Of the primary data collection techniques available to the researcher (survey, observation, experimentation), why is the experiment

Question:

Of the primary data collection techniques available to the researcher (survey, observation, experimentation), why is the experiment the only one that can provide conclusive evidence of causal relationships? Of the various types of experiments, which type or types provide the best evidence of causation or non-causation?