Massingham2012.pdf

The current issue and full text archive of this journal is available at www.emeraldinsight.com/1176-6093.htm

QRAM 9,1

Q methodology: is it useful for accounting research? Peter Massingham

66

Centre for Knowledge Management, University of Wollongong, Wollongong, Australia, and

Rada Massingham and Kieren Diment School of Accounting, University of Western Sydney, Campbelltown, Australia Abstract Purpose – The purpose of this paper is to evaluate the usefulness of Q Methodology for business research, as an alternative technique for accounting researchers. Design/methodology/approach – Q Methodology is an innovative technique that provides quantitative structure to individuals’ opinions via factor analysis. The authors present the results of a case study where Q Methodology was used to examine attitudes towards an on-line wiki, a Technology Encyclopaedia (TE), amongst 35 engineers and technical employees at a manufacturing company. Management wanted to understand whether employees were willing to embrace social conversational technology as a way of sharing knowledge. The aim of the case study is to demonstrate how Q Methodology works in a practical setting. The authors also examine a published journal article to assess how Q Methodology might be used to enhance accounting research. Findings – The results show that Q Methodology may provide advantages in data gathering (less respondent burden), data analysis (deeper insight into respondent sub-conscious), and results (better respondent “ownership” of organisational problems and solutions). However, it also has weaknesses in terms of managerial application. Research limitations/implications – A limitation is that the discussion is based on a single case study. Practical implications – When working with an industry partner, researchers may need to consider a more positivist approach and be prepared to explain context behind the statements. Originality/value – Q Methodology appears to offer most value as a data gathering technique. It may also be used to capture respondents’ subconscious views on a topic. While the limited time involved will be attractive to practitioners, there is also the potential benefit of increasing respondents’ awareness and understanding of the topic under investigation (i.e. action research), enhancing change management and other sensitive organizational issues. Keywords Accounting, Accounting research, Data management, Data sources, Research methodology, Accessing primary data, Q methodology Paper type Research paper

Qualitative Research in Accounting & Management Vol. 9 No. 1, 2012 pp. 66-88 q Emerald Group Publishing Limited 1176-6093 DOI 10.1108/11766091211216114

Introduction The purpose of this article is to evaluate the usefulness of Q methodology for business research, as an alternative technique for accounting researchers. Q methodology is a mixed methods technique that combines elements of qualitative and quantitative methods in gathering and processing data that requires respondents to perform a ranking task (Brown, 1993, 1996). By requiring the respondents to sort statements into a forced quasi-normal distribution, many of the problems associated with questionnaires (e.g. central tendency, leniency) can be avoided (Kendall and Kendall, 1993). A factor analysis of Q sort data, i.e. interpretive structural modelling

(Warfield, 1976), provides quantitative structure to individuals’ opinions via factor analysis. Proponents argue that it introduces more validation into qualitative research with the inclusion of statistical analysis. It could be used for example to identify groups of employees with similar problems. In this way, Q methodology promises a great deal: innovative data gathering, better validity, and statistical modelling. It has created considerable excitement in the business research community as a method that offers the best of qualitative and quantitative research, i.e. a leading edge mixed methods approach. It may be applied by managers to identify change agents, i.e. people who feel positively about the change initiative, and barriers to change presented by those who feel negatively. Researchers gather data to support or reject theories (Neuman, 2006). Data are the empirical evidence or information that one gathers according to scientific rules and procedures, referred to as research methods. The type of data is typically classified as quantitative (i.e. expressed as numbers) or qualitative (i.e. expressed as words, pictures, or objects) (Neuman, 2006). Data comes from a primary (first-hand) source or a secondary source (second-hand). Data is a representation of reality that we cannot grasp empirically (through the senses) or theoretically (through abstract thought; Baumard and Ibert, 2001). Reality can be seen as both a “discovery” and an “invention”. This can be explained by research epistemology. Positivist researchers aim to maintain objectivity by distancing themselves from data and discovering the truth or reality of the phenomena under investigation. Interpretivists or constructivists, on the other hand, immerse themselves in their data and believe that reality is “invented” or created through the interaction between the data and the researcher (Girod-Seville and Perret, 2001). This debate is important because it influences our view of two themes in this article: (1) that primary data is valuable; and (2) that the interpretation of this data may influence practitioners’ perception of its usefulness. This second theme is relevant because we discuss management’s response to the outcomes of Q methodology research, in our case study. This article is, therefore, grounded in this discussion about data gathering. We present the results of a case study where we used Q methodology to examine attitudes towards an on-line wiki, a technology encyclopaedia (TE), amongst engineers and technical employees at a manufacturing company. Management in the company wanted to understand whether employees were willing to embrace social conversational technology as a way of sharing knowledge. The case study explains how Q methodology may be used to surface problems, the type of data that may be gathered and how it is analysed, and management’s response to the findings. While the case study is not about accounting, it meets criteria for a critical accounting perspective, i.e. it is inter-disciplinary and engages and borrows from other disciplines (Laughlin, 1999). We then examine an accounting article to determine whether Q methodology might enhance its findings. Finally, we make preliminary observations about the implications for accounting researchers. The case study has broad applicability for accounting researchers. Given that the case study focuses on a tool to enhance knowledge sharing and organisational learning, accounting researchers may see value in using Q methodology to explore knowledge flows associated with accounting practice and tools to enhance these flows. Further, the case examines employee attitudes to the TE

Q methodology: is it useful?

67

QRAM 9,1

in terms of organisational change. Accounting researchers might use Q methodology to surface attitudes towards change. Alternatively accounting researchers may use Q methodology to examine any sensitive issue that lends itself to brainstorming techniques and a range of issues. Q methodology has advantages and disadvantages as we will see. However, it offers an interesting alternative for researchers seeking to find truth in organisational behaviours and attitudes (Kaplan, 2006).

