Likert scale versus Q-table measures – a comparison of host community perceptions of a film festival

ABSTRACT

Current literature on impact assessment stresses the need to focus on understanding respondents and suggests Q methodology as a suitable instrument. This paper aims to compare Q methodology and the commonly used Likert scaling, to evaluate the impact of residents’ perception at a film festival. A set of 23 impacts were offered to respondents and the same respondents were interviewed twice over the course of one month, first using Likert scaling, and second using Q-table. Finally, a factor analysis was conducted. The factor analysis revealed five factors for both survey techniques. However, the reliability analysis using Cronbach's alpha showed weaker internal consistency for all factors using Q-table. The research results also showed that Likert scaling provided better statistical relevance and easier interpretation of results. Likert scaling also proved that respondents tended to indicate items as “strongly agree” or “agree”, so lower impact variability is shown. Using Q-table forced respondents to diversify their impact assessments and thus provided valuable information for practical application in decision-making processes. Despite differences in both analyses, some similarities could be seen.

KEY WORDS:

• Likert scale
• Q methodology
• film festival
• event
• factor analysis

Notes on contributor

Ing. Michaela Havlíková, Ph.D. is an assistant professor at Czech University of Life Sciences in Prague. Her publication are focused on tourism economics, impacts assessment analysis, event tourism and agritourism. Her research tries to combine tourism and environmental economics.

Notes

1. The literature does not provide any agreement upon the suitable number of Q sort items. However, Addams and Proops (2000) suggest that Q-set should not include more than 40 items to ensure comprehensibility.

2. Based on consultation with psychologist.

3. Over 60% of respondents indicated festival impacts as “strongly agree”

4. Factor 1 includes increased prices. This impact is not interpreted because it had a lower loading than 0.32, which, according to Tabachnick and Fidell (1996), is less than the required minimum for interpretation.

Additional information

Funding

The findings introduced in this paper arise from research made possible by a grant provided by the Faculty of Economics and Management, Czech University of Life Sciences in Prague – [no. 20141044].