Autoethnography and Theory Testing

One of the problems in construction management research is how to harvest the living experience of people working in the profession. Surveys and interviews are the common methods, but they have obvious problems, not only with defining terminology and targeting the appropriate populations but also with the knowledge of the respondents and their willingness to reflect. Different forms of ethnography, in their conventional forms overcome most of these problems, but create a new set of problems, particularly in terms of our ability to generalise and verify the results. The purpose of this paper is firstly to advocate that one such approach: autoethnography, which allows researchers to introduce their own lived experience into their research and secondly, to show how this approach can be structured to overcome common – and justified - objections to autoethnography. To do this we use a study involving a recent research project investigating the importance of emotional intelligence (EI) in the management of large projects. The paper shows that autoethnography can be applied analytically and rigorously so that it can be used for theory testing and theory building. In doing so, it opens up an untapped source of data to researchers – their own living experience.


Introduction
The conventional format of a research project is a review of what is already known about the subject matter or phenomena under investigation -the literature review -a statement of one or more hypotheses followed by the collection of data, their analysis and organisation into useful information -quantitative, qualitative or mixed -a discussion of the findings, including, when appropriate, a validation of the results and finally, the conclusion.There are some variations in this pattern, primarily about the use of hypotheses and the method of analysing data.Different sources of data require different methods of collection.One of the problems in disciplines like Construction Management is how to access the large pool of information stored in the lived experience of practitioners in the field.Much of this information is never shared or made available to researchers and is permanently lost as the practitioner ceases to be actively involved in the profession.Attempts to retrieve this knowledge, typically using methods such as surveys and interviews, all suffer from weaknesses including: • Bias introduced as a product of the survey method that is chosen (Watt et al., 2002 ).
• Bias introduced as a result of systematic differences between respondents and nonrespondents (Richardson, 2005).• Difficulties of obtaining a representative sample ( Johnson, 2012).It is also difficult to identify the right people with the right knowledge to survey (i.e.identifying the target population).Having identified the population there may also be problems in obtaining a robust sample.Additionally, response rates are often low, making meaningful statistical analysis and control over biased responses difficult.As a result, surveys often turn into opinion polls of subjects who may, or may not, know what they are talking about -if they bother answering -where poor or non-existing definitions of terms lead to misunderstandings and answers that are superficial and badly considered.Despite these well recognised problems our discipline's reliance on surveys as a research method has confined our research into very narrow segments of construction management.
Examining current research and research practices, it would appear that to restore the faith of practitioners in our research and to move the theoretical work of the discipline forward, construction management needs new approaches to gather data, not more advanced statistics to extract even more sets of questionable findings from dubious data sets.
The purpose of this paper is firstly to advocate one such approach: autoethnography, a method that allows researchers to introduce a radically new source of data -namely their own lived experience -into their research and secondly, to show how this approach can be structured to overcome the common -and justified -objections to autoethnography.To do this we use a study involving a recent research project investigating the importance of emotional intelligence (EI) in the management of large projects (Livesy, 2016).

