I'm measuring the responses of forest spring plant communities to drought in central Europe. I have both hydrological (discharge, pH, Temp, Conductivity and anions, cations, DOC, etc...) and biotic (Abundance, Presence/absence) data.
My problem is I have quite a solid dataset that highlights patterns of responses to droughts in 2003 but when it comes to the 2018 data I only have the hydrological data but Biotic data was only collected again in 2020. So there is a 2-year Biotic gap.
The 2020 biotic data does somehow fit my hypothesis but without the response Due to lack of data, it's hard to write about it in any way that doesn't take too many liberties.
How does one go about writing about the inferred response in this data gap in a meaningful way?
It is a common problem that some data are missing during scientific research. You can just develop your model, such as abundance = f(discharge, pH, Temp, Conductivity) + error. This can be a linear model or a non-parametric machine learning model. The problem of missing data means you lose a few rows of your dataset, and the p values (the probability of the variable coefficients being 0) would be larger due to the smaller sample size. Sometime it does not matter at all.
Dear Timothy Smith
Follow these steps:
Best Regards
Dr. Fatemeh Khozaei
Hello Timothy; The two comments above are correct. I can only add that if you remain concerned about the missing records' impact, you need only note that concern in the matherials and methods section of your ms. As Mr. Li said, gaps like that are a common problem. I once lost an interval of environmental data because a land developer bulldozed the site without any warning at all...that was a big surprise! Best regards, Jim Des Lauriers
You can apply data imputing, or missing data imputation. A simple approach is to use statistical methods to estimate a value from those values that are present, then replace all missing values with the calculated statistic.There are also machine learning or random forest algorithms and packages that allow correlations over time and can automatically detect and deal with missing data :)
Article How can I deal with missing data in my study?
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics