MKTG 465 10 - scale assessment

according to churchill, what are the construct measures?

validity, reliability, dimensionality

validity includes what

face validity - judgement call by research

content validity - judgement supported by literature/studies

construct validity - discriminant and convergent

discriminant validity

the lack of association among
constructs that are purported to be different

convergent validity

the degree of association among

measures purporting to measure the same construct

reliability

repeatability, internal consistency, coefficient alpha

coefficient alpha (cronbach): alpha has to be ___ or higher to be reliable and not have too much ___

0.6, variance

multi-dimensional questions

questions have multiple dimensions

• in a question about “attitude toward this class,” statements 1 - 4 are about course delivery and statements 5 - 8 are about course content

dimensionality: factor analysis

A statistical approach used to condense information contained in the original variables into a smaller set of variables with a minimum loss of information

⧫ A “cluster analysis” of similar variables

purpose of factor analysis

summarization and data reduction

explanation of factor analysis model

• X (Observed variable): what you measured (your survey question/result)

• F (Factor): the why behind the responses

• λ): correlation, how strongly the question is connected to that why

• e: error

a good survey will be/have:

multi item (multiple item statements), content validity (verified by academic sources), sensitive, reliable (alpha >= 0.6), and multidimensional (questions have more than 1 dimension)

sensitivity

how precise your scale is at picking up differences

• A 2-point scale (Yes / No) → low sensitivity

• A 7-point scale (Strongly disagree → Strongly agree) → high sensitivity

factor analysis - SPSS

1. only look at “rotated component matrix” charts, ignore others

2. X = individual questions, F = components, λ = the numbers

3. find the highest correlations (numbers) in each row, then for each column, find the correlations between the highest correlations.

4. ex: after finding highest correlations in each row, look at each column. for each highest correlation in the column, read the questions associated with them and determine what the main theme of those high correlations are.

5. after, take your main themes and create new columns in SPSS with them and run a reliability analysis on them.

6. look at the cronbach alpha = has to be 0.6 or higher to be reliable