First, just a bit of terminology. We have said that we use experiments to find causal relationships in the real world. When we are doing a study to see if changing variable $A$ causes a change in variable $B$, we use randomness to choose experimental and control groups which are given different values of the $A$ variable and then we measure variable $B$ and hope for a change. In such as study, $A$ is called the explanatory variable and $B$ is called the response variable.
Our next discussion of experimental design was missing two terms that are a lot of fun, and also important. The first is lurking variable, which names a variable (measurement) which the experimenter did not put into their investigation. So a lurking variable is exactly the thing experimenters most fear: something they didn't think of, which might or might not affect the study they are doing.
Next is a situation which also could cause problems for learning from experiments: two variables are confounded when we cannot statistically distinguish their effects on the results of our experiments. When we are studying something by collecting data and doing statistics, confounded variables are a big problem, because we do not know which of them is the real cause of the phenomenon we are investigating: they are statistically indistinguishable.
The combination of the two above terms is the worst thing for a research project: what if there is a lurking variable (one you didn't think to investigate) which is confounded with the variable you did study? This would be bad, because then your conclusions would apply equally well (since the variables are statistically identical in their consequences) to that thing you didn't think of ... so your results could well be completely misunderstanding cause and effect.
The problem of confounding with lurking variables is particularly bad with observational studies. In an experiment, you can intentionally choose your subjects very randomly, which means that any lurking variables should be randomly distributed with respect to any lurking variables — but controlled with respect to the variables you are studying — so if the study finds a causal relationship in your study variables, in cannot be confounded with a lurking variable.
Example: Suppose you want to investigate whether fancy new athletic shoes make runners faster. If you just do an observational study, you might find that those athletes with the new shoes do run faster. But a lurking variable here could be how rich the athletes are, and perhaps if you looked at rich and poor athletes they would have the same relationship to slow and fast times as the new- vs old-shoe wearing athletes. Essentially, the variable what kind of shoe is the athlete wearing (qualitative with the two values new and old) is being confounded with the lurking variable how wealthy is the athlete. So the conclusion about causality "fancy new shoes make them run faster" might be false, and instead the real truth might be "wealthy athletes, who have lots of support, good coaches, good nutrition, and time to devote to their sport run faster."In short: experiments are better than observational studies in part because an experiment which find a causal relationship between two variables cannot be confounding the causal variable under study with a lurking variable.If, instead, we did an experiment, we would not have this problem. We would select athletes at random — so some would be weathy and some not — and give half of them the fancy new shoes (the experimental group) and half the old type (the control group). If the type of shoe was the real cause of fast running, we would see that in our experimental outcome. If really it is the lurking variable of the athete's wealth which matters, then we would see neither group would do better than the other, since they both have a mixture of wealthy and poor athletes. If the type of shoe really is the cause of fast running, then we would see a difference between the two groups, even though there were rich and poor athletes in both groups, since only one group had the fancy new shoes.
Often when doing a study, it is hard to get a true SRS. One thing that is done fairly frequently is to divide our target population into groups according to a characteristic that we think might have some influence on the subject of our study. Then we choose an SRS in each group, and combine the separate group SRSs (sometimes in a particular proportion, say by how large a fraction each group is in the whole population) into a sample for the whole. This is definitely not an SRS of the whole, but it may be what we think is a more "representative" way of picking a sample than a true SRS.
Example: Suppose we are doing a medical study of a condition which affects both men and women and we think gender might have an impact on the results of our work. Suppose also that the target population of patients is 60% women and 40% men. We might then pick an SRS of 6 women and an SRS of 4 men and combine them to give a sample of 10 people, which is definitely not an SRS of size $n=10$, but might be a sample we think is more representative of the target population.Experimental designs which use these sampling strategies based on "SRSs within blocks, then combine" strategy are called block designs.
Another non-SRS sampling strategy is to leave it up to the individuals in your target population to volunteer for your study. This is very common, and very hard to avoid, in practice, but produces absolutely terrible results. The reason is simply that the choice to volunteer may be due to some very significant lurking variable(s), and therefore the conclusions of the study are extremely unreliable. This kind of problem is called volunteer sampling bias (or sometimes voluntary selection bias or voluntary sample bias) and very easy to fall into, but almost always completely invalidates your conclusions.
Example: Some scheme where people are told at the end of a video or web page Click HERE if you think this material was useful, and HERE if not produces only garbage results, because of volunteer sample bias. Furthermore, if you offer a reward to viewers when they click on one or the other choice, or wait longer, so that you get twice as many volunteers as you did before, you are not getting results which are twice as reliable: both sets of results are equally meaningless, because of volunteer sample bias.