I’m a fan of there being many experiments, but I might be biased by my background in meta-analysis. Many good experiments are, of course, better than many poorly designed and/or executed experiments, but replication is important, even in good experiments. Even carefully controlled experiments have the potential of error. Also, having many experiments usually is a better test of the generalizability of the findings. Finally, having many experiments coming out of many different laboratories (independent of each other) increases confidence that the findings are not the result of the investigator’s preference for what the results should be. If there is conflict in findings it might be poor study design and/or execution or it might be that the field is missing something important about the truth.
I’m a fan of there being many experiments, but I might be biased by my background in meta-analysis. Many good experiments are, of course, better than many poorly designed and/or executed experiments, but replication is important, even in good experiments. Even carefully controlled experiments have the potential of error. Also, having many experiments usually is a better test of the generalizability of the findings. Finally, having many experiments coming out of many different laboratories (independent of each other) increases confidence that the findings are not the result of the investigator’s preference for what the results should be. If there is conflict in findings it might be poor study design and/or execution or it might be that the field is missing something important about the truth.