Data editing Rank Cases

Using Rank Cases one is able to assign rankings to cases. You can use is for example to create a top 10 of cases based on the values of a variable you choose.

• Example of a question
• Coding
• Using Rank Cases
• The result
• Download the sample data file

Example of a question:

In our example we have a list of exam grades for a module from quarter 2.

Coding:

The beginning of the grades in the Data View:

Using Rank Cases:

Select from the menu: Transform > Rank Cases; choose the variable to use for Rank Cases.

By means of you can add some options. In this example we have asked for Ntiles, with parameter 4. This means we want SPSS to split the data in four subgroups of equal size and indicate in the new variable Ntiles to which subgroup each case belongs.

By means of you can indicate how SPSS should handle ties, i.e. how to rank cases with the same value for the sorting variable.
The default setting is to use the average of the ranks involved. For example, if two outcomes share the first and second place, they will both recieve ranking 1.5
In our example we kept the default setting.

The result:

The output tells us that two new variables have been created:

In the data window we find:

In the picture on the right you can see how the ties for 70 (three occurences) and 71 (two occurrences) are handled. You can also see that the first quartile group ends after the grade of 70 and the second one start at the grade of 71.
The dealing with ties is the reason that the new variable RGrade is defined by SPSS as having 3 decimal places.

A frequency table for NGRADE shows us:

Note that the third and fourth groups are of unequal size. This has to do with ties. Equal grades can never be divided over two different Ntile groups. By making a frequency table of the original grades and looking at the cumulative percentages this will become clear.