People suffering symptoms of various illnesses may alter their patterns of purchase of foods in locations where some observation is possible, such as at cafeterias. For example, respiratory illness may cause an increase in sales of soups or beverages, both foods commonly considered helpful or soothing in such cases. Such a change in pattern could be a timely indicator that the population suffers symptoms not yet reported to clinicians.
Campbell and colleagues use correlation analysis to examine the purchases of beverages during the first three months of 2002 in the cafeterias at the IBM T.J. Watson Research Center, located in Yorktown Heights NY and Hawthorne NY (Li and Aggarwal, 2003, Wagner et al., 2004). They compared total beverage sales data with the same reference standard as described in previous sections for the period January-March 2002. Weekends were ignored because the cafeterias were closed on these days, and Mondays were ignored because beverages were free and sales not recorded.The maximum correlation was 0.66, with a lead time of 19 days. The investigators considered this preliminary study of cafeteria data to be inconclusive due to the biologically implausible 19-day lead time, but suggested that there appears to be sufficient evidence of potential to warrant further investigation.
Was this article helpful?