I have two Spad machines that basically give different results, I used for the measurements one for one replication and another one for the other replication.
At the end the Anova shows that there is high significant different between the two replication, meaning between the two Spad machines.
Is there any possibility to correct the data one replication and be able to use my two replications data set to perform the anova for the effect of nitrogen on Chlorophyll content of 200 genotypes?
How do you know which is teh correct spad machine? I think this would be the only way to correct for a systematic bias.
On the other hand, why not modelling the two machines as one factor and the nitrogen as a second one?
Dear Patrice
You can use the two spad machines as blocks, then you can analyze your data by Randomized Complete Block Design ( RCBD) in this case, you isolate the effect of the machines in the block source of variance from the experimental error.
Good Luck
The best way to do it, if you have resource and time is to perform a proper calibration transfer between the two systems. That would allow you to address Rainer's question of which machine (or where between them) is the 'correct' answer. This involves using the same primary standards and a set of samples that are ran on both systems (in a randomised manner to minimise operational biases such as time of day, sequence of samples, day of week, time of year etc). It is always much easier to eliminate analytical differences at source than try to untangle them with stats afterwards. If you instruments are stable with respect to operational changes mentioned above (and others) a retro calibration transfer may still be possible. I trust you have a calibration log that would allow correction of any drifts over time.
Use Khalid's answer if you just want to be done with the project. Use James' answer if you want to write a methods paper. To make this a methods paper you might need another Spad machine. Is the difference in machine due to age, calibration, manufacturer, or quality issues in manufacturing the machine? Can the problem be solved, or minimized? How large is the problem? Are the machines biased, or have more variability in measurements? Might the difference be sampling error?
Dear researcher
Please refer to:
1. Landau, S. and Everitt, B. S.(2004). A Handbook of Statistical analyses using SPSS.
2.Howitt, D. and Cramer, D.(2008). Introduction to SPSS.
Regards,
Zuhair
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Modeling
- Analyses of Variance