Generalizability

When evaluating a study, besides considering its trustworthiness, it is important to consider the generalizability of the findings and the theoretical statements related to them (Maxwell 2008). In general, theoretical statements that lack generalizability are lacking in usefulness as well (Lee and Baskerville 2003). In qualitative research, as in the case of this dissertation, the research is typically conducted in a single or very small number of research settings (Maxwell 2008), where no great emphasis is placed on generalizability beyond the case (Lee and Baskerville 2003).

While the nature of generalizability in qualitative versus quantitative studies differs, this does not mean that in the former, generalization is forbidden or lacks usefulness (Lee and Baskerville 2003). To support generalization in the context of qualitative research, Lee and Baskerville (2003) introduced a generalizability

framework, wherein four types of generalization are classified based on 1) whether they create generalizations from empirical or theoretical statements and 2) whether they create generalizations to empirical or theoretical statements. With this, they attempt to emphasize that generalization is not limited to statistical generalization but that there is value in other types of approaches as well (Lee and Baskerville 2012).

The generalizability framework has received criticism concerning its view on induction. For example, Tsang and Williams (2012) have suggested that instead of generalizability, Lee and Baskerville (2003) should use the concept of applicability.

This, according to Lee and Baskerville (2012), indicates a somewhat limited view of what generalizability can mean. For example, Lee and Baskerville (2012) state that the two classifications of generalizability allow for their mutual existence when generalizability is not considered to belong to the evaluation of statistical research.

What these two perspectives on generalization agree on is the fact that generalization cannot be extended to settings where it has not been empirically tested (Lee and Baskerville 2003; Tsang and Williams 2012). When the attempt is to generalize from rich descriptions of a single case to rich insights and not specific facts (Lee and Baskerville 2003), this line can be blurred, and the concept of applicability by Tsang and Williams (2012) seems quite reasonable.

When the insights of the case are analyzed concerning their generalizability or applicability beyond the case itself, the level of analysis needs to be considered (see Figure 25). The present case study was conducted inside a centralized IT department of a municipality and guided by the structures and values of a public sector organization. The phenomenon studied, on the other hand, is not limited to the context of one public sector organization but considers all digitally transforming organizations, which, again, can be considered a subset of all organizations.

Figure 25. Levels of generalization.

The definition above is still somewhat simplistic when it comes to generalization.

For example, the study of this dissertation was conducted in Finland, which in itself has a different governmental history and culture than other European or even Nordic countries. This means that in the context of change, the objectives but also the approaches to change considered acceptable can differ from other countries (Pollitt and Bouckaert 2011, pp. 49, 115). Similarly, the context of Finnish municipalities makes this study a somewhat special case as Finnish municipalities have comprehensive rights to self-govern their operations and decide how they produce their statutory services (Suomi.fi 2021). Moreover, because the case municipality is one of the largest municipalities in Finland with extensive IT resources, it is a rare case among Finnish municipalities. Consequently, the findings should be generalized cautiously.

While the case can be seen as special and its findings as ungeneralizable, the insights derived from the case are not. For example, the insights gained from the answer to the first research question are not limited to large public sector organizations. For instance, one of the key findings, i.e., the need to integrate agile values and incorporate ambidextrous attitudes, is recognized as critical for all types of organizations (cf. Vial 2019). The way the agile values and ambidexterity can be incorporated into the context of the public sector again limits the generalizability to the context of public sector organizations, and as the findings and insights derived from them are reliant on an organizational structure where there are independent business units, generalizability beyond the municipal context is questionable. This does not mean that other public sector organizations or even other organizations

striving for digital transformation could not benefit from the insights of this dissertation.

Whereas the answer to the first research question can be classified as a generalization from empirical findings in the generalizability framework of Lee and Baskerville (2003), the answers to the second research question and its sub-questions rely more on the use of theories to create insights. For example, while many of the identified tensions were case-specific, their classification under the three tensions of digital transformation (tensions revealed by change, hindering change, and driving change) resembles the paradox perspective’s division of tensions into latent and salient (cf. Smith and Lewis 2011). The latent and salient tensions of the dynamic equilibrium model of organizing are not bound by any specific context (Smith and Lewis 2011). The same applies here, indicating that the generalization of the three tensions of digital transformation is not bound by the context of this dissertation.

This is the case with the insights of the two sub research questions of research question two. While the identified tensions are case-specific, the insights related to the role and risks of the tensions of digital transformation are such that they can also be considered generalizable or applicable in the context of other organizations aiming to digitally transform.