Some Comments & Captured Screenshots on Using SPSS to Create an Overall Satisfaction Index

(c) 1998-2009 H. Bruce Lammers
Marketing Research Technology Lab
California State University, Northridge

The main approach taken in this guideline is that responses to a survey consisting of  several scaled (e.g., Likert format) items purporting to measure attitudes or satisfaction with some objects (e.g., brands) and also consisting of a global attitude/satisfaction item will be subjected the following analytical stages in logical sequence:
  1. Simple Descriptive Statistics--just some basics (don't forget to come back to this page)
  2. Demographics--examining some of the basic demographics of the respondents graphically
  3. Factor Analysis--of the Likert items (excluding the global satisfaction item) to determine the stucture of the items
  4. Reliability Analysis--of the structure, along with discarding the unreliable items and factors
  5. Factor Analysis--of only the reliable items to create factor scores
  6. Regression--to create the Overall Satisfaction Index with the global item as the dependent variable and the newly created factor scores as the independent variable(s).
  7. Graphs--showing the Overall Satisfaction Index for each "brand."


Demographics of Respondents(Round 1)
Example of Bar Graph selection in SPSS of relative frequency distribution on a demographic
Select "percent of cases" and the demographic to be graphed.
Bar Chart in Final Form
Another example, but this time with a pie chart.
Select Pie Chart--"groups of cases"
Select "percent of cases" and the demographic to be pie-charted.
Cute Pie Chart in Final Form
Factor Analysis (2nd round)
Factor Analysis is a data reduction technique in SPSS statistics.
The look of the opening screen in Factor Analysis.
Put the Likert items in the Factor Analysis container (window).
Descriptives Button selections
Extraction Button selections.
Rotation Button selections.
Factor Scores Button selections.
Missing Values Button selections.
Output:  Simple Descriptive Statistics
Output:  Simple Bivariate Correlation Coefficients
Output:  Communalities of each item.
Output:  Eigen Values of each factor (component).
Output:  Scree Plot of the Eign Values.
Output:  Factor Loadings (use the Rotated Matrix).
Output:  Factor 1 Loadings.
Output:  Factor 2 Loadings.
Output:  Factor Component Transformation Matrix (not needed).

Reliability Analysis(Round 3)

First, Reverse Score All Items That Negatively Load on their Factors.
Before running the reliability analysis on your factors, be sure to reverse score (recode) all items which loaded negatively on their factors (i.e., they had negative factor loadings).
Select Transform/Recode/Into Different Variables.
Select the variable to recode.
Variable for recoding has been selected.
Provide a variable name and variable label for the recoded version of your variable.
Select "Old & New Values" Button
Provide the old and new values of the variable to be reverse scored.
Be sure to push the "Add" button to make the recoding official.
Example of completely reversed scored item in the making.
Repeat for any other variables that need to be reverse scored.
Notice your newly formed variables are in your data file now.
Reliability Analysis is a Scale Procedure in SPSS Statistics.
Select the items belonging to one of the factors.
Statistics Button selections.
Compare your coefficient alpha with the "Alpha if Item Deleted" column.
Select the unreliable item for removal from the factor.
The unreliable item removed from the factor.
The mean response to the items on a reliable factor (The Likert Attitude Scale Score).
Repeat the reliability analysis on the remaining factors.
Example of a factor whose Coefficient Alpha can't be improved by deleting any more items.
The mean response to the items on a reliable factor (the Likert Attitude Scale Score).

Factor Analysis (Round 4)
Select all of the RELIABLE items for another round of factor analysis.
Factor Scores Button selections.
Factor Scores created.

Regression Analysis (Round 5)
Select Linear Regression Analysis
Select your global attitude item as the dependent variable and your factor scores as the independent variable.
SAVE button selections.
OPTIONS button selections.
Output: The satisfaction index score ("pre"  for "predicted" scores).
Rename the PREdicted score to "Satisfaction Index Score"

Graphing Results --"Mean Satisfaction Index by whatever, e.g., Age, Brand"(Round 6)

Select Bar Graph procedure.
Define bar graph as "simple for groups of cases"
Define the bars and category axis.
Output Graph:  Very rough and needs to be cleaned up a lot.
Output Graph:  Final version...looks rather nice.
Example 2 of bar graph definition.
Output Graph:  Final version of example 2---rather nice looking graph.
Example 3 of bar graph definition
Output Graph:  Final version of example 3---another beauty.
 
 

end of story.