Scaling Desin

RESEARCH METHODOLOGY MODULE 3 – PART C SCALING DESIGN

What is Scaling? • Scaling is assigning numbers (or other symbols) to a property of objects in order to impart some of the characteristics of numbers to the properties in question.

Dimensions of a Scale • Unidimensional • Multidimensional

Types of Response Scales • Rating Scales • Ranking Scales • Categorization

Types of Rating Scales 1. Simple category (Dichotomous) 2. Multiple choice, single response 3. Multiple choice, multiple response (Checklist) 4. Likert scale 5. Semantic differential 6. Numerical 7. Multiple rating 8. Fixed sum 9. Stapel 10. Graphic rating.

1. Simple Category (Dichotomous) • • • •

Two mutually exclusive options Agree / Disagree Yes / No Nominal Data.

2. Multiple-Choice, Single Response

• Multiple options provided • Only one answer is sought • Nominal data.

3. Multiple Choice, Multiple response • Checklist • Rater can select one or several alternatives • Nominal data.

4. Likert Scale • Summated rating scale • Statements express favourable or unfavourable attitudes • Each response is given numerical score • Usually 5 points • 20-25 questions about an attitude object • Interval data.

5. Semantic Differential • Measures psychological meaning of an attitude object • Semantic space • Bipolar • Usually 7 points • Interval data.

6. Numerical scales • Equal intervals • Verbal anchors serve as labels for extreme points • Rating and ranking in same scale • Ordinal or interval data.

7. Multiple rating List scale • • • •

Accepts a circled response Layout allows visualisation of results Mental map Interval data.

8. Fixed Sum scale • • • •

Helps discover proportions Easily adds upto 100% Cumbersome (involves addition!) Ratio data.

9. Stapel scale • More than 2 (bipolar) options • Set of 10 responses for each (+5 to -5) • Ordinal or Interval data.

10. Graphic rating scale • To discern fine differences • Response along a continuum • Ordinal, Interval or ratio.

Rating Scale Errors to Avoid • Leniency – Negative Leniency (“Hard rater”– higher scores) – Positive Leniency (“Easy rater” – – Low scores)

• Central Tendency • Halo Effect (systematic bias).

Types of Ranking Scales (Ordinal data) • Paired-comparison • Forced Ranking • Comparative.

Scale Design Techniques • • • • •

Arbitrary scaling Consensus scaling Item Analysis scaling Cumulative scaling Factor scaling

ARBITRARY • Scale is custom-designed to measure • No advance evidence of validity and reliability • Easy to develop, inexpensive • Subjective approach.

CONSENSUS • Judges evaluate the items to be included • Based on topical relevance.

Thurstone • Equal appearing interval scale • 50 or more judges evaluate large no. of statements • One statement per card • Judges sort cards into 11 piles • Of the 11 piles, 3 are identified as “Favourable” “Unfavourable” & “Neutral” • 8 intermediate piles are unlabelled • Costly and time-consuming.

ITEM ANALYSIS • Measurement scales are tested with a sample of respondents • Total score is calculated • Individual items are then analysed to find high and low scores • Likert scale • Large no. of statements rated • Separate the statements with high and low total scores.

CUMULATIVE • Ranking of items with ascending and descending discriminative power • Scalogram.

FACTORING • Inter-correlation of items • Uncover underlying dimensions that have not been identified by exploratory research.

Advanced Scaling • • • •

Multi-Dimensional Scaling (MDS) Conjoint Analysis Magnitude estimation scaling Rasch models.