Holding back a variable

#1
I am running data I collected between a control group (n=52) and an experimental group (n=51). The samples are trabecular bone cubes; the experimental cubes were exposed to S. aureus (bacteria), control groups were in sterile broth. All cubes were mechanically compressed for stiffness values. Using a two-tailed independent t-test, I found no significant effect on the stiffness of bone tissue in the S. aureus treatment. However, there is a problem. It turns out that the starting density (before the experiment) is significantly higher than the control group. It is well known that denser materials are intrinsically stiffer. It may be that there was an effect; it is just masked by the higher density of the experimental cubes. My ultimate question: is there a way to account for this difference in density at the beginning?

The best I can come up with is to try and reorganize the density groups into ranges (low, medium, high- as an example) to see if those ranges of density are significantly different. In the event they aren't, I would use an ANOVA to compare stiffness within these density "buckets." Any suggestions, thoughts, comments would be much appreciated. Thanks!
 

hlsmith

Not a robit
#2
Were cubes randomized to broths? If not, next time randomized. You can run a multiple linear model controlling for baseline densities and group status.
 
#3
Were cubes randomized to broths? If not, next time randomized. You can run a multiple linear model controlling for baseline densities and group status.
Cubes were randomized based on tibia number (n=9), slice number (proximal or distal), and anatomical location within each slice (medial or lateral and posterior or anterior). We thought these were the most critical considerations and, unfortunately, overlooked density differences.