Phone numbers in electronic forms

This is the third and final research post in our current series on phone numbers in electronic forms.

In our two previous articles, we looked at user behaviour regarding mobile phone numbers and the equivalent user behaviour for landline phone numbers.

Both these pieces drew on by-product data from research conducted for SEEK, Australia’s number one job site. Participants in the research were recruited using a pop-up dialog box that appeared on the website (see Figure 1).

Figure 1: Invitation used to recruit research participants.

Users were free to enter whichever phone number they wanted into the box titled “Your Phone Number”, and there was no validation.

What we’ve already found

In the previous research pieces, we found that in the absence of further instruction, the most common way to enter both mobile and landline numbers was as a single string without spaces, dashes or other non-numeric characters, and that this may be because users are wary of triggering a validation failure.

We also found that there was a great deal of variation in how users provided their phone numbers, so it was best that electronic forms be accommodating of these variations.

What we’re going to discuss here

In this final piece, we’re going to look at:

  • whether users prefer to enter their landline or their mobile number; as well as
  • how many errors users make when entering their phone number.

Users prefer to provide their mobile number rather than their landline

We had a total of 2,721 phone numbers provided via the research invitation shown above. Of these, more than three quarters were mobile phone numbers (76% or 2,081). In this case at least, the majority of users preferred to provide their mobile number rather than their landline.

This finding is particularly interesting if you consider that we included an instruction about area code next to the “Your Phone Number” box. If anything, this instruction should have led unsure users to assume that we were asking for a landline rather than mobile number. This means the proportion of users who would, in real life, prefer to provide a mobile is probably even higher than our observed 76%.

These findings highlight the importance of catering for mobile phone numbers in electronic forms and backend systems. We note that a recent estimate from the (US) National Center for Health Statistics was that 23% of American households have only mobile (i.e. cell) phones and no landlines [1]. The Center has been measuring the substitution of landlines by cell phones for many years and the figure has been steadily increasing.

Australia is a little behind the US in this regard. 2008 statistics published by the Australian Communications and Media Authority suggest that 8% of Austrailan adults — note adults not households — are ‘mobile only’ [2]. Moreover, while the proportion of US households with a landline peaked in 2000, in Australia the peak only came in 2004 [3]. Regardless, the need to cater for mobiles is here, for the short to medium term at least.

Users rarely make errors when entering their phone number

Less than half a percent of users made an observable mistake when entering their phone number. From 2089 mobile numbers in total, 8 had an observable error; from 643 landline numbers, 3 had an observable error.

From the total of 11 errors:

  • 6 were numbers missing one digit
  • 2 were numbers missing two or three digits
  • 2 were numbers with an extra digit
  • 1 was a number entered with an extra character at the front, namely a comma.

We note that aside from the above, there may have been other errors that we couldn’t detect just by looking at the numbers, e.g. a “6” typed instead of a “7”. However, there is nothing to suggest that the rate of such hidden errors is significantly different from the rate of errors we could see.

In all but the case with the rogue comma, the user would have needed to do something to get the number to a usable state. However, an observable error rate of less than 0.5% is very small and suggests that in some cases, validation may not be worthwhile. If anything, a simple check that there is an acceptable number of digits should be enough, remembering the high variability in the way that users enter their number and thus the wide range of acceptable numbers of digits.

Of course, it makes sense that users would make few mistakes when entering their own, primary contact number. This is something that is often very familiar to us and may need to be provided on many occasions. Thus we become used to the feel of typing the numbers out, and it becomes a sub-conscious, essentially automatic, process, much like driving. It’s a relief to know that for at least one form field, the task of entering one’s answer may actually be easy.

BUT … users also don’t always see/follow instructions when entering their phone number

As already mentioned, users were asked to provide their area code when entering their phone number. If the area code wasn’t provided, we could still use the number because in Australia, the first few digits of the number give an indication of the location and thus what area code to use. So numbers with the area code missing were included in the total set of usable numbers.

We found that 14% of the usable numbers were provided without an area code (i.e. 87 of the 640). There are two reasons why the area code may not have been provided:

  • users may not have seen the instruction regarding area code, which was in grey text to the right of the phone number field; and/or
  • users may have seen the instruction but not followed it.

Given that other research has shown users do, by and large, try to fill out forms correctly [4], it seems plausible that the majority of people who entered their landline number just didn’t see the area code instruction.

In hindsight, placing the instruction to the right of the field was unlikely to be the best choice, because the users would move straight from reading the question to answering it, perhaps never moving their eyes to that part of the form. A better approach would have been to include the instruction in the question itself, e.g. “Your phone number, including area code”.

Even then there would probably still have been a proportion of users who omitted their area code. This is because we also know that form-fillers satisfice, that is, read only that part of the question they think they need to read in order to be able to answer.

As such, if it is imperative that the user enter an area code, you will need to run a validation check for the number of digits provided. Had we only been interested in landlines, another approach would have been to put a separate field for area code in front of the number field.

[1] Blumberg, S.J. & Luke, J.V. (2009). Wireless Substitution: Early Release Estimates from the National Health Interview Survey, January–June 2009. Division of Health Interview Statistics, National Center for Health Statistics.

[2] Convergence and Communications – Report 1: Australian household consumer’s take-up and use of voice communications services (March 2009). Australian Communications and Media Authority.

[3] Fixed-mobile Convergence and Fixed-mobile Substitution in Australia (July 2008). Australian Communications and Media Authority.

[4] Wentland, E.J. & Smith, K.W. (1993). Survey Responses: An Evaluation of their Validity. Academic Press, Inc., California.