Integrating explanation and prediction in computational social science.

Area of Science:

• Computational social science
• Interdisciplinary research
• Data science

Background:

• Computational social science involves large digital datasets and advanced analytical techniques.
• It represents a convergence of various scientific disciplines with distinct methodologies.
• Existing approaches often focus on either explanation or prediction, leading to potential fragmentation.

Purpose of the Study:

• To clarify differences between computational social science approaches.
• To propose productive integration strategies for diverse methodologies.
• To advocate for combining explanatory and predictive modeling in research.

Main Methods:

• Development of a schema to analyze research activities.
• Categorization of research along dimensions of explanation (causal effects) and prediction (outcome testing).
• Advocacy for "integrative modeling" combining both prediction and explanation.

Main Results:

• A two-dimensional schema for evaluating research priorities (explanation vs. prediction).
• Identification of complementarity between explanatory and predictive research goals.
• Proposal of "integrative modeling" as a unified approach.

Conclusions:

• Computational social science benefits from integrating diverse disciplinary perspectives and computational methods.
• A balanced approach combining explanation and prediction enhances scientific rigor.
• Integrative modeling offers a path towards more comprehensive and impactful computational social science research.