In Psychology, one often asks questions about how the brain works, how society works, why certain people do something, and so on and so forth. A way for psychologists to be able to answer these types of questions, is to do studies, and follow different research designs. A few different research designs are "Naturalistic Observation", "Case Studies", "Correlational Designs", and "Experimental Designs".
Each research design is special and important in their own way. In "Naturalistic Observation", psychologists are able to watch the behavior of people or things or animals in the real world, and see how they naturally behave without any intervening or outside influences. In a "Case Study", psychologists will examine a single person or a small group of people for several years, documenting their lives, and further more find existence proofs, which are "demonstrations that a given psychological phenomenon can occur" (Lilienfeld, pg 51). "Correlational Designs" are also used because psychologists can use them to see how closely related two variables are such as seeing how well a person's performance is on an exam and how much sleep they got the night before.
"Experimental Designs" are special though. While the three previous research designs are often used, and are liked, none of them can allow us to infer causation like an "Experimental Design" can. The reason being is that in an experiment, as the researcher, you are the one who is manipulating variables and seeing which variables make a difference in behavior.
Like anything, it is not perfect because if you do not take extra precautions, you could make an inference on something, and it could turn out to be completely wrong due to a third variable or interference such as the placebo and nocebo effects, the experimenter expectancy effect, and demand characteristics. Which is why it is important to have blinding, and to have your independent variables set, along with your controls.
I personally know this from experience. For my 9th grade science class, I had to partake in the science fair in order to not get a failing grade. My experiment involved having a typing class, taking typing tests while listening to different songs that had different beats per minute, and then recording each person's words per minute. Each person had to take three different tests, with one time listening to a song with a high wpm, another with a slow wpm, and then one time without any music at all. However, something that I didn't account for was the demand characteristics taking place. Without thinking, I told the entire class the entire experiment and what I was hoping to happen. As a result, looking back and people's previous typing scores, they somehow dramatically improved. In the end, I had to use a completely different typing class, and this time without saying any details. Funny enough, I still managed to win a medal for it.
Despite the mistakes that can happen without precaution, experimental designs are very important because they are the only ones in which you are able to clearly see a cause and effect, and able to make those inferences.