Description
Data Analysis
Quantitative research is all about what you have measured in your data. Your results will often only be as good as the data that you collect. Your data is what your hypotheses depend on for their testing, and the data is the evidence for answering the questions you want to answer. Understanding your data and the relationships within it is crucial to successful research.
The first thing you will need to consider is how much your data will need to be cleaned and transformed to be in a useful form for analysis. It is extremely rare that data comes in a form that is directly useable for analysis and statistical testing. You will need to decide how to handle missing values, values that have impossible values, and some values that may need to be changed from categorical values to numeric ones. You might end up merging or splitting up collected data to make it useful. The end goal is to have data that can be used for exploratory data analysis, statistical testing, and creating data visualizations. Expect this step to take a significant amount of time.
You will also need to consider where you will store your data and in what form? Where you keep your data will strongly influence your workflow. Will your data be manageable in files, such as comma-separated format, that can easily be loaded into spreadsheet programs and data analysis programs? Taking this route will require careful management of the files and their backup. Or, will your data be such that it needs to be stored in a database management system? This choice will require ensuring the backup of your data, and it is necessary to have a working knowledge of how to get data in and out of the database. Will you keep your data on a local hard drive for storage, or will you utilize a cloud-based resource? Having your data in the cloud allows you to access it wherever you have internet connectivity. Cloud-based resources are appropriate whether you keep your data in files or a database.
Once you have clean and tidy data, you can proceed to explore what you have collected. You will want to know some basic things about your data and do an initial assessment of the relationships that might exist between your variables. You can do histograms of your variables to visualize the shape of their distributions. You can run tests to determine if your variables are from a normal population or other probability distributions. You can create a pairwise correlation plot of all or groups of your variables to visually determine if there are any distinct relationships between certain variables. You might find box plots or other data visualizations suitable for elucidating the secrets hidden within your data.
At some point, you will need to perform your statistical testing. Depending on the volume of data and the particular testing being done, this might be a compute-intensive task that takes quite some time to run. Or, it might involve a series of runs that soaks uptime. From your exploratory analysis, you should have a good idea of how far you from meeting the assumptions of the tests you are using and a justification for using them. Consider ways to visualize the results of your statistical tests, such as plots of residuals.
Lastly, you need to consider the tools you will use to conduct your analysis and create your visualizations. Perhaps you are already proficient with one of the numerous tools available like R, Tableau, SAS, SPSS, and Minitab. Sometimes, it is possible to do your work with a tool like Microsoft Excel. Whichever tool you choose, it is important that you are able to produce output that can be imported into your dissertation documents in an acceptable form and that your visuals are of publication quality.
Assignment: Present a Data Analysis Strategy
Instructions
For this assignment, you must develop a data analysis strategy for your research study. You must then present your design in a presentation.
Your presentation should include the following:
Title Slide
Introduction
Problem Statement
Research Purpose
Research Questions
Definition of Key Terms
Review of the Literature
Validity of Research Topic
Hypothesis
Research Design
Proposed Data Collection Strategy
Proposed Data Analysis Strategy
Resources
150-200 words of speaker notes for each slide to assist with the delivery of the presentation
Recording of your presentation of each slide using the recording feature in PowerPoint
Length: 12-15 slide presentation, with audio recording
References: Include a minimum of 5 scholarly resources.
The completed assignment should address all of the assignment requirements, exhibit evidence of concept knowledge, and demonstrate thoughtful consideration of the content presented in the course. The writing should integrate scholarly resources, reflect academic expectations and current APA standards.
Unformatted Attachment Preview
Sage Research Methods
Getting Your PhD
For the most optimal reading experience we recommend using our website.
A free-to-view version of this content is available by clicking on this link, which
includes an easy-to-navigate-and-search-entry, and may also include videos,
embedded datasets, downloadable datasets, interactive questions, audio
content, and downloadable tables and resources.
Author: Harriet Churchill, Teela Sanders
Pub. Date: 2011
Product: Sage Research Methods
DOI: https://doi.org/10.4135/9781849209229
Methods: Mixed methods, Research questions, Thesis
Keywords: software, transcripts
Disciplines: Anthropology, Business and Management, Criminology and Criminal Justice, Communication
and Media Studies, Counseling and Psychotherapy, Economics, Education, Geography, Health, History,
Marketing, Nursing, Political Science and International Relations, Psychology, Social Policy and Public Policy,
Social Work, Sociology
Access Date: January 20, 2024
Publishing Company: SAGE Publications, Ltd.
