Research Methodology
3.3 Inductive/Deductive Approach to Scale Development
Following the recommendations by Churchill (1979) and DeVellis (2003), a multi-stage development process was used to optimize the integrity of the new construct Negative Brand Personality and its associated measure. The overall stages in this process include item generation, scale development and scale evaluation.
Almost all researchers in scale development are in agreement that content validity is a minimum psychometric requirement for the appropriateness of the measurement (Hinkin, 1995; Churchill, 1979; DeVellis, 2003). Consequently, the items included in the scale need to measure the construct of interest (DeVellis, 2003). Therefore, step one entails item generation for Negative Brand Personality, as currently there is no scale to measure for Negative Brand Personality.
Within the scale development literature, two approaches, deductive and inductive, are used to generate the scale items (Hunt, 1991). The deductive method concerns prior understanding of the phenomena under investigation through the review of literature to develop construct definition, which subsequently guides the development of items (Schwab, 1980). The inductive approach, on the other hand, places less emphasis on prior theory and more on individual responses to identify the construct and item generation by asking sample respondents their perceptions/views about the construct under investigation. Consistent with Hinkin (1995) and Churchill’s (1979) recommendation, an inductive approach to Negative Brand Personality is adopted as the research topic is underdeveloped. Therefore, qualitative, in-depth interviews form the initial investigation of Negative Brand Personality. It would be difficult to examine Negative Brand personality in a focus group, due to the high possibility of individuals concurring with discussion themes and demonstrating no real individual opinion. Further, due to the sensitivity of the topic, social pressure to conform might exist which potentially could lead to socially acceptable responses.
However, the antecedents of Negative Brand Personality and the outcome variables are investigated through a deductive approach, as the variables are explored from the interview data and reinforced through prior literature. Together, the inductive (scales developed from scratch) and deductive approach (scales derived from existing measures) informed the nomological network to examine the construct Negative Brand Personality, and the likelihood of how it behaves within the network of related constructs (the antecedents of Negative Brand Personality and the outcome variables), which is important to the development of a valid measure (Cronbach and Meehle, 1955).
~ 91 ~
The overall aim of this thesis is to develop a valid, parsimonious, yet relatively short measure of Negative Brand Personality that could be easily incorporated into a wider questionnaire. It was important that the final measure captured the full domain of the construct definition, was short and concise enough to be incorporated into a wider questionnaire without taking too much of the time and energy of respondents, and was clear and understandable to working adults. By following the recommendations of Churchill (1979), a multi-stage development process was used to optimize the integrity of the new instrument. The overall stages in this process include two main studies: the first is categorized as ‘Initial Scale Development Study’, and constitutes four sub studies: (a) interviews;
(b) content validity study; (c) free and fixed sorting tasks; (d) substantive validity task. Together, the four sub studies not only investigate the Negative Brand Personality construct by first creating and validating the item measure, but also explore important antecedent and outcome variables to Negative Brand Personality.
Essentially, the objective of Study A - Qualitative interviews, was to investigate Negative Brand Personality by first investigating the Negative Brand Personality traits (items) and then identifying the antecedent constructs and outcome variables. Once the Negative Brand Personality items (traits) were identified from Study A, it was essential to assess that the face validity of the Negative Brand Personality items generated were not the mere opposite of Positive Brand Personality traits and were indeed perceived in a negative light. As a result, Study B - a content validity survey - was introduced to further investigate the face validity of the items to ensure the construct definition is captured and the items are valid. Following on from the content validity survey, Study C - Free and Fixed Sorting task - was conducted to further assess that the items generated were content and face valid by grouping the items into categories which captured the different dimensions of the construct definition. Items that were not classified into any of the categories were dropped. Study D - Substantive Validity task - further validated each dimension (alongside the items) of the construct definition. Each study was pressed into the services of the other study to ensure the scale captured the construct definition, and was valid, short and concise enough that it could help in the management of future research.
The second study, categorized as ‘Confirmatory Scale Development Study’, consists of three sub studies (Study E, F and G) to establish different validities of the developed scale. Study E: Selection of the Brand Gucci - principally investigated a single brand that was not biased to either Positive or Negative Brand Personality. This research focused on a single brand as opposed to multiple brands, to eliminate method bias (i.e., choosing an absolute brand that captures consumers’ varied
~ 92 ~
responses as opposed to a reflective brand where respondents reflect on more than one brand, which naturally encourages brand discrimination). By selecting the unbiased brand, the main study (Study F) was brought to life. Study F - the main questionnaire - was conducted to investigate the factor structure of the newly developed Negative Brand Personality measure along with the antecedents and outcome measures through exploratory and confirmatory factor analysis. Once the factor structure was established, it was essential to assess the nomological network. That is, how the newly developed brand personality measure behaves within the network of related constructs (the antecedents of Negative Brand Personality along with the outcome variables of Negative Brand Personality). Study G, therefore, focused on the Structural Equation model. A visual model of the multi-stage development process alongside the sub studies is presented in Figure 3.0.
~ 93 ~
Figure 3.0: Visual model of the multi-stage development process adopted in this research.
Stage 2: Confirmatory Scale Development Study Stage 1: Initial Scale Development Study
Note: items pertaining to Negative Brand Personality are reduced by dropping, cleaning and validating the remaining item pool subsequent to each study.
Brand that captured a high level of variance of both Positive and Negative Brand Personality. Hence, a brand that was not biased to either Positive or Negative Brand Personality traits. Brand Personality traits obtained from Interviews are indeed perceived as Negative Brand Personality traits and not Positive ones.
Identifying and categorize Negative Brand Personality items into groups (factors) to capture each dimension of
~ 94 ~
All seven studies examine several different types of validity, to adhere to psychometric rules of scale construction. The following section provides an overview of the choice to the research design for both the Initial Scale Development study and the Confirmatory Scale Development study. Specific methodologies for each study will be detailed in subsequent chapters.