Changes in Transmit Power vs changes in received SNR...

I am fairly new to wspr, I have been running the software periodically for a couple of weeks, and have started pondering how to relate this information to "real world" scenarios. I have come across a figure of -10dB SNR as a minimum usable signal level to decode PSK31 transmissions. If I look at most the spots where I have been received, I am usually well into the "minus teens or twenties". Has anybody looked at whether there is a predictable correlation between changes in transmit power and changes in received SNR? I transmit at 37dBM (5W). If I were to double my transmit power (3dB increase), would there be a predictable amount of change in SNR at the stations that receive me, or are there too many factors that affect my receive strength at any given station. Ultimately, I would like to see if we could use this wspr data to predict, for example, how much power should I expect to need to use to get a reliable PSK31 signal into Germany, from North Carolina, on 40m with enough strength to be received.

I currently transmit at 5W into what I believe is a compromise antenna on 40m. I am using a "Shorty Forty" as detailed at the following link : http://www.flashwebhost.com/circuit/shortant.php. It is mounted in my attic at approximately 25 feet above ground level. I don't know what my effective radiated power would be stated at. (also using an antenna tuner, because my SWR is about 2.5:1 at resonance, and I don't want to transmit for 2 minutes into that high of an SWR). On 20M I am using a "Hamstick Dipole" mounted in the same attic. I have been able to match this antenna to my transmitter on 30m, but I don't believe anyone has received my signal on this band, yet, so I might just be heating up my tuner. Judging from the numbers I have been seeing in most of my spots, If a 10dB increase in output power would result in a 10dB increase in SNR at the receiving stations, I would still be below getting into most of Europe below the -10dB SNR figure. Another 10dB would do it, but I don't have 500W to work with. I don't expect there to be a 1:1 relationship between an increase in my transmit power, and the SNR on another stations receiver since their SNR is comparing my signal to their background noise. I do wonder if anyone has been able to notice a predictable pattern, though. Please share your observations if you have.