The Content–Source Integration Model: An Account of How Readers Understand Conflicting Information From Multiple Documents

• In Event: 49.029 - Competing and Complementary Models of Multiple Source Use: Theoretical and Empirical Advancements

Abstract

Purpose

Today, comprehending information from multiple documents is a common reading task for many individuals. When searching for scientific information, in particular, it is likely that individuals encounter conflicts between different sources. Recent research suggests that individuals differ a lot regarding what they “take home” from reading conflicting information (e.g., Mills, Legare, Grant, & Landrum, 2011; Stadtler & Bromme, 2008; Wineburg, 1991). At the same time, the underlying comprehension processes are not yet well understood. Given the possible variation in reactions to conflicting information, a comprehensive theoretical model of how readers comprehend conflicting information from multiple texts would represent an important advancement.

Theoretical Frame

Against this background, the goal of the current contribution is to introduce the Content-Source Integration (CSI) model, which provides a taxonomy of readers’ reactions to conflicts in text. The CSI model draws on existing models of (multiple) text comprehension (Kintsch, 1998; Perfetti, Rouet, & Britt, 1999) and extends them by specifically dissecting the cognitive processes and resources involved when readers access conflicting information. Although not restricted to any specific type of conflict, the CSI model emphasizes how readers deal with conflicts between science-related knowledge claims.

According to the CSI model, readers first notice a conflict on a moment-by-moment level (Stage 1: conflict detection). We describe how conflict detection depends both on automatized memory-based reading processes and more strategic reading behaviors based on standards of coherence (van den Broek, Risden, & Husebye-Hartmann, 1995). We also describe how these processes are affected by the fact that information is distributed across multiple documents. The CSI model further assumes that in a second step, readers will try to restore textual coherence (Stage 2: conflict regulation). Using process data, we identify three prototypical ways of doing so: (a) ignoring a conflict, (b) reconciling a conflict, and (c) accepting a conflict as due to different sources. Whenever textual coherence does not lead to consistent world knowledge, readers might want to additionally resolve the conflict by assessing the validity of competing claims (Stage 3: conflict resolution). The CSI model suggests that readers may accomplish this task by making either first-hand evaluations of knowledge claims (answering the question “what is true”) or second-hand evaluations of sources (answering the question “whom to believe”). Whereas first-hand evaluations imply that readers assess the validity of knowledge claims based on their own understanding of the subject matter, second-hand evaluation can be understood as a bypass, in which readers evaluate source parameters instead (Bromme, Kienhues, & Porsch, 2010).

Data Sources and Results

To specify the cognitive resources involved at each of the three stages, we review empirical evidence from different bodies of research including text comprehension, science understanding and developmental psychology. In so doing, we forward explanations of why readers differ in their reactions to textual conflicts.

Scholarly Contributions

We close our presentation by discussing how the CSI model relates to existing theories of multiple document comprehension and suggest pathways for future research.

Authors

• Marc Stadtler, University of Münster

• Rainer F. Bromme, University of Münster