relevance of the materials.
The benefits of this drug education study derive both from the knowledge gained by trialling an optimum combination of innovative, harm reduction approaches with a large, student sample, and the resultant product. The research will provide better understanding of what benefits can be achieved by harm reduction education. It will also produce an intervention, dealing with both licit and illicit drug use that has been thoroughly evaluated in terms of its efficacy, and informed by teacher and student feedback. This makes available to schools a comprehensive drug education package with prevention characteristics and useability that are well understood.
Australia and New Zealand Clinical Trials Register (ANZCTR): ACTRN12612000079842
KeywordsPrevention Drug education Harm reduction Schools Students
The health and social costs associated with alcohol and other drug (AOD) use are considerable and fall most heavily on young people [1–3]. This in itself is an argument for early AOD prevention programs. School drug education offers the potential to prevent problems by equipping young people with the knowledge and skills to make responsible decisions about AOD use, and it has near universal reach in developed countries where the great majority of young people attend secondary school . However, historical approaches to school drug education have not been particularly successful at reducing AOD use [5–7]. This then poses the question as to whether effectiveness should be measured by abstinence or reduced use, or whether harm reduction is a more realistic and useful measure. Harm reduction programs offer greater promise of achieving worthwhile benefit because they have the flexibility to select strategies on the basis of evidence of effect. Within this model abstinence or reduced use strategies may be chosen if there is evidence that they reduce harm, but they are not goals in their own right .
The School Health and Alcohol Harm Reduction Project (SHAHRP) demonstrated the effectiveness of a skills based, harm reduction intervention. Students who received the SHAHRP program experienced 22.9% less alcohol-related harm than their control peers . A more recent study of computerised harm reduction prevention similarly reported that alcohol consumption, risky drinking and alcohol-related harms increased to a lesser extent among female intervention students, although there was no program effect among male students . A related study that targeted cannabis, as well as alcohol, found that weekly alcohol consumption of intervention students decreased, while that of their control peers increased. The frequency of drinking to excess also increased to a lesser extent among the intervention students .
Drug education has developed considerably in Australia over the past decade. However, evidence-based practice tends to be best reflected in demonstration programs that focus on a single drug type, such as alcohol [9, 12]. The most recent findings have not been brought together in a mass drug education program that targets all forms of drug use and can be readily accommodated within a secondary school health curriculum. Past classroom-based programs have not well exploited the important influence that parents have on their children in terms of AOD choices and the likely benefits of increased communication on this issue . Finally, while most Australian school drug education programs are grounded in the research literature and principles of effective practice there has been insufficient attention to the pedagogy required to deliver effective classroom lessons . Drug education is most effective when inclusive, interactive teaching strategies actively engage students in the learning process [14, 15]. However, reviews of drug education program implementation consistently identify a breakdown in fidelity when learning tasks are interactive [16–18].
This research study seeks to extend earlier Australian harm reduction education research by investigating the prevention effects of a single program for both licit and illicit drugs. The size and nature of the intervention have been tailored to facilitate incorporation within an already crowded school curriculum. It draws on the recent literature as to the common elements in effective school drug education programs [19–22]. It seeks to incorporate the social influence of parents through structured home activities that complement the classroom lessons. It emphasises the use of appropriate pedagogy through two days of training, where participatory delivery of each lesson is modelled to ensure teachers are equipped to teach the program as intended. All these elements have been trialled and refined during the course of a two year pilot program that was undertaken in four Victorian government secondary schools (three intervention and one control) .
The intervention at the heart of this study is grounded in social learning theory, which posits that human learning occurs in a social context . In this model, drug use is socially learned through modelling, imitation and reinforcement, and influenced by an individual's cognitions, attitudes, and beliefs. The corollary is that drug education can use the same learning processes to equip students with the skills to recognise these influences and develop a repertoire of counter behaviours. Social learning theory underpins all of the most effective school drug education programs, and a number of researchers have indicated that it should be the model of choice in any prevention program [7, 19, 25].
The intervention also draws on two other theoretical models, poststructuralist subjectivity and cognitive dissonance. Poststructuralist subjectivity theory provides a way of understanding how students' self concepts, and hence their approach to drug use, can change. Cognitive dissonance theory provides a way of understanding how students resolve conflicting ideas on drug use.
