SOM500: Techniques of Data Analysis

Module Provider:

Sociology

Number of credits:

10 [5 ECTS credits]

Level:

M

Terms in which taught:

Spring

Module Convenor:

Mr AD Buck

Pre-requisites:

Co-requisites:

Modules excluded:

Current from:

2005/6

Aims:
The main aim of this module is to provide students with a solid training in a variety of traditional and "state of the art" techniques of data analysis for the social sciences. Additional goals are both to help students integrate methods of collection and analysis as part of a single research process, and to help them bridge the traditional gap between quantitative and qualitative methodological approaches.

Assessable learning outcomes:
Students are expected to achieve proficiency in the use of a variety of techniques of analysis useful for sociological research. The goal is to tailor the programme’s most specialised and marketable training to students’ needs and interests. Students are also expected to gain ‘hands on’ experience and computing expertise on an appropriate balance of techniques of analysis for quantitative and qualitative data.

Additional outcomes:
Students are expected to achieve, through weekly computer labs for each option, an intermediate level of competency in the use of specialised software for different techniques.

This module complements modules SOM400 and SOM450 on Techniques of Data Collection. Students are expected to learn how to integrate different stages of the research process using data sets that they collected in modules SOM400 and SOM450 and to which they apply multifaceted methodological strategies to overcome each method’s weaknesses.

Outline content:
Short descriptions of the topics commonly on offer follow. Note that these may change from year to year. New options may also be added in the future.


Analysis of Contingency Tables
This option introduces students to simple methods for the analysis of association in two-dimensional tables, for both nominal and ordinal variables. It focuses both on the strength of the association (ratios, odd ratios and measures based on PRE) and on statistical significance (chi-square and likelihood ratios). Computer labs run with SAS and/or SPSS.

Log-linear Models and Logistic Regression
This is an introduction to log-linear and logit models for two-way and multi-way tables (uniform association, row and column effects, symmetry, etc.), with their alternative focus on association and explanation. Computer labs run with SAS and/or SPSS.

Interior Analysis & the Analysis of Influence
This option focuses on the effect of specific observations on regression results (coefficients, significance, R-square). It illustrates a set of simple Exploratory Data Analysis (EDA) tools aimed at identifying outlying and/or influential observations in a batch of data (stem-and-leaf plot, letter-value display, box plot). It also looks at various statistics that measure the effect of those observations on regression estimates (e.g., residuals, Mahalanobis distance, leverage, Cook's D statistics, DFITS, DFBETAS). Computer labs run with SPSS.

Statistical Modelling
This course introduces statistical modelling as a general and highly effective approach to multivariate data analysis. Students learn how to address explanatory questions in sociological research by examining patterns in numbers. Besides regular hypothesis testing, this course emphasises the process of comparing and evaluating alternative ways of building up explanatory structures. Students are expected to reinforce their basic knowledge of linear models (multiple regression, ANOVA analysis, logistic and Poisson regressions, etc.) and, more importantly, to develop the ability to analyse quantitative data from a coherent perspective. Computer labs run with SAS and SPSS.

Visual Analysis. Analysis of Graphic Materials
This option applies semantic, and specifically iconological techniques, to the study of visual material, such as works of art, photographs, etc. Students learn how to explore the “meaning” that informs non-literary materials, as documents of cultural expression. This meaning provides crucial data on the characteristics of the groups and societies who have created these materials, thus becoming a critical tool for performing historical and contextual analyses.

Network Models
In this module students learn the basic techniques for the study of social networks. First, they learn how to think of the social world in relational terms; then, they learn the basic tools for a statistical representation of that world (in particular, network graphs and block models). Empirical work is carried out using the packages UCINET and KRACKPLOT.

Discourse Analysis
Discourse Analysis is concerned with the description of written and spoken language in use. It aims to identify systems and patterns within discourse and to relate these features to the context in which the language is produced. This option introduces some key issues in the description of discourse, such as conversation structure, exchange structure, coherence and cohesion in text, and the organisation of information. The option comprises two complementary components: i) an outline of the main approaches to the description of discourse, and ii) workshop sessions applying these approaches to naturally occurring samples of language.

The Comparative Method
The comparative method helps bridge the gap between probabilistic arguments based on quantitative assessments of simple causal relations (the variable oriented approach), and qualitative assessments of complex causal processes inferred from a few or single case studies (the case-oriented approach). After reviewing the history of the comparative method, its strengths and pitfalls, the course focuses on the Boolean approach to qualitative comparisons. Students learn the basic features of Boolean algebra (binary data and truth tables), its basic operations (addition and multiplication), logic (combination, minimization and implication) and laws (Morgan’s law). The module also trains students to implement Boolean algorithms with the computer package QCA.

When a student can demonstrate proficiency in the contents of any of the mini-modules, it might be possible, after obtaining authorisation from the Director of the Programme, to substitute these contents with ‘independent studies’ on a technique of data analysis suited to the student’s interests and needs, under the supervision of an ‘expert’ member of staff. In this case, contents shall be similar in extent to the contents of the mini-modules they are replacing, and shall be clearly outlined in the first week of classes by the pertinent member of staff, and authorised by the Director of the Programme. Contact hours and assessment (see below) shall also be equivalent to those in the replaced mini-module.

Brief description of teaching and learning methods:
Lectures & computer labs where students learn and practice with specialised software, or perform practical assignments.

Contact hours:

Assessment:
Coursework
Coursework assessment varies by topic: from in-class tests and practicums to “real life” research practices.

Relative percentage of coursework:
100%

Penalties for late submission
As per University policy

Examinations
none

Requirements for a pass
50%

Reassessment arrangements
Students may request from the module convenor to retake failed tests or re-submit failed coursework within four weeks of the original test/deadline. Students may fail the reassessment exercise(s) and still be able to pass the module if their average mark is 50% or above.

Students who gain a final average of less than 50% may take a special examination in September on the contents of the whole module. This examination will not be necessary when the student’s average mark in all modules is 50% or more, as described in the MSc Programme Specification.