68 The case study Background The case study describes the research process and presents the findings of a study to investigate employee perceptions of a wiki, known as the TE, at a leading manufacturer. Wikis have been described as “weapons of mass collaboration” (Tapscott and Williams, 2007, p. 11). A wiki is a collaborative on-line tool that allows the creation and editing of any number of interlinked web pages via a web browser. It is commonly used as a way of sharing information on community websites and corporate intranets. Whereas collaboration in a work environment has traditionally been talking in meetings, it is now increasingly done on-line. The “new web” is about people creating, sharing, or socialising – participating rather than passively receiving information (Tapscott and Williams, 2007, p. 37). In our case study, management wanted to use the TE as an on-line tool for sharing knowledge amongst engineers and technical employees. The introduction of the TE was part of a broader strategy to improve knowledge management at the case study organisation. This was necessary to improve its capacity as a learning organisation which, in turn, aimed to improve innovation. The TE would organise (i.e. warehouse) much of the organisation’s extensive structural capital and also provide an on-line forum for knowledge creation, transfer, and collaboration. Management conceptualised this in this preliminary model of the place of the TE in its communication and information management, as explained in Figure 1 (Willis, 2004). In the past five years at the case study organisation, emails have replaced memos as the preferred mode of delivery to customers. This has greatly increased speed of delivery at the expense of thoroughness, and information storage and retrieval. The TE offered an opportunity to resolve the conflict between these parameters. The TE exists within other systems at the organisation, competing with individual information repositories and knowledge search preferences, alternative information repositories and knowledge-sharing cultures, which differ between organisational units, and a rapidly changing product offer as innovation tools are developed by individual innovators or established software suppliers. Management, therefore, was keen to evaluate employee attitudes towards the TE in order to assist its implementation as a change initiative within its broad learning organisation strategy. Objective The research objective was to identify attitudes of employees, regarding their acceptance of, or reluctance to use, the TE. Q methodology was applied as a field methodology – using its strength as a systematic way of analysing and interpreting respondents’ opinions.

Q methodology: is it useful?

High Technology Encyclopedia (Wiki) ing o g k n O Tas f ie Br ted Memo ple rief m B Co sk Ta Trip aft e aft Report Dr Not Dr por t h c Re Te ng i y pl Conference om ote c Paper lly hN Fu Tec

Email

Speed of Delivery

69

ing ply m o rc h y c sea r t l l Fu Re epo R

Refereed External Paper Low Low

High Thoroughness

Meeting the the knowledge Capture Needs of the Company (Detailed, Concentrated, Context, Indexed, Archived, Comprehensive, Complete)

Source: Massingham and Diment, 2009, p. 127

Relevance for accounting researchers The case study has broad applicability for accounting researchers. There are two key themes in the case study context that are transferable to accounting. First, the case study examines a research intervention, i.e. the TE, as a tool for improving knowledge sharing between employees. Accounting, like most other organisational functions, is increasingly recognising the importance of knowledge as an organisational resource, e.g. refer to accounting research on intellectual capital measurement (Guthrie et al., 2006). Accounting researchers may see value in using Q methodology to explore knowledge flows associated with accounting practice and tools to enhance these flows. Second, the case study examines employee attitudes to the TE in terms of organisational change. Accounting researchers might use Q methodology to surface attitudes towards change. For example, researchers interested in legitimacy theory might consider how society expectations are changing and stakeholder responses to these changes, using Q methodology as set out in this case study. Respondents There were 35 respondents. They worked in a research and development organisation and were mainly university qualified professional staff. Management identified five

Figure 1. Conceptual model of the place of the technology encyclopaedia in communication and information management

QRAM 9,1

70

groups of employees and proposed varying levels of need for the TE in their daily work as follows: (1) Researchers. These staff tend to focus on the long term, such as breakthrough research, and should see the TE as a vital part of their work. Their contributions would be expected to make fundamental scientific insights available to TE users. (2) Technologists. These staff focus more on the short term, such as improvements to plant or the business, and should see the TE as part of their work. Their contributions would be expected to make product and process knowledge available to TE users (it is expected that contributions to this type of knowledge will also come from plant-based technologists). (3) Technicians. These staff focus more on day-to-day tasks and will probably be less interested in the TE. However, their contributions would be expected to make standard processes and procedures available to TE users. (4) Administrative staff. These staff will probably not see the TE as part of their work. (5) Managers. These staff will use the TE as consumers, but not as contributors. Research design As demanded by the Q methodology, a concourse discussion was held where relevant employees provided their views on what would help them contribute to the knowledge base of the TE. This concourse generated 57 statements, which were subsequently sorted by the 35 respondents in a rank-order “forced sort” according to which statements they agreed with or thought most important. Once all respondents had completed the individual sorting process, the sorts were statistically analysed to find correlations and identify factors (i.e. sets of statements) that were common to the sorts of several individuals. We now introduce Q methodology, and then examine the three main steps in using Q methodology – the concourse, the sort, and the factors – in detail. Q methodology Q methodology is a mixed methods or pluralist approach because it involves both quantitative and qualitative research aspects. Q methodology involves the respondent in a value-oriented political-social goal that aims to facilitate social change (Neuman, 2006). In this way, respondents actively and consciously participate in the research process, creating a sense of empowerment that seeks to raise awareness of the issue under investigation. The outcome is an action that often involves organisational change with the advantage of creating buy-in and support through the research process itself. Q methodology fulfils one of Stoecker’s (1999) key criteria for action research: the democratisation of knowledge by letting respondents share their views and opinions, to further explore their experiences, and expand on our knowledge of their behaviours and attitudes (Brown, 1996). However, Q methodology is not limited to action research. Q methodology involves three phases. First, there is a data gathering phase – the concourse. Second, there is an analytical phase – the sort. Third, there is a results phase – the factors. The aim is to provide a depth and breadth of understanding about a particular topic. The breadth comes from the range of views that can be gathered using this method, while the depth emerges from the forced ranking processes

of the sort. Q methodology is often associated with quantitative forms of analysis due to the use of factor analysis for the Q sort technique. This provides it with statistical rigour that makes it attractive to many researchers. It is becoming increasingly popular in the social sciences as a way to explore traditional qualitative research questions, e.g. respondents’ views, attitudes, opinions, understanding, and experiences. The concourse: step 1 How does it work? The concourse consisted of a general discussion around what people like or expect of a wiki, in this case the TE. The following question was put to employees (respondents) as the focus of the discussion: What would help you to contribute to the TE?

