A truly “representative” sample will determine if a scientific survey or research study is going to be considered valid or reliable. However, getting a true random sample is typically not really likely or affordable especially when studying human behavior.
Despite this, by the time the data is taken from the so-called random sample, manipulated then reduced to numbers, the results often may begin to take on an aura of conviction, especially in the way it is be reported to the public. In sampling procedures the constant battle is against sources of conscious and unconscious bias and the struggle NEVER ends.
Interviewers may “shade the responses” they get, to stretch, fit or pigeonhole the data or “sought after” results into some category. Those being interviewed do not always say what they are REALLY thinking, or would like to say, but shall not.
In scientific random sampling procedures, biases may go unchecked. In sampling procedures there is no absolute form of checks and balances. There is no overriding accountability in collecting a sample, and there is no need to really even “rig” or “slant” a sample one way or another, because “bias” is inherent in ANY sampling procedure due to the limited time and resources researchers have available when collecting data.
Researchers fight a losing battle every time they collect data but, in the end, may always try and come across like winners. Is there ever any wonder why two identical studies of human behavior or public opinion polls can come up with two totally different results or practically identical results. Either their sampling procedures are flawed from the outset or their “unconscious” hidden political agendas or biases are possibly out to get the same results especially in studies or surveys of human behavior.