3 chapter3 methodology_andrea_gorra

Chapter 3 - Methodology (PhD thesis, Andrea Gorra)
Thesis title: An analysis of the relationship between individuals’ perceptions of
privacy and mobile phone location data - a grounded theory study.
Andrea Gorra, Leeds Metropolitan University, UK
Comments sent to a.gorra@leedsmet.ac.uk would be most appreciated.
Chapter 3 Research Methodology
This chapter introduces the research methodology used for this study and how it
has guided data collection, analysis and development of theory. Firstly, essential
background and fundamental guidelines common in different approaches to
grounded theory methodology (GTM) are provided. The subsequent three sections
describe the data collection phases for this study, which consisted of a mobile
phone tracking pilot study, in-depth interviews and a survey. The chapter concludes
by explicating the analysis approach for the empirical data.
3.1 Grounded theory methodology - an overview
"If someone wanted to know whether one drug is more effective than another, then
a double blind clinical trial would be more appropriate than grounded theory study.
However, if someone wanted to know what it was like to be a participant in a drug
study [..], then he or she might sensibly engage in a grounded theory project or
Page 86
some other type of qualitative study."
(Strauss and Corbin, 1998, p. 40).
Strauss and Corbin’s quote above encapsulates the essence of when it is best to
use grounded theory methodology for a research project. GTM provides useful tools
to learn about individuals’ perceptions and feelings regarding a particular subject
area. Quantitative data may be useful in measuring attitudes across a large sample,
however, GTM offers a powerful methodological framework if the aim of the study is
to learn about individuals’ perceptions.
GTM shares the following characteristics with other qualitative methods, which
correspond to those of this study:
• Focus on everyday life experiences
• Valuing participants' perspectives
• Enquiry as interactive process between researcher and respondents
• Primarily descriptive and relying on people's words
(Marshall and Rossman, 1999)

Grounded theory originated in the 1960s in the United States in the fields of health
and nursing studies. Barney Glaser and Anselm Strauss' influential book from 1967
‘The Discovery of Grounded Theory’ articulates the authors’ research strategies for
studies of patients dying in hospitals. Their successful collaborative study was
perceived as a response to the predominantly quantitative research paradigms at
the time. Grounded theory methodology advocates creating new theory consisting
of interrelated concepts rather than testing existing theories. A study guided by GTM
does not seek representativeness to achieve statistical generalisability but instead
aims to explain and sometimes predict phenomena based on empirical data. The
data collection typically encompasses in-depth interviews but can also include other
sources of data such as existing research literature and quantitative data. GTM
provides guidelines for data collection and analysis consisting of coding,
comparisons between data, memo writing and theoretical sampling.
3.1.1 Data collection and analysis in grounded theory
GTM uses a form of purposive sampling, known as theoretical sampling, where
participants are selected according to criteria specified by the researcher and based
on initial findings. Early analysis of data indicates issues that need exploration;
hence the sampling process is guided by the on-going theory development. Data
collection and analysis take
place in alternating
sequences (see Figure 3.1).
This can also be described
as an iterative cycle of
induction and deduction,
consisting of collection of
data and constant
comparison between results
and new findings in order to
guide further data collections
interviews
Starting
point
develop coding
theory
transcribing
memo
open writing
(axial) focused
(Strauss and Corbin, 1990; Miles and Huberman, 1994). For these reasons the
development and identification of variables does not take place prior to data
collection but instead as part of the data collection process. Consequently, the
variables or concepts are initiated by the interviewee and further developed and
conceptualised by the researcher. Data is collected until theoretical saturation is
Page 87
Figure 3.1: Steps in developing a grounded theory

reached, in other words until no new or relevant data emerges regarding a category
and relationships between categories are established (Strauss and Corbin, 1998).
Interview questions should give as little guidance as possible to allow the
interviewees to talk about what is of importance to them regarding a given context.
The researcher then needs to extract those phenomena or experiences significant
to the interviewee by assigning a conceptual label, known as a code. Several codes
can be grouped into more abstract categories which will eventually form the basis
for the developing theory.
3.1.1.1 Coding interviews as part of the analytic process
Interview coding is used to capture what is in the interview data, to learn how people
make sense of their experiences and act on them. Coding is the first step of data
analysis, as it helps to move away from particular statements to more abstract
interpretations of the interview data (Charmaz, 2006).
Grounded theory methodology advocates using several coding techniques to
examine interviewee’s accounts at different levels. Open coding, also known as
line-by-line coding, provides a good starting point to identify initial phenomena and
produce a list of themes of importance to the interviewee. Conceptual labels are
attached to almost every line in the interview transcript to capture what has been
said. These labels can correspond closely to the interview context and when taken
from the interviewee’s own words, are known as in vivo code. Codes are assigned
to participants’ words and statements to develop concepts, constituting the start of
the analytic process.
The detailed and meticulous process of line-by-line coding helps to open up the text
and interpret the transcript in new and unfamiliar ways which also helps test the
researcher’s assumptions. Strauss and Corbin (1998) suggest using initial or
‘sensitising questions’, to help the researcher grasp what the data might be
indicating. Suggested questions are "Who are the actors involved?", “What are the
actors’ definitions and meaning of these phenomena or situations?” (Strauss and
Corbin, 1998, p. 77).
The next coding phase is more abstract than open coding and known as focused
coding or selective coding. Focused codes are applied to several lines or
paragraphs in a transcript and require the researcher to choose the most telling
codes to represent the interviewee’s voice. Using open codes as a starting point,
Page 88

the process of focused coding helps to verify the adequacy of the initial concepts
developed. As the focused codes will be applied and therefore 'tested' on further
interview transcripts.
Another subsequent phase of coding is axial coding, defined by Strauss and Corbin
as "the act of relating categories to subcategories along the lines of their properties
and dimensions" (Strauss and Corbin, 1998, p. 123). The aim of axial coding is to
add depth and structure to existing categories (see also 3.1.1.2, below). Charmaz
(2006) explains that axial coding re-assembles data that has been broken up into
separate codes by line-by-line coding. Strauss and Corbin (1998) use axial coding
to investigate conditions of situations described in the interview, their actions and
consequences. Charmaz (2006) warns that axial coding applies a too rigid and
formal frame to the data analysis. Instead she recommends the less formalised
approach of reflecting on categories, sub-categories and to establish connecting
links between these to make sense of the interview data. The most abstract level of
coding is theoretical coding, which explores the relationships that have been
established between categories. Several ‘rules’ or “analytic coding families” are put
forward by Glaser (1978) to develop an advanced analysis of the subject area.
3.1.1.2 Developing categories
The general process of how to code an interview and develop a theory is depicted in
simplified form in Figure 3.2, below. After coding several interview transcripts a
researcher can identify many issues that are of importance to the respondents.
These issues are also known as phenomena and are assigned a conceptual label to
become a code, also known as a concept by Strauss and Corbin (1998). Some
codes or concepts will share the same or similar characteristics and can be pulled
together into more abstract categories, which can typically be interlinked and build
the basis for a theory.
Page 89

