You want to collect the dates of birth of a group of people so that you can analyse and segment the group by age, but asking for a date of birth isn’t necessary for any specific reason and many people in the group may balk at giving you this private information.

Discussion

It’s a common conceit among people designing forms and surveys that if you ask a question, you’ll get a useful answer, but when it comes to personal information the wise designer treads very carefully.

A general rule for form design is that every extra question reduces your response rate and diminishes the chances of the form being completed accurately. Every time you add something you take something away.

As people become more conscious of the real or imagined incidence of identity theft and the predations of overzealous marketers, they are increasingly reluctant to part with personal information even where they might otherwise be happy to provide information for a purpose with which they agree.

As consumers, we understand that there are times when our date of birth will be required for official purposes. We don’t expect to be able to open a bank account or obtain a passport without it. But when we’re completing a marketing survey or simply registering with an online shop it can often feel like too much information. I don’t get asked my date of birth when I shop on the high street — why would I need to disclose it just to use the online equivalent?

Having age information about a group of people is often useful or even necessary for researchers and marketers. They might want to produce an age breakdown of a group of people using their service or participating in their research, or validate that people in a group qualify to be there where there is an age restriction.

One solution to the problem is to ask respondents for their age group rather than their date of birth:

How old are you?

• under 18
• 18-25
• 26-39
• 40-49
• 50-64
• 65 and over

This is less intrusive for respondents and more likely to gain a response than asking for a date of birth. It allows a report to be produced grouping the respondents by frequency under these headings.

Unfortunately, this approach is very inflexible. Once the respondents are placed into these groups you cannot produce a different age segmentation. In the above example, if there were a subsequent requirement to count all the people aged under 30, you couldn’t.

This approach also fails if the data are to be used over time. The age segmentations will be correct at the time of data collection but will rapidly obsolesce after that. It works for a one-time snapshot such as a survey but doesn’t work for a group who will be participating over time such as members of an organisation or customers of a business.

Resolution

These issues can often be better resolved by asking respondents for their ages and then calculating an estimated date of birth for them which is then stored. Ages are less specific than precise dates of birth and asking for them is more likely to gain an accurate response.

How old are you? _____________ (write in your age; leave it blank if you’d rather not say)

If a person responds that they are 30 today, we cannot possibly know their birthday. It could be any day between today (they’ve just turned 30) up to a year’s time less a day.

So calulate the estimated date of birth to be:

today – respondent’s age in years – six months

While there is only a 1 in 365 (ish) chance of estimating the date of birth correctly for each individual, by assuming that each person is half way towards their next birthday you prevent skewing the overall age breakdown for the group either too low or too high.

Ensure that your database or spreadsheet column is called estimated date of birth to prevent subsequent users assuming that these DOBs are accurate.

Using these estimated dates of birth you can produce sufficiently accurate age segmentations at any time in the future regardless of how you define the segment boundaries.

As a caveat, be aware that while this technique allows you to estimate the number of people in the group, say, under 30, if there’s an absolutely specific reason to find the actual members that are under 30, you can’t. You’ll get a few false positives that you’re estimating to be under 30 even though they’ve already reached their 30th birthdays.