City: London
Online ISBN: 9781849209229
© 2011 SAGE Publications, Ltd. All Rights Reserved.
Sage
Sage Research Methods
Contact SAGE Publications at http://www.sagepub.com
What to do With your Data
What this Chapter Includes:
▸
From data to findings
▸
Types of data and data analysis
▸
Preparing data transcripts and field notes
▸
Labelling, recording, storing and archiving
▸
Shall I use a software program?
▸
Analytical strategies and coding your data
▸
Moving towards overall research findings
▸
Thinking through the claims you can make: issues of validity and reliability
From data to findings
A clear distinction between data collection and data analysis is difficult to maintain as researchers can begin
to interpret data as it is collected (hopefully capturing such hunches and impressions in a research diary or
log). However, data analysis refers to a more concentrated and systematic period of analysing of your data,
with the view to generating avenues for further data collection or your overall research findings. The process
of data analysis involves making sense of your data in relation to your research questions. Before considering
some important stages and approaches, we need to distinguish between types of data.
Page 2 of 15
Getting Your PhD
Sage
Sage Research Methods
Contact SAGE Publications at http://www.sagepub.com
Types of data and data analysis
Data can be:
• Numerical (e.g. numbers and measurements).
• Textual (e.g. in written form such as documents; transcripts of individual or group interviews; observational field notes; diary entries or historical records).
• Visual (e.g. such as video recordings; photos).
• Audio (e.g. sound recordings; recorded conversations).
• A mixture of any of the above.
These different types of data offer different ways of capturing the phenomena you are investigating. They can
be grouped into quantitative or qualitative forms of data, with the former involving the generation of numerical
data and the latter involving interpretations using words, pictures or sounds. Data analysis approaches also
involve quantitative, qualitative or combined frameworks.
BOX 5.1 Examples of quantitative and qualitative approaches
Quantitative analysis can involve the production of:
• Descriptive statistics on your data (i.e. variable frequency/averages).
• Inferential statistics (i.e. variable significance/association significance).
• Multivariate analysis (i.e. relationships between two or more variables).
Qualitative analysis also takes many forms including:
• Interpretative thematic data analysis: generating themes from interview, visual or audio data.
• Framework analysis: a thematic approach that seeks to generate concepts from data and a prior evidence base for applied policy research.
• Grounded theory: an approach to generating conceptual frameworks from your data rather than prior
academic theory.
Page 3 of 15
Getting Your PhD
Sage
Sage Research Methods
Contact SAGE Publications at http://www.sagepub.com
• Life story approaches: analysing personal narratives and life stories using theories and biographical
frameworks.
• Analysing biography and social context: Biographical Narrative Interpretative Method, for example,
seeks to generate a picture of life histories and social contexts.
• Reflexive readings: autobiographical approaches seek to analyse the researcher’s life experiences
or input into the research process.
At the design phase of your study, we are hoping that you have already considered your analytical approach
and established a complementary connection between your research questions, data collection and analysis
methods and approach. Box 5.1 also gives some further indications of approaches on which the suggested
reading at the end of this chapter provides more information (also see Chapter 2). Not having a clear plan
of how you propose to analyse your data runs the risk of being limited in the kind of analysis you can do.
Planning your analysis approach after data collection increases the likelihood of being restricted to the kinds
of analysis that suit the data you have collected. For example, a qualitative researcher is unable to conduct
a biographical life history approach to analysis if they have limited data on the respondent’s life history; and
a quantitative researcher will find it difficult to conduct particular types of statistical analysis of their data if
their questions or sampling method do not fit the analysis approach. However, with many existing texts that
detail particular approaches, this chapter tends to refer students on to more specialist texts for more detailed
accounts and here considers some steps and strategies for moving through your analysis in a more general
way. The chapter begins with some experiential reflections on getting started and then goes on to consider
issues for analysis such as planning your approach, preparing your data, indexing and coding, and establishing relationships, patterns and findings.