The history of autoethnography
Autoethnography has been used as a qualitative method that combines the techniques of autobiography and ethnography.This method allows practitioners to use their experience to reflect on how they work, the skills they have found necessary and those events that brought the need for these skills to their attention.The problem is that, in many applications, this approach lacks external verification and is not therefore sufficiently rigorous to be accepted by many researchers.
Researchers engaging with autoethnography tend to use different approaches in the sense that they place different emphasis on auto-(self ), -ethno-(the cultural link) and -graphy (the application of a research process) and different examples of autoethnography fall at different places along the continuum of each of these three (Reed-Danahay, 1997).Ellis et al. (2011) have suggested differences in autoethnography may be judged by the emphasis the authors place on the researcher and interaction with others, the context in which the research takes place, the emphasis on the power of a position in a relationship and finally by the degree to which traditional methods of analysis are employed.They identify eight styles: • Reflective, in which the researchers, in addition to simply reporting stories resulting from their research, place a heavy emphasis on their own reflections including the interaction of their values, beliefs and experiences whilst doing the research (Ellis (2004); • Indigenous works, in which the researchers concentrate on power differences and their effects (Denzin, 2008); • Layered accounts, in which the researchers place a considerable emphasis on the data collection and literature review as well as their experiences (Doloriert and Sambrook, 2011); • Narrative accounts, in which the researchers incorporate their experience into the research concerning the group being studied (Tedlock, 1991); • Interactive works, in which the researchers deal with highly emotional issues which are probed by in-depth interviews as exemplified by Adams (2006); • Community autoethnographers, in which the researchers involve themselves as part of a community and discuss how the community deals with the social and cultural issues associated with that community (Toyosaki et al., 2009); • Co-constructed writing, in which there is a shared work between the parties involved in the relationship or a particular situation as in Berg and Trujillo (2008); • Personal narrative style autoethnography, which is used when the author considers some aspect of himself to be worthy of research as in Berry's work (2007).Doloriert and Sambrook (2011), however, have suggested that autoethnography can be considered as having developed into three epistemological alternatives viz: • Evocative interpretivism that uses an emotive style typified by Ellis and Bochner (2000); • An analytical style typified by Anderson (2006a); • A style dealing with power conflicts typified by Jones (2008).Writers using the evocative style tend to use an emotional style and aim to fully engage the reader in their narrative.A definition of this form of autoethnography is provided by Denzin who says that users of this style "bypass the representational problem by invoking an epistemology of emotion, moving the reader to feel the feelings of the other" (1996).
However, entertaining this may be, it has limited use for generalisations.Mitch Allen (publisher of Left Coast Press) said in a personal interview in a May 2006 report: An autoethnography must look at experience analytically.Otherwise [you're] telling [your] story-and that's nice-but people do that on Oprah [the U.S.-based television program] every day.Why is your story more valid than anyone else's?What makes your story more valid is that you are a researcher?[The answer is, ] You have a set of theoretical and methodological tools and research literature to use.That's your advantage.If you can't frame it around these tools and literature and just frame it as "my story," then why or how should I privilege your story over anyone else's I see 25 times a day on TV? (Ellis et al. (2011, p. 276) This may be compared with Anderson's statement regarding analytic autoethnography: The definitive feature of analytic autoethnography is this value-added quality of not only truthfully rendering the social world under investigation but also transcending that world through broader generalization " (2006a, p. 388).
When the objective of the research, as in the paper forming the background to this study, is improving theoretical understanding -which involves theory building and/or testing -the conventional critisism of qualitative, non-generalisable stories, which are generally leveled at research using the evocative style, can be avoided by using the analytical style suggested by Anderson (2006), in which he notes the following key requirements: • The researcher is a complete member of the social event under study.There is awareness of the researcher's connection to the situation under investigation and their impact on it (analytic reflexivity).• There is visibility of the researcher's own experiences.
• There is dialogue with informants beyond the self.
• There is commitment to theoretical analysis requiring not simply the documented experience of the event but also to provide some broader understanding of the situation under investigation.A further characteristic of analytic autoethnography is its commitment to an analytic agenda.The purpose of analytic autoethnography is not simply to document personal experience, to provide an "insider's perspective," or to evoke emotional resonance with the reader.Rather, the defining characteristic of analytic social science is to use empirical data to gain insight into some broader set of social phenomena than those provided by the data themselves.
Finally, it can be seen that the analytical style described by Anderson is similar technique to the layer account discussed by Doloriert and Sambrook (2011) The use of analytical style is supported by several researchers including: (Atkinson, 2006, Buckley, 2015).For a more detailed review of the issues surrounding analytical autoethnography see (Anderson, 2006a, Atkinson, 2006, Anderson, 2006b, Burnier, 2006, Denzin, 2006, Ellis and Bochner, 2006, Hunt and Junco, 2006, Vryan, 2006) The approach used in the case study described below meets the requirements laid out by Anderson in that: the researcher had worked in project management for over 35 years of which 10 years were spent in managing Mega-Projects, the work demonstrates a connection to the research and the researcher's impact on the subject matter, experiences, are clearly visible to the reader and dialogs with others than the researcher are clearly revealed.Lastly the work was used to broaden the understanding of the skills required in project management and the findings moved EI from a nice to have skill for project managers to a necessary requirement.
All this makes it possible to utilize data not available in any other approach to data collection.Ellis et al. quoting Mitch Allen (2011 p 276, suggests that a way to further overcome the objection against autoethnography of a lack of an analytical approach, is for the researchers to use their experience to develop a framework, to compare that experience and framework with existing research (Ronai, 1995, Duncan, 2004) and to remember the guidelines provided by Anderson (2006a) as to what constitutes analytical autoethnography.
The research meets this suggestion by following Yin's s recommendation of using triangulation as a method for validating qualitative work.In research, triangulation generally refers to "collecting evidence from different sources" (Yin, 2010, p79).In his research the author used two additional methods of generating data.In common with most forms of quantitative research, it started with a comprehensive literature review that was used to establish the framework for the autoethnography.
The literature review generated the theoretical framework for the research and suggested the hypotheses to be tested.For verification, the author then employed a Delphi study to test the findings of the autoethnography by incorporating the lived experience of other practitioners.
Additionally, Creswell (2013) recommended using mixed methods to meet the challenge of verification often levelled at single method approaches such as autoethnography.In the research in question the use of autoethnography (qualitative method) and a Delphi study (quantitative method) met the requirements of a mixed method approach and thus strengthened the verification of the results.
Satisfying these requirements doesn't automatically guarantee good research.There are many general challenges to autoethnography and the next section discusses how they were overcome in the case study research project and in doing so indicating general guidelines for using this method.

