I am currently working with a functional database but there are 3 traits (1 categorical, 1 continuous and 1 ordinal) in which there is a lack of data.
If a just include the individuals that have all the traits complete then I have a lost of 31% of my total individuals.
But if I don´t remove the individuals with NAs values I can't run any analysis.
What do you suggest?
I'm currently working with functional traits for amphibian species, but the principals are the same. Can I ask, what are you trying to determine by using the traits and what are the three traits you are measuring?
I have the same question that Michael. Also, are the 3 missing values of the same species? or are traits of different species? Are you trying to calculate a specific index?maybe if you specify a little more we can help you find an index or a program which accepts NAs
Depending on the question you are trying to answer, and on the type of analysis you intend to run with these data, there are a few ways to circumvent the issue.
-First, is there any way you could perform analyses dealing with NAs? This would without doubt be the best way out.
-If not possible, would you consider using trait information generalized at the genus- or family- level ? You can sometimes find these in the literature.
-If using taxon-generalized trait values might induce spurious results, or if introducing literature-based data in a data set including field or lab observations only might add a bias to your analyses, you can also infer trait values based on other types of information, e.g. phylogeny, body size. If you have access to phylogenetic data (which can be available online, through GenBank for instance), there are a few recent methods that offer a solution to tackle such issue. Of course, carefully choose your method based on the type of trait(s) you wish to infer.
- Badges
- Science topic
- Similar topics
- Mathematics