How can we be sure that we are seeing real patterns in data and not merely seizing upon that which meets our biased prior expectations?
Can We Crack the Popeye Statistical Challenge?
There is a widely accepted story that spinach growers in Crystal City Texas erected a statue of Popeye in 1937 in recognition of the comic character’s creator E. Segar increasing spinach consumption in the USA by 33 per cent at a time when farmers were hard pressed. The statue still stands in Crystal City but can the story stand up to statistical analysis?
It has long been noted that humans see patterns in data and attribute them to causes. Hence people see a man in the moon, or Kate Middleton's face on a jelly bean, and concoct conspiracies from random data.
It is hypothesised that we are pattern seeking animals and that this pattern seeking was essential for human progress and survival (Shermer, 2007 p. xxiv). Seeing patterns in random data is compounded by the fact that people have a tendency to try to prove their current hypothesis is correct by picking out only examples that confirm it, while ignoring all that do not (Sutherland, 2007. p. 99). Availability error (Sutherland 2007) may lead us to focus upon only spinach production data for 1936 – the year when Popeye was a box office smash in the cinema. Then there is the problem of Mertonian self-fulfilling prophecies, which have a general tendency to convert little effects into big effects (Gilovich 1991. p.6) which means we need to consider also whether spinach farmers might have planted and harvested more spinach in anticipation of increased spinach demand driven by Popeye.
To gather data with an aim to check out the Crystal City Popeye statue story, in 2010 I wrote to every single United States Department of Agriculture station in the USA asking for any historical data they had for annual spinach production.
The data are presented below in Table 1. As my simple annual per cent change analysis reveals, it turns out that Texas did indeed have a 33 per cent increase in production (not consumption) in 1936.
Was Popeye’s spinach eating the cause of that increase? Only the right statistical analysis can tell.
Can we tell from examining all the historical data whether this is likely?
Set out below is the data I collected form the USDA and briefly analysed by way of simple percentage change analysis.
What statistical method is best suited for helping us decide whether this data is random or whether there is a pattern related to Popeye? Knowledge in this area might help to inform what we currently know about effective media use in bringing about nutritional attitude change.
I will of course publish an attribute to whoever helps to crack the Popeye Challenge and will include it in a book I am writing on Dysology.
Additional possibly relevant information:
· 1930 - 36 the dustbowl years focused on the panhandles of Texas and Oklahoma, and adjacent parts of New Mexico, Colorado, and Kansas. The Winter Garden area where spinach was grown would have been relatively safe.1930 Texas and Oklahoma, and adjacent parts of New Mexico, Colorado, and Kansas.
· 1930 something unusual happened in 1930 to significantly reduce spinach production in Texas - but it continued to rise in other states.
· 1931 Popeye strip cartoon reveals for the first time that spinach is the secret of Popeye’s super powers.
. 1932 (December) Popeye delivers a political message in comic strip "If ya don't give the farmers relief, I'll knock ya all out from in between yer ears and lay ya among the swee' peataters like nobody's business!
· 1936 FDR’s New Deal introduced. Soil Conservation Act encourages farming other then wheat, rice, cattle in favour of non soil eroding crops. This may have had some impact. USA also introduced Agricultural Adjustment Act to control supply of 7 basic crops (not including spinach) - may have led to more supply of spinach as farmers encouraged togrow more diverse crops.
· 1936 - First two reel colour movie Popeye the Sailor V Sinbad the Sailor. Either posted alongside or above as the main feature at cinemas.
· 1936 Last year of the dustbowl that began in 1930
· 1936 Largest spinach crop for all states.
· In Texas for the 5 YEARS BEFORE Popeye even ate spinach (which was first published in a newspaper strip in 1931) (1926-1930), production increased 135 % on the previous 5 year mean (1921 -1925) acreage. Average acres production increased 51 % over the next 5 years.
THE POPEYE EFFECT ON USA SPINACH PRODUCTION
50 years of acreage of Spinach Harvested 1928 - 1978 by Dr Mike Sutton (2010)
I'm not statistical expert (obviously), but I think your analysis at least shows that the apparent pattern is not random fluctuation (something caused it). Am I right in interpreting your results this way? Or is there some knowable and measurable probability that it could still be random?