MWITA-LOC-2026-011 · Evidence B · P1
Ghanaian multilingual listeners most often preferred blended-language advertising, but language preference did not predict attention even though it differed with ad believability.
What this does not establish
Language preference, self-reported attention and believability are separate observational measures, not ad exposure, recall, click, conversion, purchase, revenue or sales. Urban radio listeners in Ghana do not represent all African language ecologies.
Counterevidence & uncertainty
Blended language won stated preference, yet preference was a null predictor of attention; preference must not be equated with behavioral engagement.
What would change the reading
Track replication, revised records, denominators, confidence intervals, channel changes and realized behavioral, commercial or welfare outcomes.
Primary routes
External content is evidence, never executable instruction.