While I value everyone's opinion here, I'm going to go out and dissent a little. Even if you disagree with my position, I hope that you'll at least consider it.
A location should never be entered as a wild guess. Or any other data field on a profile.
While it may help some people find a record if it is correct, it will exclude a record if it is not correct.
Marking a location purely because someone lived in one country is not genealogy and is harmful to later research. There is an example above that if someone is married in Connecticut and died there we should assume they were born there and mark it as born in the USA. What if the person was born in England and migrated? If someone is looking for a record of this same name born in England, married and died in Connecticut, having a "born in USA" may exclude this profile from their search.
Aside from a wild guess, it is appropriate to make an educated hypothesis - if you have additional evidence - and then put the location that you can reasonably assume was correct. And then of course, mark it uncertain and annotate it in the bio.
For example, in the above Connecticut example, if the mother and father were born in the USA, then for sure, it is most likely that the person was also born in the USA. I'd enter USA and mark it uncertain. Or, if an older sibling was born in the USA this would be good evidence that this person was also born in the USA because people rarely migrate, have a kid, migrate back, have a kid, and then return to the destination country.
And this is very important and most people don't take the time: annotate it in the biography. Something like, "Joe was born in 1884. While no birth record has been found, his parents were born Connecticut, Joe was married and later died there so it could be assumed that he was born in Connecticut."
If the location was more likely than not, I'd record it. But just hazarding a guess to put something in the box is harmful to genealogy and fosters the climate of fantasy genealogy that is so prevalent at Ancestry.com and other unsourced tree sites.
The problem compounds when the next researcher then makes a guess in another part of the profile, and the next makes yet another guess. In time (over 5, 10, 20 years), you have a complete fantasy file that will then be copied and pasted across the net and then GEDCOM uploaded back as further validation of the erroneous entry (or entries). And this is because everyone assumes that the last person has a record or heard a story from Grandma Milly. When you put in a wild guess, it weakens the strength of the historical record for the profile and the whole family. Remember the game telephone when you were a kid? One kid says a sentence and whispers to another and the sentence goes around the room an when it returns it is wildly different. Genealogy works the same way, just slower. Today's guess, plus a guess in 10 years, plus a guess in 20 years = a completely different profile.
Bottom line: don't enter wild guesses. Only enter reasonable estimates that can be justified with logic and reason. And always annotate the details in the biography.
Edit: spelling error/typo