How to Analyze Ranking Data

It is often useful to represent ranking questions in multiple ways, as:

• Numeric - Multi variable sets, so that the average can be displayed.
• Binary - Multi variable sets, to show, for example, the top 3 ranks.
• Nominal variable sets, so that the proportion in each rank can be seen.
• Ranking variable sets if you are planning to use the data as an input to segmentation (note that in general, this Structure is not able to be understood by people that do not have substantial technical expertise, as it displays outputs from a complex statistical model).

Two cautions when dealing with ranking data are:

• Ensuring that you know what the numbers mean (e.g., is 1 the lowest or highest rank, and has the variable set been named to reflect this.
• Dealing with missing values, in situations where missing values contain information (e.g., where people have missing data because something is known to be unimportant to them).

The best way to do basic analyses of ranking data in Displayr depends upon the structure of the data.

Please note this requires the Data Stories module or a Displayr license.

Method

One variable for each option being ranked and only some of the options are ranked (e.g., top 5)

1. Go to the Data Sources tree and select all the variables.
2. If not already done, select them all, right-click and select Combine and set the Structure to Nominal - Multi.
3. Click Data > DATA VALUES > Values.
4. Put some sensible descriptions in the Label column if they are not already there. E.g., replace 1 with First.
5. If any option that was not ranked is shown as Exclude from analyses in the Missing Values column:
   • Set Missing Values to Include in analyses for this option.
   • Give the Label a better name (e.g., Not top 5).
6. Proceed to the next section.

One variable for each option being ranked and all of the options are ranked

1. If you have not already done so:
   • Go to the Data Sources tree and select all the variables.
   • Select them all, select Combine in the toolbar and set the Structure to Nominal - Multi.
2. Drag your variable set onto your page to create a summary and, provided you have followed the steps above, you will see the percentage of people choosing each option. You can merge and create NET categories.
3. If necessary, create a Filter so that the table is only computed amongst people that saw the question.
4. Follow the steps for whatever analysis you want to perform below. If you want to perform more than one of these analyses then you can do so by first making a copy of the question. You can do this by highlighting all the variables in the Data Sources tree, right-clicking, and pressing Duplicate.

Computing top few box percentages (e.g., ranked top 3)

Change the Structure to Binary - Multi and ensure that the Count This Value settings are appropriate.

Computing averages

Change the Structure to Numeric - Multi.

Computing Probability %

Change the Structure to Ranking. Note, ranking questions treat the highest value as most preferred.

One variable for each rank

1. Go to the Data Sources tree and select all the variables.
2. If not already done, select them all, select Combine in the toolbar and set the Structure to Nominal - Multi.
3. If any option that was not ranked is shown as Exclude from analyses in the Missing Values column:
   • Set Missing Values to Include in analyses for this option.
   • Give the Label a better name (e.g., Not top 5).
4. Drag your variable set onto your page to create a summary and, provided you have followed the steps above, you will see the percentage of people choosing each option at each rank.

Computing top rank box percentages

1. In the Data Sources tree, select the first variable.
2. Right-click > Duplicate.
3. Change the Structure to Nominal (it may already be this type).
4. Drag your variable set onto your page to create a summary and it will show the percentage who ranked each option as first.

Computing top few box percentages (e.g., ranked top 3)

1. In the Data Sources tree, select the variables that contain the relevant data. For example, if wanting to show options ranked in the top three, you would select the first three variables.
2. Right-click > Duplicate.
3. Right-click > Combine.
4. Change the Structure Type to Binary - Multi (Compact) and click OK.
5. Drag your variable set onto your page to create a summary and it will show the percentage who ranked each option in the top few boxes (e.g., top 3).

Next

How to Show Average Rank in a Table

How to Create a Table of Ranks Using R

How to Highlight Table Cells Based on Top and Bottom Ranked Values