I think In case you manipulate one factor and fix all other factors to observe the result, you might conclude that the change in the factor is the cause of the result. Perhaps this is why experimental design is so important.

Is there any different opinion?

"Perhaps this is why experimental design is so important." It is true. In that case, analysis would be much simpler.

But unfortunately, a lot of analysis cannot be put under the experimental design framework. Even in a simple medical research project like smoking it is immoral to force people  to smoke. Most of social and ecinomic study data are observational. This makes the analysis mush more complicated, since the observed data are the results of individual rational behavior. So the data itself is sort of "contaminated" under the view of  experimental design.  It is truncated, self-selected, etc. That comes the major difference between statistics and econometrics.

sometimes, we can not control the factors which affect the results, we can only observe its development for the lack of technology

8楼的人太牛了，佩服佩服。