On Thursday, May 26, 2011 5:09:09 AM UTC-4, Steven Gibbs wrote: > "Bob LeChevalier" <[email protected]> wrote in message > news:[email protected] > > > > That is the difference in my approach. I generally don't add someone > > to my data base unless I have connected them to at least one other > > person in my data base. Unlinked individuals are better dealt with in > > a flat table (spreadsheet) than in a relational data base. > > How do you do that when the data in the record is inadequate to provide > linkage? I used to keep my parish register extractions sorted in text > files, but it became impossible to find things once the files became > significantly large. > > Imaginr that you have the will of a John Smith which names his sons as > William and Thomas. Imagine also that you have a marriage certificate for a > Thomas Smith that names his father as John Smith. Clearly on the evidence > I've presented they may or may not be the same people. Can you search your > text files easily to find all candidates for the Thomas Smith who married, > subject to the constraint that his father is called John? If, not having > looked at the family for a few years, you later come across a document which > confirms that Thomas has a brother William, or a document which suggests > that Thomas has no brothers, can you rearrange your thought processes to > take this into account? > > I certainly couldn't. I was going through old bits of paper thinking "I'm > sure I decided that these two definitely weren't the same bit I haven't a > clue why I thought that at the time". It was only when I started to build > up possible persons from the available data that the more obscure > coincidences or contadictictions started to jump out at me. So often > previously I would build up a person, realise that this person wasn't > relevant to my research and discard it. Then I'd do all the same work again > a few years later. Now that person stays, fully formed, in my database, and > if I can link an otherwise vague document to that person, I can tell, within > a trivial amount of time, what the relevance of that document to me is. > > Steven Steven, That's the strongest argument for the persona method as the way to store evidence that I think I've seen, and you didn't use the word even once! It also sounds like you have solved the merge problem by just not doing any. So I surmise that when you decide that one of your personas ("fully formed" in your terms) is one of your persons of interest you add info from the persona to the person. So when you decide you were wrong you always have the original persona around to help you back out your change to the person record. I think this qualifies as another good approach to answering this thread's question. Tom