cair2003 willard

An Updated Elementary Peer Grouping of California Community Colleges : 

An Updated Elementary Peer Grouping of California Community Colleges Willard C. Hom, Director Research & Planning Unit, Chancellor’s Office California Community Colleges

Overall Study Goal: 

Overall Study Goal Enable each community college (CC) to compare itself with other CCs of similar qualities.

Specific Study Goal: 

Specific Study Goal Update a prior “peer grouping” matrix that the Chancellor’s Office had created more than four years ago (known as Matrix C-1).

Method: 

Method Fall 2001 student headcount per CC. Census 2000 “radial” population per CC. Cluster analysis.

Student Headcount: 

Student Headcount Based on COMIS. Credit and Non-Credit Students. Part-time and Full-time students.

Census Data: 

Census Data Census 2000 population within a 5-mile radius of the CC. Developed by the Calif. Dept. of Finance.

Clustering Technique: 

Clustering Technique Hierarchical. Average linkage between groups. Euclidean squared distance measure. Search target of 12 groups. SPSS 11.0 software.

Slide8: 

Multivariate Relationship Distinguishing Each Peer Group

Slide9: 

Peer Groups by Levels of Enrollment and Population Density     Note: Numbers in the cells are the cluster identification numbers for the different peer groups.  

Smallest Cluster: 

Smallest Cluster City College of San Francisco (9 or ‘i’) 776,000 population 31,200 students

Clusters of Two: 

Clusters of Two ’10’ or ‘j’: DeAnza & Palomar ’11’ or ‘k’: East L.A. & L.A. City ’12’ or ‘L’: L.A. Trade & L.A. Valley

Cluster #8 or ‘h’: 

Cluster #8 or ‘h’

Cluster #7 or ‘g’: 

Cluster #7 or ‘g’

Cluster #6 or ‘f’: 

Cluster #6 or ‘f’

Cluster #5 or ‘e’: 

Cluster #5 or ‘e’

Clusters #3 (‘c’) and #4 (‘d’): 

Clusters #3 (‘c’) and #4 (‘d’)

Cluster #2 or ‘b’: 

Cluster #2 or ‘b’

Cluster #1 or ‘a’: 

Cluster #1 or ‘a’

Points to Remember: 

Points to Remember Different clustering techniques will produce different clusters (“peer groups”). In one cluster, the interval (min to max) for a classifying variable (population or enrollment) may overlap the corresponding interval of one or more other clusters. These clusters may become obsolete in several years; we will need to update them.

Conclusion: 

Conclusion A different clustering process, using different or additional classification variables, may be necessary, depending upon the policy needs of the analyst.

Further Reading: 

Further Reading Lorr, M. (1987). Cluster analysis for social scientists. San Francisco: Jossey-Bass. Gore,P.A. (2000). Cluster analysis. Chapter 11. In H.E.A. Tinsley & S.D. Brown, Handbook of Applied Multivariate Statistics and Mathematical Modeling. San Diego: Academic Press. 297-321.

Contact Information: 

Contact Information E-mail: whom@cccco.edu Telephone: 916-327-5887 Mail: Chancellor’s Office 1102 Q Street, Sacramento, CA 95814-6511

Thank you.: 

Thank you.