M ETHODOLOGY

This section explains and justifies the applied research method for the empirical study to successfully develop an answer to the main research question and achieve the research objectives as outlined in chapter one. Firstly, the overall research design will be outlined, followed by an explanation of the sampling procedure. Lastly, the methods and tools applied in the data collection and data analysis process will be discussed.

3.2.1. Research Design

The researcher suggests a single-case study as the overall research design followed by the guidelines developed by Yin.

Case studies try to ‘(...) illuminate a decision or set of decisions: why they were taken, how they were implemented and with what result.’ (Yin, 2009, p. 17) As Yin explains, empirical studies with a strong explanatory nature directly lead to case studies as the ideal research method. The method perfectly fits to the research objectives and questions, as case studies are a great tool to derive an answer to the how and why of the occurrence of certain events and circumstances. (Yin, 2009, p. 9; Saunders et al., 2009, p. 146)

As Yin suggests, the basis for a case study is a thorough literature review to develop a theoretical framework from which research questions and objectives will be derived.

(2009, 40) Both primary and secondary literature were used in the literature review and the development of the theoretical framework. The literature review was mostly focused on academic journals, as they are easier to access, are usually covered by a wide range of tertiary literature and offer great detailed insights into specific topics.

(Saunders et al., 2009, p. 69 f.)

Despite that literature suggest that case-studies should be conducted and findings validated through a combination of data collection methods (multi-method / mixed method), the researcher decided to only apply a mono-method due to time-constraints and a limited scope of the thesis. (Eisenhardt, 1989, p. 534; Yin, 2009, p. 3)

3.2.2. Data Collection

In the context of a case-study research design, Yin discusses six different sources of evidence. (Yin, 2009, p. 102) In scientific research, interviews are seen as one of the most important sources in the development of new knowledge, and it is among the best

fitting method for a case study. Yin describes three different types of interviews: in-depth, focused and formal interviews. The researcher decided to conduct in-depth interviews in the empirical study, as they allow the researcher to gain detailed insights into the opinions of interviewees. (Yin, 2009, p. 106)

The interviews were conducted in a non-standardized way, either face-to-face or through telephone calls. All interviews were recorded, strictly with the agreement of the interviewees, enabling the researcher to focus fully on the interview, unbiased information to be collected and all information to be precisely transcribed without any loss of data. (Saunders et al., 2009, p. 341) A semi-structured questionnaire was developed aligned with the research questions and tested prior the qualitative study with consultants from rpc. The questionnaire can be found under Appendix A.

3.2.3. Sampling

To develop a sample for the single-case study, purposive (judgemental) sampling was applied to pick the right candidates for the qualitative study. With purposive sampling the researcher has the opportunity to freely select cases which he believes will enable him to develop an answer to his research question. Saunders et al. suggest this method when researchers are working with small samples. (Saunders et al., 2009, p. 237 f.) As the overall research design is a single-case study and in-depth qualitative interviews selected as the primary data collection method, the researcher worked with a small sample of twelve experts to conduct the interviews. Deviant sampling was applied, as it allows the data collection of extreme cases and enables the researcher to learn the most about the field of study to answer to Main-RQ. (Saunders et al., 2009, p. 239) In practice, retail digitization experts from different companies were selected that are working closely in the field of study. The researcher placed further emphasis on sufficient work experience of the experts in the field of study to ensure that relevant data could be collected from the interviews. An overview about the selected sample can be found in Appendix B, but as many interviewees only participated under the aspect of full anonymity, little data can be presented.

3.2.4. Data Analysis

The empirical data was analysed with the Qualitative Content Analysis approach developed by Philipp Mayring. ‘The goal of content analysis is the systematic examination of communicative material (...).’ (Flick et al., 2004, p. 266) It is a method

with its origins in communication sciences and used is nowadays in combination with the objectives to understand the formal aspects of texts and identify hidden meanings.

To date, it is the most popular method to analyse qualitative material. (Flick et al., 2004, p. 266; Mayring & Fenzl, 2014, p. 543; Ramsenthaler, 2013, p. 30)

Figure 4: Qualitative Content Analysis Approach (adapted from Flick et al., 2004, 268)

The method is based on a very systematic procedure, with a category (code) system at its core. Codes represent relevant topics mentioned in the raw qualitative data in a short form. At the beginning of the analysis, main and sub-categories will be pre-formulated (inductive) based on the theoretical framework of the topic. In the second step, more gradual categories will be developed (deductive) and applied in a first analysis. After having processed around 10-50% of the material, categories will be revised and adapted in several iterative steps to ensure validity of the selected codes and categories. (Mayring & Brunner, 2009, p. 678; Flick et al., 2004, p. 288 f.; Mayring

& Fenzl, 2014, p. 544; Ramsenthaler, 2013, p. 24 ff.) Possible quality criteria for this method are replicability, reliability and triangulation. (Ramsenthaler, 2013, p. 25) The researcher selected this method as it is the most established procedure to analyse fairly large amounts of quantitative data. Due to its systematic and rule-governed

Issue, research questions

General definition of categories, fixing the selection criterion and level of abstraction for category formation

Gradual category formation from the material with reference to definition and level of abstraction; subtraction under old categories or formation of

new categories.

Revision of categories after about 10-50% of the material processed

Final processing of material

Analysis, eventually qualitative analysis

Check of formative reliability

Check of summative reliability

nature, the researcher was able to identify key information from the empirical data and reproduce it in chapter four. (Flick et al., 2004, p. 288) The program MAXQDA was utilised to define the codes and summarize the data in an organized way. The code system can be found in Appendix C.