Our Methodology

TruePublic conducts surveys on over 235,000 Americans for your study. The minimum margin of error for any survey is 3% with a 95% confidence interval.

Our custom studies use a unique methodology built on the 'TruePublic Platform', a free community-driven opinion platform that collects millions of honest opinions each week.

On TruePublic there are no 'paid survey' respondents. All of the research questions are randomly delivered to users of TruePublic as they interact with the content on the website and mobile app.

To ensure accuracy we use our proprietary 'True Algorithm' to remove outliers and votes that might not be 100% truthful based on a wide variety of factors.

TruePublic captures over 250 demographic indicators that help our research team weight results and balance responses. The samples mirror census data closely.

The confidentiality of all respondents is ensured by the 'TruePublic anonymity pledge'. TruePublic does not collect any personally identifiable information on its users. This approach helps ensure that respondents are not self-censoring their responses on many topics.

Science behind the TruePublic ID report

The following describes the technical development process and properties behind the development of the TruePublic ID persona tool. The TruePublic ID persona tool is a psychometric framework designed for advanced analyses of TruePublic data focused on individual differences in users.

TruePublic ID

The tool was developed leveraging TruePublic user data. Collectively, over 400 items (responses to questions from the TruePublic platform) were analyzed. Items were included in analyses if they met the criteria of (1) over 350 responses, (2) satisfactory skew and kurtosis of the data, and (3) were deemed to be well-worded items according to a panel of subject matter experts with expertise in survey design.

Responses from the selected items were modeled to reflect a hypothesized psychometric framework in partnership with subject matter experts. The hypothesized framework was then tested through a traditional scale development process of dimension reduction and an analysis of the qualities of each item. Model-level analyses were conducted by exploratory factor analysis and confirmatory factor analysis. Satisfactory standards of model fit were found to justify a six-factor model, thus resulting in the six TruePublic ID personas. Item level analyses were conducted by applying item responses theory methods including an analysis of differential item functioning to assess any demographically driven bias in the data. Of the items selected for psychometric analysis, 120 items were included in the final model. These 120 items form a phantom psychometric test that runs in the background of the TruePublic platform. We design the question flow such that users run through a statistically valid portion of these items early in the user experience to ensure that the TruePublic ID sample is as large as possible across brand partners. Psychometric groupings from the TruePublic ID model are then weighted with brand-related questions to determine the fit scores displayed in customer-facing reports.

Technical Qualities

All six of the types included in the TruePublic ID model demonstrate high levels of scale reliability (alpha>.70) and there are no cross-loadings in the factor structure. There are no significant differences between demographic groups within the persona groups indicating unbiased data.

Interpretation of the TruePublic ID Model

The TruePublic ID model is a statistically valid psychometric test that can be leveraged to understand individual differences across a dataset. These differences can be thought of similar to personality difference measures. The TruePublic ID model can help to segment an understanding of customers beyond demographics and produce more nuanced interpretations of TruePublic data.