R ESEARCH STRATEGY

A research strategy tells how to get an answer to a research question (Robson 1993). The aim of this research and the detailed questions of this research were introduced above.

These were:

1) What kind of theoretical framework could be used to describe the customer experience in a landlord-tenant relationship?

2) According to the theory, what are the main elements of the customer experience?

3) According to the theory, what dimensions do customers use when they assess these elements?

4) How is it possible to study the customer experience empirically in the context of a landlord-tenant relationship?

5) How do the theoretical dimensions outlined meet the empirical evidence?

These research questions are studied in descriptive terms. This means that the aim is to describe the phenomenon in question as it is, without trying to find correlations between

different variables. A descriptive research requires extensive previous knowledge of the phenomenon in question. (Robson 1993.) In order to gain the necessary knowledge, a thorough literature review was carried out prior to designing and implementing the empirical part of this study.

It is typical for a descriptive research approach that the study is non-experimental. This means that it is conducted in natural environments, not in artificial environments such as laboratories or other strictly controlled situations (Denscombe 1998). This is quite a typical situation in the human sciences, as it is difficult and sometimes unethical to study human behaviour in laboratory or otherwise controlled settings. (Robson 1993.) The non-experimental nature of descriptive research has some implications for the methodological choices of this study.

To get descriptive information about the research question, the conclusions are made inductively. Inductive reasoning begins with specific observations and measures from which patterns and regularities are detected. From these patterns it is finally possible to try to draw some general conclusions and/or theories. (Patton 2002.) This directs the methodological choices to some degree, as quantitative research is more often associated with a deductive approach, while qualitative research is more often inductive in nature. It is still important to remember that this is not an unbreakable “law” - quantitative research may also be based on induction and qualitative research on deduction (David & Sutton 2004).

The intention is to find an answer to the research question through qualitative research methods. Qualitative research is an umbrella term that covers a variety of styles of social research (Denscombe 1998). Hirsijärvi and her co-authors (1993) introduce seven features that are typical for qualitative research:

1) The data is gathered in natural and real situations. This was also typical for the descriptive research (Denscombe 1998).

2) Humans are used as instruments of data collection. The data of this study is also collected by individuals and from individuals.

3) Inductive reasoning. This logic is applied in this research.

4) Gathering the data with methods such as observations, group interviews and thematic interviews. This study uses both thematic interviews and group interviews.

5) The target group is selected for the purpose, not by using a random sample. The target group in this study is selected so that the limitations are kept in mind. On the other hand there were also other case specific factors that affected the case selection. These factors are explained in the articles.

6) The research plan goes through iteration during the research process. In this study this means, among other things, that the research questions were reformulated after the literature review. This iterative nature of the research becomes clear in the section about the research process (see 1.7).

7) The cases are seen as unique and the data is interpreted accordingly. This study uses cases that are unique and individual, and the case-specific features affect the findings.

The main reason for choosing qualitative methodology is that qualitative research methods offer more depth and detail than quantitative methods (Patton 2002). As research on the topic so far is scarce, there is a need to understand the phenomenon in a detailed and wide manner. Still, there are some problems associated with qualitative research. For example, Suonperä (2002) lists five problems; 1) it is possible to study only a limited number of cases; 2) generalizability is an issue; 3) it is difficult to collect and compare data; 4) the researcher influences the research process and the findings and; 5) the research process may affect the object of the study.

The first problem of qualitative research that Suonperä (2002) mentioned was that it is (1) possible to study only a limited number of cases. In quantitative research, it is possible to study thousands or tens of thousands or even more cases. In qualitative research, the number of cases varies from 1 to dozens, but qualitative studies with 100 or more cases are rare. In previous decades, at least in the Finnish context, it was typical to have 30-50 cases in qualitative studies, but the “invention” of saturation by grounded theorists decreased the number of cases. (Koskinen & al. 2005.) This study follows this trend as there are 11 cases in this research.

The main reason to see the small number of cases as a problem is that this raises the (2) question of generalizability. In quantitative research, the lack of generalizability is a major concern. For many, research without generalizable results is no research at all. For example, Spencer and Dale (1979; see also Cummings & al. 1985) argue that research based on a small number of cases does not allow for generalizations, and thus makes only minimal contributions to the body of scientific knowledge. On the other hand, it could be argued that this logic is based on a realist view of reality – stating that there is an objective truth behind everything.

The ontological and epistemological basic assumptions of this study make the issue of generalizability more complex. If reality is socially constructed, and changing with space and time, the possibility of vast generalizations may be questioned (Cuba 1990). The aim of this study is to build theory and test it qualitatively in case study settings. The empirical findings are interpretative in nature. This means that the results are representative of the interpretations of those experiencing the phenomenon under study. (Corley 2002.) It is not the goal to make any vast generalizations from the empirical data but to describe the customer experiences, as they were perceived in our case companies and to test the theoretical construct.

The problem of comparison is also linked to the question of generalizability. Comparing the findings of qualitative data to findings of other studies is often hard to do, and even comparing cases of the same research may prove to be problematic. It has been suggested that replication could help in this respect – replicating the research gives generalizability and allows for easier comparisons. (Yin 1984; 1994; 2003.) The replication may be done with similar cases, but on the other hand, it has been claimed that using totally different cases in replication adds theoretical strength (Kennedy 1979; Koskinen &

al. 2005).

However, this is easier said than done as the situations, instances or even the population in a qualitative study may affect the results. (Dul & Hak 2008.) The traditional response to this is sampling, but even that is not flawless (Firestone 1985). Lincoln and Guba (1985) suggest that the researcher should give the necessary contextual information so that the reader may contextualise the findings and make comparisons with this crucial knowledge (see also Firestone 1985).

The next problem of qualitative study mentioned by Suonperä (2002) was that data (3) collection and comparison of data are difficult. The difficulties in qualitative data collection are well described by, for example, Koskinen and his co-authors (2005). According to

them, in some qualitative research an extra case may mean 3 months of work, whereas in a quantitative survey one more case may mean that one needs to add one new e-mail address to a mailing list. This is closely linked to the problem of the low number of cases in qualitative research and generalizability. To get enough cases to have statistically generisable results would require an immense amount of time and money.

The fourth (4) and fifth (5) problems mentioned by Suonperä (2002) were related to the fact that both the researcher and the research process influence the study and the object of the study. Koskinen and his co-authors (2005) speak of reactivity when the researcher and the research process influence the object of the study. According to them, reactivity may be reduced by getting good prior knowledge about the object, having thorough ethical discussions with the object of the study, being entirely honest to the object and by thinking of what kind of picture the researcher gives of himself as a researcher. A constructivist would say that the research builds up as an interaction between the researcher and his research object (Guba 1990). Guba’s solution to this problem, again, is to report the contextual information about the research and the researcher so that the reader may make his own conclusions about how the researcher has influenced the research and its findings.