Hi everybody. I'm comparing the phenolic content in four different anatomical parts of a seaweed. I made one extraction per part and then I quantified in triplicate. In this case, is it still ok to perform a one way ANOVA analysis (obviously after to check normality and homocedasticity)?
Dear José Avila-Peltroche,
Please excuse me for not agreeing with the previous answers to your question. The correct would be NOT to perform an ANOVA, since you quantified thrice, but extracted only one seaweed part. Therefore your phenolic quantifications are not true replicates of algal parts, you have only one replicate, although measured 3 times or more. Each quantification allows you extra confidence on the values, but it does not represent a true assessment of phenolic compounds in three seaweed individuals, or even of three parts from the same individual. You could run one if you could extract three pieces from the same thallus part of three or more seaweeds. Even then, you would need to first check for assumptions such as normal distribution and, if achieved, variance homogeneity. If the assumptions were not met you could then run a non-parametric analysis such as Kruskal-Wallis. The more true replicates you have, the more likely it will be that you achieve enough statistical power to demonstrate a difference between different parts of the thalli, in case there is a difference, i.e., you will increase statistical power and decrease the type II error rate (= accept a false null hypothesis).
It will be nicer if you can do at least two times extractions ( 2 x 3), then you can evaluate with ANOVA
Many thanks! But in case I have all the data in that way it is still possible to do a one way ANOVA analysis?
You can go for one way ANOVA, but it will asign you only one variable difference wich will remain doubtful, for two atelast 6 replicates are be needed for good analysis.
Dear José Avila-Peltroche,
Please excuse me for not agreeing with the previous answers to your question. The correct would be NOT to perform an ANOVA, since you quantified thrice, but extracted only one seaweed part. Therefore your phenolic quantifications are not true replicates of algal parts, you have only one replicate, although measured 3 times or more. Each quantification allows you extra confidence on the values, but it does not represent a true assessment of phenolic compounds in three seaweed individuals, or even of three parts from the same individual. You could run one if you could extract three pieces from the same thallus part of three or more seaweeds. Even then, you would need to first check for assumptions such as normal distribution and, if achieved, variance homogeneity. If the assumptions were not met you could then run a non-parametric analysis such as Kruskal-Wallis. The more true replicates you have, the more likely it will be that you achieve enough statistical power to demonstrate a difference between different parts of the thalli, in case there is a difference, i.e., you will increase statistical power and decrease the type II error rate (= accept a false null hypothesis).
Agree with bernardo, you have a problem called confusion (or pseudoreplication)... there is no way to evaluate your original hypothesis...
Yes I agree with previous writers, if you can do at least 2 extractions more (as I described before), and from each extraction you 3-6 analysis, so you will have at least totally 3 extractions (3 x 3-6), than it will be nice. For preliminary work, I think it is OK at first, you analyze your data from (1) extraction only.
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Modeling
- Analyses of Variance
- ANOVA