There’s More to the Story 729 Conclusion Statistical and Practical Significance The preceding results use a variety of different approaches, and they are all in agreement that the population does not have a mean body temperature equal to 98.6°F. Because there is sufficient evidence to warrant rejection of the common belief that the mean body temperature is 98.6°F, the results do have statistical significance. Based on the sample mean of 98.20°F and the preceding confidence intervals, the discrepancy between 98.6°F and the actual mean appears to be somewhere between 0.28°F and 0.55°F, and that difference is fairly substantial, so the difference does appear to have practical significance. Replication The preceding example led to the conclusion that the mean body temperature is not 98.6°F. Also, the preceding results suggest that the actual population mean is likely to be somewhere between 98.08°F and 98.32°F. These results are based on the 1992 study conducted by Mackowiak, Wasserman, and Levine. A more recent study conducted by Jonathan Hausmann et al. used smartphone crowdsourcing to collect a sample of 5038 body temperatures from 329 subjects. The subjects used a variety of different thermometers and an app to manually report their temperatures. The results from this 2018 study suggest that the population mean body temperature is 97.7°F. (See “Using Smartphone Crowdsourcing to Redefine Normal and Febrile Temperatures in Adults: Results from the Feverprints Study,” by Jonathan S. Hausmann et al., in Journal of General Internal Medicine.) Therefore, we don’t have agreement on the true value of the mean body temperature, but there is agreement that it is not 98.6°F. There is also agreement that there isn’t a single value that should be used for the “normal” body temperature. Body temperatures vary throughout the day, and they vary for different people. Where Did 98.6°F Come From? The value of 98.6°F came from the nineteenth century German physician Carl Wunderlich. He collected data from 25,000 patients by holding a foot-long thermometer in patients’ armpits for 20 minutes. His value of 98.6°F was used until the 1992 study by Mackowiak et al. Current thermometers are more accurate, and they are typically placed in the patient’s mouth instead of the armpit, so results are more reliable. Even though we now have better data showing that the value of 98.6°F is wrong, many people continue to believe that 98.6°F is the correct value. Key Takeaways The preceding analyses illustrate the different approaches that can be used. Here are some important takeaways for conducting an important statistical analysis: ■ Preliminary Analysis Start by questioning what the data represent, and identifying the goal of the study. Question the source of the data and the sampling method. Explore the data with relevant graphs, identify any outliers and important statistics. ■ Go Beyond the P-Value! It is wise to not rely solely on a conclusion suggested by a P-value. The P-value leads us to a binary conclusion equivalent to either rejecting the null hypothesis or failing to reject the null hypothesis, but a confidence interval can give us information about the size of an effect, and graphs There’s More to the Story
RkJQdWJsaXNoZXIy NjM5ODQ=