Interview
Phenomena Initial / focused
interlink categories
to build theory
It should be stressed that categories have to ‘earn’ their way into an emerging
theory (Glaser, 1978). Grounded theory methodology typically does not use
quantifying data to obtain meaning. However, counting the frequency with which
categories occur in interview transcripts can be useful to confirm their importance
for the interviewees. Categories can carry so-called properties and dimensions. A
property is a general or specific characteristic of a category, whereas a dimension
denotes the location of a property along a continuum or range (Strauss and Corbin,
1998). For example, the category 'phone ownership' could have the property
'spending' with dimensions ranging from 'low' to 'high spending'.
The central or core category is a distinctive category that sits at the heart of the
developed theory and summarises what is happening. All other major categories
should relate to the core category, which ought to appear frequently in the data
(Strauss and Corbin, 1998). The development of the core category for this study is
described in Chapter 4 - Presentation of Findings, section 4.2.4 to demonstrate the
development of categories from codes.
Coding shapes the analytic frame and provides the skeleton for the analysis
(Charmaz, 2006). Charmaz sees coding as an important link between collecting
data and developing theory but also as a connection between empirical reality and
the researcher's view of it. Coding highlights problems, issues, concerns and
matters of importance to those being studied. Strauss and Corbin (1998) refer to
categories as having 'analytic power', due to their potential to explain and predict.
‘Constant comparisons’ between collected data, codes, categories and initial
findings help to crystallise ideas to become part of the emerging theory.
Page 90
Theory
may have properties
and dimensions
share properties
for example event,
object, action or idea
codes
Open
coding
group similar
codes
Categories
add label
Figure 3.2: Coding steps in grounded theory (after Straus and Corbin, 1998)

3.1.2 Use of grounded theory methodology for this study
The data collection and analysis for this study followed a cyclical process typical for
GTM, by using early findings to shape the on-going data collection (see Figure 3.1).
The pilot study involved tracking participants’ mobile phones and conducting eight
interviews. This data collection phase was followed by five more in-depth interviews
that helped to explore issues raised in the pilot study. The survey aimed at exploring
the research area on a wider scale than would have been possible with interviews
being conducted and analysed by a sole researcher. More interviews were
undertaken after the completion of the survey and these could address issues
brought up by initial survey findings. Details about the sampling approaches for
each of the three data collection phases can be found in the relevant sections in this
chapter (3.2 Pilot study: Mobile phone location tracking , 3.3 Interviews, 3.4
Survey).
For this study, open coding was used for the pilot study interviews with the help of
qualitative analysis software NVivo. However, using the software did not produce
meaningful results and hence the pilot study interviews were re-coded with pen and
paper after conducting the survey (see section 4.1.4 False starts with initial coding
in NVivo). Focused coding was utilised for the next two interview phases, which
used the initial codes as a basis. Axial coding was not used in this study because
the method of specifying properties and dimensions for each category seemed too
prescriptive and did not help the analysis of the data (see section 4.2.6 Importance
of mobile phone settings). For the same reasons, theoretical coding was not
adopted. Instead, careful comparisons between respondents’ statements, as well as
between codes and categories were undertaken, without being restricted to interpret
participants’ words within a framework of properties and dimensions.
The researcher’s decision to use grounded theory methodology was only taken after
conducting the pilot study. An initial analysis of the pilot interviews showed that it
was not suitable to base the overall research project on existing theoretical models.
The focus needed to be on participants’ experiences and views of privacy and
mobile phone location data. Hence an inductive approach was chosen to explore
the subject area through the participants’ eyes. The decision to use grounded theory
methodology was further supported by the lack of existing theory regarding mobile
phone location data and privacy.
Page 91

Grounded theory scholars have different opinions about the most suitable time at
which to review the literature. Glaser (1978) advocates waiting to conduct the
literature review until initial findings have been made in order to not influence the
researcher with preconceived ideas. This study has followed the advice of Charmaz
(2006) and has carried out an initial review of the literature before the first data
collection in form of the pilot study took place. The reason for an early review of
literature was, on the one hand, to learn whether any similar research had already
been conducted in this area and, on the other hand, to satisfy requirements of the
university's research committee for the research proposal.
Page 92

3.1.3 Substantive and formal theory
A grounded theory is directly related to the data from which it has been generated; it
is therefore grounded in the data. Two types of theory are distinguished: substantive
and formal theory.
Substantive theories provide a theoretical interpretation or explanation for a
particular area, in other words this type of theory is used to explain and manage
problems in a specific setting. Formal theories, however, are more abstract and
provide a theoretical dealing of a generic issue which can be applied to a wider
range of disciplinary concerns and problems (Strauss and Corbin, 1998). For
example, a substantive theory can be about a limited area such as family
relationships or professional education while a formal theory might deal with the
construction of culture or the development of ideologies or stigma (Charmaz, 2006;
Glaser, 1994). Charmaz (2006) suggests combining and conceptualising the results
from several substantive grounded theories to develop a more general formal
theory. Each substantive theory can help to refine the formal theory, in other words,
a formal theory can relate to or 'cut across' several substantive ones. Charmaz
(2006) points out that most grounded theories are substantive theories as they
focus on particular problems in a specific, substantive area.
This study has developed a substantive theory as the collection of data and their
interpretation focus on the explanation of a particular area, that is the relationship
between mobile phone location data and individuals' perceptions of privacy in the
UK. This PhD thesis does not provide the scope to raise the very specific,
substantive theory to a formal theory that would be generalisable across a wider
area, such as additional types of digital data or mobile phone users from other
cultural backgrounds.
3.1.4 Memo writing
The process of coding and developing categories is supported by writing memos.
Memos are a set of notes, that kept continuously, support the researcher by
providing a record of thoughts and ideas. Memos enable the researcher to reflect on
the interviews and given codes to enter into a dialogue about the collected data.
Initial thoughts are of high relevance as they often spark the best ideas. Hence it is
important to write the memo immediately when reading and coding the interview. At
later stages in the research process, initial thoughts are represented through
Page 93