Getting started, getting stuck and pushing forward
Undertaking a PhD study will inevitably involve periods of uncertainty, difficulty and doubt. A common sentiment at the beginning of the data analysis stage is an overwhelming feeling of having a lot of data and wondering what to do with it, and whether it will provide some illuminating answers to your questions. One of our
contributors, Joseph Burridge, undertook a study of media representations of Iraq. He was all set to begin
his PhD research at the end of September 2001 when the World Trade Centre was struck earlier that month.
Page 4 of 15
Getting Your PhD
Sage
Sage Research Methods
Contact SAGE Publications at http://www.sagepub.com
This event changed the significance of his project. Joseph then spent three years collecting a huge amount
of newspaper archive data:
My data collection began a fortnight before the official start of my PhD registration (12th September 2001),
and continued up until about a fortnight before my thesis was submitted. The availability of masses of material
can be viewed as either extremely exciting or as potentially overwhelming – I constantly oscillated between
these two positions. In reality, I spent several hours each day collating and categorising media reports.
Several of our contributors found that they had under-estimated the time it would take to analyse the data
they collected. Harriet Churchill realised she had collected more qualitative interview data than she was able
to analyse, which contributed to a delay in submission by six months. Due to time constraints, John Roberts
decided to drop an area of historical data archive material in his project, but was able to return to analyse
this at a later date. Often the time taken to ‘prepare’ your data for analysis such as with inputting data into
software programs, can be grossly under-estimated. Emily Tanner felt she had:
Under-estimated the amount of time it would take to prepare the (quantitative) data for analysis. It took much
more time than expected to spot the errors in data entry to SPSS, to deal with missing data and to derive the
variable for subsequent analysis.
Another concern in the early stages of data analysis was one of doubt and uncertainty over whether ‘you have
the right data or will find anything out’. This can sometimes be even more the case in survey research, as
you may not have read through your survey responses until you begin analysis. In her quantitative study of
maternal employment outcomes, Emily Tanner was concerned about ‘whether I would find the relationships
that I was expecting or if my results will yield anything to write about’. With no scope to further probe your
participants’ responses in postal survey research, Sheelah Flatman Watson was hugely concerned whether
her ‘questions were right’ or understood appropriately. She waited with ‘nervous anticipation’ for her surveys
to be returned and to find out if ‘her questions were appropriate, and if there was enough scope for individual
Page 5 of 15
Getting Your PhD
Sage
Sage Research Methods
Contact SAGE Publications at http://www.sagepub.com
comment in the open questions’ she had devised.
A further common challenge within the data analysis phase of a research project is that of moving from a
detailed picture of your data, perhaps aided by coding, sorting and devising charts and tables of your overall
data, towards some overall general explanations and findings. Harriet Churchill discusses this stage of her
thematic qualitative data analysis:
Trying to make connections across all the transcripts was extremely difficult. I found it hard to move towards
a more general picture as there seemed to be exceptions to each overall explanation. It wasn’t until I started
writing up, writing papers and trying to explain the dominant themes in writing that I began to move towards
a more general picture of the data. I also had to drop many interesting themes, and really focus on some
particularly pertinent areas. Here I was concerned about prematurely limiting my explanatory possibilities.
In moving from data collection towards data analysis you may feel:
• Unsure if you have ‘enough’ of the ‘right’ data.
• Unsure if you have ‘too much’ data.
• Uncertain about how to get ‘started’.
• Excited to establish the central findings from the data.
• Unsure about the details of how to put your approach into practice.
• Overwhelmed by the need to learn new data analysis software packages.
To get yourself started with data analysis, it may be worth spending some time reflecting on what your concerns are and what you are expecting to find out, and turning these into questions and issues to think about
further and seek advice on. Whether studies employ a qualitative, quantitative or mixed methods approach;
analysis requires preparing the data for analysis, employing analytical strategies and moving towards research findings.
