Hi everybody. I am facing an issue about reproduction data analysis in chronic toxicity test with C.dubia. US EPA Method 1002.0 ( https://www.epa.gov/sites/production/files/2015-08/documents/short-term-chronic-freshwater-wet-manual_2002.pdf) instructions tell:
" The response used in the statistical analysis is the number of young produced per adult female, which is determined by taking the total number of young produced until either the time of death of the adult or the end of the experiment, whichever comes first. An animal that dies before producing young, if it has not been identified as a male, would be included in the analysis with zero entered as the number of young produced "
Including dead animals in the count gives me back too high standard deviation values which don't fit well in my analysis. I would like to know if alternatives are available: feel free to link me to other papers, works, methods etc.. which could help. Thanks a lot to anyone will spend time to help me.
Exclude animals that died before producing young because they are not relevant to the study
Hi,
Take a look at the guidance in the OECD version of the guideline:
https://www.oecd-ilibrary.org/environment/test-no-211-daphnia-magna-reproduction-test_9789264185203-en
How to consider mortalities was added in detail when it was revised in 2011.
Best,
James
Hi, Mattia, due to outliers it should be supposed that in yours Var list existing two distributions. One (I hope) is "normal" or you tried to transform it into "bell-shaped" (log normal origin?) and other is for dead units that you are speaking about! How even 1 result from different distribution can affect the general investigation is demonstrated in "Anscombe effect": F.J. Anscombe, "Graphs in Statistical Analysis," American Statistician, 27 (February 1973), 17-21.
So, one of the best method to exclude such data is interquartile range or IQR.
Example - Suppose you have a sample of values 3, 1, 5, 3, 6, 7, 2, 9, which variation has the form: 1, 2, 3, 3, 5, 6, 7, 9. Since the numbers 1 and 2 make up 25% of the variation series, the numbers 1, 2, 3, 3, 5 and 6 represent 75 % of the values of the variation range, the 75th percentile for this series has the value 6, and 25th percentile – 2. Then IQR = 6 - 2 = 4. So, Median is 4 as well therefore 4 ± 4 – 9 number should exclude.
Look eg http://eclinpath.com/test-basics/reference-intervals/print-17/
- Badges
- Science topic
- Similar topics
- Cell Biology
- Methods