Do **not** mess with the data!!!

Found this by accident, while looking for some U.S. Census 2000 stats on family structure in the U.S.


From the Wikipedia article on the 2000 U.S. Census:

Another issue that concerned gay rights advocates involved the automatic changing of data during the tabulation process. This automatic software data compiling method, called allocation, was designed to counteract mistakes and discrepancies in returned questionnaires. Forms that were filled out by two same-sex persons who checked the "Husband/wife" relationship box were treated by the Census computers as a discrepancy. The Census Bureau explained that same-sex "Husband/wife" data samples were changed to "unmarried partner" by computer processing methods in 99% of the cases. In the remaining 1%, computer systems used one of two possibilities: a) one of the two listed sexes was changed, making the partnership appear heterosexual, or b) if the two partners were more than 15 years apart in age, they might have been reassigned into a familial parent/child relationship.[12] The process of automatic reassignment of same-sex marriage data was initiated so that the Census Bureau would not contravene the Defense of Marriage Act passed in 1996. The Act states:

In determining the meaning of any Act of Congress, or of any ruling, regulation or interpretation of the various administrative bureaus and agencies of the United States, the word 'marriage' means only a legal union between one man and one woman as husband and wife, and the word 'spouse' refers only to a person of the opposite sex who is a husband or wife.

With allocation moving married same-sex couples to the unmarried partner category, statisticians lost any data that could have been extracted relating to the social stability of a same gender couple who identify themselves as married.


Gaahhhhh!!!! To change the raw data from a survey is such a cardinal sin in the world of social science research!!! If someone writes down ''Jedi'' for ''Religion'' -- gosh darn it, you record ''Jedi''. If, for the ''Race/Ethnicity'' question, someone writes down ''I'm a moose'' -- gosh darn it, you record ''Moose'' as their race!

Sure, you can clean it up in later stages -- but to nuke the original information such that it's **wrong**...??? GAHHHHHH!!!!!!!!!!!!!!


You're supposed to keep the original, ''real'' data -- and if you're worried about the same-sex ''married'' couples messing up your results... you just do the runs for ''Married'' couples, but throw in a ''IF'' statement that excludes all instances where the two partners are of the same sex. Easy.

And then, in the percentage table (when you write up your results), you include a footnote stating that "Same-sex couples who self-identified as 'Married' were excluded from this analysis.'' Ayep.


(I know why they had to do it: political/funding reasons. But it still stinks.)


--GG