Hi，
I am analyzing my experiment data, but I have some problems.
My experiment has three replications, which are microbial data from three different filed samples. These fields are under the same treatment.
The problem is the microbial results of 3 replication are very different. What should I do? Just average？It doesn't seem right.
If you have any good suggestions, please come up with them.
Thank you all.
Dera Mengying Ni ,
The information, supplied by you, are not sufficient for prescribing suggestion for statistical analysis of the experimental data you have mentioned.
You have only mentioned that there are 3 replications in your experiment.
However, you have not mentioned about what are those which have been replicated.
It is required to be known about the treatments or factors or treatment combinations involved in your experiment.
Only then it will be possible to identify the type of the experiment you performed.
Therefore, provide the detailed information on the experiment you performed along with the data you have.\
As far as I can guess, the experiment conducted by you may be a Randomized Block Design. However, I will be able to be confirmed only after getting the detailed information on the experiment.
Calculate the std. Deviation of your data based on the three replicates, repeat your experiment with a fresh set of three replicates, if for the new set you compute a very different, much lower std. Deviation, it could be that some systematic error, e.g contamination problems as mentioned by prof. Booth, could be the cause.
However, if you keep getting comparable, but large std. Deviation for separate sets of three replicates it imply that your experiment generating the data has a large error variance, probably due to some significant process variable is left uncontrolled and is fluctuating in its levels during the experiment. In that case, to reduce the process variation you must identify such process factors and ensure they remain at constant factor levels throughout the experiment process.
Standard deviation is a tools that you can check your data error. In your data three replicates, so repeat experiment to finite times, and calculate too low standard deviation, then does not matter. Otherwise, system error. Also check through graphical view for miss reporting data.........................
What ever you do DO NOT combine suspect data with non-suspect data. THERE are NO statistical procedures that make bad data good. David Booth PS why not just repeat the experiment?
Sound like you need to do a sample size calculation. Due to high variability, you're triplicates are insufficient sample size per treatment. I'm guessing you'll need a sample size of like 30 per treatment if they are really that variable. Please explain the experimental design better and what you are interested in studying.
Dhritikesh Chakrabarty
Thank you for your answer! My sampling in each sample site is random. I used these samples for DNA extraction and then calculated the diversity index. And a bit big differences in the indicators of diversity between replications.
- Badges
- Science topic