1
$\begingroup$

I have several different experimental conditions and for each condition I measure a value at Day 0, 4, 8, 12, and 16. (Control N=15, Condition 1 N=8, Condition 2 N=4...etc.)

Given that the starting values are not always the same, (example: Control Day 0 values varied from 0.5 to 25) to compare the data I took the percent change (Day X – Day 0)/(Day 0) for all of the averages.

I need to see if there are statistically significant differences not only between the experimental conditions and the control conditions, but also between the Day X and Day 0 values. What test would be best for this? Also, is there a way to get error bars for my graphs? Is this even a logical way to approach my data?

Example :

Control: Day 0: 0.00%,
Day 4: 0.84%
Day 8:-3.31%
Day 12: -5.07%
Day 16: 11.42%

Exp Cond #1
Day 0: 0%

Day 4: -8.64%

Day 8: -67.44%

Day 12: -74.78%

Day 16: -72.8%

$\endgroup$
7
  • $\begingroup$ Maybe you should take a look at stats.stackexchange.com/questions/166434/… $\endgroup$
    – user83346
    Commented Aug 11, 2015 at 16:41
  • $\begingroup$ (I was too late to edit the comment) You might model the day as random effect and check whether the explanatory variables are significant $\endgroup$
    – user83346
    Commented Aug 11, 2015 at 16:49
  • $\begingroup$ Are all of your original data values greater than 0? Do you expect the magnitudes of measurement errors to be proportional to the data values, or independent of the data values? $\endgroup$
    – EdM
    Commented Aug 11, 2015 at 17:02
  • $\begingroup$ Are there any differences in the Day0 values among the different experimental conditions? Was there a reason why there were different numbers of cases in each Condition? Do you expect, based on your knowledge of the subject matter, that there should be trends in the data over time within each Condition? $\endgroup$
    – EdM
    Commented Aug 11, 2015 at 17:12
  • $\begingroup$ All of my original data values are greater than 0. I expect the magnitudes of measurement errors to the independent of the data values. The day 0 values from different experiment conditions had similar values to the controls' Day 0 values run with the experiment. $\endgroup$
    – anr
    Commented Aug 11, 2015 at 18:11

0