Efa, Cfa

Download as docx, pdf, or txt
Download as docx, pdf, or txt
You are on page 1of 3

EFA:

Is the most longest procedure in Factor analysis as it gives much inputs/results.

Bartlett's Test of Sphericity:


Significant results (p-value < 0.05) shows that there is relationship between variables, making them
suitable for factor.

Kaiser-Meyer-Olkin Measure of Sampling Adequacy:


Higher KMO values (close to 1) indicate that factor analysis is appropriate.it checks that if CFA can be
applied or not.

Communalities:
This table helps us understand how well each variable is represented by the factors extracted in the
analysis. Higher values indicate better representation, while lower values suggest the need for further
investigation or adjustment.
Normally, the value of extracted communality above 0.4 is considered acceptable. It means that each
variable plays an important role in explaining the variable

Component matrix:
Generally, cumulative percent greater than 60% are generally seen as satisfactory ,and commulative %
shows that how much variation that variable explains.

Rotation Component Matrix:


Rotation in Exploratory Factor Analysis (EFA) simplifies interpretation by clarifying which variables are
associated with which factors. It reorganizes the factor structure so that each factor is defined by a subset
of variables that are highly correlated with each other. This process aims to achieve a clearer and more
meaningful pattern of relationships between variables and factors.
Rotation methods like Varimax or Promax aim to "rotate" these factors in a way that maximizes the
variance of loadings within each factor, making it easier to see which variables are primarily related to
which factors. This rotation simplifies interpretation by grouping variables that measure the same
underlying construct together under one factor, thus providing a clearer understanding of the relationships
between variables and factors in your data.

Scree Plot:
The scree plot in EFA helps us decide how many factors to keep. It shows us a graph where the factors
are ranked by importance. At first, the plot drops steeply, showing each factor explains a lot. Then it
flattens out. This flat part is where we stop adding factors because they don't explain much more. It helps
us find the right number of factors to understand our data well, without making things too complicated.

CFA:
Table1:
 Estimate" typically refers to the standardized loading of each item on its factor.
 "S.E." stands for standard error, indicating the precision of the estimate.
 "C.R." denotes the critical ratio, which is a measure of the significance of the loading (how
much larger the estimate is relative to its standard error).
 "P" represents the p-value associated with each loading, indicating whether the loading is
statistically significant.
 "Label" may provide additional information about each item or factor.

In Confirmatory Factor Analysis (CFA), researchers aim to understand how observed


variables (like survey items) relate to latent constructs (or factors). Here's a general explanation
of the concepts mentioned:

 Regression Weights: These are coefficients that represent the strength and direction of
the relationship between each observed variable (item) and its corresponding latent
factor. In CFA, these weights indicate how much each item contributes to defining the
underlying construct.
 Standardizing Weights: Standardizing weights (often by keeping them below one) is a
common practice in CFA to make the model more interpretable and comparable across
different scales of measurement. It ensures that the influence of each item on its factor is
measured consistently.
 Setting a Reference Category: Sometimes, one item within each factor is chosen as a
reference category. Its regression weight is typically set to zero for comparison purposes.
This allows researchers to assess the relative importance of other items within the same
factor. Items with higher regression weights compared to the reference item are
considered more significant for that factor.
 Interpreting Significance: The significance of regression weights (often assessed
through statistical tests like critical ratios or p-values) indicates whether the relationship
between an item and its factor is statistically meaningful. Significant weights suggest that
the item is a good indicator of the underlying construct.

Table2
By using the plugin: “model fit measures” we’ll get the table of fit. This table tells about the ftness of
model.

 CMIN: This represents the minimum discrepancy between the observed and predicted
covariance matrices. A lower CMIN value indicates a better fit, but the interpretation is
specific to the context and complexity of the model.
 DF (Degrees of Freedom): It indicates the number of independent pieces of information
available to estimate parameters in the model. It helps assess the model's complexity
relative to the data.
 CMIN/DF: This is the ratio of CMIN to DF. It provides a normalized measure of fit,
where values between 1 and 3 are typically considered acceptable or excellent, depending
on the context and conventions in the field.
 CFI (Comparative Fit Index): CFI compares the fit of your model to a baseline model
(often a null model). Values closer to 1 indicate better fit, with 0.95 or higher generally
considered acceptable.
 SRMR (Standardized Root Mean Square Residual): SRMR measures the average
discrepancy between observed and predicted correlations. Values less than 0.08 indicate
good model fit, with lower values indicating better fit.
 RMSEA (Root Mean Square Error of Approximation): RMSEA estimates the error of
approximation of the model to the population. Values less than 0.06 indicate good fit,
with values up to 0.08 considered acceptable.
 PClose: This assesses the probability that the RMSEA is 0.05 or less. Values greater than
0.05 suggest that the model fits well.

You might also like