So I have several plant extracts and I want to measure their effect at different concentrations on cell viability.
Each experiment was repeated 3 times so I used ANOVA with each concentration level as a "group" eg. to compare between 800ug/ml and 400ug/ml effect then I used Duncan's post hoc test to test for differences between concentrations effects.
Is this ok or should've I used another post hoc test? or a different statistical test altogether?
Which post hoc test is more appropriate in the following case? As has already been said, you have many options. The decision is yours, taking into account that some post hoc tests are more conservative than others.
Dear Khaled Tarek ,
If you repeated your experiment several times you would usually treat this as the random sample. Then your two concentration levels could give you two groups. This would lead to a t-test or a Mann-Whitney U test. But maybe I don't understand the design of your experiment. In any case your sample size sounds too small. Please have a look at my study guide: Technical Report One-Way Analysis of Variance (ANOVA)
.
Thanks for the answer I will have a look at your guide.
I have a range of concentrations from 12.5ug/ml all the way up to 1000ug/ml.
Each experiment were done using all these concentrations, that gives me 3x 8 results.
So I have the cell viability results for each concentration 3 different times.
My goal is to find which concentrations differ significantly in effect than others. Hope that makes it clear.
@Khaled Tarek In this case, you should an ANOVA follow Bonferroni compare all pairs of columns or Tukey compare all pairs of columns.
Khaled Tarek , it sounds like you have conducted a one-way anova. That is you have one factor (concentration) with 8 levels, and three observations for each.
In this case, you can use one of many different post-hoc tests. Some of these are Tukey, Tukey-Kramer, Dunnett, Games-Howell.
Which would you choose? The best approach is to understand the assumptions, advantages, and disadvantages of each, and choose the one that suits your study.
But sometimes people choose the one that is common in their field.
Given the nature of your IV, I would add polynomial trend analysis to Salvatore S. Mangiafico's list of possible approaches.
Mr Khaled, it sounds like you have many plants extracts and many concentrations (two factors). If it is the case, do not limit analysis in a side.
Look for all extracts combined, the effect of concentrations using Tukey as post hoc. Duncan is too sensitive in this case (cell viability). If you want to compare just the concentrations 800ug/ml and 400ug/ml, then use t test.
Thereafter, do same analysis within each extract. This, to see how viability varies within each extract.
To compare the extracts, make a regression (Finney, 1971) to look for effective dosages for each extract, and compare them.
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Modeling
- Analyses of Variance
- ANOVA