Comparing two image research instruments: The Q-sort method versus the Likert attitude questionnaire

https://doi.org/10.1016/j.foodqual.2008.02.007Get rights and content

Abstract

Despite the attention for corporate, brand and product images, only few studies focus on methodological comparisons of image research methods. This article presents a comparison of two current instruments: the Q-sort method and a Likert attitude questionnaire. The study applies both methods to measure the image of beef, using the same assertions in similar samples of consumers. The two methods produce consistent results, but differ in the possibilities of data analysis and interpretation. An advantage of the Q-sort method is that it offers straightforward insights in the underlying structure of image within audience segments. On the other hand, the Q-sort method does not give overall indications of an image, and limits occur for analyzing the relationships between image and other variables.

Introduction

With the rise of image as an object of communication and marketing research, many different instruments have been developed to measure corporate, brand and product image among stakeholder groups. Most of these instruments stem from psychological research traditions. The diversity of research approaches available reflects the ambiguity of the image concept itself, which has been defined in many different ways (Christensen and Askegaard, 2001, DeFleur and Westie, 1963, Poiesz, 1989). There is no universally accepted image definition, and the more recent introduction of the equally intangible and strongly related concept of reputation only seems to add to the confusion.

Poiesz (1989) categorizes the various image definitions by placing them on a consumer elaboration continuum. He thus distinguished three views on image, which can be easily connected with possible research approaches. In the case of high elaboration, an image represents a complex network of meanings stored in memory. This implies that the measurement of an image must aim at revealing and exploring associations people have with the image object, which calls for qualitative and open methods like the Kelly repertory grid or laddering. In the case of medium elaboration, an image is a theoretical and operational equivalent of an attitude: it is the overall evaluation of an artifact based on salient beliefs held by consumers. Consequently, the measurement of image resembles attitude scales, involving the evaluation and weighting of beliefs. In the case of low elaboration, an image is merely a holistic impression of the relative position of an object among its perceived competitors. Measurement must then focus on differences and similarities between image objects, using multi-dimensional scaling.

This classification offers a fruitful basis for comparing image research instruments. Within each particular image definition, researchers have several methods at their disposal to measure image. A relevant research question is whether or not competing research methods are equally suitable for measuring image and whether they produce similar results. So far, only few researchers respond to this challenge (Van Riel, Stroeker, & Maathuis, 1998).

This article focuses on the measurement of image in the case of medium elaboration. Image is supposed to be the equivalent of an attitude: image is an overall evaluation that is a consequence of a series of beliefs and importance weightings. Two competing methods that are potentially useful in this context are Likert attitude questionnaires and the Q-sort method. Both methods can be used to measure attitudes and both present the respondents with a predetermined set of items (beliefs), which must be judged on an x-point scale, so that respondents must make trade-offs between scale positions.

Our study investigates the suitability of the two methods for product image research. First, the rationale of the two methods is discussed. After that, the possibilities of the methods are analyzed in an empirical study into the image of beef. Finally, the article addresses the strengths and weaknesses of both methods for image research.

Section snippets

The Q-sort method

The Q-sort method is rooted in Q-methodology, an inverted technique of factor analysis. Developed by Stephenson (1953) as an alternative measurement technique to existing scales and tests in psychology, the method can be used in any situation in which subjectivity is at issue, including attitude measurement (Stephenson, 1965, Stephenson, 1968). Mainly due to the work of Brown (1980) and McKeown and Thomas (1988), who further developed the principles and procedures of Q-sorting, the Q-sort

The Likert attitude questionnaire

Attitude rating scales are the most widely used research approach within marketing and opinion research (McDonald and Vangelder, 1998, Zikmund, 1997). One of the best known rating scales is the typical five-point agreement scale developed by Likert (1932). Although its principles are very different from Q-sorting and both procedures were developed to serve different ends (Stephenson, 1953, McKeown, 2001), Likert scaling is a well-accepted technique for attitude measurement. Mainly due to its

Image object and target population

To explore the characteristics of both instruments in an image research setting, the Q-sort method and a Likert attitude questionnaire were used in a study into the product image of beef. The study focuses on beef for two reasons. Beef is a well-known product category and recent food scares concerned the safety of beef products, such as the BSE-crisis in 2000, and the 2001 outbreak of foot and mouth disease among cattle in The Netherlands. Moreover, the ambivalent thoughts and feelings of

Results Q-sort method

The Q-sort data were analyzed using PCQ for Windows (academic edition, version 1.41). The data were analyzed using centroid factor analysis with varimax rotation. Of the various results the program produces, the factor analysis—including representative sorts per factor and differentiating items—and consensus statements are the most important. Due to the relatively large sample of respondents used in this study, the factor analysis revealed up to nine factors with eigenvalues greater than 1,

Results Likert attitude questionnaire

All statements were coded on a scale from −2 (strongly disagree) to +2 (strongly agree). The respondents’ importance weightings of the nine clusters of variables were also coded on a scale from −2 (very unimportant) to +2 (very important). By multiplying each statement with the importance weight of the cluster of items it belongs to, weighted average scores were computed, which could range from −4 (extremely negative) to 4 (extremely positive).

