There are measures that must be carried our properly in order to create a reliable research design, as you cannot rely on statistical analysis to compensate for major design flaws. Samples are primarily used to make inferences about a population, but you must ensure it is valid. If not, then you will be at risk of lowering your ecological validity, which means you cannot generalise your results to you target populaiton. For instance, selecting a sample of children from one school and then generalising their IQ scores to every child in the world is bonkers. I am sure you clever people can see why it is! So, how do we prevent this form happening? Well, randomly sampling is a common technique used to help increase reliability. This involves randomly choosing participants from a sampling frame, such as doctors lists and records. However, you must ensure that these do not lead to you misleading results because some lists can be out of date or inaccurate.
The use of clinical trials help solve many of the problems i have just stated. In order to use this method, the design must be concrete as each condition must be rigidly controlled so that they can be reliably compared. This meant that research under clinical trials often experience ethical problems. This is because in order to test one medical treatment over another and placebo effect, one of the groups will subject to uneccessa risk. However, some people argue that it is unethical not to use this method for new treatments, as we would never know the true outcome.
Comparing new treatments with placebos is another major design flaw, as they should be comparing it with a pre-existing treatments. Any significant result will be due to the fact that the other group was a placebo, thus not really showing us anything.
I have only briefly talked about some of the major design flaws when concerning clinical trials, but overall, the main issue I have with design flaws are the ethical problems. The researchers must ensure they have the best interest in mind for the participants as well as making sure the results they produce are reliable.
psuca2 
http://www.apa.org/ethics/code/index.aspx?item=3 sums up what im about to say in case your bored and want to read something. integrity is a major issue in research, psychology’s must seek to be honest and accurate when dealing with research, if your participants dont trust you, your data is void and if researchers dont trust other researchers then this will just stunt scientific progress from one and others ideas.
I found your blog very interesting and its very true, design flaws lead to bigger issues such as ethics, and this is why we must get our designs right from the start. Although you have covered a lot i your blog I feel i could add to your point on placebo testing.
You have mentioned that placebos should be compared to preexisting treatments, the severity of this issue is actually quite deep as it leads individuals to believe an improvement in treatment when this is not what is being tested. What clinical trials should tell us is whether a new treatment is better than the current standard of care. this means we should be comparing new treatments to previous treatments, and a placebo condition should always be an ‘extra’. as it is to show if the idea of a drug works as treatment, but is unethically interpreted as a way of showing how effective noneffective drugs are, this is a clear example of Type 1 error.
Testing placebo conditions agains new treatment conditions does have a use that is highly ethical and contribute to society. Once a new treatment is viewed as less effective than the current treatment, testing on a placebo can say how much greater the drug is than the placebo if this is sufficient it can be justified to use this drug regardless of its standing among the other drugs of its kind.