Survey Data Analysis II: Introduction to Linear Regression Modelling
Venue: Highfield Campus, University of Southampton, UK
Presenter: Dr Nikos Tzavidis
Dates of Course: Wednesday 16th - Friday 18th January 2013
This course has already run. Please check the course listings for a future course.
Summary of Course:
The course focuses on the use of statistical modelling to study association between variables. The course covers simple and multiple linear regression, where the dependent variable is a continuous variable. (This course is designed to follow Survey Data Analysis I; course participants who have not attended this course can of course attend).
Course Objectives:
- To develop a practical understanding of the basic statistical principles and methods of modelling relationships between variables with focus on simple and multiple linear regression.
- To enable participants to employ appropriate methods in analysing their data.
- To develop a practical understanding of the principles of covariance, correlation, and simple and multiple linear regression.
- To go beyond basic principles in order to understand the assumptions behind the models, the impact of violations, how violations can be diagnosed and possible solutions.
- To develop an understanding on the reasons, use and implications of modelling categorical variables and interactions in linear regression analysis.
Course Content:
- Covariance and correlation
- Simple linear regression modelling (the dependent variable is a continuous variable)
- Interpreting computer output
- Model checking and model selection
- Multiple regression modelling (the dependent variable is a continuous variable)
- Handling categorical variables and interactions (as independent variables)
The course will have a strong practical emphasis, with regular sessions on computers enabling participants to work through examples in SPSS. The course will not cover logistic or multinomial regression (i.e. regression methods where the dependent variable is a binary or categorical variable). This will be the focus of the CASS course 'Regression Methods for Survey Data', which is also part of the CASS short course programme.
Target Audience:
The course is aimed at researchers who need to perform basic statistical analyses on data from sample surveys, especially those in the social, economic, educational and medical sciences. Participants may be researchers working in academia, local or central government, survey agencies, market research, the voluntary or the private sector.
Pre-requisite:
Participants are expected to have a basic knowledge of simple statistical methods. They will previously have taken a course in introductory statistics and/or have completed Survey Data Analysis I (another CASS course). Ideally, course participants should be familiar with all or at least some of the following before undertaking Survey Data Analysis II.
- estimation based on random (probability) samples;
- formulation of testable hypothesis;
- standardised comparisons between groups using means and proportions;
- standard errors and confidence intervals;
- analysis of variance;
- SPSS statistical package.
The course will have a strong practical emphasis, with regular sessions on computers, using SPSS and real survey data, to enable participants to work through examples. The pace of presentation will be faster than on Survey Data Analysis I, so that more ground is covered. For course participants not very familiar with SPSS there will be a prerequisite handout available to work through in their own time. For people not very familiar with the statistical methods please have a look at the preparatory reading list below.
Please bring a calculator for the workshops as well as a USB memory stick in case you would like to save your computer workshop output.
Course Materials:
Participants will receive written course notes.
Please bring a calculator for the workshops as well as a USB memory stick in case you would like to save your computer workshop output.
The Instructor:
Dr Nikos Tzavidis is a Senior Lecturer in Social Statistics at the Division of Social Statistics and at the Southampton Statistical Sciences Research Institute at the University of Southampton. He has worked extensively on issues related to the use of quantile models in small area estimation and poverty mapping. He is specialised in quantile regression, small area estimation, robust models, spatial analysis and applications of multilevel and multivariate multilevel models in psychology and psychiatry. He is currently involved in developing quantile random effects models, M-quantile models for binary and count data, quantile contextual value added models, robust models for small area estimation and poverty mapping, models for estimating income distribution functions at disaggregated geographical levels and spatial models for poverty mapping and Mean Squared Error estimation in small area estimation. He has taught courses in small area estimation, quantile regression, generalised linear models, multilevel models, and longitudinal data analysis at postgraduate level at the Universities of Manchester, Southampton and at the Institute of Education, University of London and in external programmes of short courses. He is currently supervising three PhD students, with two PhD students having already graduated, and he is coordinating the MSc programme in Social Statistics and the MSc programme in Demography at the Division of Social Statistics, University of Southampton. Dr Tzavidis has participated in research programmes at National, European and International level. He recently completed work on an EU FP7 project on small area models for estimating income and poverty. He recently also completed a project on applications of multivariate multilevel models to psychiatry data, which was funded by the British Academy. He is currently participating in the Southampton/Office for National Statistics (ONS) methodological contract (small area estimation). He is also currently advising a government research department (COLMEX) in Mexico on the implementation of small area models for poverty estimation.
Course Fees:
£30 per day for UK-registered students. £60 per day for staff from UK academic institutions (including research centres), ESRC funded researchers and UK registered charitable organisations. £220 per day for all other participants. The course fee includes course materials, lunches and morning and afternoon refreshments. Travel and accommodation are to be arranged and paid for by the participant.
Location and Accommodation:
The course will be held at the Southampton Statistical Sciences Research Institute, Building 39, University of Southampton, Southampton, SO17 1BJ. Participants will need to make their own accommodation arrangements. Further information on accommodation and course location is available here.
Duration:
On the first day, the course will start with registration and coffee at 9.30 with formal teaching starting at 10.00 a.m; it will finish at about 5.45pm. On the last day, the course will start at about 9.30am and formal teaching will end at about 3.30p.m. Afterwards there will be an opportunity for participants to ask questions about the course and to discuss with the instructor how to analyse their own data (until about 4pm). (You can bring your own data to the course if you wish).
Preparatory Reading:
For those who would like to do some preparatory reading the following references may be useful. For participants particularly rusty or new to the methods covered preparatory reading is recommended.
- Neter, J., Kutner, M.H., Nachtsheim, C.J. and Wasserman, W. (1996). Applied Linear Statistical Models, 4th Edition. McGraw Hill, ISBN 071145672
- Kleinbaum, D., Kupper, L.L. and Muller, K.E. (1996). Applied Regression Analysis and Other Multivariate Methods, 3rd Edition. PWS-Kent Publications, ISBN 0534209106.
- Draper, N.R. and Smith, H. (2003). Applied Regression Analysis, 2nd Edition. Wiley, ISBN 0471170828.