Please access the new Skills for Learning website at https://libguides.leedsbeckett.ac.uk/skills-for-learning/. Please note this site is no longer being updated and will be removed by 2021.

Workshops / Get Help | A-Z Index | Publications | Printable Guides | Media Library | Staff

Research

CONTENTS

Analysing and presenting data

«BACK | NEXT»

Qualitative and quantitative data

The nature of your data (i.e. quantitative / qualitative) will be influenced by the ontological, epistemological and methodological concerns that have Guided your research. It is probable that your research will have generated both quantitative and qualitative data, particularly if you have approached your methodology in a pluralistic way or have triangulated your methods. If this is the case you will be presenting your data in statistical and text format.

Qualitative Data Analysis

Miles and Huberman (1994) point out that there are diverse approaches in qualitative research and analysis. They suggest, however, that it is possible to identify "a fairly classic set of analytic moves arranged in sequence." This involves:

  1. Affixing codes to a set of field notes drawn from observation or interviews
  2. Noting reflections or other remarks in the margins
  3. Sorting and sifting through these materials to identify similar phrases, relationships between variables, patterns, themes, distinct differences between subgroups and common sequences
  4. Isolating these patterns and processes, commonalties and differences, and taking them out to the field in the next wave of data collection
  5. Highlighting generalisations relating them to your original research themes
  6. Taking the generalisations and analysing them in relation to theoretical perspectives.
    Miles and Huberman (1994)

Patterns and generalisations are usually arrived at through a process of analytic induction (see above points 5 and 6). Qualitative analysis rarely involves statistical analysis of relationships between variables. Indeed, many qualitative data analysts resist the use of the term 'variable'.

Techniques for coding differ according to the particular method(s) you have used to collect your data. For example, coding qualitative data will involve assigning conceptual categories to the text, whereas quantitative data is more likely to rely on the assignment of numeric codes.

Interview text may be coded as follows in this extract using colours and categories:

"I was brought up with the idea that women got married, had children and that was their life. I kept looking for self satisfaction. You didn't go into further education; you got married and somebody supported you. It's a continuous struggle with myself"

(Kay)

Examples of possible codes:

  • Life history
  • Emotions

Quantitative Data Analysis

Introduction

A statistician's wife had twins. He was delighted. He rang the minister who was also delighted. 'Bring them to church on Sunday and we'll baptise them,' said the minister. 'No,' replied the statistician. 'Baptise one. We'll keep the other as a control.'

Gary C. Ramseyer's first Internet gallery of statistics jokes

Web Based Resource

Gary C. Ramseyer's first Internet gallery of statistics jokes

Please note that availability, functionality and content of web-based resources are outside our control.

The word 'statistics' is used in different ways. The term 'statistics' sometimes refers to calculated quantities regardless of whether or not they are from a sample. For example, one might ask about 'government statistics' which refer to any numerical indexes calculated by a governmental agency.

In the broadest sense, 'statistics' refers to a set of methods, tools and techniques used to collect, organise and interpret data. The goal of statistics is to gain understanding from data. Therefore we need to know how to:

  • Produce data - for example by handing out a questionnaire or by doing an experiment
  • Organise, summarise, present and analyse data
  • Draw valid conclusions from findings

There are a number of statistical methods that you can use to analyse your data but it is important to remember that selection of an appropriate statistical method follows naturally from the research design you have chosen. Therefore, the design of a study will govern how the data is to be analysed. You need to think about data analysis and/or consult a statistician at the early stages of your study design. Statistical tests are not magic formulae to turn to when you have collected your data. As R. A. Fisher said:

To call in the statistician after the experiment is done may be no more than asking him to perform a post-mortem examination: he may be able to say what the experiment died of.

Tips for Working with statistical data

  • Plan so that the data you get has a good chance of tackling successfully the problem that you want to get to grips with. This will involve not only reading the literature on your subject of study but also literature on what makes a good study.

  • To find effects we need to reduce uncertainties i.e. 'noise'. This requires a sufficiently large sample of data. Will you have enough? A lot of data will give good precision but unnecessarily high precision will involve unnecessary 'costs' if you have to collect it.

    Will there be problems in getting data of sufficient quality e.g. accuracy, trustworthiness, completeness? Consider the logistics.

  • Statistics works on random samples - will yours be? Are there biases in the collection?

    How will you deal with the fact that some data will be not recorded for some reason? (These are called missing values). These can result from either gaps in a record or whole records being missed out.

    Invent / generate data of the sort that you think you might get and see if you can analyse it! Make sure that you can, before getting any real data. Think what the output of an analytic procedure will look like before you get the computer to do it.

    (Note: it is actually rather difficult to generate very realistic data and there are methods to detect whether data have been fabricated [i.e. to detect fraud!] so get rid of your practice data before analysing the real stuff).

  • When analysing data, start simply by looking at each variable separately and do initial/exploratory data analysis, especially using graphical displays. Do this before looking at variables in conjunction or anything more complicated. This can help locate errors in the data, for one thing, as well as to get a 'feel' for it.

  • Look out for patterns of 'missingness'. They are likely to tell you something and if the 'missingness' is not random then they will have an impact on the results.

  • Be vigilant and think through what you are doing at all times. Be critical and nosy. Do not think of statistics as a set of mathematical tricks that a computer will sort out. Think of it rather as a logical path from conception to conclusion which the human mind must fathom!

Measurement and sampling

There are two types of measurement: categorical and numerical.

Categorical measurement
Categories within categorical measurement may be characterised by distinctiveness in which case the measurement scale is referred to as a nominal scale; or the categories may be both distinctive and ordered, in which case the measurement is referred to as an ordinal scale.

Examples:

Nominal scales
response yes no
coding 1 2

this is known as the coding frame for code 1 2 question.

Ordinal scales

An example of a coding frame for questions of opinion:

strongly disagree disagree neutral agree strongly agree
1 2 3 4 5

Reference: Bell, J. (1987, p. 104).

This data can then be put into statistical format and displayed using charts, graphs, tables etc.

Numerical measurement

Examples:

  • Ratio scales
    These may be based on a true number scale with a meaningful zero point, such as age or height.
  • Interval scales
    Or they may only imply equal intervals between numbers but have no true zero point (IQ scores, temperature scales).

These are the most basic examples of coding.