Scale Development

A scale is a measurement instrument which contains a number of slightly different operational definitions of the same concept

Guidelines for Scale Development

• Determine what you want to measure.
• Generate an item pool.
• Determine the format for measurement.
• Have the item pool reviewed by experts.
• Consider including validation items.
• Administer to a development sample.
• Evaluate the items.
• Optimize scale length.

1. Determine clearly what it is you want to measure.

1. Determine clearly what it is you want to measure.

2. Generate an item pool.

• Each item is a "test" in its own right of the strength of the latent variable.
• What other ways can an item be worded to get to the construct?
• Multiple items will constitute a more reliable test.

3. Determine the format for measurement.

Thurstone Scaling
measures attitudes of individuals on an equal interval scale

Guttman Scaling
Do you smoke?
Do you smoke more than 10 cigarettes/day?
Do you smoke more than 1 pack/day?

Forced Choice

Graphic Format

• What is your opinion on changing the level of defense spending?
• 1 = it should be increased
• 2
• 3 = it should remain the same
• 4
• 5 = it should be increased

Itemized Format

• Decreasing the level of defense spending is absolutely essential to the wellbeing of our country.
• Keeping defense spending at its current level is the best thing to do now.
• Increasing the level of spending is absolutely essential to the wellbeing of our country.

Comparative Format

• Which of the following ways to increase the defense budget is most preferable?
• Increase the size of the national debt.
• Increase income taxes.
• Increase other types of taxes.
• Decrease money spent on social welfare programs.
• I prefer not to increase the defense budget.

4. Have the initial item pool reviewed by experts.

Are the items relevant?
Are they clear?
What have I missed?

5. Consider including validation items.

Detect flaws.

Including measures of relevant constructs and the pattern of relationships can support claims of validity or indicate why the set of items does not perform as anticipated.

6. Administer items to a development sample.
Use a large sample (> 300) to test your scale.
Use a split sample if your sample is large.

7. Evaluate the items.

Reverse Scoring

• Consider mood items on a Likert scale from 1 to 7, where higher numbers indicate agreement
• Endorsing positive mood items would yield a higher score
• Endorsing negative mood items would yield a lower score
• Reverse the scores for negative items from 7 to 1, 6 to 2, 5 to 3, etc.

Item Scale Correlations

Item Variances

Item Means
Our inclination is to write statements that do not offend respondents, which is probably good, but will lead to items that nearly everyone will find agreeable. It may be that a belief or opinion is completely absent from the response categories.

8. Optimize Scale Length.

Shorter scales are generally better.

Redundancy is not a bad thing. Be over-inclusive, especially in the initial development phase.

Guidelines for Scale Development

• Determine what you want to measure.
• Generate an item pool.
• Determine the format for measurement.
• Have the item pool reviewed by experts.
• Consider the inclusion of validation items.
• Administer to a development sample.
• Evaluate the items.
• Optimize scale length.

ACTIVITY