Table 2 Sampling methods employed by International Study of Childhood Obesity, Lifestyle and the Environment (ISCOLE) sites

Australia

A stratified probability sampling frame was used, aiming to ensure that each 5^th grade child in the school system has an equal chance of being selected. Schools were initially stratified into tertiles based on the Index of Community Socio-Educational Advantage (ICSEA). Schools were randomly chosen from within each ICSEA tertile, with the probability of being chosen proportional to the estimated enrolment in 5^th grade. Once the required number of children was enrolled from a tertile (200 children), enrolment continued exclusively from the remaining tertiles.

Brazil

There is variability in socioeconomic status between schools in the region of Sao Caetano do Sul. Public schools represent the lower socio-economic strata, and private schools reflect the higher socio-economic strata. Random lists of public and private elementary schools in the region were generated, and schools were selected for each list at a ratio of 4 (public) to 1 (private). If a school refused to participate in the study, it was replaced by the school next on the list. Twenty schools were sampled (16 public and 4 private), and 5^th grade students continued to be sampled in each school in order to have between 25–30 students in each school.

Canada

Schools were drawn from the Ottawa Region. Schools were stratified into four groups with proportional representation (English Public, French Public, English Catholic, French Catholic). All schools within each stratum were invited to participate and the first to respond were included into the study until each stratum was filled. Children in 5^th grade were selected from the schools to participate.

China

Three regions (2 urban districts and 1 suburban district) of Tianjin city were selected and stratified according to three levels of socio-economic status. Within each stratum, 2 schools are selected randomly from a list of all public schools, with a total of 6 schools to participate. If the selected school refused to join the project, it was replaced by the next randomly selected school. Each school ensured an average sample size of 85–90 grade 4^th grade students.

Colombia

A list of public and private schools in Bogotá were selected according to the following inclusion criteria: 1) schools in urban area; 2) including boys and girls; 3) having a morning schedule; 4) enrolling students from elementary, middle and high-school; 5) belonging to January-December calendar, and 6) not being a school for a disabled population (blind, deaf, etc.). Given the distribution of students who attend public (76%) vs. private schools (24%), 15 public schools and 5 private schools were selected randomly. Schools were sorted into high SES, middle SES and low SES. The sample will result in a minimum recruitment of 600 4^th, 5^th and 6^th grade children to obtain 500 children in 20 schools assuming an 83% response rate.

Finland

The sampling frame was a complete list of primary schools in the capital region (cities of Helsinki, Vantaa and Espoo, total population about 1 million or 19% of Finnish population). Schools were first stratified by city and then by area SES (high, low) based on socio-economic characteristics of their geographical location (educational level, if available, otherwise income level). From each of these 6 strata (city/SES), three to six schools were randomly selected to represent the distribution of pupils and SES within the total sampling area. A reserve list was used to account for school withdrawal. Children in 4^th grade were selected from the schools to participate.

India

Fee structures of different private schools catering to different socio-economic status were obtained. Based on this, a classification was made into low, middle and high socio-economic status. Three to four consenting schools were selected from each stratum. If a school declined the invitation to participate in the study, another school of the same fee structure was selected. The children from 5^th grade were sampled to have at least 60–70 students from each school.

Kenya

Non-boarding primary schools from Nairobi County were stratified into public and private (boarding schools were not sampled). The schools were then selected proportional to the distribution of public and private school attendance. Non-compliant schools were replaced with the next conveniently selected school from the group. Children in 5^th grade were selected from the schools to participate.

Portugal

There is little variability in socio-economic status at the school level in Porto; thus schools were randomly selected from a list provided by the North Regional Education Directory Board. If a non-compliant school was found, it was replaced by the next random school selected from the group. Twenty two schools were sampled, and from each, 5^th grade students were sampled in order to have 25–30 students in each school.

South Africa

The sampling frame was a list of all public schools within the geographic area of study eligibility. The list was stratified according to SES quintiles and at least 4 schools were randomly selected from each stratum for a total of at least 20 schools. If a school declined the invitation to participate in the study, another school of the same socio-economic quintile was randomly selected. Children in the 4^th and 5^th grades were selected from the schools to participate.

United Kingdom

The sampling frame was a complete list of primary schools in the Bath & North East Somerset and West Wiltshire regions. Schools were stratified according to size and socio-economic characteristics (e.g., free school meal entitlement). From each stratum, a proportional cluster was selected. Specifically, schools were randomly selected using the probability proportional to size approach and a reserve-list compiled to account for school withdrawal. Children in the 5^th and 6^th years were selected from the schools to participate.

United States

A complete list of public and private schools enrolling 4^th grade students in East Baton Rouge Parish was assembled. Private schools (collectively a stratum) were sampled separately. The public schools were sorted into quartiles (strata) according to% free and/or reduced lunch. Thus, there were five strata to sample from (4 public and 1 private). All schools were placed in random order within each stratum. Each school was approached according to the random order established within each stratum until a minimum of 4 schools were selected from each stratum, for a total minimum of 20 schools across strata, resulting in a minimum enrollment of 500 4^th grade children.