(Warning: The following post will probably bore you unless you geek out on discussions about research methodology.)
"One should ask: What is the essence of the type of theory I have constructed, and how can that be tested? even if it requires some new experimental method to do so." - Francis Crick
You get the impression (while reading the Houtkooper article reviewed awhile ago) that psi researchers thought the Observer/Observational Theory approach to psi research was promising, but that they were stymied by the fact that traditional experimental methods left them unable to carry the research to more definitive conclusions.
Observational Theory is still talked about...
"Thus Observational Theory became one of the first theories of psi to predict and successfully confirm an outrageous time-reversed effect. Incidentally, the delayed-choice experiments discussed in the previous chapter provides exactly the same prediction as these 'retro-PK' experiments. The only difference is that those experiments are considered mainstream in physics." - Entangled Minds, by Dean Radin, (2006), p. 252
.... but why isn't it being actively pursued as a valid research approach to solving the observer problem?
The key to testing for observer effects is to be able to isolate the system from all observers save the one you are interested in. (Duh, you say.) One way to do this is to be both subject and experimenter yourself. This does wonders for your ability to learn about generating observer effects, but does little to generate 'publishable' data. Were your research methods invalid? No, but they weren't independently verified, so good scientists are hesitant to accept the results as genuine data unless you already have a good reputation in the field. (I understand Houtkooper and Dick Bierman both used themselves as subjects at times while testing Observational Theories.)
Another way to do this is to devise a test framework that has a number of key parameters 'open' - a range of possible values, from which the subject can select the ones to be used. This gives the subject more information about the system than the experimenter. (Information leads to the ability to generate explicit expectations.) This does not completely resolve the problem of the experiment or data checker ultimately selecting the outcome, but with proper computer coding and analysis of the data, it can help.
The final problem is that ultimately someone will see the results in an all-or-nothing framework while asking Did we get the effect? At this point they too have the ability to influence the ultimate outcome, according to Observational Theories. The proposed resolution?
Well now, the answer to that question might be the key to taking the Amazing Randi's one million dollar prize. ;)