Challenges to the use of autoethnography
Autoethnography as an approach has attracted some criticism from more conventional researchers, as for example Delamont (2007).These criticisms challenge autoethnography on the grounds that: • It lacks an analytical approach.
• Its focus on the self is self-indulgent and it uses an overly emotional style.
• It is unethical.
• There is a risk of false memories.The method of dealing with these challenges in analytical autoethnography are discussed below.

IT LACKS AN ANALYTICAL APPROACH AND CANNOT BE VERIFIED.
While this criticism might be true for most forms of autoethnography it misses the point in analytical autoethnography.The analytical approach and the triangulation ensures that the approach suggested here corresponds to the analysis and verification of more conventional quantitative research.The same argument applies also to another frequent criticism: that autoethnography tends to be experiential and so lacking in analytical outcomes (Delamont, 2007).

IT LACKS OBJECTIVITY.
As the autoethnographer is personally involved in the research, the method has been challenged as lacking objectivity.This issue is answered by Ellis et al. who describe autoethnography as "one of the approaches that acknowledges and accommodates subjectivity, emotionality and the researcher's influence on research, rather than hiding from these matters or assuming they don't exist" (2011, p. 274).While it's true, that it's difficult to avoid subjectivity in any form of research, suggesting that a method is no worse than others, as Ellis et al. does, may be true, but is not a very positive argument.It ignores for instance that due to the literature review, the understanding and the knowledge of the issues by the researcher should be far superior to the normal understanding of survey participants, who, as outlined in the introduction, are often called on to respond to questions which they do not understand or where they have a very limited knowledge.It's likely, therefore, that the answers are more considered and reflective, and therefore more objective and thoughtful.

IT IS UNETHICAL
Autothnographers are forced to involve others who were involved in the experiences discussed and who may not wish to be identified.They may also have different memories of the events but are not afforded the opportunity to put forward their point of view.This problem has been highlighted and discussed by a number of authors (Ellis, 2007, Adams, 2008, Trahar, 2009, Etherington, 2009).Whilst it is very difficult, if not impossible, to ensure anonymity in this form of research the following steps can be taken to reduce the degree to which an individual could be identified.
1. Rely on a significant time interval having occurred to reduce the likelihood of others identifying the event.2. If the event occurred more recently, the place and time where the event occurred should not be identified.If 1 and 2 are not sufficient, engage in creative non-fiction to make changes to the narrative surrounding the event that, while not affecting the substance of the experience, will render identification of the individual(s) that the researcher was interacting with more difficult.
Finally, by dealing with the impact of the event on the researcher's thoughts and behaviour and not involving the feelings and thoughts of others, the potential to impact on the character(s) involved in the situation is reduced.

