Scales Flashcards
Nominal scale / nominal variables
Labelled. Unordered categories.
Dichotomous scale /dichotomous variables
Labelled, binary data. 0-1. E.g. sex. Dummy variables
Ordinal scale/variables
Meaningful order without measurable distance. Ordered categories. E.g. Likert scale (can under certain conditions and arguments be treated as interval scale)
Interval scale
Measurable distance. Continuous with arbitrary zero.
Ratio scale
True zero starting point. Continuous with absolute. E.g. income.
Discrete scale
Distinct, separate values. Data measured on a discrete scale can only take on specific, distinct values, often whole numbers. There are no intermediate values between these points.
Continuous scale
Values that can take on any value within a given range. Continuous data can be measured and can include fractions or decimals.
Unipolar scale
Measurement in a single direction. E.g. not satisfied to highly satisfied. The scale often includes a neutral or zero point that indicates the absence of the attribute being measured.
Bipolar scale
Responses are measured in both directions, allowing respondents to express varying degrees of a trait that has both positive and negative dimensions. Neutral point in the middle
Reflective construct
Reflective constructs are latent variables that are indicated by their observed variables (indicators). In this approach, the underlying construct is presumed to cause the observed measures. The indicators reflect the same underlying concept.
Example: A reflective construct for satisfaction might include indicators such as “satisfied with service,” “satisfied with product quality,” and “satisfied with customer support.” Each of these indicators reflects the underlying satisfaction construct.
Analysed using factor analysis, such as EFA.
Formative construct
Formative constructs are latent variables that are formed by their observed indicators. In this approach, the indicators collectively define or create the construct, without reflecting the same underlying phenomenon.
Example: A formative construct for socioeconomic status might include indicators such as income, education level, and occupational status. Each of these indicators contributes uniquely to the overall concept of socioeconomic status.
Analysed using regression analysis or structural equation modeling (SEM), f.eks. CFA.
Exploratory factor analysis (EFA)
A factor analysis is used to test the variability among observed items in terms of a
lower number of unobserved constructs, called factors. Factors are the latent variables that are inferred from the observed variables. They represent underlying constructs that explain the patterns of correlations among the variables.
In an EFA, the number of factors and the relationships among items is not fixed but “explored”.
Confirmatory factor analysis (CFA)
A confirmatory factor analysis (CFA) tests the hypothesis that the items are
associated with specific factors.