Page 6 of 15
Getting Your PhD
Sage
Sage Research Methods
Contact SAGE Publications at http://www.sagepub.com
Preparing data for analysis
There are a number of tasks involved in preparing your data for further analysis. Data will need to be organised, labelled and stored in ways that are suitable to your purpose and approach. For qualitative data, preparing your data may involve typing up field notes, transcribing interviews, anonymising data and inputting data
into a software package. For quantitative data, also you will need to input data into a software package alongside the cataloguing of survey responses, survey response rates and missing data (numerically cataloguing
how many surveys responses were received and how many questions were answered in the appropriate
way). As we have already highlighted, the time required for these activities, which often overlap between data
collection and data analysis, is often under-estimated. However, ultimately thorough and thoughtful preparation of your data will potentially save you some stress and time later on, as you will be able to manage your
data in an efficient way. We will now turn to consider four key aspects of data preparation: transcribing qualitative data; labelling and storing data; utilising software packages and preparing quantitative data for analysis.
All four of these activities may be carried out as you collect your data and further as a distinctive predata
analysis stage in your research project.
Preparing data transcripts and field notes
Transcribing a research interview or recorded field notes generates a typed verbatim transcript of the interview or observed interactions and activities. This can help the researcher to analyse interview or observational data in some depth. However, in some studies time constraints may mean that interviews or field conversations are purposefully more partially transcribed, selectively recording data that is deemed to be particularly
relevant to the research topic and questions. Alternatively there are very detailed approaches to the transcription of qualitative data that seek to record a vast array of non-verbal and detailed verbal communication. Your
transcription approach will really depend on what type of analysis you wish to pursue, your timescale and the
resources available.
BOX 5.2 Some commonly used transcription conventions
Italics or bold or underline = emphasises a word
Page 7 of 15
Getting Your PhD
Sage
Sage Research Methods
Contact SAGE Publications at http://www.sagepub.com
(…) = pauses
= for simultaneous speech
CAPS = loud sounding words
(??) = words that the transcriber cannot hear
/ = a rising intonation
full stops, commas and paragraphs at points that seem to indicate these kinds of pauses and endings.
As you type out your interview be generous with your spacing and layout to leave room to write in notes/label
segments of text.
Labelling, record keeping, storing and archiving
Good practice here may be to devise a ‘data record sheet’ for each interview, field visit, survey participant
or archived raw data; and to do this before you collect any data so that the details can be filled in as your
data comes in. This sheet can include personal details, the date that data was collected, an anonymous code
referring to each participant or case and immediate impressions. Alternatively the details can be recorded
during data collection in note form and transferred onto a data record sheet in these preliminary stages of
data analysis (see Chapter 3 for researcher responsibilities).
Joseph Burridge found it took some time to devise an appropriate system of storing newspaper reports for his
study. He ended up devising a manual archiving system, with newspaper reports filed according to a thematic
main story approach. Alternatively a number of software packages now exist — although we cannot strongly
enough recommend that regular printed out hard copies and back up files of all data files are also securely
kept throughout the data analysis phase.
Shall I use a software program?
There are many software programs designed to aid quantitative and qualitative data organisation and analysis. An example of a quantitative program is SPSS (Statistical Package for the Social Sciences) and qualitative ones (called Computer Assisted Qualitative Data Analysis Systems — CAQDAS) are packages such as
Page 8 of 15
Getting Your PhD
Sage
Sage Research Methods
Contact SAGE Publications at http://www.sagepub.com
Atlas. ti, NUD*IST, NVivo 7, Ethnograph and Hypersoft. SPSS is used for the manipulation of numerical data
and can compute a number of descriptive, inferential and multivariate statistical tests. CAQDAS can offer a
range of indexing, retrieval, organising and note-taking facilities that can be used by a researcher to enhance
their data organisation and analysis. Both quantitative and qualitative packages are aids to cataloguing, storing and analysing large amounts of data that would be much more time-consuming and cumbersome to work
with manually. SPSS offers massive scope in computing statistical formulae across large data sets in a matter
of seconds. CAQDAS also offer more efficient and accessible ways of retrieving and storing qualitative data,
as well as offering hyperlink facilities so that different sections of data within or across transcripts or even
across different types of data such as photos, video clips, background research notes and field notes can be
linked within the program. If these questions appear to be bewildering, you may need to spend some time
familiarising yourself with the software packages available for your data analysis purposes. It is common for
students to need to supplement their research methods training with more specialist training in particular software packages. Again this requires time and effort. Sheelah Flatman Watson offers an encouraging example
of taking the opportunity to enhance her professional development as a researcher when she took on to learn
how to use SPSS and NUD*IST6 during her mixed method research study:
My computer skills were limited. I had a very basic understanding of SPSS. I knew what the package had
potential to do having attended a methods class, but I had never applied the learning. My first supervisor was
not familiar with SPSS but at least my second supervisor, an SPSS user, helped with how to set up my data.
