Sunday, August 24, 2025

Duncan’s Multiple Range Test (MRT)

 Introduction

An Analysis of Variance (ANOVA) indicates whether there is at least one significant difference between group means, but it does not specify which groups differ from each other. Therefore, ANOVA is considered a global or omnibus test. To identify the specific differences, it is necessary to perform comparisons between the group means after the global test.

The methods used to compare means after an ANOVA are called a posteriori tests or post hoc comparisons. Among the most well-known tests—and widely available in statistical packages—are the Tukey test (covered previously) and the Duncan test, which will be detailed here.

The Duncan test, also known as Duncan’s Multiple Range Test (MRT or DMRT), is applied after a significant ANOVA to identify which pairs of means differ statistically in experiments with three or more groups. Unlike the Tukey test, which performs independent pairwise comparisons, the Duncan test is a sequential procedure based on the calculation of minimum significant ranges.

Although more laborious to perform manually—as it requires calculating different critical ranges—the Duncan test is widely used in some fields. Its operation will be demonstrated with an example. However, it should be noted that DMRT is not widely accepted as Tukey or SNK procedures and is declared for some statisticians to perform poorly.

Example

Consider the fictional data on blood pressure reduction (in mmHg) presented in Table 1. This data was subjected to a one-way ANOVA, the results of which are in Table 2. Since the F-value was significant at the 5% significance level, the null hypothesis is rejected, concluding that there is at least one difference between the group means. The sample means for each group are in Table 3.

          Table 1 – Blood pressure reduction (mmHg) by treatment group

Table 2 – ANOVA table for blood pressure reduction data

Table 3 – Mean blood pressure reduction (mmHg) by group

Question: Which means are statistically different?

To answer this question, the Duncan test is applied. First, it's important to understand its logic.

How the Duncan Test Works

The Duncan test compares the range (the difference) between sets of ordered sample means with a calculated minimum significant range (Rm).

·        If the observed difference between the largest and smallest mean in a set is greater than the calculated minimum range (Rm), it is concluded that the corresponding population means are different at the chosen significance level.


·        The test is sequential (stepwise). It begins by comparing the largest and smallest mean of the entire set (k means). If the difference is significant, the algorithm proceeds, comparing subsets of means. Crucially, if a difference at a particular step is not significant, all comparisons within that specific subset are considered not significant and are halted.

Steps of the Test

1.               Order the means in ascending or descending order.

 Table 4 – Means ordered in descending order (blood pressure reduction, mmHg)

2.             Calculate the minimum significant range (Rm) for different values of m, where m is the number of means spanned in the interval being compared (e.g., when comparing the 1st and 5th mean in a list of 6, m=5). The formula is:

Where:

· Rm= critical value of the studentized range for a given m and for the residual degrees of freedom, found in specific tables (e.g., Harter, 1960);

· MSE = mean square error (residual) from the ANOVA;

· r = number of replications per group (assuming groups of equal size).

Application to the Example

·  Number of treatments (means): k = 6

·  MSE = 36.00 (Table 2)

·  n = 5 replications (Table 1)

·  Residual degrees of freedom = 24


Standard Error Calculation:


Comparisons:

1.                   m = 6: Compare the largest (D, 29.0) and smallest (Control, 2.0) mean.

Comparison:

               Difference: 29.0 - 2.0 = 27.0

                                       27.0 > 8.79 → Significant.


2.                   m = 5: Compare the ranges spanning 5 means.

                                                                     

Comparisons:

               Difference: D (29.0) vs. B (9.0): 20.0 > 8.66 → Significant.

               Difference: A (17.0) vs. Control (2.0): 15.0 > 8.66 → Significant.


3.                   m = 4: Compare the ranges spanning 4 means.

                                                        

Comparisons:

                    Difference: D (29.0) vs. C (13.2): 15.8 > 8.48 → Significant.

                    Difference: A (17.0) vs. B (9.0): 8.0 < 8.48 → Not Significant.

*Since A and B do not differ, their subset (A, C, E, B) is homogeneous. Internal comparisons (e.g., C vs. B, E vs. B) are not performed for m=4, m=3, m=2.

                     Difference: E (12.6) vs. Control (2.0): 10.6 > 8.48 → Significant.


4.                   m = 3: Compare the remaining ranges spanning 3 means.

                                                                      

Comparisons:

               Difference: D (29.0) vs. E (12.6): 16.4 > 8.23 → Significant.

               Difference: A (17.0) vs.C(13.2): 3.8<8.23 → Not  Significant. (Consistent with the previous halt).

5.                   m= 2: Compare the remaining intervals for adjacent pairs or pairs of interest.

                                                                     
          Comparisons:

              Difference: D (29.0) vs. A (17.0): 12.0 > 7.83 → Significant.

*Other comparisons within the homogeneous block (A, C, E, B) are not necessary.

Presentation of Results

The standard way to present the results is to assign the same letter to means that do not differ significantly from each other.

Table 6 – Mean comparison by Duncan's test


      Alternatively, means can be presented in order with an underscore connecting those that           do not differ.

                          Control      B          E            C            A          D

                              2.0 d    9.0 c   12.6 bc   13.2 bc   17.0 b   29.0 a

Final Notes

1.If the groups have unequal sizes (different number of replications), n in the formula is replaced by the harmonic mean of the group sizes.

2. DMRT is not widely accepted as Tukey or SNK procedures; it is declared for some statisticians to perform poorly. We will discuss these issues in a next post.

 Further reading

1.   Montgomery DC. Design and Analysis of Experiments. 4ed. New York: Wiley; 1997.

2.    Zar, J.H. 4ed.Upper Saddle River, Prentice Hall.

3. University of York. Tables for Duncan's multiple range tests Critical values https://www.york.ac.uk › maths › tables › duncan.

4.    Critical care. National Library of Medicine. http://www.ncbi.nlm.nih.gov/pmc/articles/PMC420045/



No comments: