«Comparing to Improve, or Simply to Assert? A Case Study of the Application of the Benchmarking Theory within the Public Sector Lii Lindgren Wictor ...»
3.1 Qualitative Scientific Theoretical Approach This study was conducted using a qualitative scientific theoretical approach, since it was most eligible with regards to the aim and data accumulation. The aim was to receive an in-depth understanding of the benchmarking process in the public sector, and qualitative interviewing can provide insights regarding existing complexity. The qualitative approach perceives the world as a social construction that is constantly shifting (Bryman and Bell, 2011). To generate the empirical data, two research methods were conducted. Primary data5 was obtained through interviews, and secondary data6 was acquired from document analysis. The document analysis complemented the primary data, and it was a recurrent aspect in the interviews, relevant for the obtained result.
Through an examination of how the active participants in Network Södertörn interpret their world, it is possible to attain a clearer understanding of the social world that they are part of. This concept is known as interpretivism, which is an epistemological position in qualitative research (Bryman and Bell, 2011). When ontological considerations are regarded, a position referred to as constructionism is acknowledged. This states that social characteristics results from interactions between individuals, as opposed to be regarded as a "phenomena ´out there’" (Bryman and Bell, 2011, p. 386), which indicates that the respondents can be seen as individuals who partially constructs the social world.
The empirical results presented from the interviews are based on the respondents’ perceptions of the network, and these arguments serve as a base for interpretations. By conducting interviews the ability to gain insights regarding the functioning of the work prevails, and deeper knowledge can be established regarding how individuals act within this context. Thus, drawing on a Primary data can be described as using data produced by the researchers themselves (Bryman and Bell, 2011).
Secondary data can be described as using data produced by someone else (Bryman and Bell, 2011).
hermeneutics approach regarding making interpretations of human actions (Bryman and Bell, 2011).
3.2 Research Methodology 3.2.1 Case Study - Network Södertörn Contemplating the stated aim of the study, Network Södertörn constituted a suitable subject to scrutinize. The network has been a recurrent project for many years, and can in this study be regarded as a solely organization, with a structure and purpose of its own. This made it appropriate to use a qualitative case study, which commonly is associated with a geographical location or a bounded organization (Bryman and Bell, 2011). Case studies tend to make thorough assessments of its complexity and settings, which is in line with this study. This study can further be viewed from an instrumental perspective, meaning that focus is aimed on “understanding a broader issue…” (Bryman and Bell, 2011, p. 61).
The general theory of benchmarking will be used, which can describe the case in question.
During the study both the empirical findings, as well as the presented theory, has been altered and refined. As stated, the empirical findings made from conducted interviews have been complemented with a document analysis. This indicates an abductive research approach, which includes inductive and deductive aspects, but also addresses additional elements, e.g.
understanding (Alvesson and Sköldberg, 2008). Furthermore, internal validity is argued to prevail when theories are in congruence with the findings (Bryman and Bell, 2011), which this strongly indicates. This also draws towards hermeneutics, which “wants to understand and not just comprehend” (Thurén, 2007, p. 103).
3.2.2 Semi-Structured Interviews A number of 14 interviews were carried out, with 15 respondents. Interviews were conducted with each municipality within the network to receive a comprehensive view where multiple considerations had been acknowledged. The network was approached from two different perspectives to receive insights from different levels. Seven interviews were carried out with CFOs, and seven interviews with officials (see Table 1). For a further presentation of the respondents see Appendix III. After conducted 14 interviews the empirical findings where perceived to be saturated. Further, indicating that the information presented, were not of new or relevant kind (Bryman, 2002).
The interviews conducted were of semi-structured nature, and a general interview guide, containing three themes with a few main questions in each theme, was used (see Appendix IV).
An interview guide is essential to retrieve adequate information for analysis. However, when attempting to reach in-depth insights regarding the unknown it should be general (Gerson and Horowitz, 2002). This approach gives the interviewees ability to reply with large wiggle room, and questions that arise from those answers can be asked directly, implying a flexible process (Bryman and Bell, 2011). This enabled the possibility to discuss additional areas not addressed in the guide, further making each interview different in their kind. However, since they all emanated from the same interview guide this provided the ability to compare them against each other for analyzing.
3.2.3 Sampling Strategic choices should be made regarding the selection of respondents (Bryman, 2002), since it is argued to be decisive for the obtained result (Gerson and Horowitz, 2002). Therefore a strategic sampling strategy was adopted, where respondents with relation to the network where contacted. Contact details to the officials were attained from the networks webpage, and for the CFOs through each municipality’s website. Although the majority respondents were selected through strategic sampling, two were selected using snowball sampling. These respondents where obtained by asking two interviewees about the knowledge of any additional prospective participants apposite for the study. These subjects were presumed to be entitled, since they possessed desirable competence and thus, could generate in-depth analysis.
The first interview was conducted with two respondents, where a better understanding of the network was established. This further provided insights regarding the networks structure and responsibilities, which paved the way for the choice of interviewing multiple levels of participants, i.e. CFOs and officials.
3.2.4 Interviewing, Transcribing and Coding The remainder interviews were conducted with one respondent at the time, to avoid that respondents influenced each other’s answers. The interviews were conducted face-to-face during a three-month period, and they lasted between 30-90 minutes. Thirteen of them took place at the municipalities, and one was conducted in a public place, since the interviewee recently retired.