Poststructuralist subjectivity theory refers to the ways in which the social world is involved in shaping an individuals' sense of self [26, 27]. It posits that cultural norms and expectations shape the individual's sense of who they can or should be, and identifies the importance of a 'sense of belonging' to identity formation . In this process of subjectification, identity is continuously under construction, maintenance and re-construction, and hence open to change, particularly change that is social and collective in nature. Thus participatory and collective learning methods become the vehicle for the creation of new norms and possibilities in the peer group [14, 15, 29].
Festinger's theory of cognitive dissonance posits that contradictory cognitions, or cognitive dissonance, cause mental discomfort . This serves as a driving force that compels a person to adopt new thoughts or beliefs, or to modify existing beliefs, so as to reduce the amount of dissonance between cognitions. The education intervention used in this study does not tell students what they should think or how they should behave in terms of drug use. Rather, it provides a social learning process whereby the students reach their own conclusions through exploration of drug use issues. By guiding students through an interactive discovery process that involves articulation of responsible drug use behaviour their ownership of that behaviour is reinforced and they are less likely to behave in a contrary manner.
Aims and hypothesis
The aim of this study is to develop and assess the effectiveness of a comprehensive, evidence-based, harm reduction focused school drug education program for year eight and nine students by undertaking a cluster randomised trial with 21 secondary schools in the Victorian state of Australia. The specific hypothesis is that while the intervention students may not be any less likely to take up use of alcohol, tobacco and illicit drugs they will be more likely to consume in a less risky manner and experience less harms associated with use.
Twenty one Victorian government high schools have been recruited to the study. The schools were then allocated to four strata: six schools to metro location, high SES; three schools to rural location, high SES; six schools to metro location, low SES; six schools to rural location, low SES. This approximated the proportion of Victorian secondary schools in each category. Within each strata the schools were then randomly allocated to intervention or control conditions on a two to one proportion to allow more precise statements about the effects of the intervention . Random allocation and allocation concealment were achieved by one member of the research team writing the names of each school on identical paper chits which were then folded to conceal the name. The same researcher also created seven chits with the word control and 14 with the word intervention. These were similarly folded to conceal the words. The chits with the school name and those designating allocation were placed in separate containers. A different member of the team then simultaneously drew out a chit with the name of a school and a chit designating either control or intervention allocation.
Sample size calculation
The ability to detect small changes in substance use, knowledge, attitudes and harms experienced is central to this study. The primary outcome is a change in alcohol consumption patterns and associated harm, and accordingly sample size estimations are based on detecting a small effect size of .15 in relation to these measures. This effect size was chosen on the basis of previous school drug education studies [9, 32].
The target sample size has been estimated using G*Power v.3.1.3 software where α = 0.05 and 1-β error probability = 0.95 . Assuming simple random sampling, a total sample size of 364 is required at the end of the study. However, there is a design effect due to the loss of effectiveness created by cluster sampling. The design effect for the SHAHRP study, which took into consideration the effect of clustering by school and a 15% annual rate of student attrition, was calculated to be 1.48. Using this correction, a total sample size of at least 539 is required to test the effectiveness of the intervention.
The participating students attended the 21 Victorian government secondary schools that had agreed to participate in this project. Four schools are located in metropolitan Melbourne; ten are located in metropolitan fringe and major regional areas; and seven are in regional and rural areas. The student populations of six schools are in the medium high/high range in term of socioeconomic status, as measured by the Department of Education and Early Childhood Development's (DEECD) Student Family Occupation (SFO) index. The student populations of ten schools are in the medium range. The student populations of five schools are in the low range. Written active consent was sought from the 2700 year eight students in the 21 participating schools and their parents. Of this total population 1746 or 64.7% agreed to participate in the research. Intervention students numbered 1230. Control students numbered 516.
