Some people understand how scientists use numbers. Some simply like the cachet of lots of digits.
I wish I had a pointer to the article, but it was in a computer industry trade publication on-line.
A figure of 52.4% was claimed for something. This is stating the value to three significant digits. Was the sample really that large?
The author noted that that meant that there were at least 262 of the somethings in the sample. He apparently recognised that 524 / 1000 could be reduced to 262 / 500, but missed that it could be further reduced to 131 / 250.
And that is not all. Someone noted in the comments to the article that 11 / 21 rounded to three decimal places is 52.4%. It does not look nearly as impressive now, does it?
But it would never do to say “About half …” when one can whip out something like 52.4%.
How to lie with statistics.