Re: Is there research for spelling out numbers in documentation?

[Dick Margulis <ampersandvirgule -at- WORLDNET -dot- ATT -dot- NET>]

Robyn Grover, QTI, (414)246-7934 wrote:
>
> Hello,
>
> Has anyone come across any research regarding the readability of digits (4)
> compared to spelled out numbers (four)?  Our company produces technical manuals
> and we were wondering if we should use digits in instructions rather than
> words.  (We are aware of the general rules about when to spell out numbers, but
> we would make an exception to the rule if it would improve readability.)
>


Yes, there actually is research on this, and someone will probably be
able to point you to published sources; but in the meantime, I can offer
my somewhat imprecise memory of a test that was done on this very
question circa 1967 or 1968 in Peoria, Illinois.

Newspapers and magazines routinely contracted then (and perhaps still
do, if the company is still around) with the Starch organization to
measure the degree to which their publications were actually read. The
Starch interviewers would fan out on a given day and interview
subscribers by pointing to a particular article or ad and asking
questions that started with, "Do you remember seeing this
[page/article/story/advertisement]?" and end with "Did you read all the
way to the end?" (I'm paraphrasing, but this is the gist of the
questions.)

They would then collect and organize their data and give the publisher a
marked up copy of the issue with "Starch Scores" plastered all over it,
indicating, for example, 37% of your readers remember seeing this
headline, 25% of them read the lead paragraph, 15% read to the bottom of
the column on page one, 8% finished the article on page 8.

The test that was done with the Peoria Journal Star was this: The paper
printed two editions (something they did anyway); and on the day in
question, they ran one edition with numerals for money amounts and other
quantities. In the other edition, they spelled the numbers out as words.
(I don't think this extended to dates, but perhaps it did.)

Then they "Starched" that day's paper.

Scores were higher across the board for the edition with the numbers
spelled out. And they were _dramatically_ higher for female readers.
(Disclaimer: this was thirty years ago and the same test repeated today
might find a less pronounced difference between male and female
readers.) The Starch interviewers found that the majority of female
readers got as far as the first dollar sign or digit and went on to
another story immediately.

---------------

Canto the Second

The Rochester Democrat and Chronicle, one of the many weak links in the
Gannett chain, just changed their text font this past week to one with
oldstyle figures. To me this seems like a daring departure for a daily
newspaper--er, um, I guess just a daily paper would be more accurate,
there not being much news in it, he digressed--but it does give a warmer
and friendlier appearance that may overcome the math-phobic resistance
of some portion of the population. So this may be an approach to
consider as well.