5. Method
5.5 Validity and reliability
capa-bility framework resulting in a proposition for a revised pricing capability framework (section 7.2).
Table 5.3 Operationalization.
Level Constructs and phenomena
Unit of analysis (main concept)
Pricing capability
Theoretical concepts (sub-concepts)
Pricing capability ele-ments
Pricing activities Pricing policy
Type of phe-nomenon used to guide empirical study
Assets/routines or dis-crete bundles of assets and routines enabling the development and implementation of pricing policy
Activities aimed at the development and im-plementation of pricing policy
Price level and price variance over products, customers or time
Preliminary framework
Pricing organization Pricing information systems
Pricing skills
Pricing policy development Demand analysis Cost and profitability analysis
Competitor intelligence Communication and negotiation
Price discrimination Price elasticity leverage Operating leverage
The process of arriving at empirically grounded definitions of key con-cepts is only the first part of designing a valid empirical study that as-sures that empirical observations corresponds to theoretical concepts.
The second part involves the relationship between these definitions and data collection procedures, and the consistency of this relationship over time. The special measures taken in order to increase the likelihood that definitions of concepts are consistent with the type of data collected was largely accounted for in section 5.3.2. Such measures involve the use of different types of data sources to allow for data triangulation (to avoid systematic errors accruing from the use of one particular data source) and the use of a generic case-study protocol (to avoid inconsistencies in how concepts are represented).
5.5.2 Reliability of data sources
The reliability of the data collection procedures used in this study can be viewed from two perspectives. First, reliability requires that the re-search is conducted and reported in a transparent way, allowing other researchers to evaluate results and possible test conclusions in a second
study. This has been accomplished by accounting in detail for how theoretical concepts have been generated and operationalized into em-pirical indicators that are accessible to the selected data collection pro-cedures. Specific measures taken in this area include the procedure for developing the preliminary pricing capability framework, the use of a case-study protocol derived from the preliminary framework to guide data collection, and the explicit account of data collection procedures given in this chapter. Second, reliability requires that the different data collection methods used in the study generate accurate empirical obser-vations. As mentioned in section 5.3.3, this study has relied on two main types of data sources, semi-structured interviews and documents.
In addition, direct observation and physical artifacts were, to a limited extent, used as complements to the two main forms of data sources.
To large extent, the reliability of the results of this study are dependent on whether the studied phenomena have been captured by the two main forms of data sources, semi-structured interviews, and documents.
As discussed in section 5.3.3, interviews constituted the main form of data source, while documents were primarily used for validating inter-view statements of a more tangible or factual nature (primarily quanti-tative statements about prices, profitability, sales, market share, etc.).
This naturally puts a lot of emphasis on the use of interviews as a means of data collection.
The notion of the interview as a method of collecting valid empirical data has been criticized by several authors (e.g. Silverman, 2001; Alves-son, 2003). Much of the critique concentrates on the extent to which interview statements can provide an accurate representation of “facts”
outside the interview situation,36 and whether respondents can be as-sumed to give an accurate account of facts.37 Naturally, the researcher’s judgment of the truthfulness of a statement can never be completely certain. The problem consists of three parts; whether a respondent is actually giving an accurate account of facts, being able to detect insuffi-ciently accurate accounts once they appear, and examining whether
36See, for example, Silverman (2001) on the use of interviews compared to study-ing naturally occurrstudy-ing talk and events (observation).
37See, for example, Alveson (2003) on the different ways of interpreting interview results and the potential pitfalls of a naïve and non-reflexive approach to inter-preting verbatim statements by respondents.
statement by different respondents, or data obtained from other types of sources, are coherent.
Issues concerning the reliability of interview statements have primarily been addressed by means of triangulation between the statements from different respondents, and between different types of data sources (pri-marily interviews and documents). In addition to triangulating differ-ent data sources, an effort was also made to increase the likelihood of getting accurate accounts from respondents and increasing the chance of detecting ambiguities or inconsistencies. This was done by conduct-ing longer interviews that allowed respondents to speak relatively freely about the subject at hand, and returning to the same subject several times from different perspectives while explicitly questioning any form of inconsistencies with critically oriented questions or rephrasing of prior questions.
5.5.3 Causality and generalization of results
The aim of this thesis should be seen as primarily driven by a theory developing and explorative ambition. This aim has strongly affected the research design of the study in a direction allowing specifically for con-ceptual integration and the empirical exploration of different aspects of pricing capability. Naturally, such an ambition limits other aspects of the research design. Hence, the study should not be seen as testing or corroborating certain hypothesized causal relationships in a sample of cases, but rather as an attempt to develop the concept of pricing capa-bility, and integrate it with mainstream strategic management theory.
The result of this process is thus a set of empirically and theoretically grounded propositions that should be suitable for future corroborative studies and statistical generalization. Following the logic outlined above, the generalization of results should be seen as analytical rather than statistical (Yin, 2003).