I have done Radial arm test for rats of 5 groups, with 10 animals in each group. I am trying to analyse the results through graph pad prism. On what basis we have to go for selection of either Bonferroni or Turkey test?
use Bonferroni
you can use Newman-Keuls test also
both theabove tests generally used for multiple comparisons than Tukey
First of all, what is your measurement? Time taken to complete the maze? It is helpful to first get a feel of what your data looks like. Most researchers ignore the most important first step in statistical analysis, descriptive statistics, and instead opt for going straight into inference on assumptions which may not hold i.e that your data is normal. I highly recommend that you learn to use R. You should plot out your data and see what it looks like: is it skewed, are there outliers, what is the mean and standard deviation? I presume you wish to conduct ANOVA? Remember that the tests rely on an important set of assumptions which you should test (easily in R), otherwise your results will be invalid. I recommed that you find a good textbook or document that will guide you through the process. Don't rush in a bid to get "good" results. Statistical analysis takes time and thought, and you will be better off as a researcher for knowing how to do it properly.
Dear Nehru,
I can't really see why this is a 2-way ANOVA as you have mentioned having 5 groups but no other factor. Angela's advice is really important.
If in fact you have a one-way ANOVA and your group sizes are equal and the ANOVA assumptions (or their robust exceptions) are met and you get a significant result then Field recommends the following post hoc tests:
- Tukey, REGWQ or Bonferroni (more conservative) if your variances can be assumed to be equal (based on your homogeneity test)
- Games Howell if they can't (and as a back-up)
@Peter
For all the groups (different treatments) i have taken readings at 4 different intervals.
for example - each group: 10animals
Group1: Reading(1) Reading(2) Reading(3) Reading(4) -
Group2: Reading(1) Reading(2) Reading(3) Reading(4)
Group3: Reading(1) Reading(2) Reading(3) Reading(4)
Group4: Reading(1) Reading(2) Reading(3) Reading(4)
Group5 :Reading(1) Reading(2) Reading(3) Reading(4)
As i need to compare efficacy in terms of at different intervals between the groups, i have considered two way ANOVA. Is my way of approach is correct?
Hello Neshru,
a agree with taking the Bonferroni, because in literature, it seems to be the best method.
Dear readers,
please don´t forget that we are speaking about living creatures. Probably they died for this experiment. I hope, that the result are outstanding, otherwise all the creatures sufferes and died for nothing.
Best regards
Tanja
@ Tanja - Thank you for suggestion.
Dear all,
From the literature me too understand that Usage of Bonferroni is more.
Just i would like to know, what is rationale is selecting it rather than Turkey or any other, as i am not with stats back ground. I tried to understand selection criteria through various methods but i couldnt.
Please note than John Tukey is the correct spelling. I suspect he would turn in his grave knowing that he is referred to as TuRkey, as they are regularly eaten over Thanksgiving!
@ Angela Kirykowicz could you please suggest me how could I check the distribution of data is good or not. Sorry I am very bad in statistics.
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Graphs