Can data that includes less than values (below the limit of detection in chemical analysis) be used in ANOVA if the data is re-coded in some way? Perhaps using the actual detection limit or half of the detection limit is possible?
No. Reason is that Below limit of detection (BDL) or as per ICH Guidelines (LOD)Limit of Detection define as "The detection limit is the lowest amount of an analyte in a sample which can be detected but not necessarily quantitated as an exact value"
Where as Quantitation Limit (LOQ) define as "The quantitation limit is the lowest amount of analyte in a sample which can be quantitatively determined with suitable precision and accuracy" also denote as BQL and can be taken.
Hopefully answered your question, feel free to ask if question any.
I have seen many discussions about this topic, there's no real truth. Purists will tell you you cannot include those samples. However, from an application point of view, I would say you can. But my advice is to put yourself in the worst case scenario, so the conclusions are robust.
Let me explain: if the compound you are analysing is of interest, for example, antioxidants in food, and they are below LOD, I would put 0 instead, as it's the worst case scenario (this is, there's no antioxidants in your food). However, if it's cyanide in food, I would substitute values with the LOD, as it is the worst case scenario (there might be some cyanide in my food, even if it's low concentrations). In this way you will get the worst result possible, and cover your back.
However, remember that if you are in a continuous scale of concentrations, you also have the LOQ, but I'd still go with LOD in my example.