How accurate are birth dates on the 1939 Register?

Question

One of the most useful pieces of information on the 1939 Register is the date of birth, since there are no other country-wide records containing birth dates other than civil registration records. Finding a birth date on the Register has solved a brick wall I had for one of my great grandfathers with a very common name; without this information I was unable to determine which birth certificate was the correct one.

But, how accurate is the date of birth field on the 1939 Register? I am not talking about FindMyPast's transcription accuracy – which does leave something to be desired. Rather, has anyone systematically assessed how accurate is the original data?

I understand that no evidence of date of birth was required when householders filled in the National Registration schedule. In browsing the records I have come across a few records with incorrect birth year or birth days. However, given that the Register was later used as the basis for NHS registers, I would hope that errors in the data would be reasonably small.

I think it is important to assess the data accuracy because a lot of birth dates on family trees will be drawn from this dataset – it is a much cheaper alternative to obtaining a birth certificate. If 1% of birth dates are incorrect, that may be acceptable, but if 20% are incorrect then we may need to treat this birth date data with extreme caution.

Answer 1

For my one-place study of a parish in the East Midlands, I transcribed all entries in the 1939 Register. I then individually cross-referenced the birth date of each entry with a secondary source (either civil registration or parish record). This enabled me to assess how frequently the birth date given on the 1939 Register matched with the secondary source. It is essential to note that in this analysis, I was only able to assess whether birth dates from the two sources matched. In cases that mismatch I cannot say which birth date is the correct one.

For this parish I was able to compare the birth dates of 211 individuals. There were 4 entries for whom I was not able to perform a comparison. Any entry of a person born less than 100 years ago (after 1916) was excluded from the analysis.

Of the 211 entries, 38 mismatched on birth day, month, or year, or a combination thereof:

• 22 mismatched on birth day
• 6 mismatched on birth month
• 17 mismatched on birth year

To break down the data further, I determined that:

• Of entries that mismatched on birth day, 36% were off by only 1 day
• The most likely birth day to mismatch was the the 13th of the month
• The month was never off by more than 1 month away from the expected month
• Of entries that mismatched on year, 82% were only off by 1 year
• Nearly 50% of those with a mismatching birth year were born in the 1880s (age 50-60)
• Men and women were equally likely to have a mismatching birth date

So what does this mean?

We have to treat birth dates on the 1939 Register with caution. According to this analysis, nearly 1 in 5 entries has a birth date that does not match the secondary source. Interestingly the birth month was usually accurate, but it was not uncommon for the day or year to be off by 1. It is worth keeping in mind that this is a non-random sample of individuals from a single parish; whether these statistics would be the same elsewhere in the country is impossible for me to say.