Rounding
Message posted by Jen on November 4, 1999 at 12:00 AM (ET)

We've been having a discussion on rounding rules & seem to have 2 differing opinions. Assume we are rounding to 1 decimal place.

The first method says that if the second decimal is 0-4, the second digit stays the same; if the second digit is 5-9, round up.

The second method is basically the same except if the second digit is 5. In this case, you would round the last digit to the closest even number. "Thus the number would 3.55 would round to 3.6 & 3.45 would round to 3.4".

The first method appears be fair - round up 50% of the time & round down 50% of the time. However, some contend that 0 doesn't count, &, therefore, the first method is bias. Others contend that 0 is a valid number (particularly when rounding a number like 2.104 - where the 0 takes on a definite value) &, therefore, makes the second method bias.

What do you think?

READERS RESPOND:
(In chronological order. Most recent at the bottom.)

Re: Rounding
Message posted by JG on November 4, 1999 at 12:00 AM (ET)

However, you have another problem, while 0.9 can be rounded to 1, 0.2 can not be rounded to 0 . Maybe that is what they mean by zero is special.

Re: Rounding
Message posted by JG on November 5, 1999 at 12:00 AM (ET)

If x = many digits then the problem evaporates, since 0000000000 and 5000000000 will not occur very often.

Re: Rounding
Message posted by j-bamdad on November 5, 1999 at 12:00 AM (ET)