SPSS & PLS Exam Prep - Hair et al. Chapters

Two or more independent variables are tested in a single experiment.

A statistical test used to determine if the variances of the dependent variable across different groups are equal.

A statistical test for equality of variance-covariance matrices. It is used in multivariate analysis of variance (MANOVA) to check if the variances of the dependent variables are equal across all groups.

Occurs when two or more independent variables have a high correlation with each other. This can inflate the effect of the individual variables and make it difficult to interpret the results of an analysis.

A type of ANOVA that uses both metric and categorical variables.

A measure of how much a variable is explained by the common factors in a factor analysis.

A loading of 0.30 or higher is considered acceptable in exploratory factor analysis with large samples.

A loading of 0.50 or higher is accepted in confirmatory factor analysis with smaller samples.

When a variable loads highly on multiple factors, making it difficult to determine its true underlying construct.

A measure of how well the factor model reproduces the observed correlations. Aim for a residual of less than 0.2.

A model where there are more parameters to estimate than unique terms in the data.

A model where the number of parameters to estimate equals the number of unique terms in the data.

A model where there are fewer parameters to estimate than unique terms in the data.

A simple method where a single variable with the highest loading represents the entire factor. It's easy but doesn't capture all aspects of the factor.

A mistake made when you reject the null hypothesis when it is actually true. Imagine you're testing a new drug, and you conclude it works when it actually doesn't.

A mistake made when you accept the null hypothesis when it is actually false. Imagine you're testing a new drug, and you conclude it doesn't work when it actually does.

A proposed explanation for a phenomenon, often used in research to guide experiments and analysis.

A specific, testable prediction about the relationship between variables. A well-defined statement about the relationship between variables.

A type of hypothesis that doesn't predict a specific direction of the effect between variables; it simply states that there is a relationship.

An abstract concept that represents a phenomenon of interest in research. It's something you can't directly measure, but it influences what you observe.

A technique grouping cases with similar characteristics. Think of sorting people into groups based on shared traits.

When two measures of the same concept are correlated. Think of two different tests measuring the same thing, how well their results match.

A broad way to ensure a set of items accurately represents a concept. Like making sure various questions accurately tap into the concept you want to measure.

Evaluates how well items in a scale match what they are supposed to measure. Imagine checking if a test's questions actually cover the topic

A part of construct validity focusing on how distinct two concepts are. Think of separating two things that shouldn't be related.

The variance explained by a factor. Imagine how much a factor contributes to overall variation in the data.

A threshold for eigenvalue of 1, indicating factors that explain a significant amount of variance. Think of a factor needing to be strong enough to matter.

A measure of how well a correlation matrix is suitable for factor analysis. Values above 0.5 are good. Imagine it's a check on the quality of the data.

Specifying the number of factors and variables that load on each construct in a model.

Parameters in a model, like variances and loadings, can be fixed (determined by the researcher), free (estimated from data), or constrained (estimated with limitations).

Deductive approach assumes a pre-existing theory, and model parameters are used to explain observed data. Inductive approach uses data to infer new theories and model structure.

How variables differ from their average values.

How variables change together.

Estimating model parameters to minimize the difference between the observed and predicted covariance matrices. Methods include Maximum Likelihood (ML), Unweighted Least Squares (ULS), Generalized Least Squares (GLS), and Distribution-free (ADF).

Evaluating how well the model fits the data. Measures include descriptive fit, statistical inference, approximate fit, comparative fit, and information theoretical measures.

A goodness-of-fit measure indicating the proportion of variance and covariance explained by the model. Values above 0.90 are considered good.

Degrees of freedom reflect the complexity of a model. Adjusted Goodness of Fit Index (AGFI) adjusts for model complexity relative to the number of data points.

A statistical inference test where a significant result indicates a mismatch between the model and data. Aim for an insignificant result.