This section details the methods that were used to investigate the research hypotheses as outlined in Chapter One. Firstly, the design of the study is reported. Secondly, the characteristics of the participants and the recruitment methods are outlined. The measures and their psychometric properties are then discussed and consideration is given to the relevant ethical issues for this study. Finally, the procedure for conducting the research is described and the intended statistical methods for data analysis are introduced.
2.2 Design
Primarily, this study employed an experimental mixed within subjects design, investigating mental health nursing students’ attributions and emotional reactions, and the effect of these on intended behaviour, when presented with an individual labelled with personality disorder or schizophrenia.
To investigate hypotheses one to three, an experimental mixed within subjects design was used. Participants viewed a total of two videos: one depicting antisocial behaviour and the other depicting prosocial behaviour. It was felt that it was important to portray both types of behaviour as the type of behaviour with which an individual presents may in itself influence attributions and emotions. It is recognised that one individual can potentially present with both prosocial and antisocial behaviour with these changes in behaviour possibly being attributable to their mental state and, therefore, including both types of behaviour may increase external validity.
Participants were randomised by the online survey provider into one of two conditions: group one viewed the videos having been told that the person in the videos had a diagnosis of personality disorder; group two viewed the same videos, but were told that the person in the videos had a diagnosis of schizophrenia. To reduce order effects, the prosocial and antisocial videos were administered in a counterbalanced manner. The counterbalancing was also automatically conducted by the online survey provider. After viewing each video, participants then completed an adapted version of the Attribution Questionnaire – 27 (AQ – 27; Corrigan et al. 2003), the Social
Distance Scale (Link, Cullen, Frank & Woznaik, 1987) and the Perceived Dangerousness Scale (Angermeyer & Matschinger, 2003). They were asked to complete these questionnaires based on the person they had seen in the videos.
Therefore, for hypotheses one to three, the independent between subjects factor is the diagnostic label attached to the video and this has two levels: personality disorder or schizophrenia. The repeated within subjects factor is the behaviour seen and this, once again, has two levels: prosocial or antisocial. The dependent variables were the participants’ total scores on the adapted AQ-27, the Perceived
Dangerousness Scale and the Social Distance Scale.
To investigate hypothesis four, a correlational design was used to explore relationships between participant attributions, emotions and intended behaviours across all groups (personality disorder prosocial, personality disorder antisocial, schizophrenia prosocial and schizophrenia antisocial).
2.3 Participants
The participants in this study consisted of 87 mental health nursing students. Participants were recruited from a variety of universities across the United Kingdom,
via social media and through the use of NHS email bulletins. Participant demographics are detailed in section 3.2.2.
2.3.1 Inclusion and exclusion criteria.
To be eligible to take part, at the time of the study, participants must have been enrolled on an undergraduate course in mental health nursing. No exclusion criteria were set.
2.3.2 Rationale for selection of participants.
Much of the previous research into staff attributions towards clients with personality disorder has focussed on qualified inpatient psychiatric nurses (e.g. Markham, 2003; Mason, Hall, Caulfield & Melling, 2010). However, there has been little research into the attributions of student mental health nurses. As previously discussed in Chapter One, it is evident that qualified staff across a variety of
disciplines hold attributions towards the personality disorder client group which tend to be more negative than towards clients with other psychiatric diagnoses. It is also important to be aware of the attributions of students in this field, as this will enable a comparison to be made with the current findings amongst qualified staff. If
differences are found between the attributions of pre-qualified and qualified staff then this may suggest that further research is warranted to investigate the way in which attributions towards client groups develop over time and with experience.
Whilst it is acknowledged that a variety of different professional disciplines contribute to client care, nursing staff are likely to have more direct and frequent contact with clients than members of other disciplines, leading to their being more likely to disseminate information about the client to other members of the care team.
Given this increased contact, the role of disseminating information, and the fact that nursing staff are often the ‘front line’ clinicians responding to clients (particularly at times of crisis), it may be argued that it is crucial to understand the attributions and emotional reactions of student metal health nurses and the way in which these impact upon their intended behaviour towards clients.
2.3.3 Sample size.
A priori power calculations were conducted for all hypotheses using G*Power (Faul, Erdfelder, Lang & Bunchner, 2007). Effect sizes were calculated on the basis of findings by Strong (2010). This was deemed to be an appropriate study on which to base effect sizes as the same measures have been used to investigate the differences between attitudes and emotional reactions to a specific personality disorder subtype and an Axis I diagnosis (as determined by DSM-IV-TR; APA, 2000).
For all analysis where it is anticipated that mixed ANOVA will be used, effect size calculations, and thus sample size calculations, were completed for all variables, although only the one to return the largest sample size will be reported here. Based on research by Strong (2010), an effect size of f = .4 was calculated on the controllability variable. Therefore, to ensure that hypotheses one to three achieve 80% power at a 5% significance level, it is anticipated that a sample size of 40 will be required for each group.
For hypothesis four, which is correlational, a separate power analysis was conducted. In order for hypothesis four to achieve 80% power at a 5% significance level, 44 participants will be required in each group. This was based on a medium effect size of r=0.4. This choice of effect size was informed by the small to medium strength correlations reported by Strong (2010). Strong used vignettes in order to
investigate research questions similar to that of the present study. Given that the present study utilises videos, which is a method not previously employed when
investigating staff attitudes to personality disorder, the effect sizes used were based on the more conservative end of the range of effect sizes reported by Strong.
2.3.4 Recruitment.
Participants were recruited from the population of student mental health nurses in the UK. The course directors of 52 mental health nursing courses were approached directly by the researcher by email, requesting dissemination of the study information to their mental health nursing students. The contact details for the course directors were obtained via the UCAS university admissions website. To assist the course directors in making an informed decision, the email contained copies of the information and consent sheets (Appendix A) and a copy of the letter of ethical approval (Appendix B). Of these 52 courses, 14 agreed to disseminate the
information, with some requiring that the study also be approved by their internal university ethics board.
In addition to direct recruitment from university courses, the study was also advertised via; (1) social media: The study was advertised in four Facebook groups for cohorts of university students and also on generic mental health nursing interest Facebook pages; (2) internet forums: The study information was posted on three forums with a specific nursing focus; (3) nursing regulatory bodies: The invitation to participate was included as part of an e-bulletin to all members of the Royal College of Nursing; and (4) communications from NHS Trusts: The study information was disseminated to all employees of Norfolk and Suffolk NHS Foundation Trust and Cambridgeshire and Peterborough NHS Foundation Trust, once again as part of a
communications e-bulletin. Finally, the study details were also disseminated by word of mouth.