This is very interesting. First, there must be some discussion about this in the literature about soil moisture probes since they basically conductivity as a function of soil moisture. Second, I think that unless you have a reason to test soil just collected from the ground, you'd need to "treat" all the samples prior to testing.
Since you don't specify any limits on the initial conditions, I suggest you might try starting by saturating the samples and then taking readings as they drain to field capacity and dry further. If you kept the samples in a temp and humidity controlled room (or container), the process would be repeatable.
The one "problem" I see with my suggestion is that the initial quantity of moisture in the soil at saturation will differ based on soil characteristics. However saturation and field capacity are soil-relevant properties, so this might actually be a benefit.
The only other way I can think of is to oven dry each of the samples, then add a measured volume or mass of water to each sample. Then you could mix the sample by shaking in a larger container for a specific amount of time. Basically - as long as you define the steps you take to make your samples, it should be repeatable.