Google searches for "cheap cigarettes" spike at tax increases: Evidence from an algorithm to detect spikes in time series data

Abstract

Introduction
Online cigarette dealers have lower prices than brick-and-mortar retailers and advertise tax-free status.1–8 Previous studies show smokers search out these online alternatives at the time of a cigarette tax increase.9,10 However, these studies rely upon researchers’ decision to consider a specific date and preclude the possibility that researchers focus on the wrong date. The purpose of this study is to introduce an unbiased methodology to the field of observing search patterns and to use this methodology to determine whether smokers search Google for "cheap cigarettes" at cigarette tax increases and, if so, whether the increased level of searches persists.

Methods
Publicly available data from Google Trends is used to observe standardized search volumes for the term, "cheap cigarettes". Seasonal Hybrid Extreme Studentized Deviate and E-Divisive with Means tests were performed to observe spikes and mean level shifts in search volume.

Results
Of the twelve cigarette tax increases studied, ten showed spikes in searches for "cheap cigarettes" within two weeks of the tax increase. However, the mean level shifts did not occur for any cigarette tax increase.

Conclusion
Searches for "cheap cigarettes" spike around the time of a cigarette tax increase, but the mean level of searches does not shift in response to a tax increase. The SHESD and EDM tests are unbiased methodologies that can be used to identify spikes and mean level shifts in time series data without an a priori date to be studied. SHESD and EDM affirm spikes in interest are related to tax increases.