The setting for the concourse was the main meeting room at the case study organisation and consisted of one group over a period of nearly two hours. The session was led by a researcher and included the use of wireless keyboards distributed to the respondents and a common projected screen which displayed the contributions as they were given. The concourse is an interactive discussion where respondents make comments and others respond. The country setting was Australia and respondents were normally encouraged to have opinions in meetings. The sharing of multiple keyboards meant that it was very difficult to know who was typing what on the screen projection, enhancing anonymity and empowering respondents to make comments. What are the outcomes? In all, 57 statements were collected reflecting the range of views that the respondents held on what would help them to contribute to TE. Table I provides the complete statement list from the concourse and the number of the statement and the category number (explained next) for each statement. How do we analyse the statements? To help us understand the statements, and to later understand the results of the sort, the statements were reviewed by the researchers and broken down into seven descriptive categories as follows. Case study management were not asked to comment on this process because it was felt that that role required qualitative research skills possessed only by the researchers. Each statement was placed into one of the seven categories. This represented an additional type of coding to identify common themes across the statements. The categories were not presented to the respondents and were not part of the sorting process. Table II presents the title of the category and total number of the 57 statements assigned to the category. Table III consolidates these descriptions into summary themes. Discussion. A concourse can generate a large volume and wide range of views about a particular topic. The use of wireless keyboards and projection of comments onto a screen in real time was a success. This method of statement collection is particularly effective as the technology is somewhat novel and is particularly easy to use and instantly displays comments for all to see. This generated an active discussion and a large number of useful ideas expressed as statements. Perhaps the greatest value in using this technology was respondent confidentiality. Unlike verbal comments in brainstorming sessions, the use of keyboards meant respondents could type comments in response to other comments with a sense of privacy. This virtual discussion created an honesty and rigor in the discussion that generated a rich set of statements representing the group’s attitude towards the TE.

Q methodology: is it useful?

71

QRAM 9,1

72

Table I. Illustrative list of statements produced by a Q methodology concourse

Category no.

Statement no.

1 1 1 1 1 1 1

1 2 3 4 5 6 7

1 1 1 2 2 2

8 9 29 10 11 12

2 2 2 2 2 2 2 2 3

13 14 15 16 17 18 56 57 20

3

21

3 4 4 4

22 24 25 26

5 5 5 5 5 5 5 5

27 30 31 32 33 34 35 36

5

37

5 6 6

38 39 40

6 6

41 42

What would help you to contribute to the wiki? If it was of more value If I could see tangible benefits to customers If customers could access the information If it gave something back to the organisation If the objectives were made clear If its usefulness was apparent Knowing that this type of system is going to be around “for the long haul” and not be a “flavour of the month” If I thought the information was useful to the users, i.e. technologists If I thought that customers wanted information added as part of their project If I thought the system was not going to be redundant in a couple of years If it provided the ability to make anonymous entries If contributions were recognised and rewarded If contributions were tracked to me so that my boss can see my contributions If it had a high priority My professional pride Acknowledgement of co-authoring and responsibilities for articles If I knew it would not make me redundant If it told me what to do when I get stuck or do not know what to do If it had viewing stats to see who was interested in my additions If there was a wiki newsletter If there was a wiki award If the system captured info requests – so you could write on a topic for a known audience If we went completely electronic and stayed away from paper, e.g. paperless office If it provided more feedback from readers of each page If I was not limited by my ability to contribute If I had the time to contribute If a faster server was provided (multi second updates, lost pages on preview) If I could use it in focus groups with limited team members If it accepted dot points/not essay If confidentiality issues are resolved If I could easily get attachments in right format before entering If it had an improved authentication process Knowing info demand hot spots More logical structure to location of topics If it provided support for equations, I cannot put (in any reasonable form) all of the pertinent information into the wiki If another form of reporting was removed (e.g. technotes so this becomes the new format) If it provided better security If it had a specialist entry person/editor If the system allowed direct entry of existing data without the need to reformat If I had an easier to use, simpler interface Simple – the kiss theory (continued)

Category no.

Statement no.

6 6 7 7 7

43 44 45 46 47

7 7 7 7 7

48 49 50 51 52

7 7 7

53 54 55

Category no.

What would help you to contribute to the wiki? Easier more logical access If guys in the control room could browse it in the middle of the night Having people who could capture information for me as its produced If I had more training and practice If there were specific requests for information and individuals allocated to answering it If it was the primary source of information storage If it was universally regarded as a necessary job function If there was someone to maintain it If there was a higher level of commitment to wiki from management Integration with an issue resolution system so such info is automatically built into the wiki If it had someone managing it and asked people to contribute specific areas If it was linked to head office information technology If the managers allowed and supported more open sharing of sensitive information

Q methodology: is it useful?

73

Table I.

Quantity Category descriptions

1 2 3 4 5

9 9 5 4 11

6 7

6 13

Customer can access/more value/clear objectives/long term value Acknowledge/reward/anonymous/responsibility to the document Communication/audience Time/limitation Better organised/less duplication/security/without reformat/better format/ better structure Easy to use/better interface/accessibility Primary resources/necessary job function support/commitment from management/training integration with other systems/resources

Category no.

Themes

1 2 3 4 5 6 7

Mainstream/clear contribution/ongoing Privacy/recognition/job security Feedback Time Security/structure Ease of use Organisational

Q methodology is an innovative data gathering technique for two reasons. First, it provides the opportunity to capture respondents’ thoughts on the topic under investigation. This is done by asking respondents to write down their opinions or type it into a computer. This is then captured and shared amongst the group. This has the advantage over other qualitative data gathering techniques such as focus groups

Table II. Category descriptions and total number of statements assigned to the category

Table III. Summary themes

QRAM 9,1

74

or face-to-face interviews because it may capture thoughts that might otherwise not be articulated. For example, focus groups can be dominated by a vocal few, irrespective of the skills of the facilitator, particularly if some respondents are shy or unwilling to voice their opinions. In an organisational context, respondents may feel the need to agree with others, e.g. peer group pressure, or to let others speak for them, e.g. their supervisor. The second innovation of the concourse as a data gathering technique is its interactive nature. In this way, it resembles a brainstorming session. This is related to another advantage of Q methodology, that it does not require a large sample to produce meaningful results. A sample of 30-50 individuals can produce an accurate picture of the range of views on a topic (McKeown and Thomas, 1988). During the concourse, respondents are encouraged to produce as many statements as they can. This captures their full range of thoughts. However, the thoughts expressed in the statements are not limited by personal experiences. Respondents are encouraged to reflect on the statements put forward by others, as a way of reflecting on their perceptions, or considering thoughts not in the forefront of their mind. In this way, the concourse may capture some of respondents’ sub-conscious understanding of a topic. The sort: step 2 How does it work? The process of a “forced sort” begins with providing the sorting group with a statement of instruction. This statement has an impact in that it reflects the purpose of the sort. In this case, only one sort was conducted under the following instruction: What would from (your point of view), help you to contribute to the TE?