ITS FOCUS ON THE SELF IS SELF-INDULGENT AND IT USES AN OVERLY EMOTIONAL STYLE.
Autoethnography has been criticised as self-indulgence (Holt, 2003) and romantic in construction (Atkinson, 1997) and lacking in scholarship (Parks, 1998).The emotional writing style that is intended to produce an overly emotional response (Reed-Danahay, 1997, Ellis andBochner, 2000, Holt, 2003) has been ruled out in analytical autoethnography.The writing style should be the same as for more traditional research.

THERE IS A RISK OF FALSE MEMORIES.
It is not uncommon for people who have experienced the same event to have different accounts of what they observed (Owen et al., 2009).
This was overcome in this study by the researcher dealing with what he had learned from an event and how that learning affected his behaviour.To a large extent, the details of the event recalled were therefore insignificant in comparison to the learning and ongoing behavioural changes they brought about.

Study methodology and results
The study discussed in this paper involves a major research project to establish the need for Emotional Intelligence (EI) in the management of large projects.It utilised data from the researcher's 35 years of experience with managing construction projects including three mega projects, based on the theoretical framework of the Goleman-Boyatzis model (Goleman et al., 2013).
As its basis the research used a triangulation method as recommended by Yin (2010), the first stage of which involved a literature review, which identified the problems in project management, the general relevance of EI and the specific significance of competencies in the Goleman-Boyatzis model (Goleman et al., 2013).Tables 1 and 2 below present a summary of the problems identified as a result of the literature review and the Goleman-Boyatzis model.

Table 1
Problems identified in project management as a result of the literature review

Fundamental problem
Resultant impact of the problem considered in the evaluation The project's limited time frame.
Building a cohesive team.Building trust within the team.Developing rapport with stakeholders.Developing a working control system.Obtaining organisational support.
The team members' diverse backgrounds.
Ambiguity and change.
Lack of a clearly defined project scope.Scope changes as the project progresses.Lack of information to make fully informed decisions.Team member changes.
Changes in the external environment (e.g., legislative and economic).
Changes in project team and stakeholder personnel.
Loss of a cohesive team.Loss of trust between team members.Loss of with key stakeholders.
Conflicts (the disagreements that arise prior to a formal dispute).
Those arising internal to the team.
Those arising external to the team but internal to the parent organisation.
Those arising with subcontractors.
Those arising with other stakeholders.Self Other

Awareness
Self-awareness: reading you own emotions together and recognising their impact.
Social awareness: attuned to how others feel.
Emotional self-awareness: recognising our emotions and their effects on personal performance including recognition of a tendency to avoid issues or situations that cause us discomfort.Accurate self-assessment: knowing our strengths and limits.Self-confidence: a strong sense of one's self-worth and capabilities.
Empathy: understanding others' feelings and perspectives and taking an active interest in their concerns.Organisational awareness: understanding the organisation's issues, dynamics and politics.Services orientation: recognising and meeting customer needs.

Management
Self-management: focused control towards the achievement of goals.
Relationship management: the ability to guide the emotional tone of the group.Conflict management: resolving disagreements when they occur or preventing a disagreement from happening or growing.
Teamwork and collaboration: working with others towards shared goals and guiding the group to achieve a collective goal.
In the second stage an autoethnography was developed which discussed the problems the author had encountered during his career and identified which of the competencies in the Goleman model would have been relevant in dealing with the issue identified, all of which were found to be of relevance.Finally, a Delphi group was established involving over 25 participants.Members of the Delphi panel: • Must have had over 20 years of experience in the management of construction projects; • Must have experience in managing a project greater in size than $500 million.There was no requirement for the panel member to be degree qualified, as it was thought that any potential participant meeting the above criteria certainly qualified as an expert in the management of large construction projects.Participants who had either worked for contractors or acted for the client were sought.
In the Delphi study panel members were asked to rank the usefulness of the specific EI competencies identified in the Goleman-Boyatzis model (2013) in dealing with the project problems presented in round 1.The results are presented in table 3 Table 3 Relevance of the Goleman-Boyatzis competencies to the project management problems identified in the literature

Competency Times found not relevant Relevance
Emotional self-awareness 3 98.3% Accurate self-assessment 3 98.3% Self-confidence 5 97.3% Emotional self-control 0 100.0%As can be seen the majority of the competencies were found to be 100% relevant with no competency had a relevance below 95%.
The Delphi members were also asked to rank the importance of the competencies in dealing with the project problems under discussion.These results were compared with those from the autoethnography in table 4.  The difference in the rankings obtained in the autoethnography to that of the Delphi panel is explained by the frequency of the competencies in the autoethnography being based on the author's developmental needs and their emotional impact on him (the higher the impact the more likely he would have been to remember the event) rather than their relative importance in successfully managing a project.The relative ranking obtained from the Delphi panel members reflects their view of the importance of the competencies in managing large projects.

