Fish article Q1. The article reports that fish sold at restaurants and super markets are often mislabeled. This leads to health concerns and fraud against consumers. In the end the article does warn that the sampling was not done properly and the source of the mislabeling is undetermined. Q2. http://oceana.org/sites/default/files/reports/National_Seafood_Fraud_Testing_Results_FINAL.pdf +0 if they did not find the original study Q3. Yes. Overall the article did explain some limitations and biases of the study. The article quoted an expert to state the sampling did not follow proper sampling methods and how the data from Boston might be biased. Q4. The 94% fraud tuna from New York really stood out! This statistic seems too large to be true. The article was trying to show the magnitude of the issue by using this extreme statistic as a specific example. - Students could have also discussed the "17/28", "1/3 of the 1215 sample", or "52% fish tested" statistics as long as they explained their choice well. Q5. The study reported the same statistic but the statistic was based on a sample size of 18 non-representative samples. The study itself did not warn against the small sample size so technically the article was faithful but the generalization from 18 (non-representative) samples to the whole NY seems unfair and exaggerated. This falls under the "Little Figures that are not there" in the text How to Lie with Statistics. Comment: The sample was not chosen randomly. Therefore, it is hard to know whether the sample was representative or not. As a result, these statistics might or might not reflect the actual rate of mislabelled fish. Comment: There's another interesting statistical error in the article (and the study). The article says that Southern California was the worst -- worse than any other region, including New York. To be fair to the reporter, the study made the same claim. However, when we look at the details, this conclusion looks awfully dubious. If you look at the detailed figure for Southern California (p.38 of the report) and compare it to the detailed figure for New York (p.47 of the report), in each category the rate of mislabelled fish is approximately the same for both regions: 32% vs 29% for grocery stores, 39% vs 39% for restaurants, and 84% vs 76% for sushi. The only reason that the overall number for Southern California is so much worse than for New York is because in Southern California they sampled more sushi places (36% of the sample in SoCal, vs 18% in NY), whereas in New York they sampled more restaurants (38% of the sample in SoCal, vs 63% of the sample in NY), and mislabelling is more common in sushi restaurants than in restaurants. In other words, there is no evidence that mislabelling occurs more often in Southern California; all we have is evidence that the researchers chose their samples differently in the two regions. Subway article Q1. New York city officials are worried about the 6 subway deaths so far in 2013 and are calling on the Metropolitan Transportation Authority to investigate and come up with solutions. The article says that this recent spike in subway deaths has the city on pace to more than double last year's death toll of 55. Q2. http://online.wsj.com/article/SB10001424127887324329204578270292806234834.html (many other sources are possible, +1 for any other source) Q3. No. There was no original study for the information in the CBS news article, but the article makes it seem like subway deaths are a major problem. Other internet sources, like the Wall Street Journal article I found, show that subway deaths remain extremely rare. The CBS article only provides the number of deaths; it doesn't give the reader a sense of how many people ride the New York subway system, which would show that the deaths are rare events. Q4. The article says that there have been 6 subway deaths in 2013 so far, and that this recent spike in subway deaths has the city on pace to more than double last year's death toll of 55. The article uses this statistic to make it seem like subway deaths are a much bigger problem than they were last year, and that increased safety measures are therefore needed. Q5. The article is very misleading when it says that the number of subway deaths are on pace to double in 2013. Possible responses (not all are needed) - Just because there is a cluster of deaths in January doesn't mean that the rate of subway deaths will actually go up in 2013. - The article did not say anything about the total number of riders in the New York subway system. If the article had included these numbers, they would show that subway deaths are extremely rare. - The article does not include much information about deaths in prior years and their causes (i.e., suicide, falling onto tracks, etc.). This contextual information would help the reader figure out whether the safety measures being considered are really needed. - Of course, the number of deaths in any given month will vary from month to month. Because of how randomness works, sometimes it will be higher than average, and sometimes it will be lower than average. The fact that there were an above-average number of subway deaths in January 2013 is not really evidence that the problem is getting a lot worse; the above-average number for January could have arisen due to random chance.