The sorts were held in several groups on a single day. In each sort, the respondent rank-ordered the set of statements into an array of numerical data. In this study, as is the case with a “forced sort”, there was a single position for each statement. The statements placed in the spaces to the extreme right were assigned with the highest level of agreement, in this case a þ 4, those placed at the extreme left were assigned the highest level of disagreement of 2 4, with those in the centre receiving the neural ranking of 0. The Q sort method allows each participant to rank the statements presented in terms of their comparative importance (Brown, 1980). The sort is the method used to shape or present a picture of individual views on a topic by making decisions in regard to the statements presented. The participants of the sort are asked to choose amongst the statements, in this case to the extent of their agreement or disagreement with them. For example, they may receive these instructions with the sort: You are being asked to sort statements in accordance with your degree of concurrence/agreement with the statements. Where þ4 is high agreement and 24 is high disagreement and the scales between 24 and þ4 reflect shades/levels of agreement. You will find the statements on a pack of cards that will be given to you. You are asked to sort the cards in accordance with the rating given to each card. The largest number of statements will be placed in the centre and the least amount of statements at each extreme point (Meloche et al., 2009, p. 40).

In its purest sense, the sort for our case study would work by asking participants to choose the statement they felt they most strongly disagreed with and place it to the far left, and then choose the statement they felt they most strongly agreed with and place that on their far right. Given that we had 57 statements, the statement on the far left

would be 2 28, and the statement on the far right would be þ 28, and there would be a midpoint and a statement would be given a zero rating. Then the participant would choose the next statement that they most strongly disagreed with and place it one card closer to them on the far left and give it a number of 2 27, and so on until all statements are force ranked. The problem in dong the sort this way is the time involved. Participants may find it difficult to differentiate between statements at that level of detail and take too much time to complete the sort. The Q sort triangle we used in our case study aimed to make the sort efficient in terms of participant time and is recommended if your participants have time restrictions. The way our sort worked was that participants used the 2 4 and þ 4 scale as a proxy for the 2 28 and þ 28 scale of the complete sort, outlined above. In this way, they could place more than one statement as a 2 4 (strongly disagree) or þ 4 (strongly disagree) but the aim was to differentiate as much as possible by using the full range of the 2 4 to þ 4 scale as much as possible, i.e. spread the cards across the scale. This abbreviated version of the sort retains the integrity of the method but makes the sort quicker. The diagram shown in Figure 2 was given to respondents to record their ranking of the statements. What are the outcomes? Once all respondents had completed the individual sorting process, a set of cards are produced which represent the sort’s raw data. Each card represents each respondent’s rankings for each statement. These are then input into the Q methodology software to compare each respondent’s individual sort. This is called the factor analysis, which is the third step in the Q methodology process. Discussion. The sort has the advantages and disadvantages of any forced ranking exercise. From a researcher’s perspective, the ranking technique of the Q methodology sort is probably the most appealing aspect because it overcomes the clustering effect of typical rating scales. Clustering occurs when the results hover around the middle of the scale, e.g. when the sample mean sits around three on a five point scale. This happens due to respondent tendency to take the middle ground in rating single items. Therefore, when asked to rate one item, each response is biased towards the centre. This tends to dilute the results and make it difficult for researchers to interpret any real meaning from the findings, e.g. does a difference of mean rating difference of 0.5 between two items represent a difference of practical significance? Q methodology overcomes this clustering effect by forcing respondents to rank factors into a quasi-normal distribution, and therefore highlights the most extreme items. There may be negative consequences of this force sorting by artificially differentiating statements, for example, respondents may wish to group several statements as equally important or unimportant. However, recent research has found that numerical weighting of qualitative data may be as rigorous or “scientific” as any statistical analysis of quantitative data (Samkin and Schneider, 2008).

Q methodology: is it useful?

75

Figure 2. Q sort triangle sample for ranking of the statements

QRAM 9,1

76

The factor analysis: step 3 How does it work? Once the Q sort is completed, the process of factor analysis is then undertaken. The analysis provides a correlation matrix that shows which respondents sorted the Q statements in a similar order. The correlation matrix is then subjected to factor analysis to obtain groupings of data arrays that are highly correlated. It is this level of correlation that determines the factors. The factors represent clusters of respondents with similar opinions (Stephenson, 1953). In this way, we can develop a qualitative judgement or weighting about the relative importance of each statement based on how many respondents load into factors. The number of factors identified depends upon the degree of agreement amongst respondents, and in how much detail the researcher feels is useful to analyse. The factors are not necessarily mutually exclusive because a given statement or a given individual may appear on more than one factor. However, it is important to be cautious with clustering of similar statements, otherwise there is a risk that significant “outlying” statements might be overlooked. The researcher needs to carefully consider the relative significance of all statements and consider including themes even if mentioned by only a few respondents. The weakness in Q methodology in this area is that quantitative structure may dilute some of the richness in the data if outlying statements are excluded. Outcomes. The process of analysing the Q sort results involves correlation and by-person factor analysis where the analysis is performed not by variables, such as traits, or statements, but rather by persons, where people correlate to others with similar views based upon their sorts. Thus, the individuals are not grouped by traits such as age, gender, or years of experience, but upon the groupings of their expressed opinions. The outcome is a number of factors that reflect the grouping of respondents with common views (Cottle and McKeown, 1980). The main outcome from the Q sort analysis that we conducted with our case study organisation was a typology of three opinion groups, i.e. factors. The factors represent clusters of respondents who feel similarly about a topic. The factor analysis also tells us why they feel this way about the topic. The summary factors were: . Factor 1. Contains respondents whose statements are most aligned with a progressive “corporate knowledge worker”, concerned with the permanence of the TE, its value in terms of its usefulness and its ease of use. . Factor 2. Shares a number of the views of those expressed in Factor 1, the corporate knowledge worker, plus this factor has a strong customer focus in its selection of “usefulness” statements (note this is a reflected negative factor). . Factor 3. Contains the individuals whose statements are both concerned about how mainstream the TE is, whether it will be fully supported by management and about permanence (note this is a reflected negative factor). This indicates to the case study organisation that staff feel differently about the TE, and summarises these different views into factors (three in this example). The factors identify the characteristics of these different employee groups by showing how they share perceptions about the topic. The positive and negative scores explain, in more detail, how the factor groups differ in their opinions about the topic. Discussion. When we presented the results of the Q methodology study to management, they raised several queries about the data. By analysing these queries, we can develop further understanding of the value of Q methodology for practitioners as

well as its weaknesses. Management’s response to our report is summarised by this statement:

Q methodology: is it useful?

