Data Integrity

In the preface to the Georgia 1850 Census Index, the editors summarize twelve important factors to keep in mind when considering the integrity of census data. They discuss why family names may not appear on certain state schedules as well as the accuracy of the information collected. The conditions under which data was collected vary from one census to the next and can range from the competency of the census enumerator to the archival conditions. The following list makes a social historian stop and think about the impact that the availability of government records has on recording social history.

Unfortunately, census records are not perfect; to our knowledge, there does not exist today any census record that can be considered complete. Occasionally, people know of an ancestor who lived in a particular place, but they cannot find his name in the census records. To help understand why a name may not be found, consider the following facts:

1.  All of the original census manuscript of a particular state may be missing.  
    Examples:  Delaware 1790, Washington D.C. 1810, Georgia 1790.
2.  Parts of a census record may be missing for a particular state or county.  
    For example:  Tennessee 1810, 1820; North Carolina 1790; Ohio 1800, 1810, 
    and Georgia 1800.
3.  Sometimes the census taker missed names through neglect, laziness, or 
    drunkenness.  Example:  Baltimore Co., Maryland, 1800.
4.  Names were omitted if people were not home at the time the census taker 
    came.  This is particularly true in the larger cities.
5.  People who lived in remote rural areas were occasionally not reached and 
6.  Many census takers spelled names phonetically and not as they were spelled 
    by the family.  For example:  Fisher instead of Phisher or Pfisher.
7.  The census taker may not have known how to spell the name at all, and may 
    have guessed.
8.  People searching for a name are sometimes not aware of all the spelling 
    variations that a name may have.  For example:  Wyatt, Wiatt, Viatt, 
    Wyett, etc.
9.  Many enumerators wrote poorly or used their own style of writing which 
    may be impossible to read.
10. Names may be blocked out or lost from the original record by ink smears, 
    damaged pages, tape repairs, improper cutting, faulty bindery work, faded 
    ink, or low quality photography.
11. Names may have been omitted when the census workers made copies for the 
    state and county governments.
12. In the case of the census index, the indexer may not have been able to read 
    or perhaps misread the name due to some of the above reasons.  When an error 
    is found of this type the publishers take the necessary steps to correct it.8	

A look at this list makes a family historian feel grateful to discover any records at all. Fortunately the spelling of the surname did not prove to be a factor in my search. The information about the age of various family members recorded on census records does not match the information found on death certificates. According to the 1900 census record Oliver was born in June of 1845 and is listed as fifty-four years old. The year of his birth is difficult to decipher but is legible on the Soundex index card entry. This appears to be correct assuming that Oliver's birthday occured sometime after June 20th, 1900. Lillie is listed as being born in August of 1848 and fifty-one years old. The 1910 Soundex card shows that Lilly, whose name is spelled differently in 1910 than in 1900, is now only thirty-seven years old while her husband Oliver is only fifty-five years old. Amazingly Lilly became younger and Oliver did not age in the ten years since the previous census. Albert was 18 in 1900 but only 25 in 1920. How can this be? My family and I can only humorously speculate about why this information is so blatantly incorrect. Did the enumerator in 1910 ask any questions about age or did he guess at the ages of the family members? Did my relatives have a twisted sense of humor and purposely gave the enumerator incorrect information? Who knows? The recorded age data from the schedules is not terribly significant to me since I was able to access death certificates to verify birth dates.

The names of the children are not consistent from one census to the next. In 1900 the Hollis children are listed as Joseph, Lizzie, Gertrude, Rosa, John, and Oliver Jr with John Scott who is Lilly's nephew. Four boys and three girls with one boy as the nephew. Yet in 1910 the children are registered as Joseph, Lizzie, Gertrude, Felicia, Willie, Edwinah, and baby Garland. Three boys and four girls? The addition of Garland should make it four boys and three girls. Other questions include where is Oliver Jr.? Lizzie and Gertrude do not seem to age either. Are Rosa and Felicia the same person? Are John and Willie the same person? This proves to be confusing. By 1920 Lillia's children are identified as Felicia, Willa Mae, Edwinah, Garland, Albert (who is Joseph), and Elizabeth who is now married and has a little girl of her own. Elizabeth and her husband James Williams live with mother Lillia and Elizabeth's five siblings. And who is Corrine Lamare who is listed as a 19 year old married daughter of Lillie's? Several mysteries surface looking at the 1920 record.

