"I recently for instance had a male who was described as widowed and a few lines down the page was his dead wife in a different household after also having been widowed."
Sorry to keep this going, but many people "tell little white lies" when it comes,for example, to the census. In many many cass when an older male marries a much younger female the first few census recordings show their ages not that far apart, but as they get older the later cenuses show their ages getting further and further apart. I guess they initially didn't want to let the census taker believe the elder male was "robbing the cradle", but as they got older they didn't seem to care what the census taker thought
One has to be quite careful[as your example also shows]in accepting the data shown on a census as correct.
Another example of this "LWL' is when one of the spouses has been married more than once and yet all the children [from both marriages] are listed in the census under the same surname, when in fact they are of different surnames and haven't been adopted------keeping from being embarrased [sp?]seems to be very important, but causes problems for the genealogist