Granted, indexes are invaluable tools for accessing original information; however, they are subject to error and can be deceiving. In the process of transcribing information from the original source, mistakes are made. Many early records were handwritten and can be very hard to read, even for experienced transcribers, and sometimes assumptions are made that may or may not be correct. In addition, spelling was not an issue in earlier times, so even where information has been transcribed correctly, varied spelling and regional pronunciations that influenced spelling may create another issue for researchers. Just the act of transcribing can bring about error. As information is copied from one source to another, errors happen; numbers, for example, can easily be transposed and slip by unnoticed, which could affect dates.
Recognizing this margin of error, organizations indexing today often have more than one person indexing a record with conflicts resolved by a moderator reviewing the information and making a decision. This process, however, takes a lot of time and, unless the indexers are volunteers, it can be expensive -- some well meaning individuals and organizations may not utilize the process. It is then left to the individual researcher to sort it all out and find those family members "lost" in the index.
Among the most commonly used indexes, of course, are census indexes, but all indexes and compilations are subject to transcriber error.
Here is a case in point. Searching the 1910 census for Mose Henry Goodman and his wife Manerva Cole, it was suspected that the couple married in Texas, but which county was not known. The index was searched using all variations of Mose's name, including initials, but yielded no results. Neither did a search for Manerva Goodman or "Manervy," as she was known by family. The search was then abandoned. Later, the couple was found quite by accident when conducting research aimed at identifying Manerva's parents. Although the family of Ellis G. Cole had not yet been confirmed as Manerva's family, in reviewing the original image, the researcher noticed a strangely familiar name below the Cole household and, sure enough, it was Mose and Manerva Gooman. Going back to see what names were listed as neighbors, it was found the the Goodman entry had been misinterpreted as Morie H Gashttin and Miaranvey Gashttin.
There is no way under the sun this couple could have been found through the index, but because the researcher was looking at the original image, being familiar with the correct name, she was able to discern this to be the Goodman couple. Of course, finding this entry adjacent to the Ellis G. Cole family, then confirmed it to be Manerva's family. The moral here is simply, there is no substitute for the original, and it sometimes takes a bit of luck or serendipity to see beyond the index.
Often it's just one name that becomes convoluted: Morris instead of Marion; Dever instead of Daves; Lola instead of Lula. First name errors can typically be overcome by surname searches. Surname errors may require narrowing the search with known information such as a birth date and place, death date and place, or the name of a spouse or parent. Wildcard searches may or may not work when researching names. A wildcard search uses the question mark ? or asterisk * as a substitute for one or more of the letters in a word, but because the error in a name might be much more than a case of a missing letter, as when searching a regular word or term. Also, one cannot do a wildcard search on the first letter of a name; typically, at least three letters are required before the wildcard. So the best help is probably supportive information and, where possible, researching the names of others in the family, especially those with less common names . . . maybe.
Now, uncommon names are usually a benefit in family history research, as less common names help researchers differentiate families and individuals. But uncommon names can be an issue for transcribers, particularly if the script is archaic or difficult to read. For example, an ancestor with the first name Quinnie, the researcher felt confident such an usual first name would be easy to find, but it was not so. Although Quinnie was, indeed, enumerated in the census, she was not found by that name in any census index, and other methods were employed to find her. As it turns out, in the 1900 census index Quinnie was found listed as "Iuinie" or "Ininie." In 1910 she was listed as "Jimmie" or "Guinnie." She has yet to be found in the 1920 census, but in the 1930 index she was listed as "Aninnie." Whoever transcribed these records missed the "Q" in the name entirely, and we can be fairly certain it was more than one person.
And yet, when the original image for each of these census records was reviewed by the family researcher, the name was easily identifiable, which brings up two important points. First, transcribers work within their own frame of reference; if the transcriber were not familiar with the name "Quninnie, interpreting the first letter of the name as a "Q" might not even come to mind. Second, there is no substitute for examining the original document. Transcribers do not have the researchers familiarity, and there may be other clues in the record that confirm an entry to be or not to be the right family or individual. Thus, it is important to investigate any conflicts of information; a date that does not agree may well be transcriber error.
Index error is a significant problem, but awareness is the first line of defense, keeping it may take a bit of ingenuity and downright luck to filter it out.