memos and can be revisited, reflected upon and considered for the overall analysis.
Memos can be used to ask questions, philosophise about potential meanings of
interviewee's statements and compare concepts identified in interview transcripts to
each other and to the literature.
3.1.5 Criteria for grounded theory studies
Charmaz (2006) gives the following criteria that grounded theory studies should aim
for. She highlights that a combination of credibility and originality enhances the
other two criteria resonance and usefulness.
a) Credibility
- Are there strong links between gathered data and argument?
- Are data sufficient to merit claims
- Do categories offer a wide range of empirical observations?
- Has the research provided enough evidence for the researcher's claims to
allow the reader to form an independent assessment?
b) Originality
- Do the categories offer new insights?
- What is the social and theoretical significance of this work?
- How does grounded theory challenge, extend, refine current ideas, concepts
and practices?
Page 94
c) Resonance
- Do categories portray fullness of the studied experience?
- Does the GT make sense to the participants?
- Does analysis offer them deeper insights about their lives and worlds?
d) Usefulness
- Can the analysis spark further research in other substantive areas?
- How does the work contribute to knowledge
- Does the analysis offer interpretations that people can use in their everyday
lives/worlds?
(Charmaz, 2006, p. 182)
Chapter 6, Conclusions and Recommendations, revisits these four criteria of
grounded theory research and addresses how each criterion has been met by this
study (section 6.4).

3.1.6 Objectivist and constructivist approaches to GTM
Grounded theory methodology has evolved since its inception in the 1960s in the
United States. Particularly, the writings of Glaser (such as 1967, 1978), Strauss and
Corbin (such as 1990, 1998) and Charmaz (e.g. 2000, 2006) are seen as influential
for the development of GTM. The original work of Glaser and Strauss from 1967,
‘The Discovery of Grounded Theory’, suggests that the researcher should start
collecting data with a 'blank mind', meaning without reviewing the existing literature
in order to carry out a truly inductive study. Consequently, theory is built from
observation and based on the understanding that the theory is already contained in
the data and only needs to be dug up or 'discovered', as Glaser and Strauss' book
title (1967) suggests. This perspective assumes that every individual will see and
understand the data from the same point of view, making the same observations
and therefore will come to similar conclusions. The researcher should take a
passive stance and 'let the data emerge', which can be seen as a characteristic of
an objectivist or positivist paradigm (Bryant, 2003; Charmaz, 2000). The alternative
view in social sciences is the so-called constructivist or interpretivist view.
Constructivist grounded theory methodology is for example advocated by Kathy
Charmaz in her book ‘Constructing Grounded Theory’ (Charmaz, 2006). This strand
of grounded theory methodology emphasises the research participants’ experience
and how they construct their view of reality. Knowledge, and hence the grounded
theory, are constructed by both researcher and research participant and aim at
interpreting the empirical evidence within the research context.
A divergence between the two authors of 'The Discovery of Grounded Theory'
occurred in the 1980s, after which Glaser postulated his understanding of grounded
theory methodology (Glaser, 1992). The late Strauss, however, together with his co-author
Page 95
Juliet Corbin, developed a different perspective on grounded theory
methodology (Corbin and Strauss, 1990; Strauss and Corbin, 1998). The major
difference between Glaser's and Strauss' views is Glaser's stance that "data
emerges" and thus presents the same picture of facts to every researcher in form of
some objective truth. Strauss' viewpoint on the other hand stresses that a
researcher has to actively obtain theory from data. Hence, it is likely and even
expected that each researcher will place the focus on different aspects of the
collected data depending on their background, beliefs and values. Stern (in Morse
ed., 1994) calls the former the Glaserian School and the latter Straussian School, as

both approaches to GTM differ in process and product and in her view represent
completely different methods.
Charmaz (2000) argues that both Glaser's and Strauss and Corbin’s approaches to
GTM assume an objective external reality and hence take a positivist and objectivist
stance. She, on the contrary, advocates a constructivist approach to GTM that
assumes multiple social realities. Charmaz does not support the view that theories
are discovered but believes that the studied world needs to be portrayed in an
interpretive way because interviewee and researcher embark together on the
process of constructing reality (Charmaz, 2006).
This study has been inspired and guided by Strauss and Corbin’s and Charmaz’
interpretation of grounded theory. The researcher disagrees with Glaser’s stance
that reality is objective and neutral, particularly regarding the intangible and
personal subject area of privacy.
3.1.7 Limitations of the grounded theory methodology
Grounded theory methodology has limitations like any other research methodology.
Some point out that GTM is very complex and time-consuming due to the tedious
coding process and memo writing as part of the analysis (Bartlett and Payne in
McKenzie et al., 1997). This study has dealt with the lengthy process of coding by
using specialised software to help speed up organisation and analysis of data.
Others name as a limitation that the use of GTM to explain, predict a phenomenon
or to build a theory is a very subjective process, which relies heavily on a
researcher’s abilities. This study has followed the methodological guidance of
Charmaz (2006) and Strauss and Corbin (1998) to gather and analyse the interview
data. In addition, findings from the survey and pilot study were used to strengthen
those findings based on interview data with the aim of fulfilling the criteria for GTM
as described in sections 3.1.2 and 3.1.5. Many studies make use of the term
grounded theory inappropriately and Bryant (2002) points out that the flexibility of
the method can be used to provide a justification for studies lacking in
methodological strength. Stern (in Morse ed., 1994) criticises some researchers for
mixing methods such as ethnography and phenomenology and then use the label
grounded theory to explain the analysis of their research findings.
The following three sections describe the data collection process for this study.
Details about the findings and their analysis will be provided in the following two
chapters, Chapter 4 and Chapter 5.
Page 96

