Friday, February 5, 2010

What is a quasi experiment and a true experiment? and what is the difference? ?

I dont know what these are! can someone please explain them to me..and give me examples? i don't understand the examples online ahha they're too hardcoreWhat is a quasi experiment and a true experiment? and what is the difference? ?
This is a question that comes up a lot in experimental design lessons. The difference is in the independent variable. Remember that the independent variable is the factor you predict will cause a change in your dependent variable.





A true experiment will randomly assign individuals to separate conditions or levels of an independent variable or combination of variables to see what effect that has on the dependent variable. The key is in the random assignment. In a true experiment anyone could be placed in any of the conditions or levels. Due to the random assignment you can draw stronger conclusions about the causal nature of the relationship between your independent and dependent variable.





On the other hand, a quasi experiment has all the same elements except for random assignment. In a quasi experiment the independent variable is something that already exists in the population (e.g., age, gender, eye color). In this way it is similar to an observational design where two naturally occurring variables are measured and correlated. The quasi independent variable could be continuous (e.g., age) or categorical (e.g., gender). This is not a true experiment because we can never be sure that the quasi independent variable is causing the change in the dependent variable or if another third factor can explain the difference if it exists.





Let's consider a silly example of puppy ownership. I have a hypothesis that girls like puppies more than boys. I find a sample of 50 girls and 50 boys, then measure the number of puppies each group owns. I find that boys own significantly more puppies than girls. From this I can conclude that gender predicts, is associated with, or is consistent with boys preferring puppies more than girls. You could insert any naturally occurring, quasi independent variable in this example (e.g., body weight, height, etc).





One more example...I have a hypothesis that Life Saver preference causes increased puppy preference. I could make this a quasi experiment and measure Life Saver preference and the number of puppies owned, but I could also make this a true experiment. I randomly assign 66 people who were selected at random to a group that eats no Life Savers for 1 year, a group that eats 1 roll of Life Savers each month and a third group that eats 1 roll of Life Savers each week. I find that the more Life Savers eaten leads to increased puppy ownership. Because I placed people into the three conditions at random I can more strongly concluded that eating Life Savers is what is causing the difference in puppy ownership.What is a quasi experiment and a true experiment? and what is the difference? ?
In my experience, whenever you see or hear something say ';quasi';, then it is bull$hit. Basically, they legally cannot prove what they are saying, but if they could prove something using their ';quasi experiment'; or ';quasi science';, then their claim, whatever it may be, whether it is getting a bigger X in Y days, or losing Z amount of pounds without any health risks, would be true.





Just a good way for people to skim over the truth.
True experiments must sport both random selection and random assignment of participants.





Quasi-experiments typically lack one of these two aspects.





~Dr. B.~

No comments:

Post a Comment