I'm comparing the arborisation of oligodendrocytes using the Fiji Sholl plugin and have 7 different conditions (cell lines) and variable number of cells counted for each line.
Since no two cells are of the same length, there are a lot of empty values making statistical analysis difficult using traditional ANOVA.
Try a poisson regression or negative binomial regression and add a random term (mixed effects model) if you are tracing multiple cells from the same animal. You could use the lme4 package in R for this.
In order to aproximate you distribución to a continuos normal one using a non continuos range of cell sizes, you may use intervals of cell size. In that way you have a continuos data distribución, where all cell sizes should be in the range of mean +/- 3SD.
The interval size should be the squere root of difference between (máx cell size - cell size).
Hope be of help
You can use Design Expert to do general factorial experiment, where missing data can be left blank.
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Modeling
- Analyses of Variance
- ANOVA