7.6 Thurstone scales

Thurstone scales are a powerful tool in communication research, allowing us to measure complex attitudes with precision. Developed in the 1920s, these scales use carefully crafted statements to quantify subjective opinions on a continuum.

By presenting respondents with a range of pre-scaled statements, Thurstone scales offer advantages over simpler rating methods. They provide interval-level data, reduce bias, and allow for nuanced analysis of attitudes towards media, public figures, and social issues.

Definition of Thurstone scales

• Psychometric measurement tool developed by Louis Thurstone in the 1920s to assess attitudes and opinions
• Utilizes a series of statements with predetermined scale values to measure respondents' positions on a continuum
• Integral to communication research methods for quantifying subjective constructs and attitudes

Historical context

• Emerged during the early 20th century as part of the scientific movement in psychology and social sciences
• Addressed the need for more precise measurement of attitudes and opinions in research
• Influenced subsequent developments in scaling techniques and attitude measurement in communication studies

Purpose and applications

Uses in communication research

• Measures attitudes towards media content, public figures, or social issues
• Assesses audience perceptions of communication campaigns or messages
• Evaluates changes in public opinion over time or across different demographics
• Facilitates comparison of attitudes across diverse groups or cultures

Advantages vs other scales

• Provides interval-level data allowing for more sophisticated statistical analyses
• Reduces acquiescence bias by presenting statements rather than agreement scales
• Offers finer discrimination between attitude positions compared to simpler rating scales
• Minimizes the impact of extreme responses through the use of multiple statements

Components of Thurstone scales

Statements or items

• Consist of 20-30 carefully crafted statements representing various positions on the attitude continuum
• Range from extremely negative to extremely positive views on the topic
• Worded clearly and unambiguously to avoid confusion or multiple interpretations
• Reflect a single idea or concept per statement to maintain clarity

Equal-appearing intervals

• Statements are arranged along a continuum with equal psychological distances between adjacent items
• Typically use an 11-point scale ranging from 1 (most unfavorable) to 11 (most favorable)
• Aim to create a linear representation of the attitude dimension being measured
• Allow for meaningful comparisons between different points on the scale

Construction process

Item generation

• Researchers create a large pool of potential statements (100-150) covering the full range of attitudes
• Statements drawn from various sources (literature reviews, focus groups, expert consultations)
• Include both positive and negative statements to ensure balanced representation
• Edit statements for clarity, brevity, and relevance to the attitude being measured

Expert judges

• Panel of 50-100 judges evaluate and rate each statement independently
• Judges selected based on their familiarity with the topic and diverse perspectives
• Rate statements on the 11-point scale according to perceived favorability
• Provide feedback on statement clarity, relevance, and potential improvements

Scaling procedure

• Calculate median and interquartile range for each statement based on judges' ratings
• Select final set of statements with low dispersion (small interquartile range) and evenly distributed medians
• Assign scale values to chosen statements based on their median ratings
• Arrange final statements in random order for presentation to respondents

Administration of Thurstone scales

Respondent instructions

• Present respondents with the final set of statements in random order
• Ask participants to select statements they agree with or find most representative of their views
• Emphasize that respondents should choose statements based on personal agreement, not perceived correctness
• Provide clear guidelines on the number of statements to select (typically 6-10)

Scoring methods

• Calculate respondent's score by averaging the scale values of selected statements
• Alternative method involves summing the scale values of chosen statements
• Higher scores indicate more favorable attitudes, lower scores represent less favorable attitudes
• Consider using weighted scoring based on the number of statements selected by each respondent

Statistical analysis

Median and interquartile range

• Calculate median score to represent the central tendency of respondents' attitudes
• Use interquartile range to measure the spread or variability of responses
• Compare median scores across different groups or time points to assess attitude differences
• Analyze interquartile range to understand the consistency of attitudes within the sample

Reliability measures

• Assess internal consistency using split-half reliability or Cronbach's alpha
• Evaluate test-retest reliability by administering the scale to the same group at different times
• Examine inter-judge reliability during the scale construction process
• Consider factor analysis to identify underlying dimensions or subscales within the instrument

Limitations and criticisms

Time-consuming development

• Requires extensive effort in item generation, judge recruitment, and scaling procedures
• May not be feasible for rapid or low-budget research projects
• Limits the ability to quickly adapt scales for new or emerging topics
• Necessitates significant resources for proper scale construction and validation

Potential judge bias

• Judges' personal attitudes may influence their ratings of statements
• Cultural or demographic factors could affect judges' interpretations of statement favorability
• Difficulty in ensuring a truly representative panel of judges for all research contexts
• Potential for systematic bias in scale values if judge selection is not carefully managed

Thurstone vs Likert scales

Structural differences

• Thurstone scales use pre-scaled statements, Likert scales employ agreement ratings
• Thurstone scales typically have more items (20-30) than Likert scales (5-7)
• Thurstone scales aim for equal intervals, Likert scales assume equal intervals between response options
• Thurstone scales focus on statement selection, Likert scales require rating each item

Comparative strengths

• Thurstone scales offer finer discrimination but require more development time
• Likert scales are easier to construct and administer but may be more susceptible to response biases
• Thurstone scales provide interval-level data, Likert scales often treated as ordinal or interval
• Thurstone scales may reduce acquiescence bias, Likert scales are more intuitive for respondents

Modern adaptations

Computer-assisted scaling

• Utilize algorithms to streamline the scaling process and reduce manual effort
• Implement item response theory (IRT) models to enhance scale precision and efficiency
• Develop adaptive testing procedures to minimize the number of items presented to respondents
• Integrate machine learning techniques for improved item selection and scale optimization

Online implementations

• Create web-based platforms for easier administration and data collection
• Develop mobile applications for convenient participation in Thurstone scale surveys
• Implement real-time scoring and feedback systems for immediate results
• Utilize cloud-based storage and analysis tools for efficient data management and collaboration

Examples in communication studies

Attitude measurement

• Assess public attitudes towards controversial media content or programming
• Measure changes in audience perceptions before and after exposure to communication campaigns
• Evaluate attitudes towards emerging communication technologies or platforms
• Investigate cross-cultural differences in attitudes towards various communication styles

Media perception research

• Examine perceptions of media bias across different news sources
• Assess audience attitudes towards product placement in various media formats
• Measure perceived credibility of different types of social media influencers
• Evaluate attitudes towards privacy and data sharing practices of media companies