Science Buddies: "Ask an Expert"

My daughter has a bicuspid aortic valve so she did her science project to see if there was a significant difference in the amount of bloodflow out of a tricuspid aortic valve as opposed to a bicuspid aortic valve. She and her dad built 2 models of the heart. 1 with 3 valves and one with 2 valves. They measured how much water came out of each type of valve. These were the results. I have no idea how to calculate significance (even after looking it up on the internet).
She tested with 30 beats.
Experiment 1- the results are in ounces
Tricuspid 5, 3.2, 4.1
Bicuspid 3, 2.0, 4.0
Experiment 2
Tri 4.1, 4.1, 4.0
Bi 3, 2,3.8
Experiment3
Tri 4.5, 5.5.5.0
Bi 4.4, 5.0, 3.8
Your help would be appreciated.

My first question is are experiments 1, 2, and 3 different somehow or are they all the same, just repeated? I'm going to assume they are all the same, so all the tri data should be considered one data set and same for the bi data. That said, with nine data points per data set, statistical significance can be a bit dicey. Typically a much larger data set is desired for "statistical signficance". However, we have what we have. So moving forward. First thing I did was plot side by side, take the average, standard deviation, and coefficient of variation. This I believe provides enough info for guidance. Of note, the average of the tri data is 4.4 and therefore higher then the average of the bi data, 3.4. However the variability of the bi data is approximately twice that of the tri data, and both are really too variable anyway. So the standard deviation of the tri is 0.69 and the standard deviation of the bi is 1.0 which results in a coefficient of variation for the tri of 16% and the bi of 30%. These coefficient of variation values, as a general rule of thumb, should not exceed 10%. This means both data sets are too variable with the bi being almost twice that of the tri. All this means is you need to collect more data. I would also look at the testing process and attempt to figure out why there's so much variability in the data, making sure all tri results are tested in an identicle process as the bi data. Making sure the measurement instrument being used is appropriate in scale for the actual measurement being taken (ie, don't use a yardstick to measure the width of a hair...this actually happens unfortunately more often then one would think...just not necessarily with a yardstick ). Good luck and let me know how it goes!

Hello! An interesting project. I'm glad your daughter could find a topic with findings she can personally relate to.

While there are formulas to calculate statistical significance, the math is well beyond elementary school level math. Comparing the average between the bicuspid valve results and the tricuspid valve results will be fine. Once you calculate the averages, you must try to answer whether the result is significant. The difference between the averages is just under an ounce. In medical terms, is this significant? Hopefully, through your research or other experts here (I'm afraid I don't have knowledge in this area), you can find the answer.

Even if you were to do fancy statistical calculations, you would still have to answer the same question. Results sometimes are statistically significant but aren't significant within the field of study. Here, if pumping an extra ounce per heart beat has no physiological benefit, then the difference is not significant.

I hope this helps. Good luck!

Thank you, Nancy and Deana for your help. I think calculating and reportin on the averages is an excellant idea for this age level.

You're welcome! Good luck to your daughter!