TL; DR- Effect size interpretation requires more than numeric interpretation of the effect size. You need to think about what would be considered a big deal, real-life change worth pursuing, given the real-world implications for your data. For example, there is a  malaria vaccine with a 30% success rate undergoing  a large scale trial in Malawi . If you consider that many other vaccines have much higher success rates, 30% seems like a relatively small "real world" impact, right? However, two million people are diagnosed with malaria every year. If science could help 30% of two million, the relatively small effect of 30% is a big deal. Hell, a 10% reduction would be wonderful. So, a small practical effect, like "just" 30%, is actually a big deal, given the issue's scale. How to use this news story: a) Interpreting effect sizes beyond Cohen's numeric recommendations. b) A primer on large-scale medical trials and their ridiculously large n-sizes and tra...

This is a remix of a bunch of posts. I brought them together because they fit a common theme: Examples that use actual data that researchers collected but still manage to lie or mislead with real data. So, lying with facts. These examples hit upon a number of themes in my stats classes: 1) Statistics in the wild 2) Teaching our students to sniff out bad statistics 3) Vivid examples are easier to remember than boring examples. Here we go: Making Graphs Fox News using accurate data and inaccurate charts to make unemployment look worse than it is. Misleading with Central Tendency The mean cost of a wedding in 2004 might have been $28K...if you assume that all couples used all possible services, and paid for all of the services. Also, maybe the median would have been the more appropriate measure to report. Don't like the MPG for the vehicles you are manufacturing? Try testing your cars under ideal, non-real world conditions to fix that. Then get fined by the EPA. Mis...

This example from Retraction Watch illustrates how NOT to do research. It is a study that was accepted and retracted from Frontiers in Public Health. It purported to find a link between childhood vaccination and a variety of childhood illnesses. This would be a good case study for Research Methods. In particular, this example illustrates: 1) Retraction of scientific studies 2) The problems with self-report surveys 3) Sampling and trying to generalized from biased samples 4) What constitutes a small sample size depending on the research you are conducting 5) Conflict of interest This study, since retracted, studied unvaccinated, partially vaccinated, and fully vaccinated children. And the study found " Vaccinated children were significantly less likely than the unvaccinated to have been diagnosed with chickenpox and pertussis, but significantly more likely to have been diagnosed with pneumonia, otitis media, allergies and NDDs (defined as Autism Spectrum Disorder, Attenti...

Refutation to Anti-vaccination Memes  came up with this excellent illustration to explain why anti-vaxxers shouldn't claim a "win" just because more vaccinated people than unvaccinated people get sick during an outbreak. This example has a bit more credence if paired with actual immunization rate/infection rate data. For instance, in a case when an outbreak has occurred, and most infected are immunized, but there were still some un-immunized individuals. To further this case, yes, most people in America are immunized . However, here  is an example of an outbreak that has been linked to un-vaccinated folks. How to use it in class: -Base rate fallacy (which DOES matter when making an argument with descriptive stats!) -Relative v. absolute risk. -Making sense of and contextualizing descriptive statistics.

The folks at Wall Street Journal took CDC disease data (by state, by year, courtesy of Project Tycho ) as well as information on when various vaccines were introduced to the public. And the data tells a compelling story about the importance of vaccinations. Below, the story of measles. How to use in class: -Using archival data to educate and make a point (here, vaccine efficacy) -Visualizing many data points (infections x state x year) effectively -Interactive: You can cursor over any cube to see the related data. Below, I've highlighted Pennsylvania data from 1957. -Since you can cursor over any data point to see the data, you can ask your students to pull data for use in class. -The present data were drawn from Project Tycho , a University of Pittsburgh initiative to better share public health data. This resource may be useful for your classes as well. -This data is good for Stats class, as well as Developmental, Health, Public Health, etc.

This report from Pew  highlights the differences in opinions between the average American versus members of the American Association for the Advancement of Science (AAAS). For various topics, this graph reports the percentage of average Americans or AAAS members that endorse each science related issues as well as the gap between the two groups. Below, the yellow dots indicate the percentage of scientists that have a positive view of the issue and the blue indicate the same data for an average American. If you click on any given issue, you see more detailed information on the data. In addition to the interactive data, this report by Funk and Rainie summarizes the main findings. You can also access the original report of this data  (which contains additional information about public perception of the sciences and scientists). This could be a good tool for a research methods/statistics class in order to convince students that learning about the rigors of the scientif...