Audience Measurement Data Imputation for Demographic Bias Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing audience measurement techniques for online media face biases in demographic data due to incomplete classifications and misattribution, leading to inaccurate reporting of audience compositions and media ratings.
Innovation Solution
The implementation of a system that collects and processes messages from client devices to improve the accuracy of demographic data by attributing impressions and duration units to specific demographic groups, using percentages based on reliable data from other time segments and adjusting numbers of impressions and duration units to correct for biases and misattribution.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If demographic information is obtained from database proprietors for online media monitoring, then the sample population can be expanded beyond registered panel members, but the demographic information may be self-reported and thus unreliable or less reliable
Solution Approach 1:
The patent introduces an audience measurement entity as an intermediary between database proprietors and the final audience measurement system. This intermediary collects demographic information from multiple database proprietors, applies statistical techniques to assess and adjust reliability, and integrates the data into a unified measurement system. The intermediary layer enables expansion of the sample population while maintaining data quality through systematic validation and weighting procedures.
2Reliability
If traditional panel member monitoring is used with in-person, telephonic, and online interviews, then highly accurate demographic information can be collected, but the process is complex and limited to registered panel members only
Solution Approach 1:
The patent employs multiple database proprietors as alternative sources that copy or replicate demographic information functions. Instead of relying on a single complex interview-based collection system, the methodology uses parallel data collection from multiple database sources, each providing demographic profiles through their own established processes. This copying approach reduces the complexity burden on any single system while maintaining accuracy through aggregation and validation.
3Ease of operation
If demographic information is self-reported by users, then data collection becomes easier and more scalable, but the information becomes unreliable or less reliable
Solution Approach 1:
The patent implements feedback mechanisms where demographic information from self-reported sources is continuously validated against actual media consumption patterns and cross-referenced with data from multiple database proprietors. Discrepancies trigger adjustment procedures where the system provides feedback to refine demographic classifications. This feedback loop maintains ease of data collection while improving reliability through systematic validation and correction of self-reported information.
Data Source
AI summary
Methods and apparatus to generate audience measurement data from population sample data having incomplete demographic classifications are disclosed. An example method includes detecting that a number of impressions a time segment of a video occurred based on messages; detecting that no impressions for the time segment of the video were detected by the database proprietor; determining respective percentages of exposures of the video that are attributable to corresponding ones of multiple demographic groups, determining the respective percentages being based on the first demographic information provided by the database proprietor for exposures of other time segments of the video; attributing respective portions of the number of impressions to the demographic groups based on the respective percentages to determine imputed numbers of impressions; generating adjusted numbers of impressions based on the imputed numbers of impressions; and determining ratings information for the video using the adjusted numbers of impressions.