What does Duncan's multiple range test do?
DMRT is a post-hoc multiple-comparison procedure used to compare ordered group means. Statloft reports the mean and sample size for each treatment, compact-letter groupings, and pairwise comparison details. Treatments that share at least one letter are not identified as significantly different by the selected DMRT procedure.
DMRT is relatively liberal compared with methods such as Tukey HSD, so it can report more pairwise differences. Read the assumptions and consider your study plan and discipline's guidance before choosing a post-hoc test.
Prepare your data for a DMRT analysis
Provide a numeric outcome and a categorical treatment or group column. The analysis needs at least three treatment groups and at least two numeric replicate observations per group. Rows with missing or nonnumeric values in selected columns are excluded and reported.
If each replicate contains several plant measurements, first summarize those subsamples at the experimental-unit level when appropriate for your design. Counting subsamples as independent replicates can overstate the amount of independent information.
Choose the right data-set scope
When your workbook has a data-set column, Statloft can run DMRT separately within each set, analyse one selected set, pool treatment observations across sets, or compare the data sets themselves. The default is to analyse each data set separately. Comparing data sets assumes independent sets; if the same plots or subjects were measured repeatedly, use a repeated-measures or mixed-effects method instead.
Assumptions and interpretation
- Observations should be independent at the experimental-unit level.
- Residuals should be approximately normal with similar within-group variances.
- There must be non-zero within-group variation to estimate experimental error.
- Review the omnibus ANOVA result; if it is not significant, Statloft labels the post-hoc comparisons exploratory.
Blocks, repeated measures, and other dependent observations need a model that accounts for how the experiment was conducted. Statloft's current DMRT assumes independent observations.
Run a free DMRT analysis
- Upload your CSV or Excel workbook in the Statloft workspace.
- Confirm the treatment column and the numeric replicate mean or outcome.
- Select Duncan's multiple range test and set the significance level.
- Choose a data-set scope, run the test, and review group letters and assumptions.