Readings

• Perceptual Mapping with SPSS Categories: An Extended Example
  
• Introduction to Correspondence Analysis (StatSoft)

Assignments

• Article to read and critique: "Graphical Representation of University Image: A Correspondence Analysis," by Ugar Yavas and Donald J. Shemwell (1996), in the Journal of Marketing for Higher Education 7(2): 75-84. In this critique, be prepared to discuss the following:

  What is the purpose of the study?
  What are the data?
  What are main conclusions/recommendations?
  Comment on validation issues
  Comment on the author's coding of variables
  Comment on Table I (p. 79)
  Comment on Table 2 (p. 80)
  Comment on Figure 1 (the correspondence map), p. 82
  Comment on Conclusions:
  1. Comment: "High Level or Personal Attention from Instructors is so far from all the universities that this represents a great strategic opportunity...." (P. 81, bottom).
  2. Comment: "University H should emphasize other attributes with which it is closely aligned in Figure 1. For instance, Low Cost of Tuition and Convenience of Location should be stressed in promotional efforts...(also) the university should emphasize its successful retention programs."
  Comment on the Assumptions of Correspondence Analysis as listed in StatNotes, in comparison with this article.
  Discuss omissions in the discussion of eigenvalues
  What can be said about the failure to report contribution of dimensions to points?
  

• As part of your evaluation, run a correspondence analysis on their data, from Table 1, p. 79.
  • Data coding: Assign the numeric codes 1 through 10 to the ten traits (InstructionQuality, Affordability, etc). Assign the numeric codes 1 through 8 to the eight colleges, A through H.

  • Data types: In the Variable View tab, set the first three columns to numeric and label them Trait, College, and Weight. In the Value Labels field, click and enter the value labels (ex., for Trait, 1 - InstructionQuality). Enter value labels for Trait and for College.

  • Data entry: Back in the Data View spreadsheet, for the first column (Trait) enter a column of 1 through 10 eight times. That is, there will be 80 rows in 8 sets of 1 to 10. In the second column (College), enter 10 1's, then 10 2's, etc., up to 10 8's, for 80 rows. For the Weight variable, enter the corresponding cell count from Table 1 of the article.

  • Weighting cases. Select Data, Weight Cases, and weight by the variable Weight.

  • Correspondence analysis: Select Analyze, Data Reduction, Correspondence Analysis. Set Trait as the Row Variable and College as the Column Variable. For each, set the data range (1 to 10 for Trait, 1 to 8 for College). Click on the Model, Statistics, and Plot buttons to set output options you want (if in doubt, asking for more rather than less is desirable).

  • Nominal association: Also select Analyze, Descriptives, Crosstabs, and in the Statistic button option panel, ask for chi-square and all the nominal-level measures of association, in order to assess the extent to which Traits are associated with Colleges.

Explore

• Correspondence Analysis, by Diane Phillips
  
• Typical Use of Correspondence Analysis in Marketing
  
• Correspondence Analysis in SAS