Valence and attribute repetition in negative sets of online reviews : (when) can positive reviews overcome negative ones?

Review set valence (the degree of negativity or positivity of a set of online reviews) strongly determines review readers' responses. Previous research has mainly considered the mere number of positive and negative reviews to determine a review set's valence. This paper aims to study how increasing the number of important positive reviews influences readers' hotel staying intention, exploring the 'tipping point' at which important positive reviews compensate for the negative effect of a larger number of less important negative reviews. We further explore whether reader responses are more positive when all positive reviews address the same product attribute or different attributes. We present a 4 (review set valence) x 2 (attribute repetition vs. different attributes for the positive reviews) online experiment (N=408). The results show that a more positive review set leads to a higher staying intention only when the positive reviews discuss different attributes (and do not repeat the same attribute). The 'tipping point' at which positive reviews compensate negative ones is four positive reviews about different attributes in a set of 12. This study nuances the bandwagon effect, negativity bias, and truth effect by showing that negative review sets can be positively evaluated