3.2 Pilot study: Mobile phone location tracking
The pilot study explored technical aspects of mobile phone location data and
participants’ perceptions of privacy regarding this subject area. Four participants’
mobile phones were registered with an internet-based service provider that offered
identification of the phones’ geographical location 24 hours a day.
The aims of the pilot study were twofold: firstly to explore the technical side of
location tracking by locating the participants with help of the commercial tracking
service provider. Secondly, to learn about the participants' awareness of and
attitudes towards location tracking in relation to privacy. Potential disadvantages of
pilot studies were accepted and considered. These consisted primarily of the small
sample size which could not provide representative results and the danger of
making inaccurate predictions or assumption based on the pilot data.
3.2.1 Sample and ethical considerations
The participants were selected based on the researcher's judgement. This was due
to the potentially invasive nature of the research design and the explorative
characteristics of the pilot study. Two male and two female participants were chosen
for the study and it was ensured that all four UK GSM mobile phone networks (O2,
Orange, Vodafone and T-Mobile) were represented. It was a requirement of this
study that all phones were operating on the GSM network in order to be locatable by
any of the commercial mobile phone tracking providers. Out of the several internet-based
companies offering similar mobile phone tracking services, one company was
chosen for reasons of convenience and pricing. This service provider offered the
registration of five mobile phones for £5 plus VAT per month. The phones of four
participants and the researcher’s phone were registered. Due to the small sample
size and the limited age range of the participants (25 to 35 years), the study cannot
be seen as representative for the overall population. The time span of four weeks
was seen as a sufficient to generate a number of tracking results and to enable the
participants to experience location data in their daily routines.
This study was guided by the ethical principles on research with human participants
set out by Leeds Metropolitan University (Leeds Metropolitan University, 2006).
Particularly the researcher's ability to determine a participants' geographical location
24 hours per day over a period of four weeks was perceived as the most profound
ethical implication. The only possibility for a participant to not have his or her mobile
Page 97

phone’s geographical location identified would have been to either switch the phone
off or to terminate the participation in the research project. For these reasons, the
participants were informed about the study in great detail before asking for their
consent to take part. Aims and objectives of the pilot project, together with details
about the data collection process were explained to the participants. The
participants were informed that they could withdraw from the study at any time
without questions being asked. All data collected was anonymised by replacing the
participants' names with ascending code numbers (P116 - P119) in the order of the
initial interviews. Appendix A provides the email that has been sent out to the
participants before giving their definitive agreement to take part in the pilot study.
3.2.2 Data collection procedures
After all four participants agreed to take part in the study, their mobile phone
numbers were registered with the tracking service. The overall data collection had
two main aims. For one part, to compare the geographical data that was generated
by the location tracks to the participants' actual location. The second part aimed at
exploring participants' perceptions, feelings and opinions regarding privacy, mobile
technologies and particularly location data before and after the study. The data
collection process is shown in Figure 3.3, below. 'Open track' denotes a location
request known to the participant at the time of tracking and 'covert track' indicates a
location request without the participant's knowledge.
Page 98
Initial Interview
Tracking
Final Interview
Week 1: 'Open track’ – weekday, evening
Week 2: 'Covert track’ – weekend, daytime
Week 3: 'Covert track’ – weekday, morning
Week 4: 'Covert track’ – weekday, evening
Data Analysis:
'Open track’ during interview
- tracking data (quantitative)
- interviews (qualitative)
Figure 3.3: Data collection process for pilot study

3.2.2.1 Part one of pilot study: Location tracking of participants’ mobile
Figure 3.4: Example of location track
Page 99
phones
A weekly location request was submitted for each participant over a period of four
weeks and the results were stored
electronically by the researcher. The
participants were not contacted during
this time to avoid influencing their
tracking experience. The research
participants could not know when a
request for their mobile phone’s
geographical location was submitted to
the tracking service provider.
Figure 3.4 shows an example of a
location request in form of a map depicting the participants' geographical location at
the time of tracking. In order to be able to compare the participants' location
provided by the service with their actual location, it was suggested to the
participants to keep a location diary. To make it easier for the participants to recall
their location at the time of tracking, the phones’ locations were requested at
different weekdays and times (see Figure 3.3). Immediately before the first location
request only, all participants were contacted on their mobile phone. This phone call
helped to familiarise them with the study and to obtain a first valid actual location.
The participants were unaware of the three subsequent tracking requests. Hence,
the participants’ actual location had to be compared to the one supplied by the
tracking service based on diary entries and memory in the final interview. A fifth
track was conducted in the researcher's office as part of the final interview to
demonstrate the service to the participants and to be able to compare the accuracy
of tracking results between the four different mobile phone service providers.
Mobile phone tracking served two purposes, which tied in with the overall aims of
this study. Firstly, mobile phone location tracking allowed comparison of the
participants' location derived from location data with their actual whereabouts at
time of tracking. The main purpose for this was to evaluate the service regarding
reliability, validity and accuracy. Results from the location requests aimed at
bringing location data into context with claims of privacy invasion through retention
of location data. The researcher believes that police and emergency services have
access to more detailed geographical data than a commercial tracking service
provider. This assumption is based on media reports. However, using a commercial

service provider was seen as the only feasible way to get access to location data
and to visualise this hidden aspect of mobile communication. The second purpose
of location tracking was to raise the phone users' awareness of this type of data.
The 'visualisation' of mobile phone location data by using a tracking provider proved
to be a useful tool to enrich the qualitative data collection both for the participants
and the researcher.
3.2.2.2 Part two of pilot study: Interviews
Respondents' perceptions and beliefs are at the heart of qualitative research and
this was the main motivation for complementing the location requests with
interviews. Semi-structured interviews were conducted with all four participants
shortly before and after the four week tracking period to learn about their opinions
and perceptions regarding mobile phone location data and privacy. The interview
questions were of exploratory nature due to the small scale of this study and early
stage in the overall research project. The questions were designed to identify
patterns and common themes in the participants' accounts and sought to identify
the meaning of privacy in relation to mobile phone location data in the participants'
everyday lives.
During the interviews it was important not to restrain the participants but to give
them time to talk about how they understood and described their experience of
mobile phone tracking. This was particularly important as privacy is such an elusive
concept, that can be perceived differently by different people. Figure 3.5, below,
gives an overview of the interview themes and Appendix C - Interview questions for
pilot study, provides a detailed list of interview questions.
Tracking
Page 100
Awareness
of location
Technology
Legal
Framework
Control
over location
data
Access
Dissemination
Concern
Trust
Misuse
In government
In service
provider
Retention
Geographical
Location Participants'
definition
Nothing to worry,
nothing to hide?
Trade for
personal gain?
Privacy
Figure 3.5: Graphical representation of interview themes

