Kavisa

It is Tukey not Turkey

The Tukey and Duncan's Multiple Range Tests are both pairwise comparison tests which are undertaken after ANOVA has an outcome which Fails to Accept the null hypothesis (also known as post hoc tests). There are a number of these types of tests, where the probability of making a type I error can be ordered as follows: Tukey HSD Test; Student-Newman-Keuls Test; Duncan Multiple Range Test; Fisher LSD Test

The test procedure for Tukey assumes (1) equal number of observation per populations. (2) Finds a critical number w

The calculation divides the mean squared error within from the ANOVA analysis by the total number of data points for a given group. The square root of the resulting value is obtained. Finally, this result is multiplied by the studentized range statistic (can look up this statistic in a table provided by many textbooks). This result is the Honest Significant Difference, and it represents the minimum distance between two group means that must exist before the difference between the two groups is to be considered statistically significant.

The Duncan Multiple Range test uses different Significant Difference values for means next to each other along the real number line, and those with 1, 2, … , etc. means in between the two means being compared. The range value is 2 if the two means being compared are adjacent; 3 if one mean separates the two means being compared; 4 if two means separate the two means being compared 

Tukey’s is the best for all-possible pairwise comparisons when sample sizes are

unequal or confidence intervals are needed; very good even with equal

samples sizes without confidence intervals. Tukey’s test does not operate on the principle of controlling Type I error.

Duncan's multiple range test, provides significance levels for the difference between any pair of means, regardless of whether a significant F resulted from an initial analysis of variance.

Refer in the following link.

http://ykhoa.net/baigiang/Topic07.%20Multiple%20comparisons.pdf

There are a number of specialist multiple comparison tests that maintain a low overall type I error. Tukey's test and Duncan's multiple-range test are two of the procedures that can be used and are found in most statistical packages. However, each one of these tests has their own specifications.

Tukey’s test calculates one critical value that can be used to evaluate whether differences between any two pairs of means are significant.

This test is designed for a situation with equal sample sizes per group, but can be adapted to unequal sample sizes as well. In addition, you must consider the following assumptions of Tukey's test before you use it:

1)The observations being tested are independent within and among the groups.

2) The groups associated with each mean in the test are normally distributed.

3) There is equal within-group variance across the groups associated with each mean in the test (homogeneity of variance).

In contrast, Duncan's multiple range test provides significance levels for the difference between any pair of means, regardless of whether a significant F resulted from an initial analysis of variance. 

Duncan's method is not recommended when you have unequal sample sizes. Instead,  several published studies (for example, Carmer and Swanson 1973) have claimed that Duncan's method is superior to Tukey's because of greater power without considering that the greater power of Duncan's method is due to its higher type 1 error rate (Einot and Gabriel 1975).

Kavisa

It is Tukey not Turkey

The Tukey and Duncan's Multiple Range Tests are both pairwise comparison tests which are undertaken after ANOVA has an outcome which Fails to Accept the null hypothesis (also known as post hoc tests). There are a number of these types of tests, where the probability of making a type I error can be ordered as follows: Tukey HSD Test; Student-Newman-Keuls Test; Duncan Multiple Range Test; Fisher LSD Test

The test procedure for Tukey assumes (1) equal number of observation per populations. (2) Finds a critical number w

The calculation divides the mean squared error within from the ANOVA analysis by the total number of data points for a given group. The square root of the resulting value is obtained. Finally, this result is multiplied by the studentized range statistic (can look up this statistic in a table provided by many textbooks). This result is the Honest Significant Difference, and it represents the minimum distance between two group means that must exist before the difference between the two groups is to be considered statistically significant.

The Duncan Multiple Range test uses different Significant Difference values for means next to each other along the real number line, and those with 1, 2, … , etc. means in between the two means being compared. The range value is 2 if the two means being compared are adjacent; 3 if one mean separates the two means being compared; 4 if two means separate the two means being compared 

Thank you Dr. Robert James McClelland  for your explanation and also for correcting my mistake.

Sincerely

Kavisa Ghosh

From the experimental design there seem to be several unrelated hypotheses that make sense (to me). It does not look like a screening. So I wonder why the control of the FWER would make any sense here at all?

 Dear Dr. Jochen Wilhelm,

I have done animal based screening of my treatment compound. Thus, the sample size in each group is less. I am comparing the conditions/parameters reflected in  Group 2 (Toxic treated) with Group 3 (Toxic + treatment compound treated). The standard drug treatment (Group 5) used, is to compare the efficiency of my treatment compound (Group 3) with that of the standard drug. The safety of the treatment compound is studied by administrating only the treatment compound in Group 4. The Group 4 is compared to the conditions/parameters reflected in control animals.  I have used DMRT for one way ANOVA for my statistical analysis. I would like to know whether my choice of using DMRT, rather than Tukey's test is right or wrong? Can Tukey's test be used in this study? Which is better, DMRT or Tukey's test as post hoc test for one way ANOVA for my above stated work? Kindly help.

To my very personal opinion(!) among the two you listed the one better suited for your question is DMRT. But to my opinion both of them are not yet ideal. In your case someting like Fisher's LSD would be better, because you can/should not see the comparisons you can do with your data as a "family of tests". There are several more or less independent questions, possibly some of the questions arising sequentially (and not in parallel). So I don't see much value in controlling some kind of family-wise error rates, because I don't see a logical family of tests. Eventually I am not a big fan of significance tests in general, because a sensible interpretation of the results (i.e.: what p-values are considered to indicate a "significant" effect?) will still depend on the estimated effect sizes (the standard practice of automatically rejecting H0 when p

 Turkey's test was discovered by Dr. Turkey but DMRT is Dongon Multiple Range Test in Oneway ANOVA. Miss. Kavisa Ghosh ! DMRT Oneway ANOVA use for your research papers. 

                                   Thank you

Barani, it this "Turkey's test" you are referring to?

;)

http://ichef.bbci.co.uk/images/ic/448x252/p0282ckd.jpg

Dear Dr. Jochen Wilhelm,

Thank you for your explanation. Your opinion is a brainstorming exercise for me. I had never thought from this angle.

Sincerely

Kavisa Ghosh