To answer your second question first, yes it is. Sort of. The UCM answer is, in its raw form a "sizing" of the entire delivery of the use case. Thus it includes not only development, but also unit, system and acceptance level testing. However, the TCFs and ECF's that are used in the raw figures are based on some orignal (gu)estimates done some time in the latter quarter of the last century. Despite some continual effort I have not been able to find any published confirmation of those original weightings nor any refutation.
Having said that, let me now say that, as I work primarily on the testing end of this game I have developed a personal set of TCFs and ECFs and weightings that apply to those factors that have stood me in good stead regarding test estimations for the last decade or so. But, they do not estimate development effort, and I dont pretend they ever will.
So, you need to understand the context of the UCM estimation technique and more importantly how to apply it to your domain within your domain's estimation experience.
As for your first question, in fact "you dont". Ask why is unit testing causing yourself such a big worry when estimating a use case size. More to the point, you cannot take a UCM sizing and apportion detailed breakdowns to it. At best, you can get plus/minus 50% budgetary allocations at the gross levels. IOW if the entire project delivery is 6 small, 20 medium and 4 big use cases, which given your TCF/ECF mix is say 320 person days, then that should be the TOTAL effort involved in delivery - from spec to acceptance. If your answer is 32 days (i.e a raw average of about 10 days per use case) then I'd tend to suggest that the factors and costings you are using are based more on hope than experience.
When I was a young man, just coding a three level single table flat report in COBOL, took three days, 5 flints and 2 mammoth tusks to punch the cards, get overnight compiles and debug ( we were pro's then we never tested nuffink - if it compiled it must be right!) Now that I'm an old tester, my expectation for the same three level single table flat report in <insert-your-favourite-language-of-the-year-here> is about 5 days. That is, about 15 minutes for the BA to assume that "everyone obviously knows" what a "report of internet loads for budgerigar porpoises" is; 45 minutes for a recent graduate of what universities laughably call their information engineering departments to produce some sort of illegible and oftimes erroneous garbage using a reprot* {sic} generator and about 4 days and 7 hours effort on behalf of the test team and the users to convince the developer that what they needed was a "report of interbank loans for budgetry purposes".
I sincerely hope this helps
bruce
* I kid you not - this was on a trade show flier for a <cough, cough> Cr**t*l Reports.
