Query on usage research and logfiles

[Armbruster, Chris]

Some promising usage research based on logfiles of publishers, repositories etc. has been going on. 

I have two queries in respect of comparative usage research:
- If I want to compare usage across two or more repositories (or publishers’ sites), how important is the uniformity of logfile formats or production? If logfiles are configured differently, how difficult is it to achieve comparable data?
- If I want to compare usage for a specific item across repositories and publishers’ sites (i.e. a pre-print versus the published version): What might be suitable procedures for doing so?

And a more general query: 
Could specific logfile configurations make it problematic to understand the source and nature of usage?

Thanks.

Chris Armbruster