A ranking is a relationship between a set of items such that, for any two items, the first is either “ranked higher than”, “ranked lower than” or “ranked equal to” the second. – Wikipedia
Ranking in SPSS Statistics
IBM SPSS Statistics ranks cases in your data pond by automatically defining new variables to contain ranks, normal and Savage scores, and percentile values for your selected numeric variables.
New variable names and descriptive variable labels are automatically generated for you by SPSS, based on the original variable name and the selected measure(s). The ranking process also produces a summary table listing the original variables, the new variables, and the variable labels.
On the road to Ranking Cases
Transform > Rank Cases…
from the menu.
From there the Rank Cases dialog allows you to:
- Select one or more variables to rank (you can rank only numeric variables)
- Rank cases in ascending or descending order
- Organize rankings into subgroups by selecting one or more grouping variables for the “By list”.
Rank Cases: Types
IBM SPSS allows you to select multiple ranking methods and a separate ranking variable is created for each method you choose. Ranking methods include:
- Rank – Simple rank. The value of the new variable equals its rank.
- Savage score -. The new variable contains Savage scores based on an exponential distribution.
- Fractional rank. The value of the new variable equals rank divided by the sum of the weights of the non-missing cases.
- Fractional rank as percent. Each rank is divided by the number of cases with valid values and multiplied by 100.
- Sum of case weights. The value of the new variable equals the sum of case weights. The new variable is a constant for all cases in the same group.
- Ntiles. Ranks are based on percentile groups, with each group containing approximately the same number of cases. For example, 4 Ntiles would assign a rank of 1 to cases below the 25th percentile, 2 to cases between the 25th and 50th percentile, 3 to cases between the 50th and 75th percentile, and 4 to cases above the 75th percentile.
- Proportion estimates. Estimates of the cumulative proportion of the distribution corresponding to a particular rank.
- Normal scores. The z scores corresponding to the estimated cumulative proportion.
A total preorder is defined as a ranking where no pair of items is incomparable. In realistic case ranking it is not uncommon for you to encounter cases that “cannot be ranked” or, in simple terms, you have a “tie”. SPSS recognizes case ties and gives you the ability to determine what method to use to resolve the tie.
Rank Cases: Ties
From the Rank Cases dialog, you can click on Ties… The Rank Cases: Ties dialog box controls the method for assigning rankings to cases with the same value on the original variable. You can select 1 of 4 methods:
- Sequential ranks to unique values
Going back to a previous blog post which included a data pond of male and female respondents indicating a perception of married life, I expanded the possible response to make things a little more interesting. Now, a respondent can choose from 1 of 10 values:
Happy, Sad, Bored, Excited, Neutral, Angry, Blissful, Exhausted, Youthful and Lost.
I’ve also added to the data to represent these new cases. Now I am ready to perform a ranking. I’ll keep it simple, so from the Rank Cases dialog, I’ll select the variable “Happy” (which if you remember, holds the respondents response value) as the only (ranking) variable:
Now I’ll click OK. The output pane in the viewer shows my summary table:
More interestingly, my data viewer shows the new ranking variable that SPSS added (RHappy):
Finally, I can use the Data View to see the actual assigned ranking values:
I hope you find this post helpful and would enjoy any feedback you may have.