The Quantitative Research Reading Group
The Quantitative Research Reading Group is a cross-departmental group of researchers who are interested in issues around the application of statistical methods to social science questions. It represents one of several initiatives including the Q-Step and Eastern Arc partnership which aims to expand interest in quantitative methods amongst social scientists.
We meet several times a term and discuss (amongst other things) - challenges faced by quantitative social scientists in getting research published, specific methodological issues such as the use of the comparative method, and advice for early career researchers and PhD students. We also invite expert speakers who advise on how innovative quantitative methods can be used to expand knowledge in various important social science areas. This term we have Dr Andreas Murr from the University of Oxford coming to speak about the use of Bayesian statistics and hierarchical models. Speakers such as this are funded through the University of Kent Q-Step award.
For further details about this and other events please see the schedule for meetings posted below and if you wish to be added to the mailing list please click here (please make sure you login to the Mailing List Service before you clik on 'subscribe').
Owen Davis - Please email any queries to firstname.lastname@example.org
Owen has recently had a blog article published on the LSE British Politics and Policy site. In an upcoming article with Dr Ben Baumberg Geiger, the authors provide evidence which refutes governmental claims surrounding food banks and ask why did food insecurity rise the most in the UK over the course of the recession?
QRRG Meeting - Week 11 - Thursday 11th December 2015 - 11.00 to 12.00 in Maths Terminal Room B
QRRG Meeting - Week 9 - Thursday 27th November 2015 - 11.00 to 12.00 in KS2
QRRG Meeting - Week 6 - Thursday 6th November 2015 - 11.00 to 12.00 in KS2
Previous STATA Workshops
Wednesday 16th March 2016 - Room CC03
Friday 27th March 2015 - 11.00 to 15.00 - Room CC03
Friday 13th March 2015 - 11.00 to 15.00 - Room CC03
Participants are expected to have prior experience of running analysis using other statistical packages such as SPSS or R. The following are specific analyses we would hope that participants are familiar with:
- Univariate Statistics (Central Tendency, Standard Deviation)
- Bivariate Statistics (e.g. T-Tests, Chi-Square, Correlation Analysis)
- Multivariate Statistics (e.g. Linear/or Logistic Regression)
We do not teach any statistical or mathematical theory. Therefore we would already expect participants to be familiar with the rationale and logic for running these tests. The workshop is applied and is geared towards researchers using statistics in social science research.
- Introduce students already familiar with packages such as SPSS and R to the Stata statistical analysis software
- Explain the process of cleaning and managing data, whilst providing hands-on examples of how to do this
- Demonstrate how to conduct univariate and bivariate statistical analyses using Stata coding and help students run these analyses on their own
- Extend this to cover multiple (linear and logistic) regression and, for more advanced students, multilevel modelling. This will be applied using examples from both Political Science and Social Policy
- Data management: This includes the fundamentals around using Stata, saving datasets and do-files. It covers the principles of good practice including how to use Do-Files and Logs. We also cover some basics of data cleaning including summarizing data, checking for missing data and recoding variables.
- Univariate and Bivariate Statistics: Here we will use applied social science examples to show attendees how to conduct t-tests, chi square tests, and how to run correlation statistics for different measurement levels. We also go through some more specialist commands for describing data such as the 'tab', 'proportion' and 'if' commands.
- Multiple Regression: In this part of the workshop we introduce linear and logistic regression on Stata. We also show attendees how to run the relevant commands, interpret statistical outputs and present data in academic research and to clients in the private sector. Primarily, examples will be drawn from our own academic research (Far Right Voting in Europe; Anti-immigrant sentiment and Chinese Immigration; impact of social policies on health and health inequalities), public and private sector research (TNS opinion Brussels, European Commission).
- Extensions: For those with more advanced interests we will have a session later in the day in which we look briefly at some advanced statistical techniques such as factor analysis and multilevel modelling (Hierarchical Linear Modelling, Binary Logit and Multinomial Logit Analysis). We will speak about extension packages that can be run alongside the use of Stata to generate results. This session involves understanding how to apply these statistical techniques in an advanced setting. If time permits, an introduction to Bayesian Statistics and applications of Multilevel Modelling in academic and non-academic research would be introduced. This would cater for more advanced students.
Dr Anna Brown, Lecturer from the Department of Psychology ran two workshops on the program MPlus, in conjunction with the QRRG and Q-Step.
MPlus is a latent variable modelling program which can be used for exploratory factor analysis, latent class analysis and structural equation modelling, amongst many other functions (see http://www.statmodel.com/glance.shtml). Anna's class comprised of two half days with the following content:
Day 1 - Thursday 23rd April 2015 - 12.30 to 17.00 - Psychology Computing room in (Keynes, N1.04)
- Introduction to MPlus, describing data and variables
- Regression and Path Analysis
- Confirmatory and Exploratory Factor Analysis
Day 2 - Friday 24th April 2015 - 09.30 to 13.00 - Psychology Computing room in (Keynes, N1.04)
- Models for multiple groups, issues of measurement invariance using latent trait models
- The group-covariate and multi-group approach with equivalence constraints
- Latent Class Analysis