If this research has identified different groups, how do the properties of these groups relate to the ability to tailor interventions? Are there groups which require interventions? Would different groups require the same or different interventions?

This shows that practitioners require information that allows them to take action. Practitioners want the capacity to generalise from the research findings in order to find meaning. This means classifying employees in ways that allow them to be differentiated and then managed based on their differences. In our case study, management were frustrated by the results of the Q sort. If researchers are constructivist or interpretivist, they might respond to these queries with this statement: The data is what it is.

We found that management might respond with this statement: What does the data mean?

Even more importantly: How can I use this data?

Table IV looks at each question raised by management in the case study organisation and the actual responses from a constructivist or interpretivist researcher compared with a positivist researcher. The table does not place value judgment on the correctness of each response, rather it reflects the mix of epistemologies in the research team, and is presented to generate discussion. We can see from this discussion that Q methodology results may create problems for researchers working with industry partners. The interpretivist responses are weak in comparison with the positivist responses. This was reflected by case study management’s dissatisfaction with the interpretivist responses in the table above. The comments reflect different research epistemologies. We do not aim to disparage interpretivist research. However, the table highlights potential difficulties for interpretivist researchers when trying to explain research findings to practitioners. In our case study, management wanted a more positivist response to the Q methodology findings. Q methodology is an excellent vehicle for identifying a range of views on a particular issue, i.e. the sort statement(s), but can be problematic in applied research. When working with an industry partner, researchers may need to consider a more positivist approach and be prepared to explain the context behind the statements. Managers may find that knowing that a group of employees feels a particular way about an issue is of limited benefit. They need to know who feels that way, at least in terms of profiling groups of respondents with similar attitudes. Respondent profiles by factors may then help managers target actions at specific employee groups rather than all staff. It would not be ethical for the researcher to identify specific staff for action. In our case study, management wanted to know the impact on other factors (e.g. question 5 about mutual exclusivity). It would be counter-productive if an action targeted at Factor 1 respondents had a negative impact on Factor 3 respondents for example. The researchers sensed that management in the case study organisation were unhappy with the broad findings of the three factors and frustrated by the results

77

Table IV. Management queries and researchers’ response Constructivist or interpretivist response

Positivist response

78

The distinguishing characteristics of the factors are the collection of the statements favoured by the group of people on each factor. The research design did not seek to summarise the characteristics of the individuals comprising the groups, but the three factors reflect the views of the individuals in light of their positioning of the statements 2. Can we plot the three groups in a two-by-two matrix The outcome of a Q study is not really suited to a twothat clearly labels the axis (i.e. how are the groups by-two matrix. It can produce a graph that plots the different and what is the spatial degree of difference)? position of the sorts which reflects the factors

1. How do we summarise the characteristics of the three different groups, i.e. the three factors?

We can do further analysis to look for similar characteristics which distinguish the factor respondents, such as classifying them in terms of their work profile (see discussion of respondent characteristics earlier), age, experience or other areas. This might be used to generalise about respondents, enhancing the managerial application of the findings A suitable graphic should provide a spatial explanation of the degree of overlap between the three groups. This would provide management with a guideline on the inter-relatedness of the respondents, i.e. if a research intervention is focused on one group, will it also address a second group? 3. What are the needs of the three groups in terms of This research has not sought to address the needs of The brief description of each factor provides some identification of the needs of each group. This may addressing their attitudes to the TE? the individuals, as such, however the statements gain managerial application by examining the nature provided by the individuals provide the breadth of their views with regard to the question asked “What of the factor description, e.g. the underlying would help you to contribute to the TE?” and the high statements, to design solutions and low statements associated with each of the factors gives their concurrence or disagreement with the statements 4. How may management address the needs of each The statements were sorted to the extent that the The statements do provide useful insight. A next step group? respondents’ agreed or disagreed with them. Thus, the would be to take specific action designed to address positive statements in each factor are a very direct and the issues they raise. We can work with management clear indication of what the people on each factor to develop an action plan based on looking at the express as their needs strongest positive and negative statements The main issues raised by respondents (see the 5. Are these needs mutually exclusive or inclusive? In The needs are not mutually exclusive. This is other words, if the needs of one group are met, will this particularly the case given the nature and flexibility of statements above) relate to organisational issues and not the TE itself. The factor graphic suggests while affect the needs of other groups? technologies such as the TE. The processes for working with the TE can be, and in fact need to be, the three groups are close in terms of spatial differentiation, they do not overlap much. adapted and developed in accordance with the Examination of the actions for each factor (see above) expressed needs of those who use that technology. suggests that if we address the needs of Factor 1, we This process should be ongoing as the expressed address most of the issues for Factor 2. Factor 3 seems needs change with time and experience different

Question

QRAM 9,1

of the Q methodology research. Therefore, we produced further analysis by examining the statements underlying each factor in more detail. We presented recommendations for management at two levels. First, we drew general conclusions that the success of the TE will depend on whether the following issues can be explained to staff by management: . Whether continued support will be provided for the TE. . If their contributions will be acknowledged. . Whether the TE will only duplicate other forms of information provision. . If its usefulness is apparent. . If it provides tangible benefits to customers. We then designed an action plan for each factor as explained by Table V. This type of analysis provided management in the case study organisation with an action plan. While this additional step in the Q methodology process provided management with more confidence in the research, we learned that building specific details about the type of respondents in each factor into the research design would enhance the managerial application of the Q methodology results. Benefits for accounting researchers We now turn to looking at whether Q methodology can provide benefits for accounting researchers. We do this in two ways. First, we examine the application of Q methodology to published research. Second, we present preliminary ideas about whether Q methodology might help some of the current issues facing accounting research, i.e. a critical accounting perspective. Published research We review one accountancy article which uses primary data to suggest how Q methodology could be used to improve or clarify the findings. The paper selected clearly uses respondent’s subjective judgements to develop hypotheses and inform conclusions. The paper examines the effects of experience on complex problem solving in accounting (Lehmann and Norman, 2006) from Behavioral Research in Accounting. Discussion of “The Effects of Experience on Problem Representation [. . .] in Auditing [. . .]” (Lehmann and Norman, 2006) This study is related to the psychological literature concerning problem representation, heuristics and expert systems. The investigators essentially performed a single blind, prospective study comparing the problem-solving processes of experienced employed accounting professionals with “novice” graduate students. The experimental subjects were provided with a single case study outlining an auditing problem. Their written responses were analysed by a researcher examining the type and number of variables that the subjects would take into account, while solving the problem. There were eleven categories used to encompass the problem domain. These were “losses”, “cash flows”, “line of credit” “accidents/lawsuits”, “financial conditions”, “industry/competitor”, “reorganisation/chapter 11”, “labour issues” and “liquidity”. The scoring of the questionnaires were single blind in that the researchers coding the questionnaire did not have access to the demographic details of the respondents while coding

Q methodology: is it useful?

