Yay science posts are back! Today we are (finally) going to start that daunting question about how to read about science by talking a little about experimental design and what to look for when trying to read about a science topic in the news. There’s a lot going on here so take your time and leave me a comment if something didn’t make sense or you want to know more about something.
Let’s start at the beginning. All science starts with a question, such as “Why are cats afraid of cucumbers?” Then, in order to start answering the question, the scientists have to come up with a hypothesis – their educated guess for an answer. For example, “Cats are afraid of cucumbers because they are green.”
Now comes the tricky part. Scientists have to design an experiment that directly tests their hypothesis. This part is tricky because there are always a ton of potential answers and scientists need to figure out how control their experiment so that solely it tests their hypothesis and doesn’t bring any other factors into the mix. For example, if we wanted to test whether cats hate green, we’d want to control our experiment so we wouldn’t accidentally be testing the cats’ response to different shapes or smells.
Designing a good experiment is really complicated. It’s made even worse by the fact that the very systems that some scientists study are filled with differences. For example, all humans share more than 99% of the same DNA but think about how unique we all are (even identical twins who have exactly the same DNA). The term we use for this phenomenon is called “heterogeneous” and scientists are finding to this day that organisms with the exact same DNA can act completely differently from each other. So with all of this crazy heterogeneity in mind, another way scientists can be cautious about designing experiments that solely test their hypothesis is to replicate the experiment a lot or test multiple subjects (cats, people, bacteria, etc).
Replicating an experiment is really important. For example, if I put a cucumber behind my cat and she doesn’t freak out, can I really conclude that all cats are not afraid of cucumbers? Let’s add some replicates in there! I could put a cucumber behind my cat 10 days in a row and then determine if she continues to stay nonplussed by the cucumber. I could also try putting a cucumber behind my cat at different times of the day to determine if it depends on the time of day. Or I could put cucumbers behind a variety of cats to determine if my cat is just weird and likes cucumbers. All of these ideas would add replicates to my experiment and help me identify if my results are just a weird fluke associated with some other factor that I don’t care about or if they are directly related to my hypothesis.
Good science experiments have established controls and include large numbers of replicates to eliminate “weird flukes.” All of these factors should be listed in the original scientific paper describing the study but these papers are often incredibly dense and hard to follow (even for fellow scientists). However, a good science report or article written for the general public should also list these qualifications. So to test the quality of a good source, I like to see what an article says about controls and replicates.
Here is a fake article that I just made up:
Scientists determine that too much sleep causes cancer.
Scientists at Questionable Science University have completed a study about sleep and cancer. They interviewed two different people who have lung cancer and found that they sleep 7 hours every night. As these data clearly show a link between too much sleep and cancer, people should sleep no more than 6 hours a night to prevent cancer.
Yikes! Does this mean we should stop trying to get a solid 7-8 hours a night?
Well, it looks like these scientists talked to two people who already have cancer. I want to know more information about the people who were interviewed for this study. Did the scientists take care to control for other variables like age, race, or gender? What else do these people have in common (i.e. do they smoke? Do they exercise? What type of food do they eat? Are they the same age?). All of these questions could have affected their results in a way that disconnects sleep from cancer. Furthermore, they didn’t talk to anyone who doesn’t have cancer (this is called a negative control and is INCREDIBLY IMPORTANT). Plus they only asked two people! That’s like me saying cats aren’t afraid of cucumbers because my cat isn’t! It doesn’t look like they have any controls or replicate the experiment so maybe this isn’t a great source after all. Yay! Time to get more sleep! -_-
I like to think of science as “a quest for the truth.” Good experimental design is hard but it’s worth it because it helps scientists get closer to finding out the truth! It’s really important to make sure your sources report on good science so that you can learn about the truth! I tried to give you some tools that let you sift through some scientific topics you are interested in so let me know if they help! Go practice the game and then report your findings back to me! :D
Now it’s your turn: tell me about the science you are the most interested in learning about. Or tell me about something completely unscience related. That was a lot of science for one day…