The interviews that took place prior to the tracking period focused on participants'
awareness and knowledge of location data, as well as their descriptions and
thoughts about privacy. Responses to these initial interviews have helped to
develop the questions for the second set of pilot study interviews which was
conducted soon after the four week tracking period. The final interviews aimed at
learning about participants' experiences of taking part in the tracking study and any
thoughts or potential concerns regarding privacy.
The pilot study proved useful to familiarise the researcher and the participants with
the concept of location data. The pilot interviews provided valuable leads to be
pursued later on in the research project in more in-depth interviews.
3.3 Interviews
The interviews for this study were conducted in three phases: the first phase
consisted of eight explorative interviews for the pilot study and the second phase
consisted of five interviews, after which the survey was initiated. The reasons for
conducting the survey at this point was for one part to allow time to reflect on how to
approach the next set of interviews and for the other part to learn about individuals’
awareness and opinions regarding location data on a greater scale than it would
have been possible with interviews. The third and last interview phase took place
after the paper-based survey was completed.
Interviews can provide insights that are not available to researchers working with
large survey samples and are known to be the most suitable approach when
seeking rich data illuminating individuals’ experiences and attitudes. Drawbacks are
that interviews are very time-consuming to conduct and analyse. The interview
questions for interview phases two and three were based on findings from the pilot
study. The questions were asked in as non-directive a manner as possible to meet
the study’s principal aim of learning about the interviewees' perceptions. The data
collection and analysis for this project took place in alternating sequences and was
guided by the grounded theory methodology. This meant that the interviews were
transcribed and coded immediately after they took place. Hence, initial findings from
interview coding could help to shape the questions for subsequent interviews.
Page 101

3.3.1 Sample design and ethical considerations
For the second and third interview phase, the participants were initially contacted
informally, which was followed up by an email explaining the study’s aims and the
interview procedure. The email ensured participants about anonymity and
confidentiality of data collected and informed them that the interview was recorded
for transcription (see Appendix B - Email sent out prior to interviews). The interviews
were recorded with a digital voice recorder and the files transferred to a PC for
transcription. When transcribing the interviews, participants’ names were replaced
with code numbers. Participants of the pilot study were assigned the codes P116_F,
P117_M, P118_F and P119_M. "M" indicates males and "F" stands for female. The
codes for respondents of the second and third interview phase followed the same
system, starting with number P131_F. In addition to the eight pilot study interviews,
ten more in-depth interviews were conducted which were between 30 and 90
minutes in duration.
Respondents were selected based on initial findings. The pilot study had for
example highlighted different types of mobile phone users and it seemed as if
different types of users utilise their phone settings in different ways to regulate how
others can get in touch with them (see section 4.2.8). Hence it was sought to
interview male and female mobile phone users from different age groups and with
different mobile phone contracts. A list of respondents, including demographic
characteristics, can be found in Appendix F - List of Interviewees.
3.3.2 Interview preparation
The information sent out before the interview included some technical and legal
background about mobile phone location data. The email did not make any
references to potential impacts of data retention on civil liberties or privacy so as not
to influence the respondents’ views. The aim of sending out information prior to the
interviews was to familiarise the participants with the subject area. This seemed
necessary because early survey findings showed that the majority of people had
either not heard of location data or were not sure about the details.
Page 102

3.3.3 Development of interview questions
The interview questions were based on initial findings from pilot study and survey
and the following themes were addressed (see Table 3.1, below). Appendices D
and E provide a detailed list of interview questions.
Table 3.1: Themes of interview questions for second and third interview phases
Page 103
Theme Notes about question
Mobile Phone Mobile phone usage and habits in everyday life situations.
Privacy An open question as to what privacy means to the interviewee.
Deliberately trying to avoid mentioning any categories of privacy
known from the literature.
CCTV, Loyalty
Cards
Do interviewees perceive these as privacy invasive?
Phone as
tracking device
Referring to technical and legal facts of mobile phone location
data that were sent out prior to the interview.
- Is the interviewee aware of location data before this study?
- Are any issues or concerns raised?
Government &
mobile phone
service provider
- What is the participants’ opinion towards storage of
communications data (and its availability to government)?
- What else is mentioned in this context? Regarding government,
regarding data retention?
In addition to these themes, participants mentioned terrorist attacks, current politics
and ID cards.
The interview questions changed and improved over time, influenced by codes and
categories developed for previous interviews. For this reason, two sets of interview
questions are shown in the appendices: Appendix D - Interview questions March
2005 and Appendix E - Interview questions December 2005. The second set of
questions was guided by Kathy Charmaz’ approach to GTM, particularly her chapter
on how to phrase interview questions to allow respondents to express their views
without constraints (Charmaz, 2006). For example the question 'Have you ever
heard of mobile phone location data before taking part in this study?' was changed
to 'What do you know about mobile phone location data?'. A question mentioning
consumer loyalty cards and the storage of personal information by commercial
companies was removed after the second interview phase. The reason for this was