79

Table V. Recommendations to management If I had the time to contribute

These respondents are generally supportive of the TE. The actions above will specifically increase their uptake of the technology Factor 2 Commitment Value Accessibility If customers could access the information These respondents do not understand how the TE will work. The actions above will specifically address their acceptance of the technology Factor 3 Job security If I knew it would not make me redundant Safety If it provided the ability to make anonymous entries These respondents are more concerned with themselves than the organisation. The actions above might change their negative attitude toward the technology

Time

Value

If I thought the system was not going to be redundant in couple of years If its usefulness was apparent If it was of more value If I could see tangible benefits to customers

Statement

System issue

Announcement from Executive

As for Factor 1 As for Factor 1 User specification for IT solution

Focus groups/brainstorm to define its value (short term) Training sessions explaining how the TE can benefit the individual, customer, and organisation Staff required to undertake TE time each week and record it (and the output) on their time sheets

Announcement from executive of longterm commitment to the TE

Action

80

Factor 1 Commitment

Factor issue

QRAM 9,1

the questionnaire. However, these researchers may have been present during the administration of the experiment, and it is this which determines that the coding was not double blind, which is a weakness as it introduces the possibility of researcher bias. The results of Lehmann and Norman’s study demonstrate some of the limitations to hypothesis testing statistics, particularly with respect to quasi-experiment design (Nester, 1996). On a number of occasions Lehmann and Norman have had to use less stringent statistical criteria (i.e. accepting p values a little above p ¼ 0.05) in order to highlight issues of marginal statistical significance. In particular, the marginally significant result for number of concepts brings into question one of the core hypotheses of the study regarding the greater efficiency of problem representations for experts versus novices. The appropriate application of Q methodology to the problem domain of the study could help inform this finding and clarify how problem representations differ. However, care is required to ensure statistical frailties inherent in the study are not glossed over. Given that the amount of information used by experts and non-experts is similar (although as noted of marginal statistical significance), development of an appropriate Q sample could help inform the size and direction of the differences of opinion observed. While Lehmann and Norman (2006) do not give details on the content of what was discussed with respect to their case study task, we would expect that the considerations would split into generic factors which are applicable to solving all auditing problems, and case-specific factors. These could form the basis of a Q sample, but may require further discussion in one or more concourses. This is because bias may be introduced by novices (possibly silently) deferring to expert opinion. Note that it may be possible to use a Q sample to triangulate the quantity of information used by experts and novices by carefully constructing the Q sample to have some semi-redundant items (i.e. things that are closely related to each other), but this would need significant pilot testing. Determining how to develop the Q sample would be informed by the context of the study. A case-specific Q sample would be appropriate for providing a larger study, highlighting the salient issues with their existing case study. Because the “answers” would be given by the Q sample, the task would involve less effort for respondents to complete, and therefore this would help reduce the risk of sample bias, a concern highlighted by the researchers. The reduced time to complete and reduced complexity of the task would also provide the opportunity to collect a larger sample, and thus possibly allow more finely grained distinctions between experts and novices. Lehmann and Norman (2006) are also concerned that their work relates to a single case study example and may not generalise well. One possibility for investigating this further would be to cross-validate the Q sample generated from the first case study against a second case study. This would provide several benefits. It could be that the separation of generic items and case-specific items is not straightforward. Examining the differences between the factor structures on the two separate Q sorts may help to clarify this. A factorial quasi-experimental design could be used to examine the extent of the differences between the two different Q sorts. A potential approach would be to randomly assign respondents into one of two groups. The first group would perform the Q sort on the first case study followed by the second case study, while the second group would do them in the reverse order. This way the researchers could see the effect of order of exposure as well as the effect of only doing one case study (i.e. by comparing

Q methodology: is it useful?

81

QRAM 9,1

82

the factor structure of the first group’s first Q sort against the second group’s first Q sort). This demonstrates the advantage that sample size is not as crucial for Q methodology as it is for R methodology, in that lowering the sample size does not have proportionally the same effect of reducing statistical power for Q methodology as for R methodology (Nester, 1996). This is because statistical power is a function of sample size. For R methodology, sample size is the number of respondents, while for Q methodology it is the number of statements in the Q sort. In summary, the core of our argument here is not that Q methodology is a replacement for other approaches, but that it is a potentially powerful adjunct approach which can help bring answers to research questions into clearer focus. In respect to the Lehmann and Norman (2006) study, we show that while Q methodology is not a panacea for some of the logical problems that beset null hypothesis testing approaches to research (Nester, 1996), its orthogonal approach to statistical power and data reduction can help provide tools to deal with some of these problems. A critical accounting perspective Kaplan (2006) argues that accounting research needs to pursue the “truth” underlying the social reality of how accounting is practised. Critical accounting theory is a suitable lens from which to tackle this problem and also whether Q methodology might assist. Critical accounting theory aims to identify the prevailing social arrangements within a community, such as an industry, within the broader context of the society in which it exists (Roslender, 2006, p. 250). Critical accounting researchers seek to benefit society by turning learning to the advantage of a better society. Therefore, understanding what we study is not enough. Critical accounting theory uses understanding to enact change via the philosophy of praxis (Roslender, 2006, p. 264). This philosophy enables theory to inform practice and vice versa. We can see how the frustrations of management in the case study organisation, outlined earlier, could be addressed by the philosophy of praxis. Critical accounting theory aims to promote self-awareness of both “what is” and “what might be”, and how the former might be transformed to install the latter (Roslender, 2006, p. 250). This theory examines the social praxis involved in organisational change (Tinker, 2005, p. 101). In this sense, Q methodology could help by surfacing the underlying issues associated with organisational change initiatives, in this example the introduction of the TE. One implication of social praxis within the context of critical accounting theory is that theories need to change in response to changing social conditions. Well-developed and empirically grounded theories may create impetus for changing the way accounting is practiced and ultimately lead to reforms in how business and society operate (Deegan, 2009, p. 530). One practical use for Q methodology in critical accounting research is to help understand social contracts using legitimacy theory. Legitimacy theory proposes a relationship between corporate disclosures and community expectations, the view being that management reacts to community concerns and makes necessary changes (Deegan, 2009, p. 340). But how do organisations identify societal expectations? Q methodology might be used to examine stakeholder perceptions of this question. If we can understand how stakeholders such as investors, management and staff perceive social contract and why, then we may develop a theory leading to changes in behaviour that match the industry’s social contract, i.e. society’s expectations.

