AI-ordlista
Den kompletta ordlistan över AI
Manifest variables
Directly observable or measurable indicators used to represent and quantify underlying latent variables in a structural model.
Measurement model
Part of SEM specifying the relationships between latent variables and their observable indicators, assessing construct validity and reliability.
Structural model
Component of SEM describing hypothesized causal relationships among the latent variables themselves, testing direct and indirect effects.
Causal path
Theorized directional relationship between two variables in an SEM, represented by an arrow and quantified by a standardized regression coefficient.
Composite reliability
Measure of internal consistency evaluating the coherence of indicators in representing a latent variable, superior to Cronbach's alpha in structural models.
Convergent validity
Degree to which indicators of the same theoretical construct are strongly correlated, demonstrating that they adequately measure the same underlying concept.
Discriminant validity
Statistical evidence that distinct latent variables measure different concepts, assessed through moderate or low inter-construct correlations.
Fit indices
Statistical metrics (CFI, TLI, RMSEA, SRMR) evaluating the extent to which the theoretical model reproduces the structure of observed data.
RMSEA
Root Mean Square Error of Approximation, a fit index penalizing model complexity, with values below 0.08 indicating acceptable fit.
Mediating effect
Mechanism by which a variable influences another indirectly through a third variable, explaining the underlying causal process.
Moderating effect
Variable that modifies the strength or direction of the relationship between two other variables, creating conditional interaction effects.
PLS-SEM
Partial Least Squares Structural Equation Modeling, a variance-based approach favoring prediction and suited to small sample sizes or non-normal distributions.
Measurement invariance
Statistical verification that a measurement instrument functions equivalently across different temporal or population groups.