Comparison of the two methods

The use of a Likert attitude questionnaire and the Q-sort method to investigate the same image object with exactly the same items and highly similar respondent groups gave the opportunity to compare and illustrate the use of the two methods for image research.

General discussion

The comparison of the Likert attitude scale and the Q-sort method this article describes leads to mixed results. The two image research methods appear to produce both similar and different outcomes. At the level of specific ratings per statement, the results indicate a high degree of similarity between both methods. This finding provides concurrent validity of the methods: the tasks of ordering statements in a normal distribution and judging statements on a 5-point scale do not lead to

References (66)

  • S.R. Brown

    Political subjectivity: Applications of Q methodology in political science

    (1980)
  • S.R. Brown

    Q technique and method: Principles and procedures

  • S.R. Brown

    Q methodology and qualitative research

    Qualitative Health Research

    (1996)
  • S.R. Brown

    Structural and functional information

    Policy Sciences

    (2002)
  • J.M. Carlson et al.

    Perspectives on entertainment television’s portrayal of a racial incident: An intensive analysis

    Communication Review

    (2001)
  • B.A. Chadwick et al.

    Social science research methods

    (1984)
  • L.T. Christensen et al.

    Corporate identity and corporate image revisited: A semiotic perspective

    European Journal of Marketing

    (2001)
  • R.M. Cross

    Exploring attitudes: The case for Q methodology

    Health Education Research

    (2005)
  • M.L. DeFleur et al.

    Attitude as a scientific concept

    Social Forces

    (1963)
  • J. De Jonge et al.

    Monitoring consumer confidence in food safety: An exploratory study

    British Food Journal

    (2004)
  • K.E. Dennis

    Q-methodology: New perspectives on estimating reliability and validity

  • D. Durning

    The transition from traditional to postpositivist policy analysis: A role for Q-methodology

    Journal of Policy Analysis and Management

    (1999)
  • A.L. Edwards et al.

    A comparison of the Thurstone and Likert techniques of attitude scale construction

  • J.R. Fairweather

    Reliability and validity of Q-method results: Some empirical evidence

    Operant Subjectivity

    (1981)
  • J.R. Fairweather et al.

    Visitors’ and locals’ experiences of Rotorua, New Zealand: An interpretative study using photographs of landscapes and Q method

    International Journal of Tourism Research

    (2002)
  • M. Fishbein et al.

    Belief, attitude, intention and behavior: An introduction to theory and research

    (1975)
  • J. Fyock et al.

    Beneficiary decisionmaking: The impact of labeling health plan choices

    Health Care Financing Review

    (2001)
  • E.P. Gelineau

    A psychometric approach to the measurement of color preference

    Perceptual and Motor Skills

    (1981)
  • B.C. Hamm et al.

    Self-actualization and product perception

    Journal of Marketing Research

    (1969)
  • M.A. Hogg et al.

    Social psychology: An introduction

    (1995)
  • C. Kitzinger

    Introducing and developing Q as a feminist methodology: A study of accounts of lesbianism

  • A. Lecouteur et al.

    Repertoires of teaching and learning: A comparison of university teachers and students using Q methodology

    Higher Education

    (2001)
  • R. Likert

    A technique for the measurement of attitudes

    Archives of Psychology

    (1932)
  • Cited by (70)

    • Global competence in Japan: What do students really need?

      2022, International Journal of Intercultural Relations
    • Why cultural ecosystem services matter most: Exploring the pathways linking greenspaces and mental health in a low-income country

      2022, Science of the Total Environment
      Citation Excerpt :

      We used the Q-methodology to examine the outcomes of participatory video and focus groups and to investigate the viewpoints on ecosystem services and disservices held by residents of Kathmandu. The Q-methodology involves the rank-ordering, by the participants, of a set of 40–60 statements into a near-normal distribution, ranging from the least to the most agreed (ten Klooster et al., 2008). To better understand sorting decisions, we complemented the sorting task with in-depth interviews focusing on the statements on the far ends of the distribution (Table 1).

    • Are ohmic heating-treated whey dairy beverages an innovation? Insights of the Q methodology

      2020, LWT
      Citation Excerpt :

      It consists in asking participants to sort some statements into ordinal predefined categories corresponding to a continuum of agreement; these statements being defined by the experimenter in such a way that they are in relation with the given topic (Coxon, 1999). It has been used in well-succeeded way to understand the freshness of commercial orange juices by Chinese immigrant consumers (Zhang, Lusk, Mirosa & Okey, 2016) and to construct image of beef providing complementary findings when compared to Likert attitude questionnaire (Ten Klooster, Viser & Jong, 2008). As far as the authors know, Q methodology has not been used in whey beverages or to evaluate ET.

    View all citing articles on Scopus
    View full text