Lecture 1 - Scale Construction Process Flashcards
What is a summated rating scale?
— Consists of multiple items that will be summed to obtain an overall score.
— Each item is a statement to which the respondents are asked to provide a rating based on the response choices given.
— Each item measures the attribute along a quantitative continuum.
— There is no right or wrong answer.
Construct conceptualisation
How to properly define a construct?
— Always aim to provide a clear definition for the construct.
— Literature review should serve as a starting point.
— Know clearly the purpose that scale is going to be used for and its intended target audience.
Construct conceptualization
What does one look out for in literature review?
— Theories of the construct that will help to:
• Formulate a definition
• Inform on the construct’s dimensionality
• Point us to constructs that are related or unrelated to the construct of interest.
— Survey the empirical research carried out on the construct.
• Nomological net (Cronbach & Meehl, 1955)
— Think about conceptually similar constructs.
— Previous attempts to measure the construct, and strengths/weaknesses of such attempts.
What is needed in construct conceptualisation?
— Define the construct
— Define the boundary of the construct
— Explain the need for your measure to be developed
— Argue why the existing measures could not be used.
Item Generation
1. How many items to generate in the initial item pool?
The larger the item pool, the better.
Aim to begin with an item pool that is 3-4 times as large as the final scale.
Length depends on the complexity of the construct, which determines the number of dimensions. Each dimension has its own sub-scale. The minimum is to have at least 4 items (aim for 6-10 items) in each dimension or sub-scale for reliability
(30-40 for each dimension in starting item pool)
Item Generation
2. What response choices are there for Likert scale?
- Agreement
- Frequency
- Evaluation
Types of response option
— Agreement (strongly disagree slightly disagree neither agree nor disagree slightly agree strongly agree)
• I always attend the 8am lesson.
— Frequency (Rarely Seldom Sometimes Most of the time)
• How often do you attend the 8am lesson?
— Evaluation (Terrible Inferior Passable Good Excellent)
• How good is your attendance record for the 8am lesson?
Item Generation
3. How many response options should I have?
— Generally 5-7 choices are optimal for most uses.
— More choices allow for greater precision and increase variability in response. But the fine distinction must be meaningful to the respondents or otherwise, it merely contributes to random error.
— Odd or even number of response choices have implications on whether there is a neutral point.
Item Generation
4. What pitfalls to look out for when writing items?
Cardinal rules for writing items: CLARITY
i. Keep item short and simple.
ii. Each item should express one and only one idea.
iii. Avoid colloquialism and jargon.
iv. Using negatively worded items.
v. Avoid writing items that will be endorsed or disagreed by everyone.
Item Generation
5. Is it a good idea to have redundant items?
— DeVellis: Redundancy is less of an issue in early stages of item development but undesirable in the final version.
— Distinguish between useful and useless redundancy:
- Item 1: “l will do almost anything to ensure my child’s success.”
- Item 2: “No sacrifice is too great if it helps my child achieve
success.”