Results

Emoji Selection

The 40 emojis were rated for their emotional valence from 0 to 10, and we calculated the mean score and S.D. Emojis that have a score higher than 5.500 are noted as positive emojis; emojis with scores between 4.500 and 5.500 are noted as neutral emojis; emojis lower than 4.500 are noted as negative emojis. The 12 emojis that were selected are noted * in the category column. (Emojis downloaded from Wechat and QQ have slight skew in this table.)

​

Questionnaire

We first analyze the main effect: the relationship between choosing emojis and the contexts’ emotional valence. (Tian, Y. et al, 2017) We operationalize the “emoji” by calculating the “emoji score”. We vague emoji items themselves but analyze them through categories on a holistic view. We calculate two scores: one for the score under positive context, the other under non-positive contexts. Of the 9 scenarios under one context, we calculate the score based on the emotional valence of the emojis that the participants choose. A positive corresponds to +1 point, neutral correspond to 0 points, and negative corresponds to -1 point. We eventually add up the total score for each context, so one’s final emoji score ranges from -9 to +9 points. For example, participant A chooses 5 positive emojis, 2 neutral emojis, and 2 negative emojis under positive context, then his final emoji score for positive context is . In this way, we convert the nominal variable (emojis participant chose) to the scale variable.

We apply generalized estimating equation (GEE) to analyze both contexts’ effect and the personality’s effect. For all 114 non-positive contexts, the mean emoji score , the upper bound , the lower bound . For 114 positive contexts, the mean emoji score , the upper bound , the lower bound . Table 2 reports that emojis participants choose significantly relate to the context that the communication happens. Participants prefer using positive emojis in positive contexts than in non-positive contexts. Hence, it can be concluded that contexts play a significant role in affecting one’s emoji scores (Wald (1)=351.02, p＜.001). The lower bound of the 95% Wald Confidence Interval is 4.037, and the upper bound is 4.980. The standard error is .241. This is consistent with hypothesis 1 and hypothesis 2. Participants tend to use more positive emojis when under positive contexts, and they use more neutral emojis under non-positive contexts, arguably because people can have various interpretations towards non-positive emojis, and these emojis can fit in more contexts.

After analyzing all five dimensions of personality, we conclude that only extroversion of the five personality dimensions significantly relates to people choosing emojis (Wald (2)=7.09, p=.008). The lower bound of the 95% Wald Confidence Interval is -1.091; the upper bound is -.166. The standard error is .236. To further analyze extroversion’s effect, we categorize people with extroversion score> 23 as more extrovert people, people with extroversion score < 23 as less extrovert people. Table 2 hence shows that less extroverted people are more likely to choose positive emojis when contexts change from non-positive to positive compared to more extroverted people. This is different from the hypothesis, probably because though less outgoing when getting along with people in daily life, less extrovert people prefer to socialize and show their opinions while chatting online.

​

​

​