1.9: Step 5: Draw conclusions, make revisions.

Once the results of the critical experiment are in, they are pulled apart, examined, and analyzed. Researchers look for patterns and relationships in the evidence they’ve gathered from their experiments; they draw conclusions and see whether their findings and conclusions support their hypotheses. If an experimental result is not what you expected, that does not make it a “wrong answer.” Science includes a great deal of trial and error, and if the conclusions do not support the hypothesis, then you must revise your hypothesis, which often spurs you to conduct more experiments. This step is a cornerstone of the scientific method because it demands that you must be open-minded and ready to change what you think.

The results of the purse-snatching experiment were surprising. When the suspects (which, as you’ll recall, did not include the actual “criminal”) were viewed together in a lineup, the observers/witnesses erroneously identified someone as the purse snatcher about a third of the time. When the suspects were viewed one at a time, the observers made a mistaken identification less than 10% of the time.

In this or any other experiment, it does not matter whether we can imagine a reason for the discrepancy between our hypothesis and our results. What is important is that we have demonstrated that our initial hypothesis—“Eyewitness testimony is always accurate”—is not supported by the data. Our observations suggest that, at the very least, the accuracy of an eyewitness’s testimony depends on the method used to present the suspects. Based on this result, we might adjust our hypothesis to: “Eyewitness testimony is more accurate when suspects are presented to witnesses one at a time.”

We can then devise new and more specific testable predictions in an attempt to further refine our hypothesis. In the case of eyewitness testimony, further investigation suggests that when suspects or pictures of suspects are placed side by side, witnesses compare them and tend to choose the suspect that most resembles the person they remember committing the crime. When viewed one at a time, suspects can’t be compared in this way, and witnesses are less likely to make misidentifications.

Hypothesis: Echinacea reduces the duration and severity of the symptoms of the common cold. In the echinacea study, the results were definitive (FIGURE 1-13). Those who took the echinacea were just as likely to catch a cold as those who took placebo, and, once they caught the cold, the symptoms lasted for the same amount of time. In short, echinacea had no effect at all. Several similar studies have been conducted, all of which show that echinacea does not have any beneficial effect. As one of the researchers commented afterward, “We’ve got to stop attributing any efficacy to echinacea.”

Although it seems clear that our initial hypothesis that echinacea stops people from catching colds and reduces the severity and duration of cold symptoms is not correct, further experimentation might involve altering the amount of echinacea given to the research subjects or the length of time they take echinacea before exposure to the cold-causing viruses.

Hypothesis: Hair that is shaved grows back coarser and darker. In the hair-shaving experiment, the observers discovered that, after six months, it was impossible to distinguish which eyebrow had been shaved. This was not a surprise to dermatologists evaluating the study, because all of the living parts involved in hair growth are below the surface of the skin. (Plucking hairs can damage the root of the hair, potentially affecting future hair growth, but that is another story that awaits further study.)

The outcomes in all of these studies show that after the results of a critical experiment have been gathered and interpreted, it is important not just to evaluate the initial hypothesis but also to consider any necessary revisions or refinements to it. This revision is an important step; by revising a hypothesis, based on the results of experimental tests, we can explain the observable world with greater and greater accuracy.