1) initial PCA – number of components equal to number of variables, only the first few components will be retained
2) Determine the number of components to retain
a. Eigenvalue > 1 criterion (Kaiser criterion, (Kaiser, 1960))
Each observed variable contributes one unit of variance to the total variance. If the eigenvalue is greater than 1, then each principal component explains at least as much variance as 1 observed variable.
b. Scree test – look for an elbow and leveling, large space between values, sometimes difficult to
determine the number of components
c. Proportion of variance for each component (5-10%)
d. Cumulative proportion of variance explained (70-80%)
e. Interpretability – principal components do not exhibit a conceptual meaning
3) Rotations is a linear transformation of the solution to make interpretation easier (Hatcher, p.28) With an orthogonal rotation, loadings are equivalent to correlations between observed variables and components.
4) Interpret rotated solution
5) Summarize results in a table
6) Prepare paper
a. Describe observed variable
b. Method, rotation
c. Criterion to determine number of components, eigenvalue greater than 1, proportion of variance
explained, cumulative proportion variance explained, number of components retained d. Table, criterion to load on component