But mainly I had to learn quickly, and decide on an appropriate coding frame. This required hours of trial and
error.
BOX 5.3 The use of SPSS and CAQDAS needs to be researcher and methodologically driven
Software programs do not do your analysis for you but operate a number of preprogrammed functions that
can support your analysis in particular ways. You will need to have a clear and appropriate rationale for utilising a program and its functions and be aware of the possibilities and limitations built into it. For those using
SPSS, familiarity with the rationales underpinning a variety of statistical tests is required as is careful inputting
Page 9 of 15
Getting Your PhD
Sage
Sage Research Methods
Contact SAGE Publications at http://www.sagepub.com
of data in order to make sure the raw data is correct before the software conducts any tests. The different
CAQDAS available are underpinned by different explanatory logics. In considering the use of software for
your analysis ask yourself:
• Am I clear what the software can do? For example, some CAQDAS claim ‘theory building capacities’.
But it is the researcher who decides what codes are devised, what functions are operated and what
relationships or patterns are significant. For SPSS users, while the programme will work with the
data inputted, it is the role of the researcher to ensure the data has been inputted into the program
correctly.
• What is useful about the program in the context of your research project and approach? For example,
will you use software as a means of storing, indexing or manipulating data? Or all three?
• What information can I get about the epistemological and explanatory logic of the software from the
published information, demonstration guides and formal/informal electronic networks?
• Does the software suit my IT, time and training resources?
• What are my reasons for not using CAQDAS? Am I being technology phobic?
Analytical strategies
Coding your data
Coding involves labelling and categorising your data. For qualitative data, codes represent ways of categorising chunks of data and offer a shorthand label for larger sections of data. For quantitative data, codes can
represent different types of responses to social survey research questions. The codes chosen can be derived
from conceptual frameworks relevant to your research question and area, devised ‘a priori’ at the stage of
questionnaire design so that answers to questions already have a code attached to them and the task of data
analysis involves numerically investigating the coded results. Codes can also be attached to data after data
collection, as a researcher devises a label to categorise open-ended responses to a questionnaire or sections
of transcript data. Overall, there are three types of codes:
• A priori coding: codes that refer to concepts generated from the wider research and theoretical debates related to your research topic.
• Open coding: codes representing your interpretation of respondents’ viewpoints.
• In-vivo coding: codes generated using terms that are used by respondents themselves.
Page 10 of 15
Getting Your PhD
Sage
Sage Research Methods
Contact SAGE Publications at http://www.sagepub.com
Sheelah Flatman Watson describes devising the majority of her codes prior to administering her questionnaire, all of which were recorded in a ‘codebook’. Her questionnaire design involved using closed questions
(where a respondent chooses between different answers already set out) and open-ended survey questions
(where a respondent can write out their own answers using their own words). Some of her ‘codes’, therefore,
were developed through open and in-vivo coding:
I drafted a codebook in the survey design phase, with codes emerging from the possible range of answers
given to the questions I had devised. In the data analysis phase I then had to re-develop this coding system
to incorporate the open-ended survey answers. This required similar considerations but in relation to different
sources. For the a priori coding, I referred to issues raised in the literature and practitioner debates. For the
open coding I was concerned – Were the codes a fair and reasonable description of the meanings within the
data? It was a very time-consuming exercise requiring patient readings of each contribution to decipher appropriate categorisation.
Qualitative approaches often involve developing codes inductively. Here codes represent central aspects of a
sentence, phrase or a few sentences within a transcript. Codes can be developed in relation to respondents
or observational settings viewed as ‘cases’ or as themes emerging across cases or transcripts. Strategies towards this process include: generating an initial set of codes, re-working coding towards a coding framework,
coding your data and analysing relationships between coded data. Initially, researchers often take a cross
section of their transcripts or documents, read through the transcripts and begin to label sections of the data
using open, in-vivo or some a priori derived codes. If coding is performed manually on printed out documents,
different coloured highlighter pens can be used to denote different codes. However, beyond some fairly simple
coding this task may become very cumbersome and CAQDAS have facilities for highlighting and labelling text
as ‘open or in-vivo codes’. Other concepts derived from your reading may seem pertinent to what is being
portrayed in your interviews and these can also become codes. Harriet Churchill offers an example of how
she generated some codes related to ‘coping’ in her research on parenting:
Many interviewees discussed their commitment to coping; their experiences of coping and their difficulties.
