Quiz #4 Homework #3

  1. use  q4hw3a.sav The data set includes single male undergraduate students. Perform a cluster analysis using the variables in the data set.
  1. Transform the variables into Z-scores
  1. Obtain a hierarchical clustering solution using Wards method with squared Euclidean distances. Be sure to save the cluster memberships for 2-10 cluster solutions.
  1. Run Frequencies for the 3-, 4-, and 6-cluster solutions. How many clusters in each are “strays”?

3-cluster solution ______4-cluster solution ______6-cluster solution ______

  1. Obtain the cluster profiles for the following groups from specific clustering solutions. Arrange the order of the variables to maximize the clarity of the group similarities and differences (and use the same order for all three profiles). Copy the profiles into the empty table below.
  1. Describe how the clusters in each solution agree and differ (no stats, do this by inspection).
  • the 4 larger clusters from the 6-cluster solution (insert the profile plot below)
  • the 3 larger clusters from the 4-cluster solution (insert the profile plot below)
  • the 2 larger clusters from the 3-cluster solution (insert the profile plot below)
  1. Pick the “best” solution, and describe why you think it is the best representation of the subpopulations of this group.
  1. use  q4hw3b.sav The data set includes data from high school students. Perform a cluster analysis using the variables in the data set.
  1. Do not transform the variables. They are are T-scores -- scaled to a population (not sample) mean = 50 and std = 10.
  1. Obtain a hierarchical clustering solution using Wards method with squared Euclidean distances. Be sure to save the cluster memberships for 2-7 cluster solutions. Be sure to examine the frequencies for each and note how many of the clusters are “clusters” and now many are “strays”!!!
  1. Obtain the cluster profiles for the three most likely cluster solutions. Arrange the order of the variables to maximize the clarity of the group similarities and differences (and use the same order for all profiles). Copy those 3 profile graphs into the table below.
  1. Pick the “best” solution, and describe why you think it is the best representation of the subpopulations of this group. Remember that we are interested in both “shape” and “level” differences” between clusters.