Rainyday Insurance Adjusters Company

Only available on StudyMode
  • Download(s) : 163
  • Published : May 11, 2013
Open Document
Text Preview
Institut f. Statistik u. Wahrscheinlichkeitstheorie
1040 Wien, Wiedner Hauptstr. 8-10/107 AUSTRIA http://www.statistik.tuwien.ac.at

Benefits from using continuous rating scales in online survey research H. Treiblmaier and P. Filzmoser

Forschungsbericht SM-2009-4
November 2009

Kontakt: P.Filzmoser@tuwien.ac.at

Benefits from Using Continuous Rating Scales in Online Survey Research Horst Treiblmaier* Institute for Management Information Systems Vienna University of Economics and Business Augasse 2-6, 1090 Vienna, Austria1

Peter Filzmoser Department of Statistics and Probability Theory Vienna University of Technology Wiedner Hauptstraße 8-10, A-1040 Vienna, Austria

The usage of Likert-type scales has become widespread practice in current IS research. Those scales require individuals to choose between a limited number of choices, and have been criticized in the literature for causing loss of information, allowing the researcher to affect responses by determining the range, and being ordinal in nature. The use of online surveys allows for the easy implementation of continuous rating scales, which have a long history in psychophysical measurement but were rarely used in IS surveys. This type of measurement requires survey participants to express their opinion in a visual form, i.e. to place a mark at an appropriate position on a continuous line. That not only solves the problems of information loss, but also allows for applying advanced robust statistical analyses. In this


Augasse 2-6, A-1090 Vienna, Austria

Tel.: +43/1/31336/4480 Fax: +43/1/31336/746 E-Mail: Horst.Treiblmaier@wu.ac.at


paper we use a real-world sample and a simulation to illustrate how noise impacts our data set. A noise level of 10% has only a small effect on both classical and robust estimates, but when 20% of noise is added, the classical estimators become severely distorted. Continuous rating scales in combination with robust estimators turn out to be an effective tool to reduce the impact of noise in surveys. Keywords: Measurement, Scaling, Continuous Rating Scale, Online Research, Robust Correlation, Factor Analysis, MCD estimator

The concept of measurement is fundamental to all empirical social science research, including Information Systems and closely related disciplines such as Marketing and Psychology. Given its widespread and frequent application in countless studies, it seems peculiar that Allport and Kerler (2003, p. 356) caution that „measurement is perhaps the most difficult aspect of behavioral research‟. The classic definition of measurement was given by Stevens (1946), who described it as the assignment of numerals to events or objects according to rules. This definition has been criticized over the last few decades, as for instance by Mitchell (1999), who argues that there is a difference between the traditional understanding of measurement in the natural sciences and Steven‟s definition. The first pertains to „the discovery of real numeric relations (ratios) between things (magnitudes of attributes), and not the attempt to construct conventional numerical relations where they do not otherwise exist‟ (Mitchell, 1999, p. 17, cited in Balnaves & Caputi, 2001). Accordingly, the two main tasks of quantitative science are to (1) make sure that the attribute under investigation is in fact quantitative and (2) devise procedures to measure the magnitude of this attribute. It can easily be seen that in social science research the first assumption is an essential precondition for all further analyses. Balnaves and Caputi (p. 51) give the examples of self-esteem and extroversion, but quantifiability is also implicitly taken for granted for all the constructs being frequently used in IS research. The second task, albeit being important, is scarcely a topic of 2

interest in IS literature. A plethora of research on how to build valid and reliable constructs exists (e.g. Straub, 1989;...
tracking img