The Comparative Image Reconstruction Framework: An Empirical Pilot Study of Nigerian and US Election News

Abstract

Candidate-image studies often examine verbal evaluation and news imagery separately, which obscures how the two modes jointly construct published political representation. This study assesses the operational feasibility of the Comparative Image Reconstruction Framework (CIRF) through a descriptive comparative pilot of 16 online election-news articles: eight from Nigeria and eight from the United States, drawn from eight outlets. CIRF integrates framing, attribute agenda-setting, visual framing, and comparative media
systems perspectives. Two pre-adjudication coding records were compared across 18 textual, discursive, and visual variables. Across 288 paired decisions, 282 agreed (97.9%); variable-level agreement ranged from 87.5% to 100.0%, with Cohen’s κ ranging from .810 to 1.000. The primary record identified seven reinforcing, six qualifying, one contradictory, and two absent-image text–image relationships. After three records with source-access limitations were excluded, the sensitivity subset retained six reinforcing, six qualifying, and one absent-image classification. The six disagreements exposed decision boundaries involving textual evaluation, visible affect and multimodal interpretation. These findings provide preliminary evidence that CIRF can organise and audit published candidate representations, while identifying codebook refinements for subsequent testing. CIRF measures content outputs; claims about exposure, audience interpretation, reputation or voting require separate audience evidence.