The 1920 Soundex card and the census record declares that Lillie is now only forty-three years old. In 1900 she was fifty-one years old and now twenty years later she appears to be forty-three. Again, did she provide false age information to the census enumerator or did the person merely guess based on appearance? In any event, it seems that my great grandmother was a very young looking woman or at times she looked older than she really was. A copy of her death certificate states that she was born November 15, 1880 and died August 18, 1960. That would have made Lillie only twenty years old in 1900. According to the 1900 census Oliver and Lillie had already been married twenty years and their oldest son Joseph Albert was eighteen years old. This means that her death certificate is incorrect or else she gave birth to her oldest son at the tender age of two! Or was Lillie Oliver's second wife? My father is unaware of any previous marriage. Whatever the case, Lillie must have been a loving and strong person. Records indicate that she continues to lead her family after the loss of her husband Oliver.

Age is but one example of the mistakes found in government records. Names can also be incorrectly recorded. Comparing the federal census records to the state death records provides more clues. By requesting the death certificate for Lillie from the state of Michigan I discovered that her full name was actually Lillia Beatrice Scott Hollis. Scott was her maiden name. Did she go by the name Lillie or was that also a mistake on the part of the census enumerators? My father remembers family members referring to her as Lillia and was surprised to see that census records record her as Lillie. Her tombstone reads Lillia.

For now, this is as far as I can research the Hollis family using census records. I have not been successful in locating any other federal records for the family prior to 1900. I did request and receive copies of death certificates for Lillia and her son Joseph Albert. An essential tool in locating these records is an item published by the U.S. Department Of Health And Human Services entitled Where to Write for Vital Records: Births, Deaths, Marriages, and Divorces. This guide lists the availability of birth, death, marriage, and divorce certificates by state. The cost of requesting these documents and information such as how far back records are kept in each state is also provided. Most public and academic libraries have a hardcopy of this title in their reference collections.

The next avenue involves the use of cemetery records. The 1930 census schedules will be released in the year 2002. I look forward to the possibility of finding yet another entry for Lillia Hollis and family in Chicago. In the mean time I must try to contact other Hollis family members to continue the search into my family's history. Other possible records of value include state government annual reports. Perhaps the Georgia Department of Education publications will shed light on the education of Black in the late 1800's and early 1900's. Likewise the records of the Superintendent of Education for the State of Georgia, Bureau of Refugees, Freedmen, and Abandoned Lands, 1865-1870 may help explain how my great grandparents were educated.

The census has evolved from counting "white souls" to including all Americans. As our country gears up for another count in the year 2000 one must think for a moment about how the information gathered has the ability to communicate family history to our great grandchildren. The only difference in the future will most likely be the format of the data. An electronic record might be the means by which a future relative verifies a deceased family member's existence. Changes in the information collected may affect the level of detail in future census records. In an effort to cut costs the Census Bureau is considering a postcard format to be mailed to each household in the 2000 Census. Race information may change if a new multiracial category is added between now and the year 2000. [warning: this links to a very large text file of the 8/28/95 Federal Register notice.] While the amount and kind of information will certainly continue to change, the effort to track America's population will not. Government records are indeed an integral source of genealogical information. They can validate the oral history heard at the family table. |next section| |Table of Contents|

Links to genealogical resources

|Afrigeneas| |Colorado Genealogical Society| |National Archives and Records Administration| |U.S. Census Bureau| |National Genealogical Society| |Federation of Genealogical Societies| |social & family history links|
Date last modified: 2/98. Send comments to author Deborah Hollis. Note: Copyright 1996 Deborah R. Hollis. Unlimited permission to copy or use is hereby granted subject to inclusion of this copyright notice.