What is qual data analysis
Part mechanical, handling and sorting data, and part intellectual, thinking about and with data. In the same way we look for patterns and relationships in quant data, we examine qual for common themes and relationships.
Process of analysis is not a discrete phase as the end of fieldwork, but rather ongoing from the very stat of research. A lot of ideas will occur during fieldwork, when fieldwork is over and you have change to organise data, sort through it, you will be able to pull together ‘findings’
are there Approaches to analysis? what will impact use of them?
No standard techniques / clearly defined procedures, as there are many different approaches. The approach you might take in analysing data depends on a range of factors and their interaction. These include:
Way your mind works to sort and think about things - influenced by learning style, training and perhaps left / right brain split
Level of experience
Level of knowledge of area under investigation
Availability of relevant theories or models
Type of projects e.g. groups, depths, workshops, F2F, online
Nature of research enquiry e.g. exploratory, descriptive or explanatory
Subject matter and how respondents approach it
End us of research
Resources available - time, money and number of people
With so many factors having potential influence it is unsurprising that qual data analysis is idiosyncratic - there are almost as many approaches as there are researchers.
deductive reasoning
Deductive - speculate advance of fieldwork, about what it is we think we will find and set out research to tets this idea / hypothesis. Research is designed and analysis approach is designed in a way that allows this to happen. You move from general to specific in deductive reasoning - from general idea to general hypothesis or theory about what might be happening to specific observations to see if what we expect is happening. This can also be referred to as analytic induction.
IA works like this:
Defined research problem and idea about what you are looking for
Using understanding of issues and background to problem, you develop working hypotheses about matter under investigation
Fieldwork is started and throughout you assess with how what respondents are telling you fits in with initial ideas and hypothesis
Modify ideas about what is happening, explore some issues in greater depth and get more examples of things that fit with hypotheses and so on
inductive reasoning
Induction - using this approach means that we do not go into fieldwork to test assumptions or existing theories / ideas. Data is collected and from data we identify general principles that apply to subject under study - move from specific to general. Theory building rather than theory testing.
It is difficult to use a purely inductive approach; difficult to keep all other ideas and to have completely open mind when tackling a problem, it is likely you will have some understanding of product field / area under investigation, or at least some understanding of general patterns of behaviour / attitudes (from previous research / literature). In real world research it is an iterative process involving both reasonings - ideas and hypotheses emerge from data and are tested out within data, you might revise or change them, collect more data in which to test and develop ideas on and so on.
Grounded theory
Data is examined using constant comparative methods in order to identify themes and patterns; concepts and codes are developed in order to summarise what is in the data. These concepts and codes are used to build propositions / general statements about relationships within data. Codes and propositions are tested out in data to make sure that they hold up, to make sure that they fit categories to which they were assigned and that propositions help to understand what is being studied.
Recognising own biases
Everyone has biases - they are owed to life experience and general knowledge as well as work on projects in the same area, briefing documents and background reading. Important that these are not allowed to skew analysis and interpretation of data / limit it in any way. Own thinking may mean that you only see what you want to see, or only what fits in with your view of the problem. It is therefore important in analysis to think about alternate hypotheses, to be open to different ways of looking at / interpreting evidence, question and challenge everything seen in data.
At the outset of the project you should examine what you know or assume, what preconceptions might be bringing into fieldwork or analysis. Remember to:
Keep open mind
Do not jump to conclusions
Separate how you see the issue from how respondents see it, to avoid imposing your views and ways of thinking on data
Do not force data to fit with what theory or model suggests
Making use of theories and models
Can be used to help develop / expand your thinning, speed up analysis by giving it a framework and thus a coherence, suggest questions to ask abd lines of enquiry to follow, provide ideas for developing typologies. Use Alongside a systematic testing of ideas in data - looking for evidence that supports / refutes them - a model / theory can help produce more robust analysis. Use models / theories that are well researched and empirically based.
Aim of analysis
Extract meaningful insights from data to produce valid and reliable findings to help answer research problems. To achieve this analysis should be disciplined and rigorous; does not mean it should be entirely mechanical or prescriptive. Should be thorough, consistent, comprehensive, systematic without being rigid - open to possibilities and insights that emerge as a result. Intuition and creativity are vital.