The interviews were conducted in Swedish, and have been translated to English; this includes stated quotes and the interview guide. All interviews were audio-recorded for personal use, with approval from the respondents. For qualitative research this is recommended, since it facilitates taking every aspect of the interview into account, i.e. both what is said and how it is said (Bryman and Bell, 2011). Further, all attention can be directed towards the respondents, and on formulating complementary questions. Throughout the interviews, notes were taken continuously to enhance the understanding of the context.
Transcribing was conducted for the majority of the obtained data, although, focusing on the imperative parts, in relation to the aim. All interviews were listened through multiple times to receive a comprehensive understanding. The material transcribed provided opportunities to review the answers multiple times and enables the ability to conduct adequate interpretations (Bryman and Bell, 2011).
Coding is considered an important tool to interlink findings and organize statements (Miles and Huberman, 1994), and is conducted to categorize similar answers, in order to receive specific concepts to be analyzed (Rubin and Rubin, 1995). After the interviews had been transcribed, the answers received were coded with regards the study’s aim and research question. Coding was conducted by reading through the transcripts to get a comprehensive view of the material, and to receive an initial idea regarding important aspects. The stated themes from the interview guide were considered when subcategories where determined. The transcript was further reread;
appointing similar answers with a certain color, which further was comprised to three different themes (see Appendix V). These codes further served as a base when conducting the analysis.
3.2.5 Document Analysis A secondary data collection was conducted by studying a document, to complement the interviews. A recurrent aspect during the interviews was the produced project directive, which the work should emanate from. The document was attained from some CFOs and became relevant to study to gain insights regarding the network, in particular the stated aim and participants’ responsibilities. The project directive can, with regard to Bryman and Bell’s (2011) description, be classified as an organizational document, which often are valuable in case studies The document analysis is acknowledged through an approach termed ethnographic content analysis, outlined by Altheide (2006), which includes searching for themes in the documents (Bryman and Bell, 2011). By this position, an assessment of a project directive in the public realm was conducted, where vital features were acknowledged, and used for analysis.
3.3 Criticism 3.3.1 Subjectivity Qualitative research has tendencies of becoming subjective, where the researchers considerations and interpretations are too apparent (Bryman and Bell, 2011; Thurén, 2007), or where statements have been adjusted to better suit the topic (Gustafsson et al, 2004). The questions asked during the interviews where formulated so they could be answered in broad terms. Hence, avoiding asking leading question, easily answered by yes or no. Interpretations of the respondents’ answers have further been made, which increases the risk of subjectivity, although this was attempted to avoid, and thus striving to increase objectivity.
3.3.2 Generalizability and External Validity The ability to generalize qualitative research is constrained, i.e. external validity is low. This can be particularly prevalent when using unstructured interviews, because the number of respondents is often few and active within a specific organization (Bryman and Bell, 2011). This study did not attempt to generalize the findings to the population, since the sample was not random, nor large enough. However, the research focused interest in obtaining theoretical generalizations and understanding the complexity of the specific case (i.e. particularization), which, as argued by Bryman and Bell (2011), is common for case studies. Although, transferability, which describes how findings applies to other contexts (Bryman and Bell, 2011) is argued to be higher, since municipalities within the network are of various sizes, ranging from small to large.
3.3.3 Snowball Sampling Two respondent where selected through snowball sampling, making it important to recognize that the person who recommended them was provided the ability to impact the findings. This, since they could recommend a person who they know agrees with their opinion. Although, both respondents that were appointed where participants with long experience in the network, which could explain why they were recommended, further decreasing the potential negative impact snowball sampling could have had.
3.3.4 Disadvantages regarding Semi-Structured Interviews and Coding The most evident drawback of semi-structured interviews relates to interpretations. The researchers interpretations might be biased or the respondents’ answers might be misunderstood.
Interpretations are made when interviewing, coding, analyzing, etcetera, indicating that this could strongly have had an impact on the result. One interview was conducted with two respondents simultaneously, which could have increased the risk of them influencing each other.
This was acknowledged, and thereafter the interviews were held separately, indicating that interviewing is a learning technique as argued by Bryman and Bell (2011). A level of saturation was reached when interviews in eight workgroups out of thirteen had been conducted. However, since insights regarding all operational areas were not attained, this could impact the study’s result.
As stated earlier, all interviews were recorded. This can make the respondents more concerned regarding how correct they answer questions (Bryman and Bell, 2011), further indicating that the true aspects of certain situations were not articulated. Thus, the interviews allowed scrutinizing the respondents’ body language and behavior, which, to some extent, could reduce the concerns.
3.3.5 Secondary Data The secondary data that has been used relates to articles, books and reports. A critical approach was aimed at all data at selection. Regards were further taken to the production year, to make sure that the information still was applicable. The majority of the data has been distributed by acknowledged publications, which increases the credibility. Secondary data was also used in form of a document produced by Network Södertörn. This document was considered as trustworthy, since the networks participants communicated it. The document was used to reinsure that aspects stated was perceived correctly. Further, benchmarking is an area of extensive research, and it is therefore vital to recognize the risk that all previous data have not been identified, which could have influenced the results.