Data Analysis

#1
Hey.

I'm an undergrad in chemistry and I have never excelled in statistics. I am currently working on a project where I have 16 sample which were each measured under 2 different conditions. Each sample was measured 6 times at each condition. There are no replicates whatsoever I just repeated measurements because the method for determining the value is not very accurate. The standard deviation is pretty big and I somehow need to analyse the data and determine wich parameter has the greatest influence.

ID repeated measurements Standard deviation
1 5,028 5,685 4,763 5,484 5,037 4,736 0,385180
2 4,097 4,216 4,271 4,635 4,827 5,949 0,686590
3 4,479 5,292 5,347 5,676 5,749 6,004 0,533154
4 4,398 4,590 4,617 4,745 5,019 5,338 0,339992
5 5,210 4,252 5,398 4,708 6,232 5,635 0,696523
6 4,055 4,344 4,626 4,708 4,800 5,119 0,370037
7 4,982 5,083 5,092 5,402 5,776 5,840 0,373330
8 4,654 4,982 5,092 5,338 5,365 5,694 0,359750
9 5,511 5,767 6,324 4,736 5,575 5,913 0,528413
10 4,845 5,147 5,210 5,247 5,347 5,438 0,204591
11 4,845 4,964 4,964 5,986 6,004 6,816 0,796287
12 4,581 4,663 3,677 4,818 4,462 4,946 0,448869
13 6,013 6,242 4,973 5,831 5,320 5,987 0,481495
14 5,502 5,612 5,639 5,831 6,378 6,488 0,421403
15 5,612 5,767 5,931 6,041 6,251 6,461 0,311625
16 5,174 4,964 4,909 4,891 5,384 5,320 0,215680

Would that be possible or not? I could fit the data into a full factorial design but I think that only works with replicates. Does anyone has any suggestions?
 

j58

Active Member
#2
So, to make sure I understand the design: You have 16 samples. Each sample was measured 6 times under condition A and another 6 times under condition B; so you have a total of 12 measurements for each sample, 12x16=192 measurements in total?
 
Last edited:
#3
So, to make sure I understand the design: You have 16 samples. Each sample was measured 6 times under condition A and another 6 times under condition B; so you have a total of 12 measurements for each sample, 12x16=192 measurements in total?
Yes that is correct
 

hlsmith

Omega Contributor
#4
Does the order of the condition exposure matter? Meaning if #1 was exposed to condition A than B the effect of B would have been different than effect of B if it was exposed to it first. I get this is chemistry, but can you provide a little context on what you are working with, thanks.
 

j58

Active Member
#5
As a follow-up to hlsmith's question, were the two treatments applied to every sample in the same order (ie, always A before B or vice versa). If not, how many samples were treated in sequence AB and how many in sequence BA? If samples were randomized to different treatment sequences, and it is possible that measurements taken of the treatment applied first could systematically differ from measurements of the treatment applied second, and you want to control for that, then the analysis should be done differently than what I initially had in mind.
 
Last edited:
#6
Does the order of the condition exposure matter? Meaning if #1 was exposed to condition A than B the effect of B would have been different than effect of B if it was exposed to it first. I get this is chemistry, but can you provide a little context on what you are working with, thanks.
They weren't combined, the 16 samples were were measured once at condition a and they were measured at condition b. There is no double treatment. So it is either condition a or b, no combo
 

j58

Active Member
#7
@zxelly - If I understand your experimental design, then for each sample you can compute the average for each condition, then conduct a paired t-test on the 16 paired averages.