The drug education intervention
Year eight and year nine lesson plans
WHAT IS A DRUG? - Introduction, agreements, definitions and drug categories
PRIORITIES AND CONCERNS - Identifying what young people value and worry about and what worries they have around drugs
ALCOHOL AND EFFECTS AND STANDARD DRINKS - How alcohol effects the body, assessing harms associated with use, pouring standard drinks, understanding blood alcohol content and safer levels of use
FACING FACTS AND FINDING SOLUTIONS - Alcohol and Cannabis- guidelines on use and the research that informs them
PARTY BEHAVIOURS AND ALCOHOL - The relationship between levels of alcohol use and the risk of harm to self and others
USING YOUR RESOURCES - Pouring standard drinks, matching harms to levels of alcohol use, identifying strategies to reduce harm
PREVALENCE AND NORMS - Dispelling myths about levels of drug use amongst young people, identifying reasons for use/non -use
WINDING UP, WINDING DOWN - Learning about the effects/risks of Amphetamine type stimulants, identifying drug-free ways of achieving 'high' and 'serene' states of mind
TOBACCO - Considering gender differences in relation to smoking; the impact of media messages
DRUGS, DISINHIBITION, SEXUAL VULNERABILITY AND VIOLENCE - Discussing sexual vulnerability in relation to drug use, identifying strategies for avoiding or reducing harm
CANNABIS - Information about cannabis and its effects, identifying risks associated with Cannabis use
INVISIBLE RISKS - Information about injecting drug use, blood-borne viruses and methods of protection
RISK REDUCTION - Assessing risk and developing strategies to avoid or minimise harm
PERSONAL CONFIDENCE - AND DRUG USE - Developing and rehearsing positive self talk, refusal skills and tactics for peer negotiation
INFLUENCES - Identifying social and media influences to use alcohol
GETTING HELP AND TALKING WITH ADULTS - Information about heroin, rehearsing steps for practical first aid in situations involving overdose, rehearsing help seeking with adults
OPTIONS AND DECISIONS - Generating and rehearsing strategies to reduce harms associated with drug use
STANDING UP FOR YOURSELF - Providing peer support, using assertion skills in situations involving alcohol
The education program incorporates the elements of effective practice identified by reviews of school drug education [19–22, 25, 36]. This literature identifies the importance of basing programs on evidence of proven effect; the needs of students; provision of essential knowledge; adequate coverage of salient issues; and use of interactive learning strategies that enhance negotiation skills, involve participants in problem-solving and engage them in deconstructing the social pressures and perceived norms around drug use. The curriculum is also informed by research in the field of resilience education that identifies social competence, problem-solving, autonomy and a sense of purpose as key attributes of resilient young people , and highlights the importance of participatory and developmentally appropriate learning strategies in enhancing social and emotional learning .
Each year all teachers delivering the classroom program will participate in an intensive two-day professional learning program that provides a grounding in the evidence-base informing the study and active sampling of each of the lesson activities they will be teaching their students. Emphasis will be given to modelling and explicit leadership coaching in use of the participatory methods.
Student survey instrument and measurement of change
The survey instrument to measure change is a development of the self-completion questionnaire used in SHAHRP and was trialled in the pilot research that preceded this study [9, 23]. Self-report is well accepted practice in studies of this type and research indicates little inconsistency between self report and other measures of drug use [39, 40] The instrument will collect information on knowledge, patterns and context of use, attitudes and harms experienced in relation to alcohol, tobacco, cannabis and other illicit drug use. As was done in the pilot study, scales will be constructed to measure overall change in AOD knowledge (38 items), attitudes (4/5 items per drug type), and harms (5/10 items per drug type)
Internal consistency of knowledge, attitude and harm scales
% of variance
Attitudes towards alcohol
Knowledge and communication
Alcohol harm scale
Attitudes towards tobacco
Knowledge and communication
Tobacco harm scale
Attitudes towards cannabis
Cannabis harm scale
Attitudes towards other drugs
Knowledge and communication
Other drugs harm scale
Participants cannot be blinded to intervention in this sort of psycho-social educational intervention, nor can program deliverers or outcome assessors.
The study was approved by Edith Cowan University's and the University of Melbourne's human research ethics committees. It was also approved by the Research Branch, Education Policy and Research Division of the Victorian Department of Education and Early Childhood Development.
