ICOTS8 (2010) Invited Paper Utts In C. Reading (Ed.), Data and context in statistics education: Towards an evidence-based society. Proceedings of the Eighth International Conference on Teaching Statistics (ICOTS8, July, 2010), Ljubljana, Slovenia. Voorburg, The Netherlands: International Statistical Institute. www.stat.auckland.ac.nz/~iase/publications.php [© 2010 ISI/IASE] UNINTENTIONAL LIES IN THE MEDIA: DON’T BLAME JOURNALISTS FOR WHAT WE DON’T TEACH Jessica Utts Department of Statistics, University of California, United States of America jutts@uci.edu It’s easy to find misleading and even harmful reporting of statistical results. For example, a 2008 study titled “You Are What Your Mother Eats,” asserted that children born to mothers who eat breakfast cereal are more likely to be boys than are children born to mothers who do not eat breakfast cereal. A 2009 analysis by statistician Stan Young and colleagues showed that the result was almost surely a false positive, but by then the study had gained widespread media attention. Many students who take introductory statistics come away from the course able to compute a standard deviation, yet unable to spot an egregious example of poor statistical reporting such as the one illustrated by this example. We are doing an inadequate job of educating the next generation of medical researchers, journal referees, policy-makers, journalists, and so on. I will discuss some ways we can do a better job. INTRODUCTION The majority of students in an introductory statistics course in college will never take another statistics course. Yet they will need to use statistical information to make decisions throughout their lives. Some of them will enter professions such as journalism or medicine that require them to inform others based on statistical information. Therefore, those of us who teach introductory statistics courses have an opportunity to change peoples’ lives, and the lives of those with whom they will come in contact, for years to come. But most of us are not taking full advantage of this opportunity. Although there have been positive changes in the way introductory statistics courses are taught, there is so much more we could do. We can’t do it unless we are willing to give something up, but the trade-off is worth making. It is easy to find examples of misleading reporting of statistical studies and to criticize the media for them. But I speculate that most of the journalists who wrote those stories took an introductory statistics class as part of their education. What were they taught? Most likely, they came out of the course knowing how to construct a histogram and compute a standard deviation, but never being exposed to issues like the problems with multiple testing, how to know when a causal connection can be made, why it is important to consider baseline risk when assessing relative risk, and a variety of other topics that are critically important to decision-making in daily life. We are doing the world a disservice by clinging to the teaching of topics that only a small subset of our students will ever need to know, while ignoring topics that would benefit the vast majority of them. It is not enough to separate “statistical literacy” into a course of its own. Every student who takes one statistics course should first and foremost be learning the statistical ideas they need to make informed decisions in daily life. SOME TOPICS WORTH LEARNING There are certain statistical ideas that all educated citizens should understand. In Utts (2003) I discussed seven such topics, and in this paper I repeat one of them (because it is so important and so often misunderstood), and I introduce three additional ones. For some of these topics students will need to understand very basic probability, for others they will need to understand the fundamentals of hypothesis testing. But some of these topics require no background beyond what is covered in primary and secondary school. What Educated Citizens Should Know about Statistics and Probability The topics covered in Utts (2003) include:  Unwarranted causal connections based on observational studies,  Statistical significance versus practical importance, especially for large studies,