In case your answer is "yes," how would you represent them (in terms of weight) in the estimates?
The hypothetical scenario is that I have 3 biological setups designed to test the same hypothesis (say, 3 cancer cell cultures of the same line being exposed to equal doses of a candidate agent). The results will be measured spectrophotometrically, and I decide to run each setup 5 times in the machine to get absorbance readings (which will be the technical replicates for each setup).
I have seen studies that only consider the biological replicates (making the effective n = 3), but a concern here is choosing 1 of the 5 readings as estimate for each 1 of the 3 setups. Other studies, on the other hand, consider both the biological and technical replicates in reporting variability (making the effective n = 15), but at least one issue here is "weighing" those estimates.
If you were facing such a scenario, how will you go around these issues? Thanks!
If you have a spectrophotometer where you need to average 5 reads of the same sample to get a stable reading, then it's time to buy a new spectrophotometer. You should test the reliability of your instrument now and then, and make sure they're all in top condition. For common lab instruments you don't need to report it. Your readers can assume that you do that and know how to use it.
There are measurements where technical variability should be reported (but still often isn't). For example if you have something like a hybridization microarray, or qPCR, or animal life span study, it can be really hard to get seemingly the same experiment give the same result each time. With those techniques you'd have a section of the paper characterizing technical variability. That should be separate from your biological questions.
In your example, reading the same sample 5 times might not be the best way for technical repetition. I think for each cell culture of the three ones, you need to have 5 or three repeats, then you would read those repeats and take the average. This would be considered as one biological repeat. You would then do the same for the other two biological repeats.
Also, in case of possible high variability between biological repeats you can normalize the data (as fold change or percentage, for example).
- Badges
- Science topic
- Similar topics
- Clinical Pharmacology
- Dosing