State codes

Please pay attention to the state codes you are inputting. When you finish inputting a manuscript line, look over what you typed. Make sure the data in the fields makes sense. The cleaners are noticing a lot of incorrect state codes. They don't check every state code. If it seems like a reasonable place for a person in our sample to be born, then they won't check it. So, please review your work. It only takes a few seconds.

Here are some of the recent mistakes:

• New York is NY not NT (Northwest Territories)
• Illinois is IL not UK
• Virginia is VA not VI (this mistake is extra disappointing, since most of the inputters live in or previously lived in Virginia)

Here are some tips:

• DO NOT make up state codes. Look it up in your manual. If a code doesn't exist, type the location in the remarks.
• Pay attention to where your fingers are on the keyboard. NT and UK in the examples above are probably because the inputter had her fingers in the wrong place on the keyboard.
• Review your work. That means, look at what you just typed. If you had accidentally typed UK when you meant IL, that would jump out at you.

We would like our data to be as accurate as possible. The cleaners can't catch everything, and sometimes a user needs raw/uncleaned data. So, please do your best to type in the correct state code.

Feel free to review previous posts about state codes.

Age fractions

This is just a reminder that when entering the age field from the census manuscript that you only enter a fraction for children under 1 year old. If the child is 1 1/12, enter 1 in the field. There is no need to write the fraction in the remarks. Once the child turns one, she is one until she is two. She is two until she is three. So if the child is 2 6/12, enter 2 in the field.

Recently, the cleaners have been noticing that inputters are doing math (sometimes incorrectly) and creating decimals to deal with these fractions. Please do not do this.

Here are some examples: There were two entries typed in as 25/12. For one of them the manuscript said 1 1/12, and for the other the manuscript said 2 1/12. That's quite a difference. These should have been entered as 1 and 2, respectively. Another person did some conversions from fractions to decimals, 1 9/12 to 1.75.

Much of this goes away in the cleaning, but if you enter the wrong information that could make the rules clean the data to the wrong age.

To sum up. Use fractions only when the child is under one year of age. After the child turns one, use only whole numbers.

Reminder - Do your corrections!

We have noticed that sometimes people get their checks back, and they forget to make all the corrections the checker said were needed. I don't mean corrections that the inputter didn't agree with. I mean the inputter entirely ignored the checking document she/he received.

We don't do checks to annoy everyone. We do them for two reasons:

1. Quality control of data
2. Ongoing training

When you get your checks back, you don't have to immediately stop what you're doing. But you must complete your corrections by the NEXT DAY. If you're in the middle of a soldier, you may do your corrections before you resume work on him the day after you receive your checks.

Also, you must READ EVERY LINE that the checker wrote. Ask questions if you need to. Make ALL necessary corrections. The checker should have made that easier for you by highlighting them in yellow.

If you don't read the checks or make the corrections, then everyone is just wasting a lot of time. Sandy will probably start randomly opening soldiers in the screens to see if the corrections have been made.

Link rates for daughters - a comparison

Recently, we've been working on our annual report. As part of that, we report on the link rates, that is, how successful we are at finding people in the census. For the most part, our link rates are pretty consistent from year to year. There is one group of children, for whom our link rates have increased a lot from 2015 to 2016. This group consists of the daughters that are found in the pension but whose married names are not listed.

The following tables for both Project 1 and Project 2 show what percent of the daughters without married names in the pensions that we were able to link to census or death records during both 2015 and 2016.

Project 1

	1870	1880	1900	1910	1920	1930	1940	Death
2015	91.9	90.8	79.4	71.8	67.6	64.5	56.3	69
2016	90.7	93.7	84.7	77.2	73.7	70.7	62.5	75.6

Project 2

	1870	1880	1900	1910	1920	1930	1940	Death
2015	72.4	87.5	61.4	41.4	26.4	18.7	13.5	26.8
2016	79.6	92.7	70.3	49.5	35.5	28.0	20.1	36.0

For Project 1, our link rates increased by an average of 4.7 percentage points from 2015 to 2016. For Project 2, or link rates increased by an average of 7.9 percentage points from 2015 to 2016.

We attribute this increase not only to the greater availability of marriage records online, but specifically to the fact that all of you are going to Familysearch.org and searching for those marriage records. It is well worth the effort to find those marriage records.

The take away here is, don't forget to use Family Search.

Any questions?

Irene and Janice are busily preparing for our upcoming (in March) census inputting conference. It will be excellent.

To help them prepare, we'd like to know if you have any questions or confusion.

• Do you have a specific inputting-related question?
• Are there any searching tools or websites (such as Family Search) you'd like to learn more about?
• Do you have any questions about the census?
• Do you have any questions about death records?

If you have any questions, please submit them in the COMMENTS. That way, Janice and Irene will be able to access them easily.

During the conference, we are hoping to have time for small-group or one-on-one work. Over the next few weeks, make a note to yourself if you have a specific family you'd like help with. If there is time, Janice or Irene might be able to assist you. Please don't list these in the comments or email Irene and Janice. Just keep a note for yourself.

We hope this will be an excellent meeting and everyone will learn something new.

M or 7 in marital status

Anne found the following information about the marital status column in the 1940 census.

To summarize, if the enumerator left the marital status column blank, then the Census  Bureau staff in Washington D.C. would attempt to complete the field based solely on other elements they read on the page. If no spouse or children were living in the household, the coder assumed the individual was single. If, however, the person had no spouse in the home and there were children present, then one of two codes was entered, either 1) widowed, if the individual was 55 or older, or 2) 7, if the person was under age 55. Even when a marital status of M[arried] was recorded, clerks were instructed to interline the M and replace it with 7 if no spouse was included in the household, thus resulting in this entry: M 7. This distinction with respect to age is important. In a nutshell, if you find a marital status of M crossed out and replaced with Wd, don’t assume the spouse is deceased; look for that individual in another household. If you find a 7 in the marital status column, or an M crossed out and a 7 penciled in, then look for the spouse in another household.

The full article can be found here.

January 2017 checking stats

In January, we checked 41 soldiers under our system. I've reviewed all of the checks, and I've tallied the number of differences. Some of these are errors, and some are judgment calls. Here are the categories and the total number of differences for each category.

GRID Errors

MILIN?/MAR? - 3
Missing HH member - 8
Duplicate people - 1
Wrong person - 1
Other - 4

Inferred Relationships

Incorrect relationships - 4

Census Errors

Name - 25
Typo/Reading/Wrong - 88
State Code - 0
Missing/Wrong URL - 4
Missing data - 35
Additional finds - 24
Quality Code - 13

Death Errors

Typo/Reading/Wrong - 9
Missing data - 15
Missing/Wrong URL/Source - 11
Quality Code - 4
Additional finds - 32

Tree Errors

Missing/Incorrect information/relationships - 10

The total number of differences for all 41 soldiers is 291. This is 82 fewer differences than December, when we checked 42 soldiers. Some differences/errors affect the data more than others. If we checked other pensions, we'd probably find similar differences.