Further details of the autoethnographical method employed
In order to establish a firm foundation for the research and to ensure that it is not viewed as a series of short stories the criteria suggested by (Eisner, 1991) for use in qualitative research provided the backdrop.These criteria consist of: • instrumental utility; • coherence; • consensus.To this suggested list may be added the production of a scholarly account.

INSTRUMENTAL UTILITY
It was this identified by (Eisner, 1991) as the most important test for qualitative work.It requires that the research must have an identifiable usefulness.To meet this requirement Eisner suggests that a study should achieve one or more of the following: • Assist its reader to understand a confusing situation, • Provide some guide as to the behaviour to be expected in similar situations, • Bring to light factors that had not been previously noticed.In the study project, the analytical approach, aimed at testing a theory, satisfied all of these requirements.The project also had an identifiable usefulness for the profession in that it highlighted factors involved in project management that are not normally considered part of the project management body of knowledge.While not providing a firm predictive model, it provided a framework that, if followed, gives a greater possibility of success.By achieving the first two points, the research will bring a further level of understanding to the problems involved in managing projects.

COHESION
It is suggested by (Eisner, 1991) that coherence in research can be identified by cohesion of the arguments presented, that is, they are internally consistent and supported by evidence.In this research the researcher used a literature review and a Delphi study to either confirm or rebut the results of the autoethnography.

CONSENSUS
To achieve consensus, it is not necessary to ensure that others find that the conclusions of the work agree with their own lived experience.It is only necessary that they agree that the conclusions reached are consistent with the evidence presented in the research.(There is an analogy here to legal argument presented in court proceedings.)

PRODUCTION OF A SCHOLARLY ACCOUNT
To ensure the production of a scholarly account the objectives of providing insight into phenomena under investigation and providing guidance to others as to where future research maybe carried out, were made a major goal of the research.

Summary and conclusion
Autoethnography has, some would say justifiably, a reputation for just creating nice or interesting stories without much scientific value in that the findings can't be generalised or verified.This paper has presented an approach that makes it possible to use autoethnography to test and/or build theories by structuring the research in a conventional way: creating a theoretical framework -a literature review gathering and analysing data -in this case from past experience of the researcher -and finally verifying the findings -in this case through a Delphi study and a comparison with the original theoretical framework.The paper starts by outlining different styles and approaches to autoethnographic research.The requirements of analytical autoethnography are outlined and discussed.It is then confirmed that the studythe testing of the Goleman-Boyatzis model satisfies all of these requirements.A Delphi study, using 25 experienced project managers over five stages examined and confirmed the validity of the test.The results confirm the Goleman-Boyatzis model and the importance of Emotional Intelligence in project management.It also confirms the use of analytical autoethnography for theory building and testing and thereby opens a way to use one of the most extensive but so far inaccessible sources of data in our discipline: the knowledge that the professionals apply every day in their work.
Emotional self-control: keeping disruptive emotions and impulses under control.Transparency: maintaining integrity and acting congruently with one's values.Optimism: persistence in pursuing goals despite obstacles and setbacks.Adaptability: the ability to adapt to change and work effectively as circumstances change.Achievement orientation: the drive to meet an internal standard of excellence.Initiative: the readiness to act in order to seize an opportunity.Developing others: sensing others' development needs and bolstering their abilities.Inspirational leadership: inspiring and guiding others either as a group or an individual.Influence: the ability to persuade others.Change catalyst: initiating or managing change.

Table 2
Summary of the Goleman Boyatzis model

Table 1 continued
Autoethnography and Theory Testing Construction Economics and Building, Vol. 18, No. 3, September 2018