Results Findings

Data analysis

Sort Data analysis

Data gathering

Data gathering

Data gathering

Concourse Data gathering

Technique

It provides a solution for change management and creates buy-in through the action research process

It has statistical validity as well as capturing the range of subjectivity on a topic It identifies respondents’ perceptions through forced ranking technique rather than artificial agreement created by traditional rating scales

Small sample size and time of the concourse (1 hour) limits the burden on respondents making them more willing to get involved Limited time required from respondents provides better opportunity for access to organisations A non-threatening and interactive way of capturing the full range of views about a topic A method to capture the three dimensions of the tacit triangle: focal awareness, subsidiary awareness, and the knower’s perspective (Polanyi and Prosch, 1975)

Value

Practitioners

Researchers

Researchers

Researchers

Practitioners & Researchers

Researchers

Practitioners

Target

Q methodology: is it useful?

83

Table VI. Q methodology value proposition

QRAM 9,1

84

In sensitive areas, such as the role of accounting in legitimising social contract, Q methodology could offer an alternative method to surface organisational culture issues, particularly when we consider the “confidential” nature of the concourse method presented in this article. Summary comments Q methodology has value as a tool for gathering data and analysis of data. Table VI summarises the range of benefits identified by this article. The exhibit shows that the Q methodology appears to offer greatest value as a data gathering technique. The notion that it can capture respondents’ subconscious views on a topic is interesting. While the limited time involved will be attractive to practitioners, there is also the potential benefit of increasing respondents’ awareness and understanding of the topic under investigation (i.e. action research), with opportunity to examine critical accounting themes. Value for industry As we saw from our discussion of management’s response to the Q methodology report, a significant problem is that researchers need to understand how the data can be used. This addresses the heart of debate over conceptual versus applied research. The solution is a simple paradigm shift. To understand this shift, we need to delve into the world of research epistemology. In our view, the problem with many proponents of Q methodology is that they are constructivists or interpretivists. We state this based on the influence of action research and our observations of working with such researchers. For these researchers, reality is “unknowable because it is impossible to reach it directly”; rather it can only be summarised, i.e. interpreted, or represented, i.e. constructed (Girod-Seville and Perret, 2001, pp. 16-17). This view of knowledge privileges the status of understanding (interpretivism) and construction (constructivism). Researchers with these approaches seek to make sense of their world through the social construction of reality. This means trying to identify the contextual and subjective setting of the entity under enquiry. As a result, these researchers are reluctant to commit to solutions, retreating instead into the safer world of problems. The only tangible outcome is the varying combinations of opinions expressed by the group of respondents. But practitioners may ask [. . .] so what? On the other hand, positivists believe “reality exists in itself (and) it has an objective essence, which researchers must seek to discover” (Girod-Seville and Perret, 2001, p. 15). This view of knowledge privileges the status of explanation. Positivists seek to make sense of their world by finding universal truth. For management researchers, this means trying to identify a “best way” that others may follow. A positivist view provides the meaning lacking in current approaches using Q methodology. If we can switch the focus of the Q approach from the production of a range of views to the identification of people with similar views, then we may be able to better engage practitioners. Rather than selling practitioners the idea of getting deeper insight into the problem, we should be selling them the notion of better understanding their employees – and ways to manage these different views. If we can identify people who are for and against a change management initiative, we can then manage the change process. We can do this by using those who are positive as change agents, and working

to address the barriers to change (i.e. roadblocks) presented by those who feel negatively. The Q statements provide us with this ammunition by giving insight into the reasons people feel this way and, therefore, an action program. For example, if the “negative” group felt most strongly about lack of communication from management, this can be immediately addressed. The solution is then: improved communication by management. But we feel Q methodology researchers need to go one step further. We need to assess the characteristics of the respondents in each factor group in detail. What is their age, seniority, work history, qualifications, business unit, work activity, and so on. This information can be used in a way that still protects respondent anonymity but classifies people in meaningful ways. Managers can then generalise about the characteristics of change agents and roadblocks and develop strategies that specifically address these employees. This switches the focus to a positivist view, and solutions, rather than problems. However, perhaps the most interesting aspect of Q methodology is its potential to identify respondents’ sub-conscious views of a topic. This allows researchers and practitioners to delve into the complexity of employees’ “black box”: the triangle between focal awareness, subsidiary awareness, and the knower (Polanyi and Prosch, 1975). It is here that we think Q methodology might be helpful for practitioners and accounting researchers, particularly when examining sensitive issues from a critical accounting perspective. Value for researchers Q methodology can be interesting for accounting researchers, particularly in pure quantitative/experimental studies, for example, Lehmann and Norman’s (2006) study, which is essentially a classic R method experimental design in the mould of psychology research. Some qualitative research has problems with defining the different kinds of subjectivity, but Lehmann and Norman’s (2006) study has the opposite problem. Due to the limitations inherent with quasi-experimental design and its statistical analysis, the researchers’ conclusions are somewhat muddied, and do not fully support their central hypothesis: that the heuristics used by experts are more efficiently represented compared with those of novices. We demonstrated how Q methodology could be used to make the experimental task quicker and easier to administer, and thus is potentially able to provide a bigger and more representative sample. We also demonstrated how appropriate use of Q methodology for sample development and experimental design has the potential to elucidate differences in problem solving between different groups. While R methodology enables the researchers to see some properties of the differences between groups, their ability to define how individuals’ perceptions differ and change across the groups is limited. While the use of Q methodology to assess the efficiency of problem representation is somewhat difficult, as discussed previously, it is probably not insurmountable through careful design of the Q sample. Due to the relative lack of information loss inherent in the Q factor analysis compared to the classification process used, the Q methodology approach would have better ability to demonstrate how the representation of problems progresses from expert to novice, rather than presenting them as discrete categories. This could be done by validation through predicting months of experience from individual scores on latent variables. This demonstrates how Q methodology and its

Q methodology: is it useful?

