These experiments are not as useful nor controlled as you may think. Let me break down into some experiment design principles here:
#1 You can only generalize to the population studied
--Thus your demographics will be that of a more computer savvy user, with leisure time. This is NOT a representative sample of a normal population. When using regression methods you will get a homogenous result only from 'game performers'. These experiments are not valid unless you can prove that there is no purposeful difference between this population and the general population.
--It still can provide interesting insights, but any quantified data must be taken with a grain of salt
--The population may have one net effect, but perhaps a different type of agent/actor would have an OPPOSITE or equalizing effect (games are an artificial setting)
#2 Process defined by agency
--Here the markets are designed by an all-seeing game developer. I don't know about you, but many MMO's I have participated in had lackluster markets due to poor UI or the mere fact that it is a new market.
--Product innovations are limited or non existent, and users cannot refine markets based on their experiences. These days the most interesting economic studies are looking at PROCESS which is understood much less than outcome. Neo-classical economic theory does a great job at explaining outcome, but is horrid at process. That is why many market failures are not forecasted, but instead studied post-mortem.
The Apple App Store is worth $200M monthly, whereas the Android market is worth a paltry $5M. As sad as that comparison may be, from our experience the total is probably much lower. We know from experience that below position 25 on the top selling games the earnings drop off to almost zero. To back that up a bit, we're going to release our latest Android sales data.
Software production is assumed to be a line function, but it is run like a staff function. -- Paul Licker