Types Of Knowledge

Each is an order of magnitude better than the next. That said in each class you have a range of believability. You shouldn't trust the opinion of certain people; you shouldn't believe theory that doesn't make sense; and you should only believe a well carried out experiment. All other information is derived from these three things.

The way research should work is that someone (an expert in the area) forms an opinion based upon other work he has seen. He then performs further research to see if any work has been done in this area. This body of previous work is collated to produce a theory something on the lines of "because of these things that we know (from repeated experiment) it makes sense that ___ is true".

Once you have the theory you can move on to the experiment. This should be repeatable in both senses of the word. One, the method should be written so that it can be copied (and ideally should be submitted for peer review before any tests are done). Two, the experiment should give the same results when done again. This should be written up. The results should not be interpreted here. All you know is that when you follow the Experiment Plan you get ___.

From here you have a new piece of evidence to update your theory; this is the conclusion. It should follow the form "I think the reason we got ___ from the experiment is because". This should be treated as any other piece of theory.

Once this is done other researchers should step in. If you want to expand the work, keep everything the same apart from the one area you want to look at. This will greatly enhance the reputation of the previous work by showing the repeatability. It will also make people trust your work.

If you disagree you can probably say "I think the reason you were seeing these results was actually ___", you then copy the experiment other than the thing that you think was the problem. You shouldn't disagree with the findings you can disagree with the interpretation.

I would consider this basic knowledge, yet I see many pseudoscience and alternative medicine break these rules all the time. You don't need a Ph.D. to produce good scientific findings. In fact, even if all other rules were broken I could still see this method working.

1. Bob thinks that better lighting will make people more productive. He thinks this because at his last job they had better lights and people worked faster.

2. Bob does some research (he can't find much) but some of it seems to agree. He writes up his literature review and experiment plan and sends it to all his employees. Bob intends to turn up the lights every other week for a month and log productivity.

3. Bob sees much better productivity when the lights are higher, especially in the first week. He sends the results (raw data) to everyone.

4. Jane thinks that people were only working faster because they were being tested. People were trying harder when the lights were higher to be helpful. (new theory from same data). She proposes the they turn down the lights instead of up. Sends and email saying what the new test is.

5. Peoples productivity went up. This new piece of evidence convinces both Bob and Jane that people work harder when they are tested.

6. They both start reading up on how to eliminate bias in experiments. Think up a much better experiment and implement it.

The initially flawed experiment was fixed on the next iteration because someone disagreed with it and worked to show why. If Jane had instead tested something else or tested in a different way she wouldn't helped nearly as much.

I see research like a forest, where each new experiment is a new tree. If you instead build of others work (especially if you disagree) you add a new branch and the tree gets bigger and stronger. To me psychology is a shrub-land with everyone starting their own trees rather than building on each others work.

The science we are taught in school tell us to write the following sections. I hope that now you can see why. I also hope that you can see the point in separating this work out so that people see every step before the next.