that initial interview findings and the survey showed that respondents did not
perceive this area as related to privacy.
3.3.4 Use of software for data management and analysis
Weitzman (2000) advocates the following benefits of using software in research
projects: writing up, editing, coding, storage, search and retrieval, data 'linking',
memoing, content analysis, data display and graphic mapping. However, he warns
of 'false hopes and fears', pointing out that no software will be able to actually carry
out the analysis process for the researcher. Software can support the research
process but ideas and intellectual efforts have to come from the human being
conducting the research and analysis.
For this study a word processor, Microsoft Word, and the qualitative analysis
software NVivo were used to support the analysis and to help manage the interview
data. The computer was utilised as efficiently as possible to reduce the amount of
time spent on organising data and findings, to increase the speed of tiresome tasks,
resorting the material and redefining codes. Qualitative analysis software facilitated
following potentially promising analytic routes but also enabled these routes to be
discontinued with ease. Dynamic and real-time representation of the findings
considerably assisted reflection on data and connections between the data.
3.3.4.1 Transcribing interviews and importing into NVivo
The use of the NVivo software has significantly facilitated the process of organising,
re-arranging and managing the considerable amount of data. For example, after
coding the interviews in NVivo, all passages assigned to a specific code could be
viewed on screen and printed. In the same manner, searches for specific text
strings could be conducted across all interviews and relevant paragraphs containing
the search string could be compared on screen or printed. The interview transcripts
were formatted in a particular way in Microsoft Word to facilitate importing the
transcripts into NVivo. This meant for example, that the interview questions were
assigned a 'Heading 1' format. When importing the transcript into NVivo, this
resulted in the questions being displayed in the content panel in the NVivo explorer.
Hence when selecting a question, it was possible to jump to this section in the
interview transcript. Furthermore, meaningful information about the interview was
placed into the first two paragraphs of each transcript. This enabled this information
to be automatically put into the properties of the interview document when importing
the interview into NVivo.
Page 104

The appropriate formatting of the interview transcripts from the beginning helped to
organise the data efficiently and thus facilitated the analysis of the interviews. In
addition, by using the NVivo software much tedious and time-consuming work for
managing and resorting the data could be avoided which freed time for more
meaningful tasks, such as analysis and interviewing. Different sets of interviews
could be assigned with different colours in NVivo for easy distinction.
An advantageous feature of NVivo is that the software keeps a log of all data that
has been entered, which means that all codes and memos are automatically
assigned a date and time stamp. This feature helped to trace the development of
codes. After coding the interviews in NVivo, all passages assigned to a specific
code could be viewed on screen and printed. In the same manner, searches for
specific text strings could be conducted across all interview sets.
3.3.5 Interview coding
The first set of interview transcripts were coded with the help of the qualitative
analysis software NVivo, which however did not prove to be the most beneficial
approach (see section 4.1.6.1 Some initial concepts not confirmed). After setting
aside several months for the survey, a fresh and more suitable approach to the
analysis of interviews was adopted. This time interview codes were not devised on
screen with help of software but with pen and paper. Following the GTM guidance
on coding (see section 3.1.1.1), the researcher worked through each of the
transcripts and used line-by-line coding to take note of themes and phenomena on
the margins. The codes were not devised strictly microscopically and some more
abstract categories came into view; some codes were very close to the
interviewee’s accounts and others more abstract or conceptual. Keywords and
phrases were noted on differently coloured post-it notes and stuck onto a blank A2
flip chart sheet (see Appendix I). The post-it-notes were arranged in a logical order
on the paper sheet. As more and more interviews were coded, this sheet started
looking less like a random collection of labelled post-it notes but more like a brain
storm map or a tree where branches of thought grew from certain categories.
Memos were written throughout this exercise to keep track of thoughts and ideas
regarding the data analysis.
This system of creating codes, combined with reflection was maintained for coding
all interviews. At the end of this coding exercise, three A2 sheets were covered with
post-it notes containing categories and codes. In addition, a matrix had been
Page 105

devised, which held codes, properties and dimensions, as well as some comments
and quotations. This list of codes was revised continuously as more interviews were
coded. The codes were modified and verified by being applied to further interview
transcripts but stayed alike for the most part. Subsequently, the codes were keyed
into the NVivo software to allow searching the interviews, re-sorting of material and
consistent redefining of codes in order to support the analysis process.
Figure 3.6, below, shows a summary of the coding process. More details about this
analysis process are discussed in the subsequent findings and analysis chapters.
Appendix J provides a summary of the development of codes.
Page 106
Codes on Post-it notes Matrix with properties &
dimensions in MS Word
(Re-)apply codes to
interviews in NVivo
Final codes in NVivo
Categories and initial
theoretical framework
Writing Memos
‘
‘ = when developing the final
categories and links between them, the
writing of memos became particularly
important.
Figure 3.6: Graphical representation of coding process

3.3.6 Memoing to develop and clarify categories
The extended coding process, as described in the previous sections, has facilitated
reflections on codes and categories, which were captured by writing memos (for
examples of memos see next chapter, section 4.2.2 Situational map as a memo to
consider wider social context). The memos were consulted when establishing links
between categories and setting up the initial theoretical framework. The following
chapter details how initial codes have developed into focused codes and finally to
abstract categories (section 4.2 Development of categories based on interview
data). The writing of memos ('memoing') was particularly useful as this helped the
researcher to keep a note of thoughts without the pressures of having to
immediately determine how ideas fitted within the overall research findings and
analysis. Memoing allowed the freedom to jot down ideas so that these could be
sorted, categorised or discarded at a later point in time. The writing and reflecting on
memos has been a crucial step in the development of the final categories based on
initial and focused codes. Some examples of memos and how they have influenced
the analysis process are shown in the next chapter (for example section 4.2.4
Memo ‘Balancing security and privacy’).
3.4 Survey
A survey was conducted in form of a questionnaire and distributed after the pilot
study and some further interviews took place. The survey complemented findings
from the interviews and helped to obtain a better idea of individuals' opinions
towards privacy and mobile phone location data. Additionally, the survey highlighted
interesting cases which could be approached in further data collection stages and
thus supported the purposive sampling approach of interviews. It could be
understood as a weak point of questionnaire-based surveys that they only capture
surface opinions, seeing that respondents will not necessarily report their beliefs
and attitudes accurately. These are easier to identify in interviews, as also prompts
can be used. In addition, the use of mainly closed questions in a questionnaire
merely allows respondents to choose between a limited number of responses
(Robson, 2002). However, some of the respondents of this survey have overcome
this limitation by making comments on the questionnaire’s margins to express
further opinions and thoughts. A significant advantage of using a survey for this
study was to collect larger amounts of data in a shorter time scale than would have
been possible with interviews.
Page 107

