Table 1.

Sampling decisions.

Goals		Criteria or process used	Decision
Identify cities where sampling will take place.		Include cities that have a large population; include cities that are geographically dispersed throughout the country.	The following five cities were identified: Moscow, St. Petersburg, Novosibirsk, Yekaterinburg, and Kazan.
Identify diverse socioeconomic zones in each city.
	Identify and map zones in city based on socioeconomic status.	Include areas of different socioeconomic status; without available census data on resident education and income, a proxy value of property value is used.	Study mapped 3 types PVZa. PVZa was classified as being: •High property value (>1.25x average property value) •Average property value (>0.85-1.25 x the average property value), or •Low property value (<0.85x the average property value).
Identify sampling areas.
	In each property-value-zone, identify areas of the city with significant retail activity (retail centers). Create a catchment area around the retail center, this area is known as the sampling area.	Identify three different types of retail centers in each PVZa: 1. Shopping mall, 2. Major intersection, and 3. Transit station.	From each retail center, a 3 km radius was drawn, creating 3 different sampling areas in each of the 3 property-value-zones. [9 sampling areas in each city]
Identify retail venues to be sampled in each sampling area.
	Identify which types of tobacco retail venues (POS) to be included.	Chain supermarkets [both up market (luxury) and mid-low market]; independent markets/convenience stores; Kiosks (street stalls).	Sampling goals included an equal number of: 1. Chain supermarkets, 2. Independent markets/convenience stores, and 3. Kiosks. Chain supermarkets included equal number of luxury/high market stores, and mid-low market stores when possible. Independent markets include both convenience stores and gas stations. [6 chain supermarkets, 6 independent markets, and 6 kiosks per sampling area-18 total retail venues per sampling area]
	Identify locations of specific retailers within the sampling areas.	Use available databases that list and include the location of chain supermarkets (both high end, and mid and low end) in each study city. Map locations of chain supermarkets to identify which stores are in the physical sampling areas. If there are more than 9 chain supermarkets in the sampling area, number stores and use a number generator to randomly identify stores to be included in the sample.	In each sampling area, 6 chain supermarkets were identified and mapped in each of the cities' 9 sampling areas. An additional 3 “back-up” chain super markets were also included in the sampling areas. [54 chain supermarkets per city, and 27 back-up chain supermarkets]
	Identify other retailers nearby supermarkets.	There are no available lists of licensed retailers, so a walking protocol is used to identify near-by independent markets (convenience stores and gas stations) and kiosks.	After data are collected at a chain supermarket, data collectors will exit the supermarket and, using the walking protocol, identify a nearby independent market and kiosk. [54 independent markets and 54 kiosks per city]

^aPVZ: property-value-zones