--Original published at Ashley's Psyche
What were the strengths of the research design?
One of the strongest parts of the research design for the Coke vs. Pepsi experiment was the random assignment of the independent variable. The independent variable in this case was if the taste tester received Coke or Pepsi in each of their five cups, as this was able to be manipulated by the Fluid Dispensing Engineer. The random assignment of Coke or Pepsi in each of the cups was done by flipping coin and pouring the corresponding drink for when it was heads or tails. Due to this process, whether Coke or Pepsi was in each of the cups can also be referred to as a true independent variable, due to its ability to be randomly assigned, which lessens the chance of a confounding variable being present.
Another strength of the research design was the prevention of bias as a confounding factor by keeping the Fluid Dispensing and Randomization Team separate from the Data, Logistics, and Analyst Team. The separation of these two groups was crucial as it prevented the Fluid Dispensing and Randomization Team from making any faces or saying any comments that would influence the Executive Soda Analyst while he or she tasted each drink, which would result in a confounding variable.
What were the limitations of the research design?
Two of the limitations that stood out in the research design were that there was neither a random sampling, nor a random assignment in regards to the person who held the position of Executive Soda Analyst.
There was a lack of random sampling in regards to the Coke vs. Pepsi research design as people who drank the drinks were only chosen from Dr. MacFarlane’s General Psychology class that occurs at 2 p.m. on Mondays, Wednesdays, and Fridays. Due to this fact, the results of the experiment cannot be generalized to the entire population of soda drinkers, as one class cannot be representative of the “people” addressed in the research question.
As well, there was a lack of random assignment in regards to the Coke vs. Pepsi research design as the Executive Soda Analysts were chosen by volunteer. Dr. MacFarlane asked those who believed very strongly that they could tell the difference between Coke and Pepsi to raise their hands, and proceeded to pick those people to be the taste testers. These was not an effective way to assign students to that role as confounding variables could then influence the experiment and provide invalid results. The question for the experiment was “can people tell the difference between Coke and Pepsi,” not “can people who think they can tell the difference between Coke and Pepsi actually tell?”
What potential confounding variables were present in the study?
Potential confounding variables in the study could have been the amount of time it had been since the Executive Soda Analyst drank a Coke or Pepsi, the temperature of the soda, and the confidence level of the Executive Soda Analyst.
The amount of time it had been since the Executive Soda Analyst drank a Coke or a Pepsi can be considered a confounding variable as the taste tester who has not had Coke or Pepsi over a long period of time would struggle much more between the choose of both soda than a person who drink Coke, Pepsi, or both often.
The temperature of the soda can be considered a confounding variable as most people do not drink soda warm. The taste testers may know the taste of cold Coke or Pepsi better, and could therefore identify each drink easier than they did when it was warm, as it may have tasted different.
Lastly, the confidence level of the Soda Analyst can be seen as a confounding variable. Dr. MacFarlane chose only those that felt he most confident in their abilities to tell the difference between Coke and Pepsi, meaning that each of the tester may have made snap judgement due to their over confidence.
Was the conclusion we drew valid? Why or why not?
The conclusion we drew was not valid due to the limitations and confounding variables of the research design. Due to the lack of random sampling, none of this information can be generalized to a bigger population. As well, due to the presence of so many confounding variables, there would not be a high level of confidence in the results of the experiment.
What are 5 specific changes you would make to the research design to improve the study?
- Refrigerate each of the sodas so that they are cold for when the taste testers are drinking them.
- Take a random sample from the entire campus population and not just one psychology class.
- Randomly assign who will be a taste tester using a random number generator.
- Give the taste tester a palate cleanser after each drink.
- Have more than five trials.
