Ensuring a Representative Sample

Rachel Levenstein
Rachel Levenstein
  • Updated

We are committed to achieving comprehensive representativeness across demographic categories and perspectives in our surveys. To achieve this, we implement the following best practices, continuously working to minimize bias in our survey data collection:


During Data Collection


  1. Setting Response Targets: Zencity employs national statistical data (e.g., from the U.S. Census Bureau or the UK Census) to establish response targets for each geographic area. These targets are defined based on key demographic factors such as race/ethnicity, age, and gender.
  2. Real-time Monitoring: We constantly track the demographic composition of the survey responses in real-time, comparing them to the predetermined targets. We fine-tune our distribution and advertising strategies as needed to target all demographic groups.
  3. Neutral Advertising to Minimize Bias: The advertisements, designs, images, or videos shown to potential respondents are neutral and do not hint at the survey's specific topics. We do this so that we hear from people with and without a strong opinion on the survey topic. For instance, our ads may simply announce, "We have an important new study for [community] residents." The survey's content is only revealed after participants click into the survey itself. All survey requests originate from Zencity to mitigate bias from individuals who may follow our clients' social media accounts. By displaying neutral ads and showing the survey sponsor as Zencity rather than the client itself, we hear opinions well beyond the STP - Same Ten People, whose voices often get disproportionate prominence.


Adjustments After Data Collection


  1. Removing Ineligible Respondents: After data collection, we eliminate ineligible respondents, including those who identify as underage, those who provide zipcodes or postcodes that are outside the city limits, and those who don’t provide a zipcode or postcode at all.
  2. Rake Weighting: Zencity employs an industry-standard statistical technique called rake weighting (also known as "rim weighting" or "iterative proportional fitting"). This technique assigns a unique weight to each respondent based on their demographic characteristics (age, gender, and race). This process ensures that the distribution of these characteristics in the final weighted sample aligns with the community's overall demographics to ensure a stronger representation of the whole community.
  3. Division-Level Calculation (where applicable): In many cases, our clients want to see estimates at a division level (e.g., council district or police precinct). To deliver the most precise and actionable results to our clients, we use rake weighting at both the city and the division levels, taking into account division-level demographics to ensure that no demographic group is overrepresented or underrepresented in the data at either geographic level. For example, in a city with five police precincts, we apply the rake weighting process for the city as a whole and each of the five precincts to account for their unique demographic makeup.


Zencity's unwavering commitment to best practices and our ongoing efforts to minimize bias have led to digital surveys that accurately reflect the community, going beyond the STP.

Was this article helpful?




Please sign in to leave a comment.