Planning analysis at research design stage - questions to ask self
List of questions to think through what implications each part of research process and decision has on analysis:
Problem:
Issues clear? Problem clearly defined?
Task to explore, describe, explain or evaluate
What output is expected? How will information be used?
What are your working hypotheses or ideas?
Using theory to drive or inform analysis?
Any previous research / relevant literature that might be useful?
Simple:
Who do you need to interview? How many?
Identified different kinds of respondent?
Expect to see different responses from different types of respondent?
Useful to compare responses among similar groups / diff groups?
Methods:
Observation? Depth? Group discussions? How will this affect analysis process?
Questions:
Topics covered in interview / discussion?
Questioning techniques? Projective / enabling
Implications these questions have for analysis?
why plan analysis at research design phase?
Process of analysis will be easier and outcome of better data quality if analysis is thought of up front, and how research design decisions will implicate analysis on the other end. May involve reviewing any relevant literature on topic / reviewing findings of other research projects etc. objectives of research drive research design and choice of sample, method and questions which all determine analysis strategy. Thinking of these at early stage will give way into analysis, ideas of how to tackle it, helping to develop strategy and framework for interrogating data and presenting findings.
Planning analysis at fieldwork stage
Overlap between fieldwork and analysis; collect data, think about them, collect more - perhaps using slightly amended discussion guide or reworked stimulus material as fieldwork sheds light on issues. Whole time your thinking about issues is developing - ideas, hunches, insights, hypotheses you may want to test and explore further.
Planning analysis at research design stage :Making fieldnotes
Due to this it is worth keeping a detailed log of thoughts and insights as they occur; written down ASAP as you may not remember, or remember correctly at the main phase of analysis. Make detailed notes or maps about what is emerging, what picture is a beginner to build up, write down particularly relevant or interesting questions.
Ask yourself what was unexpected or surprising in order to examine and challenge your own assumptions. Consider what is to be explored in further depth, what new areas need to be probed, consider implications of these for further fieldwork and for analysis / interpretation and make changes if necessary. Note down key themes, relevant quotations - anything that may be useful when analysis is in full swing.
Planning analysis at research design stage: reviewing fieldwork with colleagues / clients
Review fieldwork together in detail as soon as it is over and make detailed notes - if you have client observers ask them what they thought, and note down what they say
Planning analysis at research design stage: writing up summary
Write up a summary of main points made by participants under each of the questions / topics on interview guide or on contact summary form. Another approach is to mind map them.
Whichever is chosen should help settle and fix things in memory that will be useful later in analysis. Having a summary record of some sort will help you think about and develop ideas about data / decide on an analysis strategy. May also be a useful source to reference when it comes to writing up findings in detail; particularly if more than one person is involved in analysis, where other members can read them in order to get to grips with data across the whole sample
Developing analysis strategy - importanc e/ why
Having thought through research problems and partially completed fieldwork, you should have in mind /notes the basis of an analysis strategy or plan to tackle analysis. This should be formalised and made explicit.
Possible lines of enquiry in most qual studies are numerous and resources are limited. Analysis strategy should set out a way of approaching data, ensuring it is tackled in a systematic and rigorous way. Strategy that has been developed to suit aims and objectives should help you make the most out of resources, especially by helping you prioritise lines of enquiry. Strategy should be flexible - can and should be adapted and modified to fit circumstances.
Developing analysis strategy - considerations
Practical considerations How many will be involved in analysis Will client / sponsor be involved How long do you have for analysis Going to work from transcripts, recordings, notes or combination? Using a computer analysis package? Research considerations Decisions to be made on basis of research findings How detailed analysis needs to be Outputs required - presentation, summary report, full report Are findings to be published? How is it going to be tackled By country? Interview by interview or gorup by group Question by question Respondent type
Developing analysis strategy - how to
Use research briefs / proposals, write down big research questions set out to answer (objectives of research). List Qs and the type of respondents that might help to throw light on each of these and write down what it is you will be looking for in data generated by the questions / respondents that will help address research objectives. This is an analysis strategy,
As analysis / ideas develop may find body of knowledge that supports them or that will give you ideas or alternative ways of looking at data. Night find this knowledge in previous reports of research in topic, or literature about substantive topic you are investigating. Often worthwhile to make use of these models and theories, as they can help you to structure analysis, suggesting lines of enquiry and help develop thinking.
