Should I use Pearson Correlation or Linear Regression to show the linearity of two sets of numbers？

I have two sets of acceleration data in terms of time. In one set the acceleration increases more linearily than another set. How can I show the linearity difference of the two set of data best? Should I use Pearson Correlation or Linear Regression? If I get r = 0.8 and 0.98 by Pearson Correlation, how I can interpretate the result? Do they have very big difference on the linearity based on 0.8 and 0.98?

Thanks.

• My reply at stats.stackexchange.com/a/13317 addresses the negative part of this question: namely, why $R^2$ does not of itself tell you anything about linearity, especially when the independent variables in the data sets are different. Although you did not ask it, this question begs for a positive response, too (and perhaps you should mention this explicitly): given that your proposed methods don't do the job, what does work for assessing linearity? – whuber Oct 15 '13 at 14:56
• Thank you for your answer. I am trying to show how linear the data increases by time, and their difference. If my methods don't work, what else method can I use to address this? – HMLAZIO Oct 15 '13 at 15:07
• Please edit your question to include the clarification you just made. It would also help to explain more fully what you mean by "more" or "less" linear and precisely what "difference" you refer to. – whuber Oct 15 '13 at 15:23