Techniques of Scale Construction
Scaling techniques are used in research to measure attitudes, opinions, preferences, and perceptions in a systematic way.
One of the earliest and simplest methods is the Paired-Comparison Scale.
- Paired-Comparison Scale
Meaning / Definition
- In the Paired-Comparison Method, respondents are presented with two items (a pair) at a time and asked to choose the one they prefer (or which they think has more of a certain attribute).
- It is based on comparative judgment → instead of rating each item individually, participants compare pairs.
Example: “Which of the following two crops do you prefer to cultivate: Wheat or Maize?
Process
- Select a set of items (say 5 agricultural practices).
- Present them to respondents two at a time (all possible pairs).
- Respondents indicate preference in each pair.
- Tally responses → construct a preference matrix.
- Rank the items according to the number of times each was preferred.
Mathematical Side
- If n items are to be compared, the number of possible pairs is:
- Pairs=n(n−1)2
- Example: For 5 items → 5(5−1) / 2 = 10
2) Equal Appearing Interval Scale (Thurstone Scale)
Meaning / Definition
- Developed by Louis Thurstone (1929).
- A set of attitude statements are given to judges/experts, who rate each statement on a scale (e.g., 1 = extremely unfavorable to 11 = extremely favorable).
- The statements are then arranged according to their median scale values.
- Respondents are asked to agree or disagree with these statements.
Example: “Farmers should adopt organic farming practices.” (judges rate how favorable this is on 1–11 scale).
Steps
- Collect large number of attitude statements.
- Get them rated by a panel of judges on a scale (favorable → unfavorable).
- Calculate scale values (median) and interquartile ranges (Q).
- Select statements that are equally appearing along the continuum.
- Administer to respondents.
Advantages
- Provides interval scale measurement
- Systematic & scientific.
- Reduces subjectivity (uses expert judgment).
3) Summated Rating Scale (Likert Scale)
Meaning / Definition
- Developed by Rensis Likert (1932).
- Respondents indicate their degree of agreement/disagreement with a series of statements on a 5-point or 7-point scale (Strongly Agree → Strongly Disagree).
- Final score is the sum of responses across all items.
- Example: “Extension training programs are useful.”
- Strongly Agree (5)
- Agree (4)
- Neutral (3)
- Disagree (2)
- Strongly Disagree (1)
Steps
- Collect many attitude statements.
- Pre-test them on a sample.
- Eliminate weak items using item analysis.
- Administer final scale to respondents.
- Calculate summated scores for each respondent
4) Item Analysis
Meaning
- Process of testing items (questions/statements) to select the best ones for a scale or test.
- Ensures that only valid, reliable, and discriminating items remain in the final tool.
Steps
- Draft a large pool of items.
- Administer to a pilot group.
- Analyze each item using:
- Item difficulty index (p-value) → how many answered correctly.
- Item discrimination index (D) → ability to differentiate high vs. low scorers.
- Item-total correlation → consistency of item with overall test.
- Retain good items, discard weak ones.
Purpose
- Improves validity and reliability of tests/scales.
- Used in knowledge test, attitude scale, and achievement tests.
5) Scalogram Analysis (Guttman Scale)
Meaning / Definition
- Developed by Louis Guttman (1944).
- A cumulative scaling technique: If a respondent agrees with a “strong” statement, it is assumed they will also agree with all weaker statements.
Example:
- I can recognize improved seed.
- I know how to store improved seed.
- I use improved seed regularly.
- I promote improved seed to others.
If someone agrees to the last (strongest) statement, they are assumed to agree with the earlier ones too.
Steps
- Prepare items arranged in a hierarchical order of difficulty/intensity.
- Collect responses (agree/disagree, yes/no).
- Check for cumulative property (scalability coefficient).
- Construct scale if reproducibility is high (>0.90).
Quick Exam Key Points
- Paired-Comparison → Compare two at a time, rank preferences.
- Equal Appearing Interval (Thurstone) → Judges rate statements, scale values determined.
- Summated Rating (Likert) → Agreement scale, total score = sum of responses.
- Item Analysis → Selecting best items for scale/test (difficulty & discrimination).
- Scalogram (Guttman) → Cumulative, hierarchical items (if agree with strong → agree with weaker).