3.4.1 Sample design and ethical considerations
It is hardly ever possible to survey the entire population to be studied and for this
reason sampling techniques need to be employed. A representative sample
produces results which can be used to formulate generalisations. However, this can
only be achieved by using probability sampling where the likelihood of the sampling
unit - in this case an individual - to be included in the sample is known. The
sampling population for this research project consisted of British residents older
than 15 years, as the survey was not designed to explore the views of children. A
representative sample was not chosen as it had not been possible to obtain a large
enough random sample due to time and monetary constraints. For these reasons
purposive sampling, a form of non-probability sampling, was used. Participants were
not selected randomly but judged to be of interest to the researcher, which should
not be understood as a limitation since the survey was designed as an explorative
study. In addition, this approach tied in with the sampling procedures common in
studies using grounded theory methodology (see section 3.1.1).
The questionnaire was distributed to individuals of different age groups, diverse
professional backgrounds and users with different types of mobile phone contracts.
The sample size of 477 respondents was deemed to be sufficient for an explorative
survey and provided an overview of participants’ opinions at the time of data
collection. Statistical measures such as confidence intervals could not be produced
as the sample was not a probability sample and the survey did not have a defined
population. Ethical guidelines were carefully followed when collecting the survey
data: all data were treated confidentially, which was explained to the respondents
together with the aims of the study. Respondents were free to fill in their name on
the last page of the questionnaire if they were interested in taking part in a follow-up
interview. The section revealing individuals’ names could be handed in separately
from the questionnaire responses to ensure anonymity.
Page 108

3.4.2 Pilot testing of survey
The questionnaire was thoroughly pilot tested before dissemination to uncover flaws
and potential causes of confusion, such as misleading questions that could
potentially result in invalid responses. For the pilot test 15 questionnaires were
distributed to friends and other well-known contacts to verify feasibility and
compliance with objectives set out by the overall study. The following steps were
undertaken to pilot test the questionnaire:
1) Develop questions
The survey questions were based on interview questions from the pilot study.
Five questionnaires were distributed to friends and colleagues, who were
informed that the questions were developed for a survey. Their comments and
feedback was used to modify some of the questions.
2) Pilot testing
Ten revised questionnaires were distributed to other friends and colleagues
and their feedback helped to develop the final version. The number of
questions was reduced from 23 to 20 and the layout was condensed, so that
the number of pages could be reduced to four instead of seven pages.
3) Pilot analysis
Frequency tables for the responses to each question were generated to obtain
an early impression of the results. This trial analysis ensured that the survey
would fulfil the survey's aims and yield data valuable to the overall study.
3.4.3 Changes undertaken after successful pilot testing
After the initial analysis of pilot data, some changes in questionnaire design and
coding were undertaken. The most significant change was to amend the answer
choices of question 15 “Have you heard of mobile phone location data before taking
part in this study?” to:
- I have heard of location data from the media and other sources, but I am not
exactly sure about the details
- Yes, I know what location data is and how it works. I could explain it to a friend
- No, I have never heard of location data
The reason for this change was that the previous answer choices were not
meaningful enough, as the questionnaire merely asked whether people had 'heard
of' the data and if yes from which sources. However, it was acknowledged that it
would be more important to learn how much respondents would consider
themselves to know about the subject. The drawback of these answer choices was
Page 109

that respondents had to use their own judgement. Nevertheless, it was not seen as
appropriate to examine participants' knowledge, since it was more important to gain
respondents’ trust to enter into a dialogue. An additional change regarding question
15 concerned the layout. This significant question was moved to the bottom of the
page, so that the box providing explanations about location data would only appear
on the subsequent page of the questionnaire. Hence, this forced respondents to
make a judgement of their awareness of location data before being presented with
further background information on the subject area. Appendix L contains the final
version of the questionnaire.
3.4.4 Distribution of questionnaires
The survey was disseminated in two formats, paper-based and online via the
internet; the majority of questionnaires were distributed paper-based. Initially the
survey was given to individuals working and studying at Leeds Metropolitan
University, ranging from respondents with an academic background (lecturers, post-and
undergraduate students) to administrative, cleaning and security staff. Friends,
personal contacts from various leisure activities and colleagues from part-time jobs
were also invited to fill in the questionnaire. In addition, acquaintances working in
different professions (landscape architect, engineer, manager in a bus company)
were asked to distribute approximately 10 questionnaires each at their workplace,
sports club or family. This distribution method could be seen as a form of snowball
sampling, which ensured access to social groups beyond the researcher's
immediate social circle and made it possible to collect data across a diverse
population. A known limitation of the sample for this study is that unemployed,
unskilled or manual workers were only included marginally.
It was acknowledged that researcher's bias could have occurred when selecting
participants. Response rate and quality of survey data was very high due to the
questionnaire distribution on a personal basis. An additional benefit of this
approach was that respondents could receive help with completing the
questionnaire if necessary. Only a few people refused to fill in the questionnaire,
mainly due to time constraints, and one person did not want to participate as she
perceived questionnaires as an invasion of her privacy. It is acknowledged that due
to the chosen non-probability sampling method, the number of non-respondents and
their reasons for not taking part in the study could not be known.
Page 110

3.4.5 Online survey
In addition to the paper-based version, an online version of the questionnaire was
published on the internet on a website developed by the researcher
(http://www.locationprivacy.org). The survey questions were identical for both
distribution media so that the results could be pooled. An online survey was seen as
a convenient way to collect additional data with minimum involvement of the
researcher in data collection and analysis. Results of the survey were automatically
gathered in a database and could be downloaded in form of tabular data which then
could be imported into Ms Excel. Advantages of electronic surveys include being
able to direct respondents to particular sections of the questionnaire, depending on
the way they have answered previous questions. Respondents can be automatically
prompted when they provide an invalid response, such as selecting several tick
boxes when only one should be marked. However, electronic surveys also have
distinctive technological, demographic and response characteristics which were to
be taken into consideration for this study. For example, only individuals with
computer and internet access were able to fill in the survey. To account for these
differences data from the electronic survey was stored and analysed separately
from the data collected by the paper-based version. However, a comparison
between the responses from both distribution media showed that results were
similar (see section 4.3.1).
3.4.6 Email as distribution medium
Email was planned to be used as a distribution medium for the survey but discarded
due to particular problems to this approach. Two alternative methods for using email
to distribute surveys were considered; firstly, to send the survey as an attached
word document, and secondly to include the questions in the email body. However,
for the first option it would have been necessary for the respondent to download the
attachment, complete and save it and then re-attach it to the email to send it back to
the researcher. This is a rather cumbersome process and if any of these steps is not
followed correctly, the survey is returned blank. The second option would not have
allowed the researcher to control how the questions are displayed, as this depends
on the respondent’s email programme. For these reasons the use of email was
restricted to distributing the questionnaire’s website link and to contact respondents.
Page 111

