Why is it important to be able to reproduce results?

Answer 1

Let's say a drug company claims that "90 out of 100 people" respond well to a certain medication and "only 1 person out of 100 people reported a minor side effect. Statistics are used to extrapolate to the larger population. So this drug company claims that 90% of the world's population would benefit from this drug. But when the company releases the drug to the market, 50% of the people who received a prescription reported severe side effects. Obviously, the original statistic was wrong. The drug was only tested on 100 people; this small size cannot represent and extrapolate to a much larger population.

So let's say you tested whether 5 potato slices with skins would sprout "eyes" and whether the sprouts could be planted in the ground to grow more potatoes. You conclude that 4 out of 5 potatoes do sprout-- and 3 successfully produced potatoes after being planted. If I can reproduce the steps you used, it would give reliability and credibility to your study or experiment. However, if you didn't say you put the potato slices with skins in water plus sugar and placed the potatoes in direct sunlight for 3 hours each day for 10 days (e.g. state exactly what you did and the conditions you set up), another person might set up the experiment in a different way and get totally different results. For example: I put my potatoes with skins in 1-inch of buttermilk and under a lamp but only on 4 days-- so, none of my potatoes grew sprouts... I conclude your experiment wasn't helpful or was even invalid because I didn't get the same results. But---if I repeat your steps exactly, I should get the same results, which lends credibility to your experiment.

If a study is well-documented, but still produces different results, that helps define both the study's steps and outcomes. The next person can change the variables to make a different study.

Answer 2

If you can't reproduce results, you won't know if you've gotten it right or if it was just a lucky fluke.