Multi-level modelling can best accommodate hierarchically structured data and will be used in this analysis, adjusting for any baseline differences between the intervention and control groups [42, 43]. The hierarchy in this longitudinal multi-level dataset comprises level 1 units (occasions of repeated measurement), nested within the level 2 units (the individual student), nested within the level 3 units (the school). Stata version 10.0 will be used for model fitting, with a three-level mixed regression model fitted to the data to account for the repeated observations . If necessary, the knowledge, consumption and harm scales will be log-transformed to satisfy the assumption of normality. If normality cannot be achieved by log-transformation, non-parametric procedures will be used for analysis. All analyses will be conducted on an Intent-to-Treat basis. Complete-case analysis (CCA) will be complemented with multiple imputation analysis (MIA) to account for missing data.
The benefits of undertaking this harm reduction focused drug education demonstration program derive both from the knowledge gained by trialling an optimum combination of evidence derived approaches with a large, representative student sample, and the resultant product. The program targets alcohol, tobacco and illicit drug use; draws on research that consistently identifies a core set of effective practice elements and is rigorously evaluated over an extended period. The program also does not simply rely on the traditional measures of effectiveness, namely abstinence or reduced use. Harms associated with use are also measured as valid indicators of program effect. This will provide a broader understanding of what benefits can be achieved by school drug education. By the end of the study a well documented, comprehensive drug education package will have been produced. It will be of a size able to be accommodated within a secondary school's health curriculum and the evaluation findings will provide evidence as to its potency. This makes available to schools a readily useable drug education program with well understood prevention characteristics.
This study is jointly funded by the Australian Research Council and the Victorian Department of Education and Early Childhood Development through Australian Research Council linkage grant LP100100798
- Anderson P, Baumberg B: Alcohol in Europe: a public health perspective. 2006, London: Institute of Alcohol StudiesGoogle Scholar
- Chikritzhs T, Pascal R: Trends in youth alcohol consumption and related harms in Australian jurisdictions, 1990-2002. National Alcohol Indicators, Bulletin No.6. 2004, Perth: National Drug Research Institute, Curtin University of TechnologyGoogle Scholar
- Toumbourou JW, Stockwell T, Neighbors C, Marlatt GA, Sturge J, Rehm J: Interventions to reduce harm associated with adolescent substance use. The Lancet. 2007, 369 (9570): 1391-1401. 10.1016/S0140-6736(07)60369-9.View ArticleGoogle Scholar
- Midford R: Is Australia 'fair dinkum' about drug education in schools?☆. Drug and Alcohol Review. 2007, 26 (4): 421-427. 10.1080/09595230701373842.View ArticlePubMedGoogle Scholar
- Foxcroft DR, Ireland D, Lister-Sharp DJ, Lowe G, Breen R: Longer-term primary prevention for alcohol misuse in young people: a systematic review. Addiction. 2003, 98 (4): 397-411. 10.1046/j.1360-0443.2003.00355.x.View ArticlePubMedGoogle Scholar
- Giesbrecht N: Reducing alcohol-related damage in populations: rethinking the roles of education and persuasion interventions. Addiction. 2007, 102 (9): 1345-1349. 10.1111/j.1360-0443.2007.01903.x.View ArticlePubMedGoogle Scholar
- Midford R: Drug prevention programmes for young people: where have we been and where should we be going?. Addiction. 2010, 105 (10): 1688-1695. 10.1111/j.1360-0443.2009.02790.x.View ArticlePubMedGoogle Scholar
