In this example, the following scale is used to compute quantitative summaries for responses:


ValueAbbrevDescription
 5  SA  Strongly Agree
 4  A  Agree
 3  N  Neutral
 2  D  Disagree
 1  SD  Strongly Disagree

NA (Not Applicable responses and responses left blank are not included when computing quantitative summaries.


The median is the middle observation in a sorted list of data. Half of the values in the data set are less than or equal to the median and half are greater than or equal to it. The Group Median (which is an optional statistic on the Evaluation and Survey Intelligence reports in Course Evaluations) adjusts the median slightly upward or downward.


For example, any group median between 3.5 and 4.5 indicates that the actual median rating for the question was 4. A group median between 4.0 and 4.5 also indicates that there were more ratings above 4 than below 4. Similarly, an group median between 3.5 and 4.0 indicates that there were fewer ratings above 4 than below 4.


Why use group median?

To illustrate the usefulness of the group median, consider two questions with 20 responses for each question. The table below lists the number of respondents for each question that gave each response to a particular question:


ResponseQuestion 1Question 2
5 = Strongly agree91
4 = Agree1010
3 = Neither agree nor disagree 06
2 = Disagree11
1 = Strongly disagree02
Median44
Mean4.35
3.35
Group Median*4.403.60
* The subjects of the evaluation really appreciate the higher scores typically produced using group median vs. mean!


Both question 1 and question 2 have medians of 4 for this question. However, it is quite clear that the overall ratings on question 1 were substantially higher than question 2. The group median provides a way to adjust the median to reflect this. The group median for Question 1 is 4.40. The median is adjusted upward since 9 respondents gave a rating above the median while only 1 gave a rating below the median. On the other hand, for Question 2, more respondents gave ratings below the median than above it, so the group median adjusts downward to 3.60. The group median clearly represents the differences between the two questions, while the median failed to do so.  


Group Median vs. Mean

In the example above the mean (average) is lower in both cases. While the mean statistic is fine and valid, it does not take into account the population bias toward the positive nor does it take into account the responses which no, or very few, respondents selected. The graphic below illustrates how the mean is negatively skewed by a few answers when most of the answers are positive.




How is the group median actually computed?

Define variables as follows:
M = the standard median of the responses
nl = number of responses strictly less than M
ne = number of responses equal to M
ng = number of responses strictly greater than M

The group median GM is then computed as follows:

If ne is nonzero:

    GM = M + (ng - nl) / (2ne)

If ne is zero:

    GM = M