Sara Dolnicar

SEGMENTATION RESEARCH

In the area of market segmentation research, Dolnicar has uncovered flaws in segmentation studies, both at the methodological and conceptual level (Dolnicar, 2002; 2003; 2005; 2007), recently proving that factor analysing items before constructing segments (the predominant method in tourism research) leads to inferior results compared to using the raw data directly (Dolnicar & Grün, 2008).

Dolnicar has also contributed significantly to segmentation methodology. Since her PhD in 1997, which compared nueral networks with traditional algorithms, she has been exploring the usefulness of novel algorithms for market segmentation. Later she adapted bagged clustering — an ensemble method leading to increased stability of solutions — for segmentation purposes (Dolnicar & Leisch, 2000; 2003). Most recently she introduced bi-clustering to solve the item selection problem while grouping individuals (Dolnicar, Kaiser, Lazarevski & Leisch, 2013). A comparative study of indices for determining the number of clusters in binary survey data led to a Psychometrika publication in 2002 (Dimitriadou, Dolnicar & Weingessel, 2002) which is cited by researchers from not only marketing, but also psychology, geophysics and engineering. Dolnicar was part of the research team that invented perceptions-based market segmentation, a nonparametric simultaneous analysis of segmentation, positioning and competition which prevents sequence errors from occurring when both positioning and segmentation decisions are made (Dolnicar, Grabler & Mazanec, 1999; Buchta, Mazanec & Strasser, 2000; Buchta, Dolnicar & Reutterer, 2000). As such, she has made a significant contribution to the improvement of segmentation in the broader context of marketing strategy.

MEASUREMENT

In the the area of measurement in the social sciences, Sara has conducted numerous empirical studies investigating the VALIDITY OF ORDINAL ANSWER FORMATS, raising serious validity concerns about the most popular answer format used by marketers: the ordinal multi-category scale such as the Likert scale (Dolnicar, 2003; Dolnicar, Grün & Leisch, 2004; Dolnicar & Grün, 2007). She also investigated the problem of data contamination by cross-cultural RESPONSE STYLES, and developed an ensemble-based method to assess the robustness of results from potentially contaminated data (Dolnicar & Grün, 2007a; Dolnicar & Grün, 2007b). During her research on the validity of marketing measures as well as ensemble methods for the assessment of response bias in survey data, the question of OPTIMALITY OF ANSWER FORMATS arose. Currently Dolnicar is conducting a large-scale investigation supported by two ARC Discovery grants: one specifically investigating optimal brand image measures (Dolnicar & Rossiter, 2008; Dolnicar & Grün, 2007), the other investigating general answer format optimality. Results indicate that the level-free full binary answer format outperforms more commonly used multi-category answer formats in the context of brand image measurement.

APPLIED RESEARCH

Sara has worked with many industry partners since commencing her academic career in Austria. For example, she assisted the Austrian Business Chamber (Hotel Section) in developing a research-based national hotel starring system for Austria. She advised the Austrian Ministry of Tourism on strategic matters and undertook research for the Austrian National Tourism Organization. In Australia, she has for many years worked with a consortium of not for profit organisations responsible for the placement of foster children. Currently she is working with a group of online market research companies in developing and testing improved survey measures specifically designed for the online environment.