- Lenton S, Midford R: Clarifying 'harm reduction'?. Drug and Alcohol Review. 1996, 15 (4): 411-413. 10.1080/09595239600186181.View ArticlePubMedGoogle Scholar
- McBride N, Farringdon F, Midford R, Meuleners L, Phillips M: Harm minimization in school drug education: final results of the School Health and Alcohol Harm Reduction Project (SHAHRP). Addiction. 2004, 99 (3): 278-291. 10.1111/j.1360-0443.2003.00620.x.View ArticlePubMedGoogle Scholar
- Vogl L, Teesson M, Andrews G, Bird K, Steadman B, Dillon P: A computerized harm minimization prevention program for alcohol misuse and related harms: randomized controlled trial. Addiction. 2009, 104 (4): 564-575. 10.1111/j.1360-0443.2009.02510.x.View ArticlePubMedGoogle Scholar
- Newton NC, Teesson M, Vogl LE, Andrews G: Internet-based prevention for alcohol and cannabis use: final results of the Climate Schools course. Addiction. 2010, 105 (4): 749-759. 10.1111/j.1360-0443.2009.02853.x.View ArticlePubMedGoogle Scholar
- McLeod J: Rethinking drinking: you're in control. an alcohol education program for secondary students. Reflections on the development, implementation and strengths of the program. 1997, Melbourne: Youth Research Centre, Faculty of Education, University of MelbourneGoogle Scholar
- Ward B, Snow P: Supporting parents to reduce the misuse of alcohol by young people. Drugs Education, Prevention and Policy. 2010, 17: 718-731. 10.3109/09687630902806723.View ArticleGoogle Scholar
- Cahill HW: Challenges in adopting evidence-based school drug education programmes. Drug and Alcohol Review. 2007, 26 (6): 673-679. 10.1080/09595230701613593.View ArticlePubMedGoogle Scholar
- Cahill HW: Using role-play techniques to enhance engagement in the health class: issues and strategies. Health Education Australia Journal. 2003, 3 (2): 17-23.Google Scholar
- Dusenbury L, Brannigan R, Falco M, Hansen WB: A review of research on fidelity of implementation: implications for drug abuse prevention in school settings. Health Education Research. 2003, 18 (2): 237-256. 10.1093/her/18.2.237.View ArticlePubMedGoogle Scholar
- Stead M, Stradling R, Macneil M, Mackintosh AM, Minty S: Implementation evaluation of the blueprint multi-component drug prevention programme: fidelity of school component delivery. Drug & Alcohol Review. 2007, 26 (6): 653-664. 10.1080/09595230701613809.View ArticleGoogle Scholar
- Stead M, Stradling R, MacNeil M, MacKintosh AM, Minty S, McDermott L, Eadie D: Bridging the gap between evidence and practice: a multi-perspective examination of real-world drug education. Drugs Educ Prev Policy. 2010, 17 (1): 1-20. 10.3109/09687630802228341.View ArticleGoogle Scholar
- Cuijpers P: Effective ingredients of school-based drug prevention programs. a systematic review. Addictive Behaviors. 2002, 27: 1009-1023. 10.1016/S0306-4603(02)00295-2.View ArticlePubMedGoogle Scholar
- Midford R, Munro G, McBride N, Snow P, Ladzinski U: Principles that underpin effective school-based drug education. Journal of Drug Education. 2002, 32 (2): 363-386.View ArticlePubMedGoogle Scholar
- Midford R: Looking to the future: providing a basis for effective school drug education. Drug Education in Schools: Searching for the Silver Bullet. Edited by: Midford R, Munro G. 2006, Melbourne: IP Communications, 213-230.Google Scholar
- Soole DW, Mazerolle L, Rombouts S: School-based adolescent drug prevention programs: a review of what works. Aust New Zealand J Criminology. 2008, 41: 259-286. 10.1375/acri.41.2.259.View ArticleGoogle Scholar
- Midford R, Cahill H, Ramsden R, Davenport G, Venning L, Lester L, Murphy B, Pose M: Alcohol prevention: what can be expected of a harm reduction focused school drug education programme?. Drugs Education Prevention and Policy. 2011, Early Online: 1-9. http://informahealthcare.com/doi/full/10.3109/09687637.2011.639412 Google Scholar
