As an editor, my bile rises whenever I come across the words "the data were analyzed by computer." That assertion gives me the uncomfortable feeling that the author questions his own data and hides behind the computer to still his doubts and, presumably, those of the reader. Other authors turn to the computer to clothe results in pseudoscientific respectability— often going to ludicrous ends such as performing multivariate analysis of responses to a questionnaire with a 40% reply rate or calculating percentages to four decimal places, to cite two recent examples.
There is an old (as computers go) aphorism "garbage in, garbage out." The meaning is evident: if your data base is shaky, no amount of fancy computation will shore up the end product. Computer analysis merely eases and speeds the drudgery of computation; it does nothing for ordinary statistics that you cannot do with an inexpensive electronic calculator or