Table 4 Effect of negative emotions on timeliness.

From: The reduction of effective feedback reception due to negative emotions in appeals

 

(1)

(2)

(3)

Emotion

−0.184*** (0.004)

−0.229*** (0.005)

−0.286*** (0.009)

Health

 

0.138*** (0.008)

0.009 (0.010)

Request

 

−0.006 (0.005)

−0.036*** (0.009)

Truth

 

−0.224*** (0.009)

−0.254*** (0.011)

Emotion×Request

  

0.052*** (0.010)

Emotion×Health

  

0.180*** (0.015)

Emotion×Truth

  

0.055*** (0.009)

Target

 

0.034*** (0.006)

0.033*** (0.006)

Repeat

 

−0.191*** (0.005)

−0.190*** (0.005)

Length

 

0.000*** (0.000)

0.000*** (0.000)

Department

 

0.167*** (0.005)

0.167*** (0.005)

Type dummy

 

Yes

Yes

Source dummy

 

Yes

Yes

Year dummy

 

Yes

Yes

Province dummy

 

Yes

Yes

Cons

1.712*** (0.003)

0.452*** (0.018)

0.483*** (0.019)

lnalpha

1.175*** (0.001)

0.816*** (0.002)

0.815*** (0.002)

N

2333229

2333227

2333227

Pseudo R2

0.000

0.056

0.056

  1. The independent variable, Response Time, is the days taken by a department to resolve a complaint from its receipt; Emotion equals 1 if negative emotions are included and 0 otherwise; Health equals 1 if health risk information is included and 0 otherwise; Request is a dummy variable equal to 1 if the complainant requests an intervention and 0 otherwise; Truth equals 1 if appeal content is true and 0 otherwise; Repeat equals 1 if the appeal is made more than once and 0 otherwise; Target equals 1 if the negative emotion is aimed at the appeal processing department and 0 otherwise; Length indicates the character length of the appeal content; Department equals 1 if the appeal handling department is at the city level and 0 otherwise; i.Type represents the dummy variable of different types of appeals, i.Source represents the dummy variable of appeal source; i.Year represents the year dummy variable; i.Province represents the province dummy variable; N is the number of observations; Pseudo R2 measures the fitness of the regression model. Robust standard errors are in parentheses.
  2. * p < 0.05, ** p < 0.01, *** p < 0.001.
  3. lnalpha is the logarithm of the overdispersion parameter, and H0: lnalpha = 0 indicates that the Poisson regression model should be used. The likelihood ratio test shows significant rejection of H0, therefore, negative binomial regression should be used. In the estimation, we first examined the effect of negative emotions without control variables, as shown in Column (1), because the impact of negative emotions may be influenced by other factors. Then, we added control variables to estimate the individual effects (Column (2)) and interaction effects of negative emotions (Column (3)).
Back to article page