- McAlister AL, Perry CL, Parcel GS: How individuals, environments, and health behaviors interact: social cognitive theory. Health Behavior and Health Education: Theory, Research, and Practice. Edited by: Glanz K, Rimer BK, Viswanath K. 2008, San Francisco: John Wiley & Sons Inc, 169-188. 4Google Scholar
- Tobler NS, Roona MR, Ochshorn P, Marshall DG, Streke AV, Stackpole KM: School-based adolescent drug prevention programs: 1998 meta-analysis. J Primary Prev. 2000, 20 (4): 275-10.1023/A:1021314704811.View ArticleGoogle Scholar
- Davies B: Post-structuralist Theory and Classroom Practice. 1994, Geelong: Deakin University PressGoogle Scholar
- Davies B: Subjectification: the relevance of Butler's analysis for education. British J Sociology Educ. 2006, 27 (4): 425-438. 10.1080/01425690600802907.View ArticleGoogle Scholar
- Grimson A: Culture and identity: two different notions. Social Identities: J Study of Race Nation and Culture. 2010, 16: 61-77.View ArticleGoogle Scholar
- Cahill H: Drama for Deconstruction. Youth Theatre J. 2011, 25 (1): 16-31. 10.1080/08929092.2011.569299.View ArticleGoogle Scholar
- Festinger L: A Theory of Cognitive Dissonance. 1957, Stanford: Stanford University PressGoogle Scholar
- Hendricks Brown C: Design principles and their application in prevention field trials. Drug Abuse Prevention; Theory, science, and Practice. Edited by: Sloboda Z, Bukoski WJ. 2006, New York: Springer Science+Business Media LLC, 523-540.View ArticleGoogle Scholar
- Malmberg M, Overbeek G, Kleinjan M, Vermulst A, Monshouwer K, Lammers J, Vollebergh W, Engels R: Effectiveness of the universal prevention program 'Healthy School and Drugs': study protocol of a randomized clustered trial. BMC Public Health. 2010, 10 (1): 541-10.1186/1471-2458-10-541.View ArticlePubMedPubMed CentralGoogle Scholar
- Faul F, Erdfelder E, Buchner A, Lang A-G: Statistical power analyses using G*Power 3.1: tests for correlation and regression analyses. Behavior Res Methods. 2009, 41 (4): 1149-1160. 10.3758/BRM.41.4.1149.View ArticleGoogle Scholar
- Cahill HW, Stafford K, Shaw G: Get WISE: working on illicits in school drug education. 2000, Melbourne: Department of Education Employment and TrainingGoogle Scholar
- Miller-Day M: Talking to youth about drugs: what do late adolescents say about parental strategies?. Family Relat. 2008, 57 (1): 1-12. 10.1111/j.1741-3729.2007.00478.x.View ArticleGoogle Scholar
- Wood E, Shakeshaft A, Gilmour S, Sanson-Fisher R: A systematic review of school-based studies involving alcohol and the community. Aust New Zealand J Public Health. 2006, 30 (2): 541-549.View ArticleGoogle Scholar
- Benard B: Resiliency: what have we learned. 2004, San Francisco: WestEdGoogle Scholar
- Payton J, Weissberg RP, Durlak JA, Dymnicki AB, Taylor RD, Schellinger KB, Pachan M: The positive impact of social and emotional learning for kindergarten to eighth-grade students: findings from three scientific reviews. 2008, Chicago: Collaborative for Academic, Social, and Emotional LearningGoogle Scholar
- Morgan M: EMCDDA Instrument Bank: core scales, sources and guidelines. 1997, Lisbon: EMCDDAGoogle Scholar
- Lintonen T, Ahlstrom S, Metso L: The reliability of self-reported drinking in adolescence. Alcohol and Alcoholism. 2004, 39 (4): 362-368. 10.1093/alcalc/agh071.View ArticlePubMedGoogle Scholar
- Bland JM, Altman DG: Statistics notes: Cronbach's alpha. British Medical J. 1997, 314 (7080): 572-10.1136/bmj.314.7080.572.View ArticleGoogle Scholar
- Greenland S: Principles of multilevel modelling. International J of Epidemiology. 2000, 29 (1): 158-167. 10.1093/ije/29.1.158.View ArticleGoogle Scholar
- Heck R, Thomas S: An Introduction to Multi-level Modeling Techniques. 2000, New Jersey: Lawrence Erlbaum Associate IncGoogle Scholar
- Goldstein H: Multilevel Statistical Models. 1995, London: Edward ArnoldGoogle Scholar
- The pre-publication history for this paper can be accessed here:http://www.biomedcentral.com/1471-2458/12/112/prepub