Page 11 of 15
Getting Your PhD
Sage
Sage Research Methods
Contact SAGE Publications at http://www.sagepub.com
However, the code of ‘coping’ is too broad and so I tried to devise codes that demonstrate the different facets
of coping. The codes that emerged were: coping strategy, services and coping, social support and coping and
good motherhood and coping. Some of these codes were generated from the data but others were derived
from conceptual debates in the literature around parenting on a low income. Once the codes were devised
and all the data indexed, all the data coded as related to coping could be analysed. Codes were subsequently
refined and further questions were asked about coping.
As you read through your data, keep asking yourself: what issue is being discussed here; what perspective
or concern is the interviewee presenting? Codes should also be clearly defined (CAQDAS have in-built logs
for this purpose), mutually exclusive and can be categorised according to overall theme with sub-themes/
categories. After reading a selection of transcripts, you will probably already have a long list of codes and categories. It will be worth examining this list once it gets beyond twenty or so codes and thinking about overlaps
and connections between codes. Further subsequent rounds of reading transcripts, applying and modifying
codes and thinking through the connections between codes can then be performed in order to generate a
coding framework. Once you have developed all your codes and these have been arranged into a coding
framework, you can then index all your data using these codes. This organises and categories your data into
what Mason has called ‘bags of data’ which can then be analysed further towards deriving explanations for
your research questions (Mason, 2002).
Many contributors found coding their data surprisingly time-consuming, demanding much concentration, ‘boring and repetitive’ as well as exciting and challenging. One contributor felt they ‘paid more attention to some
transcripts than others because of the time of day they were coded, their interest in the topics and their mood
at the time’. Breaking up periods of indexing with refining codes, or doing other activities such as checking
recent journals in your area for new publications or inputting your bibliography may help.
Moving towards overall research findings
Once your data are categorised into codes, analysis moves towards focusing on explaining these themes and
deciphering patterns and relationships between them in ways that return to providing some answers to your
research questions. A line of argument is hence developed, as you offer an overall interpretation of your data.
Many of our contributors refer to some helpful activities that aided the construction of an argument.
Page 12 of 15
Getting Your PhD
Sage
Sage Research Methods
Contact SAGE Publications at http://www.sagepub.com
Framework analysis, or other qualitative thematic approaches, often utilise tables, diagrams, charts or matrices to facilitate examining cases or themes. Some contributors also found writing up their data analysis in the
form of thematic papers helped them to think through their overall research questions, theoretical frameworks
and the patterns emerging in their data. Both of these strategies helped students to investigate overall patterns in their data. Sheelah Flatman Watson also found that writing a paper on her quantitative data analysis
helped her to arrive at some overall findings:
Writing a paper was a wonderful way to focus the data. I developed a series of descriptive statistics that wove
a path across phase one, the service provider survey data, and followed these findings with comparative findings from phase two, the service client survey data. The paper had to be submitted by a particular date and
the pressure was on to decide on the salient points to be addressed that would give a coherent picture of the
data. The paper got written and my data were no longer data but research findings. I now felt I had an idea
how I would approach the remainder of the data.
The process of developing an argument is often conveyed as an iterative process of moving between interpretation, data and theoretical explanations. This involves a continual search for evidence and counter-evidence
when devising explanations. Critically assessing your developing interpretations against your data evidence
and thinking about alternative explanations is a crucial aspect of this process.
