Now that you have a better understanding of how to look at graphs in scientific literature, let’s delve more into how to actually interpret those results (make conclusions – claims – based on the evidence). The purpose of this exercise is to make you think critically about graphical results. Most of the results you will find in scientific papers are reliable and tell a story, but it is important to think critically about all scientific literature.
Read through this short link, and watch the video if you like: Correlation https://whydoscientists.org/cause-correlation/
Then, check out this website about correlations, and find one that you like to share and discuss in your initial post: http://www.tylervigen.com/spurious-correlations
Read the following before answering the discussion questions for this week:
This is an excerpt from a scientific article. I have included the introduction, methods, and results. Your job is to provide the discussion – the INTERPRETATION of the results!
Thomas Ruder and colleagues discover that once again the sleuths hunches were right
There has been murder done, and the murderer was . . . more than six feet high.
In Sir Arthur Conan Doyles A Study in Scarlet, Sherlock Holmes and his companion Dr John Watson find themselves at the scene of a homicide: a mans body has been discovered in an abandoned house, with the word RACHE (German for revenge) written in blood on a nearby wall. Holmes reads footprints in the dust, scrutinises little heaps of tobacco ash with his magnifying glass, and takes a close look at the bloody inscription. Based on his observations he states:
[T]he murderer . . . was more than six feet high.
Flattered by Watsons amazement, Holmes explains:
[T]he height of a man . . . can be told from the length of his stride [and] I had this fellows stride both on the clay outside and on the dust within.
With this knowledge, Holmes was ahead of his time: it was not until the late 20th century that a group of forensic scientists rediscovered the correlation between height and stride. However, they found that the length of a stride depends not only on someones height but also on walking speed. Holmes, of course, knew about the limitations of his method and applied a second technique to confirm his estimation: When a man writes on a wall, his instinct leads him to write about the level of his own eyes. Now that writing was just over six feet from the ground. It was childs play.”
In true Holmesian fashion we set out to determine whether there is a correlation between writing level, eye level, and height. Does instinct really lead people to write about the level of their own eyes, and can a writers height be calculated from an inscription on a wall?
Each of 100 volunteers (46 men, 54 women), aged 19-66, was asked to write the word RACHE on a large glass window using a non-permanent marker (Whiteboard Marker 1741 Velleda, Bic, Clichy, France). The study supervisor used a standard retractable tape measure (Lux Basic, Lux-Tools, Wermelskirchen, Germany) to record the height of the writing (to the centre of the first letter), the font size (height of the first letter), and the volunteers height and eye level. Measurements were rounded up or down to the nearest centimetre. At the end of the experiment, the study supervisor cleaned the inscriptions off the window. No windows were permanently blemished during this study.
We used statistical software (SPSS 17.0, IBM, Chicago, USA) to assess the normality of distribution of all variables and to calculate correlation coefficients and significance levels. P values less than 0.05 were considered significant.
The volunteers mean height was 175.1 (SD 8.9) cm, mean eye level was 164.1 (SD 8.3) cm, and mean writing level on the wall was 163.3 (SD 7.5) cm)less than 1 cm below eye level (mean difference 0.7 (SD 7.0) cm). Font size ranged from 3 cm to 33 cm (mean 8.5 (SD 4.0) cm).
A strong correlation was found between the volunteers height and both eye (r=0.99, P<0.001) and writing level (r=0.58, P<0.001) as well as between writing level and eye level (r=0.61, P<0.001). A scatter plot depicts these relations. The interaction of the regression lines illustrates an additional finding, that shorter than average volunteers showed a tendency to write above eye level and taller than average volunteers to write below eye level.
Font size was unrelated to the other variables and did not correlate with body height (r=0.00), eye level (r=0.00), or writing level (r=0.24).
1. Two variables are correlated in the results of a scientific article. What other information might you need to know if one causes the other? Explain, using examples if you wish.
2. From the spurious correlations website, add the graph that you found the most interesting and discuss what the data suggest, What interested you about this graph? if one variable caused a change in the other variable. Does this interpretation of the data seem reasonable? Explain.
3. To determine FOR SURE whether your spurious correlation graph indicated causation, what other information (evidence) would you need?
4. The article shown above includes a figure representing the results of the experiment, which are also described in the text.
(a) Write an assertive figure legend for the graph.
(b) Explain whether these data support or refute Sherlock’s hypothesis, and why these data provide evidence for your claim.
(c) Are you convinced that eye level height is a direct cause for writing level height? If so, what convinced you? If not, What other experiments would you suggest for researchers to perform in order to provide the necessary evidence?