skip to navigation skip to content
Providers & themes
Find theme:
Select provider / theme

Theme: Bite-Sized Modules

Show only:

7 matching courses

Causal Inference in the Social Sciences Wed 4 Mar 2020   14:00 Finished

The challenge of causal inference is ubiquitous in social science. Nearly every research project fundamentally is about causes and effects.

This introductory session will:

  • 1. Introduce three main approaches to elucidate causal relationship: structural equation models, causal directed acyclic graphs, and the counterfactual/potential outcome framework;
  • 2. Explain the common challenges in empirical research;
  • 3. Talk through some principles and intuition of several research designs that can help researchers make stronger claims for causality.

The emphasis is on setting out applications of each approach, along with pros and cons, so that participants understand when a particular design may be more or less suitable to a research problem.

Event History Analysis new Mon 1 Mar 2021   09:00 [Places]

This course offers an introduction to event history analysis, which is a tool used for analyzing the occurrence and timing of events. Typical examples are life course transitions such as the transition to parenthood and partnership formation processes, labour market processes such as job promotions, mortality, and transitions to and from sickness and disability. The researcher may be interested in examining how the rate of a particular event varies over time or with individual characteristics, social conditions, or other factors. Event History Analysis lets the researcher handle censoring and truncation, include time-varying independent variables, account for unobserved heterogeneity (frailty), and so on. The course will rely on Stata as the main computing tool, but users of other statistical software could still benefit from the course. The course is taught through both lectures and lab sessions.

This course will introduce students to the approach called "Exploratory Data Analysis" (EDA) where the aim is to extract useful information from data, with an enquiring, open and sceptical mind. It is, in many ways, an antidote to many advanced modelling approaches, where researchers lose touch with the richness of their data. Seeing interesting patterns in the data is the goal of EDA, rather than testing for statistical significance. The course will also consider the recent critiques of conventional "significance testing" approaches that have led some journals to ban significance tests.

Students who take this course will hopefully get more out of their data, achieve a more balanced overview of data analysis in the social sciences.

  • To understand that the emphasis on statistical significance testing has obscured the goals of analysing data for many social scientists.
  • To discuss other ways in which the significance testing paradigm has perverted scientific research, such as through the replication crisis and fraud.
  • To understand the role of graphics in EDA
Issues in Measurement: Validity and Reliability Mon 1 Feb 2021   14:00 [Places]

This short two-hour course will provide an introduction to measurement issues in the social sciences. We design questions (or "survey instruments") to gain information on the concepts we are researching. Two prime considerations in whether an instrument is effective are validity (does our instrument actually measure what we want it to measure?) and reliability (does our instrument give consistent results across a range of different situations?) Considerations of validity and reliability are important across many areas of social science, including the measurement of personality and mental health; attitudes; ability tests; substance use disorders; and cultural differences and similarities between various groups. The course will discuss the importance, concepts, and types of validity and reliability. We will also briefly look at some statistical techniques for validity and reliability checks: Cronbach’s Alpha, Kappa coefficient, and Factor Analysis.

Merging and Linking Data Sets Mon 30 Nov 2020   16:00 [Places]

Merging and linking data sets are a process that researchers often encounter. In most cohort studies and longitudinal data sets, data on the same respondents who were interviewed at various times may be stored in different files. Or, data on different respondents but were interviewed at the same time, such as mothers and their children, may also be stored in various files. In either case, we may want to merge/link the files together before performing further analyses. This course will discuss two different ways of combining data files: merge (one-to-one merging and one-to-many merging) and append, and will demonstrate how to use ‘merge’ and ‘append’ commands in Stata.

Secondary Data Analysis Tue 9 Mar 2021   09:00 [Places]

Using secondary data (that is, data collected by someone else, usually a government agency or large research organisation) has a number of advantages in social science research: sample sizes are usually larger than can be achieved by primary data collection, samples are more nearly representative of the populations they are drawn from, and using secondary data for a research project often represents significant savings in time and money. This short course, taught by Dr Deborah Wiltshire of the UK Data Archive, will discuss the advantages and limitations of using secondary data for research in the social sciences, and will introduce students to the wide range of available secondary data sources. The course is based in a computer lab; students will learn how to search online for suitable secondary data by browsing the database of the UK Data Archive.

Weighting and Imputation Mon 24 Feb 2020   13:00 Finished

In order for the findings of statistical analysis to be generalisable, the sample on which the analysis is based should be representative of the population from which it is drawn. But it is well known that some groups are under-represented in social science surveys: they may be harder to contact in the first place, less likely to agree to participate in the survey, or less likely to answer particular questions even if they do agree to participate.

This short module will introduce students to the techniques used by survey statisticians to overcome these problems. Weighting is used to deal with the problem of certain groups being under-represented in the sample; imputation is used to deal with missing answers to individual questions. Students will learn how and why weighting and imputation work, and will be taken through practical lab-based exercises which will teach them how to work with secondary data containing weights or imputed values.

[Back to top]