one way ANOVA or paired t test?

#1
Apologies if this is a basic question, but my stats is very rusty these days!

Background: I'm looking at a procedure that automates FISH analysis in a genetics diagnostics laboratory. I am currently running 7 patients that are negative controls (known normal patients) on the automated system to establish a false positive rate. Each patient is run just once (so single replicates of data per patient). At the end I get percentages of normal and abnormal cells for each patient (expressed as a percentage of the total number of "informative" cells for that patient). There are also a selection of "uninformative" cells that the software has classed as neither normal or abnormal. I can look through these uninformative cells and disagree with the software classification, and manually reassign them to either the normal or abnormal category.

Question: I want to know whether manual reclassification of the uninformative cells has a significant effect on my normal and abnormal percentages for each patient. I have before and after percentages for each patient. I'm not sure what statistical test I should use to assess this?

The overall question I'm wanting to answer is whether reclassification of uninformative cells is needed as part of the analysis procedure? I'm sort of hoping not... because it is very time consuming!

Thanks in advance,
Sabrina
 
Last edited: