To transform or not to transform?

That doesn't mean that all curvy lines become straight lines if you log-transform the data. The straightening effect only occurs if the original graph involves exponential growth -- that is, growth with the constant multiplier. Or exponential "de-growth" (decay).

Nevertheless, many kinds of graphs to become "un-squished" when you log transform the data. Transforming the data can be especially useful when there are big differences in the magnitudes of the numbers you're working with. If most of your numbers are fairly small, but there are a few very large numbers, it can really help to log-transform the data.

Remember, logs are all about orders of magnitude. So, if your data differs by a couple orders of magnitude or more, then you might want to think about log-transforming it.

Just to get you thinking a little bit more about log transformation, see if you can match up their regular graphs with their log transformed counterparts in the applet below:

(if you are having problems with the applet, make sure you're dropping the normal graph right in the center of the transformed graph).