The reason for this is twofold -- first, you end up with a form of the equations you're dealing with you can use for various values, and second, you skip rounding errors.

Case in point: My Macroeconomics homework with three values for consumer price index baskets: $124 (baseline), $153 and $161.

First problem: calculate the CPIs, no decimals. The results are 100, 123 and 130. Clear as mud.

Second problem: calculate inflation between the last two. (130-123)/123 = 5.7%. Except I was doing the math in Excel, and had set it to only display the number of digits required in the answer. It displayed (130-123)/123 = 5.2%. Huh?

Well, Excel isn't a physicist, but it is a computer program so it carries a fair bit of precision. Though it

*displayed*the rounded values, it actually did all the calculations with full precision, which turns out to be more like (1.298387097...-1.233870968...)/1.233870968... = 5.2288... %

I'm not convinced this homework really taught me what the professor thought it would.