How to run an EFA & CFA in R

TL;DR

This video introduces exploratory and confirmatory factor analysis using R with practical examples.

Transcript

hi friends welcome back to the channel today we're going to be looking at how to do an exploratory factor analysis or EFA and confirmatory factor analysis also known as a CFA and R so this is an introductory video we're not going to go through all of the theory and it's going to be practical it's been mainly around how to do this NR but as we go I'... Read More

Key Insights

• 🧑‍🏭 EFA and CFA serve distinct but complementary roles in factor analysis, facilitating the development and validation of psychometric instruments.
• 🧑‍🏭 Data suitability checks are critical before performing EFA, ensuring that the data's characteristics support meaningful factor extraction.
• 🤢 The use of specific R packages, such as 'psych' and 'lavaan,' simplifies the coding processes for EFA and CFA.
• 🧑‍🏭 Researchers should interpret factor loadings to understand how well different items map onto latent variables during factor analysis.
• 🧑‍🏭 The selection of rotation methods in factor analysis can significantly influence the interpretation of factors, highlighting the importance of understanding factor relationships.
• 🫰 Fit indices like RMSEA and CFI provide insights into the overall model fit, guiding researchers in evaluating the effectiveness of their proposed models.
• 👷 A careful approach to model specification in CFA can enhance the robustness of psychometric assessments, ensuring accurate reflections of underlying constructs.

Questions & Answers

Q: What are exploratory factor analysis (EFA) and confirmatory factor analysis (CFA)?

EFA is used to identify underlying relationships between measured variables without predefined notions of how many factors exist. In contrast, CFA tests a hypothesized factor structure, confirming whether the observed data fits an expected model based on prior research or theory.

Q: Why is it important to perform Bartlett's test and check Kaiser's criterion before EFA?

Bartlett's test checks for sphericity to ensure that there is enough correlation between variables to justify factor analysis. Kaiser's criterion assesses the sampling adequacy, indicating whether the sample size is sufficient and appropriate for factor analysis with good common variance.

Q: What role does the scree plot play in factor analysis?

A scree plot visually represents the eigenvalues associated with each factor. It helps researchers identify the number of factors to retain by highlighting where there is a change in slope, commonly referred to as the "elbow," indicating diminishing returns of added factors.

Q: How do you determine the number of factors to extract from the data?

Researchers often apply a scree plot to visualize eigenvalues or rely on theoretical or previous empirical findings. The goal is to retain factors that together account for a significant proportion of variance, providing useful insights into underlying constructs.

Q: What is the difference between orthogonal and oblique rotation methods in factor analysis?

Orthogonal rotation assumes that factors are uncorrelated, producing independent factors. Oblique rotation allows for correlation between factors, which is often more realistic in psychology or psychometric testing where variables may influence each other.

Q: What might indicate that a factor model is not a good fit?

If fit indices like the chi-square statistic are significant, or other fit indices such as CFI, TLI, and RMSEA fall below acceptable thresholds, it indicates that the model does not adequately represent the relationships within the data.

Q: Why is it valuable to link EFA and CFA in research?

Linking EFA and CFA allows researchers to first explore data-driven factor structures (EFA) and subsequently test and confirm these structures (CFA) to ensure reliability and validity in constructing measurement instruments.

Q: What can you conclude from the percentage of variance explained by factors?

A higher percentage of variance explained by the factors reflects a better model fit. Ideally, researchers aim for at least half of the total variance to be accounted for, ensuring the factors meaningfully capture underlying constructs.

Summary & Key Takeaways

• The video provides an introductory guide on performing exploratory factor analysis (EFA) and confirmatory factor analysis (CFA) using R, focusing on practical coding examples rather than theory.

• It discusses important preparatory steps, including testing data suitability with Bartlett's test and the Kaiser's criterion, essential for successful factor analysis.

• A real dataset of mental ability scores is utilized to illustrate how to determine the number of factors, perform factor analysis, and evaluate model fit in CFA.