Baselines
In general, in a time-series, use a baseline that shows the data not the zero point. If the zero point reasonably occurs in plotting the data, fine. But don't spend a lot of empty vertical space trying to reach down to the zero point at the cost of hiding what is going on in the data line itself. (The book, How to Lie With Statistics, is wrong on this point.)
For examples, all over the place, of absent zero points in time-series, take a look at any major scientific research publication. The scientists want to show their data, not zero.
The urge to contextualize the data is a good one, but context does not come from empty vertical space reaching down to zero, a number which does not even occur in a good many data sets. Instead, for context, show more data horizontally! .
-- Edward Tufte, October 18, 2001 http://www.edwardtufte.com/bboard/q-and-a-fetch-msg?msg_id=0...
Ugh, the more I read of Tufte the less I like him. He loves to make blanket statements (like calling a point "wrong") in an authoritative way, and he's amassed enough adoring followers that people repeat his words like gospel instead of calling him on it.
It's valid to say that a non-zero-based scale shows the data better. But it's also completely valid to note that non-zero-based scales can be used in alarmist ways to make data extremely misleading.
I thought "The Visual Display of Quantitative Information" was terrible, and I went in really wanting to like it: http://www.amazon.com/review/R11NYC3OE3LBE
What does tell you about that type significance of (rational) data is the delta, which requires the 0 in order for you to see the scale. Now, in many interesting cases this could be represented as error bars under H0, but that's usually not done in publications, because they only have room for one plot type.
The range of the plot is not arbitrary. To suggest otherwise is deceptive.
Yeah, I'll buy that.
0 is the only number that is not arbitrary in a scalable measurement such as a graph. That's what makes it important.