Abstract

Statistical surveys face increasing problems with decline in response rates and problems with missing data. People are hard to reach and reluctant to participate in surveys, while businesses often regard responding to surveys as additional administrative burden. In addition, surveys face competition in the information market, with businesses creating new information products, that meet the new information needs with fast production times, which adds to time pressure for accelerated release of survey results. To encourage people and businesses to respond and share their data for surveys, without exceeding reasonable survey budgets or time frames, it is essential to examine all points of interaction with respondents, and analyse which modern methods and design strategies could improve the survey performance.

This presentation demonstrates on how digital service design and responsible use of artificial intelligence (AI) can empower surveys that face severe quality challenges. We examine the entire survey process and identify areas where the quality could be developed. We identify use cases for AI to provide quality improvement, such as increasing response rates, reducing survey bias, resource-efficient data collection, and enhancing the efficiency of the entire statistical survey production process.

The use of artificial intelligence increases the levels of data intensiveness which needs further supporting elements in the process. The data intensive methods increase the power of prediction and allow also tailored approaches for differential sub-populations. On the other hand, strengthening engagement of the respondents and users with implementations of AI and digital service design introduce new essential steps to statistics production and dissemination, which indicate the need for expansion of the Generic Statistical Business Process Model (GSBPM 5.2).

We present an advanced survey blueprint framework for managing the survey process and statistics production. The framework aims to help the survey producers addressing key quality issues by utilizing customer-friendly modern technologies and techniques utilising AI in a responsible manner. The framework demonstrates how focused the GSBPM model is to the producer oriented approach, and how introducing enforced service thinking to the operations can entail the potential for improved quality and cost-efficient management of statistical operations.

We also review suitable AI methodologies that can leverage the data provider centric service design together with the intensified use of artificial intelligence. The novelty of the survey blueprint is that it provides required set of new techniques and analytical check points for maintaining the statistical surveys as prominent sources of information also in the future.