3.4.7 Question design and coding
Questions were worded carefully and avoided long and ambiguous, leading, biased
questions, as well as jargon. The following question types of questions were used:
Numerical rating
A Likert scale from 1 - 6 was used. The reason for using an even number of
options was to not allow non-committal answers. In the data analysis, each
statement was assigned a separate variable.
8. How important is your phone to you on a scale from 1 to 6?
Page 112
My phone is VERY IMPORTANT to me.
It would be difficult to live without it.
1 2 3 4 5 6 NOT IMPORTANT at all to me.
I could live without it.
Multiple choice
Require respondents to choose just one response from a list of alternatives.
Only one variable per question was necessary in the data analysis.
14. Giving up some privacy is necessary to fight terrorism and crime
Agree Disagree Don't know
Check lists
List a set of items of which respondents select those that apply. Each statement
was assigned a separate variable in the data analysis.
18. In your opinion, what would be good uses for location data? [Select all that are appropriate]
Solving crime Employers to track their employees
Preventing acts of terrorism Emergency services
For parents to track their children (e.g. rescue, ambulance, fire brigade)
Open questions
Open questions allow respondents to formulate their own statements and can
lead to unexpected responses. Nevertheless, open questions are more difficult
to analyse and take respondents longer to respond. For this reason only one
question of this type was included.
9. What comes to mind when you think about the term 'privacy'? [Keywords are fine]
---------------------------------------------------------------------------------------------------------------------
---------------------------------------------------------------------------------------------------------------------

Non-committal responses
The survey questions raised many issues which respondents might not have
thought about before taking part in the study or might not hold an opinion. Hence,
alternative responses such as 'don't know' or 'no opinion' were provided where
appropriate. It would create false and unreliable answers to force the respondents to
express an opinion. Some see the danger in providing these options that some
respondents might select them out of laziness (Converse and Presser, 1986 in
DeVaus, 2002). Therefore, alternative responses were always the last one in a list
of options (see Appendix L).
3.4.8 Coding missing responses
In every survey there will be missing data occurring for different reasons, which
needs to be recorded and coded in a similar way to valid responses. In this study,
different codes were given to different types of missing data. The following types of
non-responses were identified:
1) The respondent has not responded to a question. The reason might have been
by choice or simple overlooking of the question. The code given was ‘9’.
2) The respondent was not required to answer this question. After responding to
the question ‘Do you own a mobile phone?’ individuals who did not own a phone
would proceed straight to Section B ‘Privacy’. The missing responses in Section A
‘Mobile Phone’ was coded with an ‘8’ to distinguish those from other missed
responses (see above).
3) Response was invalid, e.g. where only one answer was required but the
respondent ticked several responses. However, this was only the case with one
questionnaire and therefore in this case was assigned a ‘9’.
Page 113

3.4.9 Data entry
The survey data was firstly entered into an Excel spreadsheet and then imported
into the statistical analysis software SPSS (Statistical Package for Social Scientists).
The reason for using two different software packages was that Excel would
automatically update graphs showing the frequencies of responses as survey data
was entered. This gave early impressions and ideas about the data and helped
reflecting on findings at an early stage. Another reason for entering the data first into
Excel was that there was no character limit for variables, which was convenient for
open question and comments. Afterwards, data was imported into SPSS to facilitate
advanced analysis of data, such as correlations between variables. In addition,
SPSS could help to prevent false entries, as data ranges could be set, so that for
example no number greater than 6 could be entered for a question.
3.5 Analysis of findings
Data collected by pilot study and survey were compared to the grounded theory
categories identified in the interviews in order to support the analysis of findings.
This process of triangulation between qualitative and quantitative data was used to
confirm and validate the findings.
The interplay between induction and deduction, in other words between data
collection and interpretation, was another process of validation of findings but also
part of the theory development (see section 3.1.1). In what is known as a process of
abduction, the interpretation of observed data to the best explanation helps to form
a tentative theory, which then needs to be confirmed or disconfirmed with help of
further data collections and analysis. This procedure is repeated until the best, the
most plausible interpretation of data is found (Charmaz, 2006; Haig, 1995). The
findings from empirical data were compared to the reviewed literature, as described
in Chapter 5, which then lead to conclusions and recommendations (Chapter 6).
Page 114

3.6 Chapter summary and conclusion
This chapter has introduced and discussed the choice of grounded theory
methodology as a suitable research methodology for this study. The continuous
generation and long-term retention of mobile communications data are inherently
related to the every day transactions of every mobile phone user. The views of
mobile phone users are necessary to be captured as information about their lives is
at the centre of debates around data retention.
A grounded theory has been developed to provide an explanation for the
phenomenon under study: the relationship between mobile phone location data and
individuals' perceptions of privacy in the UK. The theory can be categorised as
‘substantive’ as opposing to a ‘formal’ theory, since the collection of data and their
interpretation focus on the explanation of a particular area. The chapter has in great
detail explained each of the three data collection phases, including sampling and
ethical considerations. A mobile phone location tracking pilot study, a survey and a
series of interviews have been conducted to learn about individuals’ perceptions of
privacy in relation to location data. The iterative cycle of data collection and analysis
is an essential element of GTM and has helped to shape the on-going data
collection.
The following chapter, Chapter 4 - Presentation of Findings, discusses in detail the
findings of all three data collection phases, pilot study, interviews and survey.
Page 115

3 chapter3 methodology_andrea_gorra

More Related Content

What's hot (20)

Viewers also liked (14)

Similar to 3 chapter3 methodology_andrea_gorra (20)

Recently uploaded (20)

3 chapter3 methodology_andrea_gorra