“Write short note : Techniques employed in measuring attitudes.” (1986)
- Measuring an attitude is not just a matter of choosing a scale type — it requires a defensible procedure for writing, sorting, testing, and validating the statements that eventually make up the instrument.
- L.L. Thurstone’s method begins from the premise that a scale is only as good as the process used to build it, which is why he built in an entire panel-based construction stage before a single respondent is ever surveyed.
- Rensis Likert, by contrast, showed that a far simpler construction procedure could still yield a reliable and widely usable instrument — a difference in how the scale is built, not merely in what it measures.
- This note therefore stays on the construction and administration side: how each technique is actually assembled, checked, and deployed in the field.
Constructing the Thurstone Scale: The Judge-Panel Procedure
- The researcher first assembles a large pool of statements — often over a hundred — expressing every shade of opinion on the attitude object, from strongly unfavourable to strongly favourable.
- A panel of judges, who need not personally hold any of the views expressed, sorts each statement into eleven ordered piles purely by its apparent favourability, not by whether the judge agrees with it.
- The scale value assigned to each statement is derived from the median of the judges’ placements; statements on which judges disagreed too widely are discarded as ambiguously worded.
- The final instrument retains a smaller set of statements spaced evenly across the continuum — a comparatively laborious construction process, since it depends on recruiting and coordinating a sizeable judge panel before any respondent data is even collected.
Constructing the Likert Scale: Item Analysis
- Construction begins with writing a large pool of candidate statements and piloting them on a trial sample of respondents, who rate their agreement with each on a graded scale (commonly strongly agree to strongly disagree).
- The researcher then performs item analysis: each item’s ability to discriminate between respondents who scored high overall and those who scored low is checked, and only the items that discriminate well between these two groups are retained in the final scale.
- Retained item scores are simply summed to produce each respondent’s total attitude score — a comparatively simple construction process, since no independent judge panel is ever needed.
- This relative ease of construction, more than any conceptual advantage over Thurstone’s method, is why the Likert format went on to become the most widely used attitude scale in sociological and market research.
Constructing the Guttman Scale: The Reproducibility Coefficient
- Items are selected and arranged so that, in principle, agreeing with a more extreme item logically implies agreement with every milder item on the same scale.
- Because real response patterns rarely fit this cumulative logic perfectly, Louis Guttman’s procedure requires calculating a reproducibility coefficient after administration — a statistic measuring how well a respondent’s actual pattern of answers can be reconstructed just from knowing their total score.
- A high reproducibility coefficient, conventionally expected to approach or exceed roughly 0.90, is taken as evidence that the items genuinely form a single cumulative, unidimensional scale rather than a loose collection of unrelated statements.
Administering the Semantic Differential
- Charles Osgood’s technique asks a respondent to rate a single concept across a battery of bipolar adjective pairs — good–bad, strong–weak, active–passive — typically on a seven-point continuum between each pair.
- Because the format is identical regardless of what concept is being rated, the same adjective grid can be administered rapidly across many different concepts in a single sitting, making it efficient for comparing attitudes toward several objects at once.
- Factor analysis of responses across many studies consistently recovers three underlying dimensions — evaluation, potency, and activity — giving the technique an empirically grounded structure beyond its surface simplicity.
Non-Scale and Indirect Techniques
- Structured attitude questionnaires with fixed response categories (agree/disagree/don’t know, or ranked options) remain the most common everyday technique, valued for ease of coding but limited by the same rigidity that affects any closed-ended instrument.
- Indirect or projective techniques — sentence completion, hypothetical vignettes, or indirect third-person phrasing (“do people of your community still believe…”) — are used specifically where direct questioning would invite a socially desirable rather than an honest answer.
- Survey-methodology work on sensitive topics in India has found that indirect and list-experiment-style questioning techniques can surface meaningfully different, often higher, estimates of prejudiced or exclusionary attitudes — on caste discrimination or communal bias, for instance — than direct questioning does, precisely because direct questions on such topics invite under-reporting.
- The four major scales differ as much in how they are built as in what they measure: Thurstone’s judge-panel procedure is rigorous but labour-intensive, Likert’s item analysis is comparatively quick and became the field’s default, Guttman’s cumulative logic is checked empirically through the reproducibility coefficient, and the semantic differential trades depth for administrative speed across multiple concepts.
- Non-scale and indirect techniques remain necessary wherever a topic’s sensitivity makes direct questioning an unreliable route to the truth.
- The choice of construction method is itself a methodological decision, not a neutral afterthought — it shapes how much labour a researcher must invest before data collection even begins.
- In the Indian context, this construction-side attention to indirect questioning continues to matter most wherever caste, communal, or gender attitudes are the object of study.