Randomized Comparative Experiment

From GM-RKB
Jump to navigation Jump to search

A Randomized Comparative Experiment is a randomized assignment comparative experiment that follow a randomized controlled experiment design (with randomized controlled experiment treatment group(s) and a randomized controlled experiment control group(s)).



References

2021a

  1. Chalmers TC, Smith H Jr, Blackburn B, Silverman B, Schroeder B, Reitman D, Ambroz A (1981). “A method for assessing the quality of a randomized control trial". Controlled Clinical Trials. 2 (1): 31–49. doi:10.1016/0197-2456(81)90056-8. PMID 7261638.

2021b

2021c

2018

2016

  • Amy Gallo. (2016). “A Refresher on Randomized Controlled Experiments.” In: HBR, MARCH 30, 2016
    • QUOTE: ere are the basic steps:
      1. Decide what your dependent variable of interest is (remember there might be more than one). In our oil well example, it’s the speed or efficiency with which you drill the well.
      2. Determine what the population of interest is. Are you interested in understanding whether the new bit works in all of your wells or just specific types of ones?
      3. Ask yourself, What is it we’re trying to do with this experiment? What is the null hypothesis­ — the straw man you’re trying to disprove? What is the alternative hypothesis? Your null hypothesis in this case might be, “There is no difference between the two bits.” Your alternative hypothesis might be, “The new drill bit is faster.”
      4. Think through all of the factors that could spoil your experiment — for example, if the drill bits are attached to different types of machines or are used in particular types of wells.
      5. Write up a research protocol, the process by which the experiment gets carried out. How are you going to build in the controls? How big of a sample size do you need? How are you going to select the wells? How are you going to set up randomization?
      6. Once you have a protocol, Redman suggests you do a small-scale experiment to test out whether the process you’ve laid out will work. “The reason to do a pilot study is that you’re most likely going to fall on your a**, and it hurts less when it’s called a pilot study,” he jokes. With an experiment like the drill bit one, you may skip the pilot because of the cost and time involved in drilling a well.
      7. Revise the protocol based on what you learned in your pilot study.
      8. Conduct the experiment, following the protocol as closely as you can.
      9. Analyze the results, looking for both planned results and keeping your eyes open for unexpected ones.

2011

2003a

2003b

Figure 1: The randomised control trial.