First versus final impression formation: cues from dynamic and continuous emotional expressions in social media profiles
通过在线实验，研究了社交媒体档案中一致与不一致的情绪表达如何影响第一印象和最终印象，发现不一致情绪表达会修正初始印象，且近因效应主导，人际喜欢比信任更敏感。
Purpose This study investigated the influence of inconsistent versus consistent emotional expressions (independent variables) in social media profiles on first and final impressions in terms of interpersonal liking and trust (dependent variables). Design/methodology/approach Consistency in emotional expressions was operationalized as the same or different valences across all emotional posts. A 2 (initial positive vs negative emotional expressions) × 2 (subsequent positive vs negative emotional expressions) between-subjects online experiment (N = 254) was employed. Joy and sadness were employed to represent positive and negative emotional expressions, respectively. Findings The findings revealed that first impressions formed based on emotional expressions at Time 1 were transitory. Both positive and negative initial impressions were susceptible to modification by inconsistent emotional expressions at Time 2, indicating the dominance of the recency effect over the primacy effect. Notably, expectancy violation valence, triggered by these inconsistent emotional expressions, significantly mediated the relationship between the displayed emotional expressions and the final impressions at Time 2. Furthermore, interpersonal liking was highly sensitive to inconsistent emotional valence at both stages, whereas interpersonal trust was less affected. Research limitations/implications On a theoretical level, this study reveals the dynamic and continuous process of impression formation based on users’ emotional expressions on social media. On a practical level, the findings can guide users to effectively manage impressions via emotional expressions in social media profiles. Originality/value This study examined the effects of inconsistent emotional expressions on impression formation through competing mechanisms (primacy or recency effects, valence biases and expectancy violation theory).