85

QRAM 9,1

86

assumptions can better reflect the underlying nature of the phenomenon under research than can R methodology. The case study we examined showed that Q methodology can be used as an adjunct to R methodology in order to better relate the often rather contrived ways that quasi experiments have to be set up in order to confirm to statistical assumptions. We see the power of Q methodology as providing a means to make social science style experiments less contrived, or at least provide a means to relate them to a more ecologically valid representation of the problem space, which nonetheless provides a quantitative backing to the experimental design. This quantitative backing is compelling in that it provides statistically valid, robust description and validation of structures that relate to the research problem. Conclusion This article aimed to present Q methodology as an alternative research methodology. As with most research techniques, Q methodology has advantages and disadvantages. We suggest that it may be useful in research settings that involve somewhat emotive or sensitive themes, such as knowledge sharing in our case study, or social contract expectations as suggested for accounting research. Here the anonymity of the concourse might help surface important issues. However, researchers need to be aware that Q methodology may require a positivist approach when dealing with applied research and practitioners, and it can tend to dilute outlying themes and perhaps produce superficial results. First, we looked at how industry perceives Q methodology based on a case study, and propose several advantages over existing methods, including time benefits and employee “buy-in”. Second, we looked at whether Q methodology may be applied to data in a way that enables deeper insight. We examined this proposal by looking at an article found in the literature review and re-examining that data using a Q methodology perspective. The results show that Q methodology has the potential to bridge the divide between qualitative and quantitative research methods. On the one hand it may better reflect the nature of the phenomena under investigation, while on the other it may be used as an adjunct to compensate for the often contrived element of R methodology. However, we also saw that Q methodology has disadvantages, particularly in terms of management application. Producing factors is not enough. Q methodology must also explain the context underlying each factor, i.e. the statements, and develop an action plan based on analysis of the characteristics of respondents in each factor group. References Baumard, P. and Ibert, J. (2001), “What approach with which data?”, in Thietart, R.A. (Ed.), Doing Management Research, Sage, London, pp. 68-84. Brown, S.R. (1980), Political Subjectivity: Applications of Q Methodology in Political Science, Yale University Press, New Haven, CT. Brown, S.R. (1993), “A primer on Q methodology”, Operant Subjectivity, Vol. 16 Nos 3/4, pp. 91-138. Brown, S.R. (1996), “Q methodology and qualitative research”, Qualitative Health Research, Vol. 6 No. 4, pp. 561-7.

Cottle, C.E. and McKeown, B. (1980), “The forced-free distinction in Q technique”, Operant Subjectivity, Vol. 3 No. 2, pp. 58-63. Deegan, C. (2009), Financial Accounting Theory, McGraw-Hill, North Ryde.

Q methodology: is it useful?

Girod-Seville, M. and Perret, V. (2001), “Epistemological foundations”, in Thietart, R.A. (Ed.), Doing Management Research, Sage, London, pp. 1-30. Guthrie, J., Petty, R. and Ricceri, F. (2006), “The voluntary reporting of intellectual capital: comparing evidence from Hong Kong and Australia”, Journal of Intellectual Capital, Vol. 7 No. 2, pp. 254-71. Kaplan, R.S. (2006), “The competitive advantage of management accounting”, Journal of Management Accounting Research, Vol. 18, pp. 127-36. Kendall, J.E. and Kendall, K.E. (1993), “Metaphors and methodologies: living beyond the systems machine”, MIS Quarterly, Vol. 17 No. 2, pp. 149-70. Laughlin, R. (1999), “Critical accounting: nature, progress and prognosis”, Accounting, Auditing & Accountability Journal, Vol. 12 No. 1, pp. 73-8. Lehmann, C. and Norman, C. (2006), “The effects of experience on complex problem representation and judgment in auditing: an experimental investigation”, Behavioral Research in Accounting, Vol. 18, pp. 65-74. McKeown, B.F. and Thomas, D.B. (1988), “Q methodology”, Quantitative Applications in the Social Sciences, Sage, Newbury Park, CA. Massingham, P. and Diment, K. (2009), “Organizational commitment, knowledge management interventions, and learning organization capacity’?”, The Learning Organization, Vol. 16 No. 2, pp. 122-42. Meloche, J.A., Hasan, H.M., Willis, D., Pfaff, C. and Qi, Y. (2009), “Co-creating corporate knowledge with a wiki”, International Journal of Knowledge Management, Vol. 5 No. 2, pp. 33-50. Nester, M.R. (1996), “An applied statistician’s creed”, Applied Statistics, Vol. 45 No. 4, pp. 401-10. Neuman, W.L. (2006), Social Research Methods, 6th ed., Pearson Education, Boston, MA. Polanyi, M. and Prosch, H. (1975), Meaning, University of Chicago Press, Chicago, IL. Roslender, R. (2006), “Critical theory”, in Hoque, Z. (Ed.), Methodological Issues in Accounting Research: Theories and Method, Spiramus Press, London, pp. 247-70. Stephenson, W. (1953), “Postulates of behaviourism”, Philosophy of Science, Vol. 20, pp. 110-20. Stoecker, R. (1999), “Are academics irrelevant? Roles for scholars in participatory research”, American Behavioral Scientist, Vol. 42, pp. 840-54. Tapscott, D. and Williams, A.D. (2007), Wikinomics: How Mass Collaboration Changes Everything, Penguin Group, New York, NY. Tinker, T. (2005), “The withering criticism: a review of professional foucauldian, ethnographic, and epistemic studies in accounting”, Accounting, Auditing, and Accountability Journal, Vol. 18 No. 1, pp. 100-35. Warfield, J.N. (1976), “Complexity and cognitive equilibrium: experimental results and their implications”, Human Systems Management, Vol. 10 No. 3, pp. 195-202. Willis, D.J. (2004), Figure 1 “personal communication” – this model of information repositories at BSR was first developed by Dr BR Morrison in 2001, and extended by Dr D.J. Willis, in 2004, to include the wiki.

87

QRAM 9,1

88

Further reading Rossiter, J.R. (2002), “The C-OAR-SE procedure for scale development in marketing”, International Journal of Research in Marketing, Vol. 19 No. 4, pp. 305-35. Xin, Z. and Huang, S. (1961), “How to construct new China’s theoretical accounting basis”, New Accounting, Vol. 1, pp. 12-16. Corresponding author Peter Massingham can be contacted at: [email protected]

To purchase reprints of this article please e-mail: [email protected] Or visit our web site for further details: www.emeraldinsight.com/reprints