Skip to main content

Changes in sugar-sweetened beverage consumption in the first two years (2018 – 2020) of San Francisco’s tax: A prospective longitudinal study


Silver, Lynn D. et al. (2023), Changes in sugar-sweetened beverage consumption in the first two years (2018 – 2020) of San Francisco’s tax: A prospective longitudinal study, Dryad, Dataset,


Background: Sugar-sweetened beverage (SSB) taxes are a promising strategy to decrease SSB consumption, and their inequitable health impacts, while raising revenue to meet social objectives. In 2016, San Francisco passed a one cent per ounce tax on SSBs. This study compared SSB consumption in San Francisco to that in San José, before and after tax implementation in 2018.

Methods & findings: A longitudinal panel of adults (n = 1,443) was surveyed from zip codes in San Francisco and San José, CA with higher densities of Black and Latino residents, racial/ethnic groups with higher SSB consumption in California. SSB consumption was measured at baseline (11/17–1/18), one (11/18–1/19), and two years (11/19-1/20) after the SSB tax was implemented in January 2018. Average daily SSB consumption (in ounces) was ascertained using the BevQ-15 instrument and modeled as both continuous and binary (high consumption: ≥6 oz (178 ml) versus low consumption: <6 oz) daily beverage intake measures. Weighted generalized linear models (GLMs) estimated difference-in-differences of SSB consumption between cities by including variables for year, city, and their interaction, adjusting for demographics and sampling source. In San Francisco, average SSB consumption in the sample declined by 34.1% (-3.68 oz, p = 0.004) from baseline to 2 years post-tax, versus San José which declined 16.5% by 2 years post-tax (-1.29 oz, p = 0.157), a non-significant difference-in-differences (-17.6%, adjusted AMR = 0.79, p = 0.224). The probability of high SSB intake in San Francisco declined significantly more than in San José from baseline to 2-years post-tax (AOR[interaction] = 0.49, p = 0.031). The difference-in-differences of odds of high consumption, examining the interaction between cities, time and poverty, was far greater (AOR[city*year 2*federal poverty level] = 0.12, p = 0.010) among those living below 200% of the federal poverty level 2-years post-tax.

Conclusions: Average SSB intake declined significantly in San Francisco post-tax, but the difference in differences between cities over time did not vary significantly. Likelihood of high SSB intake declined significantly more in San Francisco by year 2 and more so among low-income respondents.


The three waves of the study utilized a “push-to-web” data collection method, in which sampled households were sent an invitation via mail, text, and/or E-mail to complete an online Web questionnaire. Additional web completes were collected using a non-probability web panel. Questionnaires were completed in English, Spanish, or Chinese. Data from each wave were appended together--each row/observation is unique to participant ID and wave. Variables for the study were constructed using Stata. More details on methodology can be found in SSB_Sampling_and_Data_Collection_Methodology.pdf and SSB_Analytic_Sample_Creation_Flowchart.pdf.

Usage notes

Data are in CSV format. Documentation is in PDF format. Code is in Stata format (.do) and can also be opened as a text file. 


National Cancer Institute, Award: R01 CA230733-01

Division of Cancer Prevention, National Cancer Institute, Award: 1R01CA214470-01