Chapter 3: Methodology 98
4. The methods 125
4.4 Interaction 3: Relationships with others: a tool for investigating attachment and
The third Interaction comprised the completion of an online questionnaire that had been adapted and piloted during my previous Masters-‐level project (Levine and Edwards, 2014).
The questionnaire was based on the Adolescent Attachment Questionnaire (AAQ) (West et al., 1998) (see Appendix 2 for the adapted tool and introductory
text). I chose this tool because:
-‐ It has been tested within the field of psychology and is relatively robust (although there are inevitably issues with this that manifest more seriously in a conventional survey setting than in the study reported here).
-‐ It is relatively brief, and given the time commitment asked from
participants over the data gathering period, I wanted to keep each tool to a focused, core minimum of questions.
The AAQ gathers data against three attachment-‐related parameters: ‘Angry Distress’, ‘Availability’ and ‘Goal-‐Corrected Partnership’. The Angry Distress scale measures levels of anger towards the parent. For example:
My parent only seems to notice me when I am angry. I often feel angry with my parent without knowing why.
I get annoyed at my parent because it seems I have to demand his/her caring support.
The Availability scale measures the extent to which the respondent believes that the attachment figure is available to them, and vice versa. For example:
I’m confident that my parent will listen to me.
I’m confident that my parent will try to understand my feelings. I talk things over with my parent.
The Goal-‐Corrected Partnership scale measures the extent to which the
respondent reflects on others’ perspectives, differentiates those perspectives from their own, and works with others to construct the relationship. For example: I enjoy helping my parent whenever I can.
I feel for my parent when he/she is upset.
It makes me feel good to be able to do things for my parent.
My adaptation was to include parameters for online and offline friendships, thereby providing a technological and non-‐technological comparator for each participant, as well as an insight into their relationships with each parent. For
example, the opening Angry Distress question became four options: My mother only seems to notice me when I am angry.
My father only seems to notice me when I am angry.
My closest offline friend only seems to notice me when I am angry. My closest online friend only seems to notice me when I am angry.
Each dimension contained three paired sets of questions. Each question was rated on a five-‐point Likert-‐type scale. Goal-‐Corrected Partnership and Availability were reverse scored. Scores are combined to provide a sub-‐score for each
attachment for the dimension (West et al., 1998, Levine and Edwards, 2014).
I used Kwiksurveys to deliver the questionnaire. Kwiksurveys was chosen for its ease of use and clean layout options. As with my Masters-‐level work, the questions were grouped in pairs, with linked questions sharing a page.
Although the AAQ is a questionnaire that can be run in large samples, and produces a quantifiable result, I used it in a new way in this study. Given that these sorts of tools can be used in clinical psychology to evaluate clients, it did not seem too outlandish to use a survey research tool in a more intimate way. Thus, instead
of using a survey to gather data across a large sample, the survey became the beginning of an exchange about relationships with each participant, a source of discussion during the interview, and a way of looking at relationships in the analysis stage. I piloted this approach during my Masters study and found it to be both conducive to a pragmatic epistemology (i.e. outside the qualitative/
quantitative paradigm), and was also a rich source of ensuing email exchange (Levine and Edwards, 2014). As a result, I used the same tool in the same way in
this study. Limitations to using surveys in this way are outlined in Table 9 below.
Limitation Mitigation
Although I could use the tool to explore participants’ attachment types, I could say almost nothing about
why those types resonated for them, whether the types were stable or in flux, or
how they came about.
I used the survey as the beginning of an email exchange to explore participants’ perspectives of these contextual matters, and for discussion during the interview.
Surveys cannot easily demonstrate trends over time.
As with all of the Interactions described in this study, if I could have found a way of gathering continuous, or at least periodic data on all of the socio-‐cognitive parameters throughout the entire data gathering period, without being too intrusive on participants and their families, that would have been a wonderful opportunity to explore the stability of these parameters over time. It was, however, outside the scope of this initial study to do so.
Participants may not have felt ready to provide open or accurate responses to the survey. In addition, because the survey was delivered online for all bar the youngest of the
participants, there was no way of knowing whether a friend or parent had been present during the
Interaction completion, with implications for the replicability of the data.
Asking participants to respond on sensitive matters such as their relationships with parents and friends raises issues around gatekeepers, access and
accuracy of response (Cohen, Manion and Morrison, 2007). This is compounded when responses are completed online, outside of the immediate control of the researcher.
I addressed this issue in two ways. Firstly, I made it clear at the consent stage that we would be
covering some potentially problematic areas, and communicated the process for dealing with disclosure issues clearly (see Appendix 8). I also explained to participants that I would protect their anonymity outside of the disclosure sphere. I repeated this at the start of the Interaction. Secondly, at the coding stage, I introduced an ‘accuracy’ code that would flag up any possible discrepancies, that might suggest social, environmental or parental pressure across the initial survey, email exchanges and during the interview. No issues were raised.
However, if I were to run the survey again in any context, I would include a question that asks whether the respondent is completing the survey in private, with someone else in the room, or sitting together with someone else.
Table 9: Limitations of using an established survey tool in a qualitative context
The AAQ served a useful purpose in opening a discussion about
consideration of those relationships. However, it could not stand alone – it needed
the additional email exchanges and discussion during the interview to become a whole greater than the sum of its parts.
4.5 Interaction 4: Strange Stories: a tool for investigating theory of