# Introduction to multi-level modelling with Stata

Dates: Day-1: 26th November 2018 Day-2: 30th November 2018 10:00-16:00 Mohammod Mostazir (m.mostazir@exeter.ac.uk) F19 IT Suite, Medical School Building, St Luke's Campus

## Workshop aim:

The aim of this two-day workshop is to provide an introduction to multi-level modelling using the statistical software package Stata.  The two-day interactive course is designed to i) enhance participants' capacity to programming in Statistical software package Stata and ii) Statistical programming of hierarchical/clustered data with continuous outcome variables.

## Why Stata:

A recent meta-analysis found Stata to be the most favoured statistical software package in the Health service research (HSR). It is flexible and intuitive with great capacity to answer research questions.

## Requirements:

Basic statistics will not be covered in terms of conceptual teaching but will be touched through programming context. Some basic level of understanding about linear regression models is required i.e. interpretation of coefficients, Variance, Standard error, 95% CIs etc. No programming knowledge is required as the course targets to cover the programming part in Stata. No personal laptop/notebook is required as the software is installed in the lab computers.

## Programme outline:

### Introduction to Stata

- Managing Stata and data manipulation in Stata

• Opening Stata, window management, writing syntax, executing commands
• Import data (from SPSS, Excel, CSV, TEXT)
• Using/installing user written programmes
• Basic commands (generate/recode/label variables, change variable string to numeric, data storage type etc.)
• Explore basic descriptive statistics i.e. mean, standard deviation, correlation, t-test, chi-square test etc.)
• Basic graphics (scatter plot, linear/quadratic relationships/histogram, density plot)
• Loop using local macro

- Multiple linear regression with Stata

• Fitting regression models in Stata
• Exploring slopes, interaction effects
• Scalar/Matrices
• Use of Stata's "margins" command (further testing hypothesis)
• Interaction plot using "margins"
• Use of Stata's "test"/"testparm"/"lincom" commands to test interim hypotheses (exploiting matrices)
• Polynomial function

### Day-2: 30th November 2018

Multi-level/hierarchical linear regression models with Stata

• Setting data for hierarchical/longitudinal data
• Variance component models
• Intra-class correlations (ICC)
• Random intercepts multiple regression models
• Interactions
• Random coefficient multiple regression models
• Further test of hypotheses
• Exploring subject specific slopes (Time permitting)
• Further model guidance (Time permitting)