# Does it make sense to compare values by percent change or percent point?

Sorry this is a newbie question but I'm not used to working with numbers. I have graduation rate data for states over several years, showing the average graduation rate for schools in those states. I'm trying to compare the states against one another and show how they've changed over time.

Does it make more sense to do this by comparing the direct scores, the percent change over years, or the percent point change?

    California - Averages
2014: 86.69230769%
2015: 85.46153846%
2016: 85.30769231%

California - Pct. point difference
2014-2015: -1.23076923 pct. pts.
2015-2016: -0.15384615 pct. pts.

California - Pct. change
2014-2015: -1.42%
2015-2016: -0.18%


Some more context would really be helpful here, the "correct" answer is going to depend on what you are trying to express most. I guess you want to underline in particular the states trends, but even in this case there is room for choices based on taste. You can have the same data look much differently: a $+15$ pct. pts. increase will look bigger w.r.t. a $+10$ if they are compared plainly this way. But if the first applies to an initial $60\%$ (becoming $75\%$) it represents a $+25\%$, while the second could bring an initial $20\%$ to $30\%$, a $+50\%$ increase. Which appears as the most significant improvement now? And not always the smaller the starting point is, the higher the impact of a change looks. Even a $+5$ can seem even more astounding, if it brings a $94.9\%$ to $99.9\%$! Here you could say that the $+5$ brought graduation also to "almost all" (to the $98\%$) of those who were missing the previous year! Or that the failure rate dropped from $5.1\%$ to $0.1\%$ (now "almost no-one" fails!). Or even that the previous year's failure rate was higher by $5000\%$ ...