Determining the Number of Principal Components: A Comparison of Two Methods