[All learning outcomes assessed:

A: Summarize the characteristics of longitudinal data.

B: Apply linear models to analyze longitudinal data and clear the model assumptions. C: Apply generalized linear models for the analysis of longitudinal data.

D: Be proficient in SAS, R or other software to perform appropriate longitudinal analyses of data and able to interpret and clear the clinical/scientific meaning of the results.]

Background

Multilevel data arise when there is a hierarchical or clustered structure to the data. Data of this kindfrequently arise in the health sciences, since individuals can be grouped in so many diferent ways.

The Television, School and Family Smoking Prevention and Cessation Project (TVSFP) was a study designed to determine the efficacy of a school-based smoking prevention curriculum in conjunction with a television-basedprevention program, in terms of preventing smoking onset and increasing smoking cessation.

The study used a 2 x 2factorial design, withfour intervention conditions determined by the cross- classification of a school-based social-resistance curriculum with a television-basedprevention program. Randomization to one of thefour intervention conditions was at the school level, while much of the intervention was delivered at the classroom level.

Reference: Flay, B.R., Miller, T.Q., Hedeker, D., Siddiqui, O., Brannon, B.R., Johnson, C.A., Hansen, W.B., Sussman, S. and Dent, C. (1995). The television, school andfamily smoking prevention and cessation project: VIII. Student outcomes and mediating variables. Preventive Medicine, 24, 29-40.

Tasks and instructions

You will make use of the dataset titled ‘Dataset of TVSFP’, a sub sample of the TVSFP study consisting of 1,600 students, nested within 135 classrooms, and nested within 28   schools. In this coursework, you are expected to explore the three-level hierarchical structure of the data: students (level 1) in classrooms (level 2) in schools (level 3). You are expected to fit three-level multilevel models to examine the relative importance of    schools and classrooms as influences on student tobacco and health knowledge (measured by tobacco and health knowledge scale (THKS)), and to pay particular attention to assessing the effects of the classroom curriculum (CC) and Television-based programme (TV) interventions.

There is good reason to expect both school and classroom effects on students’ THKS scores. While schools were randomly assigned to thefour study conditions, implementation of the CC and TV interventions were carried out at the classroom level. It seems very likely that some schools and teachers would have been more enthused about the interventions than others and this is likely to have had a direct effect on the success of the interventions. We therefore expect to see both between- school and within-school-between-classroom variation in students’ THKS scores, even after accountingfor baseline differences in their tobacco and health knowledge.

The outcome variable is students’ post-intervention THKS. You shall treat this score as a continuous outcome variable in your multilevel models. The predictor variables of key interest are the school level binary indicators of whether each school was randomly assigned to the CC or TV interventions. The predictor variables also include students’ baseline THKS scores. You are expected to include this predictor variable in your models to adjustfor baseline variation in students’ tobacco and health knowledge.

You are expected to provide the analysis details, present the results, and interpret the findings. You must carry out statistical analysis using a standard platform such as R, STATA, SPSS and SAS . You are encouraged to produce Tables/Figures for data summary. Finally, you have to attach your command file of particular statistical package you used (either R, STATA, SPSS or SAS).

Prepare a report of no more than 600 words (excluding tables, figures, references or package commands), at least one and no more than four tables/figures.

This task comprises 15% of your total grade for the module. Your answers will be assessed using the XJTLU Assessment Criteria released to you previously.

Submission

You need to submit one document, containing an analysis report, tables/figures, and the command files, in PDF. The document should be named .pdf. For example, APH421 coursework, 123456789.pdf.

You must submit your work via Learning Mall . All penalties for late or incomplete submission apply.

Assessment