Caitlin is studying for a master’s degree in Finance and Banking. Before starting her course, Caitlin worked as a trainee accountant and also had an internship with a large merchant bank. During her time at the bank, Caitlin became particularly interested in the different factors that impacted upon the behaviour of financial market analysts. As part of her master’s programme, Caitlin has to conduct her own research project, something that she is very much looking forward to.
She is keen to bring her practical experience from the banking industry to her academic studies. Having learnt about the efficient market hypothesis as part of her postgraduate programme, Caitlin would like to conduct research about the impact of market share announcements on the behaviour of analysts. She is interested in some of the cognitive biases such as overconfidence and overreaction that she has read about in the behavioural finance literature. Given her background and undergraduate degree in accounting, Caitlin has always felt comfortable with using and analysing numbers.
However, her interest in behavioural finance has led to her making a decision to take a qualitative approach to her research (see Humphrey and Lee 2004/2008). The aim is to interview market analysts about their views regarding the impact of share announcements on their own behaviour.
Caitlin had read the relevant literature in the area (e.g. Fenton O'Creevy et al. 2004;
Willman et al. 2002) and has decided that it would be useful to study the impact of share announcements by the top ten companies in the FTSE (Financial Times Stock Exchange) index.
She is in the lucky position that she has access to market analysts in the bank where she did her internship and they have agreed to be interviewed. Caitlin decides that she will use critical incident technique as a way of asking interviewees to talk about their views of the impact of the share announcements of those companies on their behaviour. Hence each announcement will be treated as a critical incident. From hearing how interviewees talk about these specific incidents she aims to generate insights into the perceptions of analysts which can further develop her understanding of the extent to which markets behave efficiently.
When planning her research project, Caitlin had realised that she needed to familiarise herself with different types of qualitative data analysis (Symon and Cassell 2012). This is the first time that Caitlin has conducted any qualitative data analysis and she is concerned to find an appropriate method that will enable her to draw some insightful conclusions from her interview study. Initially it all seemed a bit bewildering and she was unsure where to start. In her research methods training her tutor had talked about three different types of qualitative data analysis as examples of the variety of data analytic processes available. These were Content Analysis; Narrative Analysis; and Thematic Analysis using templates. Caitlin was unsure about which of these would be appropriate in her case.
Caitlin initially decided to use Content Analysis for her interview data. The idea of counting was particularly attractive to her given her background. She decided that the best way was to categorise reactions to each of the critical incidents – or share announcements – in one of three categories: positive, negative or neutral. She then went through the interview transcripts and categorised each incident accordingly. From a total of 51 critical incidents identified through the 10 interviews Caitlin found that 17 were categorised as positive, 7 as negative and 27 as neutral. However, on reflection, Caitlin was concerned that this form of analysis provided little rich detail about her research question. The interviews had generated a range of insightful comments from the analysts, yet she felt that this detail had been lost through the analytic process. Therefore she decided to try a different approach.
When reflecting upon the data she had generated from the interviews, Caitlin realised that the analysts talked about responding differently to different announcements as a result of a range of different contextual factors. Indeed one way of understanding the data was that they were telling her stories about the different announcements. These stories drew upon their previous experience of the company and their extensive knowledge of the market. After discussing these impressions with her supervisor, Caitlin decided that Narrative Analysis might be a more appropriate technique to use.
Using Narrative Analysis meant that Caitlin had to look at the data in a different way. Initially she looked at each individual transcript and focused upon the way the analysts talked about their different responses. For example, she identified that when talking about some of the incidents or share announcements, analysts would contextualise their responses by drawing upon the history of the company concerned and the extent to which the announcement followed the usual trend of that particular company. Therefore she identified a narrative throughout their accounts which she labelled as the historical consistency narrative in that analyst behaviour and decisions were influenced by how they interpreted the historical consistency of the share announcement. Another issue that Caitlin believed influenced how the analysts talked and behaved was informed by a shared or collective understanding among their colleagues about the relative performance of different companies.
She labelled this as a consensus narrative, in that analysts would comment on the interpretations that others had of share announcements when they discussed their own evaluations and resultant behaviours. Through this Narrative Analysis, Caitlin felt that she had found an interesting way of addressing her research question. In identifying the narratives that the analysts drew upon to talk about their responses she could understand more about their resultant behaviours.
When Caitlin came to write up her research project she found herself in a bit of a dilemma.
Although she felt that the Narrative Analysis was perhaps more insightful in enabling her to answer her research question, she was concerned about what she saw to be the somewhat subjective nature of her analysis and interpretations. Whereas incidents were real, narratives were constructed. Perhaps she should go back to the content analysis where she could comfortably allocate the different incidents to the different categories and do some basic statistical analysis of patterns. In discussion with her supervisor she was told to relax. What she was facing here was an ontological dilemma, and one that was quite easy to deal with.

Questions 1 Which form of analysis do you think Caitlin should use in her dissertation write-up and why?
2 What does Caitlin’s supervisor mean by her ‘ontological dilemma’?
3 How can that ontological dilemma be addressed?