Dear Tipsters, while reading the article Mike pointed us to (http://www.nature.com/mp/journal/vaop/ncurrent/full/mp201448a.html) I stumbled upon an interesting diagram which shows a nonlinear relationship between gestational age and testosterone concentration (Figure 7 in the supplemental material, http://www.nature.com/mp/journal/vaop/ncurrent/extref/mp201448x7.jpg).
It would have come very handy as an example of a quadratic relationship, but upon closer inspection: The graphs show a peak testosterone concentration e.g. for males at ca. 15.2 weeks (ca. 0.8 nmol/L), but the corresponding regression equation sets the maximum at 0.54/0.02= 27 weeks (ca. 3.97 nmol/L). Also, curvature of the parabolas in the graph seems different, while the parameter in the equation is identical (ok, could be roundoff-error). So the regression equations in the diagram do not seem to correspond to the graphs. Maybe I miss something obvious (has happened before), but right now I see no simple way to reconcile these two separate descriptions of the same data. So if you teach statistics or methods, this could perhaps be used as an exercise in critical reading. Kind regards, Rainer --- You are currently subscribed to tips as: arch...@jab.org. To unsubscribe click here: http://fsulist.frostburg.edu/u?id=13090.68da6e6e5325aa33287ff385b70df5d5&n=T&l=tips&o=37044 or send a blank email to leave-37044-13090.68da6e6e5325aa33287ff385b70df...@fsulist.frostburg.edu