Doing analysis - 5 stages?
Organising data Getting to know data Getting to grips with what is going on in data Making Links, looking for relationships Pulling together findings
Doing analysis - organising data
Ordering materials in order to get on with analysis. Depending on size / complexity of the project, you may have accumulated a lot of raw materials e.g. recordings, filedonotes, transcriptions.
Spend time sorting materials into files, labelling and generally making it easy to retrieve. Particularly useful to make several copies of transcripts, unadulterated master copy, copy for cutting and pasting and copy to make notes. Once this is complete you can review field notes, listen to / watch recordings, read transcripts and plan how to tackle analysis.
Once these have been reviewed it is likely story will start to emerge and you will recognise themes and patterns.
Doing analysis -Getting to know data
Listen / review fieldwork recordings / transcripts. Will allow you to learn alot about your own interviewing technique, and will give you a chance to get into data and know it thoroughly. Data collected is spoken in discourse, so it is important to hear / see it in that form more than once, otherwise you may lose nuances or richness. If you are not able to prepare your own transcripts, as it is a time-consuming process, make sure you listen to or watch recordings at least once and read transcripts in full. Make notes as you do this, about how things were said, what was not said, what interpretations occur to you as you go through etc.
It is important not to jump to conclusions as you enter this intensive phase; may find that until reviewing all materials that all groups merge into one. Danger of misremembering things, give some things more importance than is actually the case. Need to protect against selectivity and decay of memory. This is why notes made at time are important; when reading notes and reviewing materials, write down any analytic ideas / impressions that occur to you and make notes about testing them out across all data to see if they hold up. Will need to systematically review to make sure you see the whole picture, not just bits stick in mind. Test ideas by looking at and comparing data from other respondents. Keep mind open to alternative explanations / ideas,
Doing analysis -Getting to grips with what’s going on
Pulling apart the stage of analysis. You may start to recognise patterns and themes after eviewing materials. Some may crop up more often, or less and there may be discernible patterns in attitudes, behaviour, opinions, experiences. Patterns in way people express themselves and language used - these should all be recorded for future reference.
Getting to grips with what’s going on: coding and summarising
Need to dissect data, pull apart and scrutinize it bit by bit. Involves working through data, identifying themes, patterns and labelling them / placing them under rheadings / brief descriptions summarising what they mean. This is known as categorising or coding data.
Coding process is not just mechanical one of naming things and assigning them to categorize - it is also a creative and analytic process involving dissecting and ordering in a meaningful way that helps to think about and understand research problems. Coding is a useful data handling tool that allows you to bring similar bits of data together and by reducing them to summary codes, making mass of data more manageable and easier to get to grips with, enabling you to see what is going on relatively quickly and easily.
Process of developing codes and searching for examples, instances or occurrences of material that relate to code ensures that you take a rigorous and systematic approach.
Also useful data thinking tool, that allow you to see fairly quickly what similarities, differences, patterns, themes, relationships and so on exist in data - helps you develop bigger picture by bringing together material related to ideas, hunches enabling you to put them in a conceptual order and make links and generate findings.
Generating codes
Can use topics or questions from interview or DG as general codes or headings. You may have asked respondents to decrease ideal airline flight - develop general code called ‘ideal flight’ and during coding bring together all relevant data under this heading. Some people may have talked about a particular topic / answered questions later or earlier than the topic was mentioned, so it is necessary to search the data record for incidences of it.
Rather than imposing codes from outside the data you can go into data (a bottom up, data driven approach) and see what terms, words, concepts respondents use to describe things and use these as codes.
Coding process
Can be tackled in a number of ways, different researchers will have different approaches - using pen and paper or computer. Easy way of doing so is to create a new document for each heading / code, and copy / paste pieces of text that relate to code into the document. This way you build up a store of relevant material related to code. Take care to label the source of each bit e.g. respondent details, fieldwork details, place in transcript, so that you can know the context from which it came and refer back if necessary. One piece of data may fit into multiple code.
Likely you will take several coding passes through data - two. First pass may be fairly general and kept to a minimum number. E.g. 4 or 5 key themes. As you work through data second time you can divide these themes under several specific topic areas. May group data extracts under each of the revenant codes as follows, in ideal flight example these may be:
Emotional aspects
Physical aspects
Facilities
service