Thinking about the claims you can make: issues of validity and reliability
When arriving at your overall research findings you will be engaging with complex issues of validity and reliability. A major debate in qualitative research fields has been recognition of the interpretative nature of research, and the need to provide evidence of how you arrived at your conclusions from the data you generated
via a transparent account of your research design, data collection and analysis and interpretation logics, including perhaps a reflexive account of the role of the researcher in the interpretative process. Making claims
in quantitative research also involves interpretation. Quantitative research can attempt to extrapolate patterns
from representative samples of wider populations. Emily Tanner, however, felt limited in the generalisation
claims she could make from her research approach, although by the end of her study she was also aware of
Page 13 of 15
Getting Your PhD
Sage
Sage Research Methods
Contact SAGE Publications at http://www.sagepub.com
the significant contribution her findings could make to her research area:
I’ve learned that it is very difficult to design a quantitative study based on new data that uses random probability sampling methods and has a sample size large enough to yield high quality reliable data. I did generate
results, but since my sample was based on convenience methods, I could not extrapolate my sample estimates to the wider population. Also I could not make reliable arguments about causation, only relationships
of association. To identify cause and effect requires a multiple stage design, which is often unfeasible within
the funding and time constraints of a doctoral thesis.
You will need to ask yourself about the possibilities of your research approach and analysis for the types of
claims you can make, issues on which the suggested reading below offers further deliberation.
Key Points to Remember
▸
You will need to think about your analysis approach during the research design phase and ensure a
logical ‘fit’ between your research questions, approach, data type and claims.
▸
Preparation for data analysis involves labelling, sorting, archiving, transcribing, anonymising data
and inputting data into software programs.
▸
Consider using software packages.
▸
Allow plenty of time for data analysis.
▸
Find out about your ethical and data protection responsibilities.
▸
Data analysis can involve coding, indexing and generating explanations.
▸
Page 14 of 15
Getting Your PhD
Sage
Sage Research Methods
Contact SAGE Publications at http://www.sagepub.com
Devising charts, tables and matrices may enhance your overview of your data and the relationships
between codes.
▸
Writing thematic papers may help develop your argument.
▸
Think about what type of claims your research design and approach allow for.
SUGGESTED READING
Cameron, D.(2001)Working with Spoken Discourse.London: Sage. Field, A. (2005)Discovering Statistics Using SPSS.London: Sage.
Fielding, J. and Gilbert, N.(2006)Understanding Social Statistics.London: Sage.
Fielding, N.G.andLee, R.M.(1998)Computer Analysis and Qualitative Research.London: Sage.
Mason, J.(2002)Qualitative Researching, 2nd edn.London: Sage.
https://doi.org/10.4135/9781849209229
Page 15 of 15
Getting Your PhD
J Vis (2019) 22:819–832
https://doi.org/10.1007/s12650-019-00569-2
R E G UL A R P A P E R
Xiaonan Luo • Yuan Yuan • Kaiyuan Zhang • Jiazhi Xia
Tianlong Gu
•
Zhiguang Zhou • Liang Chang •
Enhancing statistical charts: toward better data
visualization and analysis
Received: 18 November 2018 / Accepted: 8 March 2019 / Published online: 17 June 2019
Ó The Visualization Society of Japan 2019
Abstract Conventional statistical charts are widely used in visual analysis. With the development of digital
techniques, statistical charts are confronted with problems when data grow in scale and complexity.
Accordingly, a huge amount of effort has been paid on the enhancement of standard charts, making the
design space dramatically increased. It is cumbersome for naive users to choose appropriate design in a
specific analysis scenario. In this paper, we survey the enhancement techniques for a compact set of
statistical charts, and identify the types and usage scenarios. Motivated by the new problems, such as data
volume and complexity, we present a challenge-and-task-driven framework to guide the understanding of
the design space and the decision-making process.
Keywords Statistic charts Enhancements Design space
X. Luo Y. Yuan
Institute of Artificial Intelligence, Guilin University of Electronic Technology, Guilin, Guangxi, China
E-mail: [email protected]
Y. Yuan
E-mail: [email protected]
K. Zhang
State Key Lab of CAD&CG, Zhejiang University, Hangzhou, Zhejiang, China
E-mail: [email protected]
J. Xia
School of Computer Science and Engineering, Central South University, Changsha, Hunan, China
E-mail: [email protected]
Z. Zhou
Zhejiang University of Finance and Economics, Hangzhou, Zhejiang, China
E-mail: [email protected]
L. Chang (&)
School of Computer Science and Engineering, Guilin University of Electronic Technology, Guilin, Guangxi, China
E-mail: [email protected]
T. Gu
Guangxi Key Lab
