Defining model - are my samples too small?


I am attempting a valuation of a non-market good. This non-market good is made up of a number of component parts. I have conducted a meta-analysis of these valuations and am attempting to model these to establish which explanatory variables given about the good explain values given.

The problem is that some studies have attempted valuation of the overall good, without specifying input from components, and others have only valued individual component parts. Some of the components only have 1-3 studies valuing them so I can't split these off and do separate models. I wish to know if anyone knows of a model that can look at both levels of this, or confirmation that with such small samples in some cases, it's infeasable. I successfully ran a multi-level, mixed model on the data, grouping by component type, but a journal found the method too complex and the outcomes uncertain. "n" in the following indicates number of studies found.

Explanatory variables (categorical and numeric) potentially explaining value of:
- Component 1 (n=3) -
- Component 2 (n=10) -
- Component 3 (n=1) - all of which make up Total value
- Component 4 (n=5) - (n=5)
- Component 5 (n=19) -

I hope this makes some sense! Thanks,
Last edited: