Cross-national comparative research: Longitudinal analysis of panel data
18-19 February 2019, University of Ljubljana
This course will introduce participants to some of the many topics, data and analysis tools available for cross-national comparative research on the basis of panel data. Participants will gain hands-on experience working with some of the world’s most powerful panel data: the UK Household Longitudinal Study (UKHLS), the German Socio-economic Panel Study (SOEP) and a cross-nationally harmonised data set of panel studies from around the world (CNEF). They will learn about the many steps involved in undertaking this type of analysis, covering the conceptual bases as well as the basic data management and modelling techniques.
The course is modelled on the success of the “Introduction to Understanding Society using Stata” training course, which has been developed by Dr Knies and colleagues at ISER University of Essex. It is delivered through a series of overview lectures which are followed by hands-on sessions in the lab. Participants will be provided with a number of worked examples that they can work through during the course and use as the building block for their own analyses in the future. They have the opportunity to ask the trainers any questions they may have on the material. Examples will range in complexity from an introduction to the data management steps involved in setting up panel data to do transition analyses, to implementing and choosing between different panel estimators in within country comparisons, to testing hypotheses about effects in two different types of country, to implementing panel models across a larger number of countries
Who is the course for?
This course is aimed at new users of panel data, as well as those who have so far only made use of simpler aspects of the data such as single-country analyses. It aims to guide the analyst through the complexities of using this data for cross-national and longitudinal analysis, and ensure that they can make effective use of the data for their own research projects.
Course participants will learn about:
- Longitudinal studies for cross-national comparative research
- Basic models for longitudinal analysis: Transition models, fixed effects and random effects models (for continuous outcomes).
- Topics in cross-national comparative research and what panel data add
- Different approaches to undertaking cross-country analyses
- The design and content of some of the world’s most widely used panel data sets
- How to access the data
- How to prepare the data files for analysis using Stata
- How to use weights for producing population estimates
- How (not) to choose between different panel estimators
- Additional resources to support their own research projects
How to apply
Users will need a basic working knowledge of Stata. Without this basic knowledge of Stata the user will not be able to make optimal use of the course.
- starting Stata, Stata windows
- getting help: help, search
- working with do and log-files
- editing do-files
- comments, line breaks
- working directory, working memory: dir, cd
- opening and saving data files: use, save, compress
- inspecting data: describe, list, inspect, summarize, count, tabulate subsetting data: in, if
- generating and renaming variables: generate, replace, rename labelling variables: label define, label variable, label value
Participants not familiar with these Stata commands are strongly encouraged to use the resources recommended below to familiarise themselves with these commands before the course. Otherwise, participants may end up spending more time learning Stata than the complexities of using panel data for cross-national comparative analysis.
Resources for learning Stata: Books
- Kohler, U. and Kreuter, F. (2009) Data Analysis Using Stata, 2nd ed. College Station, Texas: Stata Press.
- Acock, A. (2014) A Gentle Introduction to Stata, 4th edition, Stata Press.
Courses and online resources
Measurement quality and correction for measurement error
The third SERISS training course – Measurement quality and correction for measurement error – took place at the University of Ljubljana in Slovenia on 26-27 February 2018. The course discuss issues related to measurement error in survey research and consequences for the results of the research. It offered different methods to estimate and predict measurement error and explained how this information can be used for the improvement of survey questions before data collection, and for the correction for measurement error in data analysis.
Course participants were trained in how to understand what measurement error is and how it can affect substantive conclusions;how measurement error can be estimated and predicted; and how corrections for it can be made in regression and structural regression models with single and multiple indicators and sum scores using R. The training session was delivered by Diana Zavala-Rojas and Wiebke Weber of the Research and Expertise Centre for Survey Methodology (RECSM), Universitat Pompeu Fabra in Barcelona, Spain.
Sampling, weighting and estimation in survey methodology
The Sampling, Weighting and Estimation in Survey Methodology training course took place on 24-25 April 2017 at the University of Ljubljana in Slovenia. It covered three interrelated topics: methods of selecting complex samples, creation of analysis weights that adjust for nonresponse and undercoverage and the analysis of data collected via complex weighted surveys.
It began with an introduction to the framework for design based inference and basic sampling designs. Common features of sampling designs such as stratification, sampling of clusters and multi-stage sampling were discussed. For each method, students learnt the relevant formulas for point estimates and variance estimates; however, the course emphasized application over theoretical proofs of the formulas.
The second day focused on estimation based on survey samples and inference. Furthermore, students were taught how complex designs and estimators alter the ways in which survey data should be analyzed. The course was instructed by Stefan Zins and Matthias Sand (Leibniz Institute for the Social Science – GESIS).
Designing questionnaires for cross-cultural surveys
The first SERISS training course took place on 24-25 October 2016 at the University of Ljubljana in Slovenia. The course – Designing questionnaires for cross-cultural surveys – was instructed by Ana Villar (City, University of London) and Dorothée Behr (GESIS – Leibniz Institute for the Social Sciences in Mannheim). You can find out more about the course and instructors by reading the SERISS 2016 Course Description.
25 participants interested in in the impact of linguistic and cultural aspects on cross-cultural survey research attended the course. Those participants were taught to:
- be familiar with different approaches to organisation and implementation of cross-cultural questionnaire design;
- be familiar with different pre-testing techniques available for cross-cultural contexts;
- be familiar with aspects to take into account when producing questionnaire translations;
- be familiar with best practice in carrying out questionnaire translation and assessment;
- be able to better account for the needs of cross-cultural questionnaire design and translation in project proposals.
All attendees were awarded a certificate for completing the course.