Automatic Weighting

  • 26 January 2023
  • 0 replies

Sometimes, despite your best outreach efforts, your raw survey results will not be fully representative of your entire community. For example, we often tend to see higher rates of participation from individuals within certain demographic groups (e.g., older adults, homeowners, women) and lower response rates from others (e.g., younger adults, renters, men).

Together, Polco’s representivity visualizations and automatic weighting functionality help address this issue. Our representivity graphs help you visualize the reach of an active survey and adjust outreach as needed to bring more people into the conversation. In addition, automatic weighting will regularly adjust the survey results to amplify the voices of underrepresented residents and provide more representative data overall.  

The automatic weighting functionality can be toggled on for each individual survey on the Summary tab. 

Although some respondent demographic data is gathered from voter registration records, most comes from respondents’ self-reporting. To facilitate that, you’ll need to add demographic questions to your survey using Polco’s list of Suggested Demographic Questions. We recommend adding the pre-built questions about age, gender, race, ethnicity, rent/own, and housing type, as these are all factored into the weighting calculations. 

After the survey is published and has received at least 10 responses, the auto-weighting calculations will run once per day. The weighting algorithm compares the respondents’ demographics to Census data for the entire population, adjusting the data to reduce discrepancies between the two. In the Results tab, you can toggle between Weighted and Unweighted results to view each dataset as needed.

We caution against making decisions based on very few responses; even with weighted results, it’s unlikely that only 10 or 20 responses are truly representative of your whole community. For this reason, we suggest expanding and adjusting outreach efforts (informed by the representivity graphs) to increase participation overall and hear from more residents, rather than relying solely on weighting to achieve greater representation. Together, these tools will strengthen your engagement efforts for current and future surveys, providing more accurate results and fostering more confident and informed decision-making. 

0 replies

Be the first to reply!