Imported overseas cases in the early stage: the rise of public epidemic awareness
Table 1 shows the correlations among total global and Taiwanese confirmed cases, the volume of the mentions on social media, number of news reports, and Google search query volume from the official press release on COVID-19 issued by the government (December 31) to the implementation of a new mask rationing system (February 6). The correlation coefficients among the seven variables were between .55 and .97. In the early stage of the epidemic spread, the correlation coefficients between the total Taiwanese confirmed cases and number of news reports (r = .92), volume of the mentions of COVID-19 on social media (r = .92), the volume of the mentions of face masks on social media (r = .92), Google search query volume of COVID-19 (r = .76), and Google search query volume of face masks (r = .90) were significant and high, suggesting that the news media and the public attached great importance to the imported cases of COVID-19 in Taiwan.
Table 1. Pearson correlation coefficients among the seven variables
|
N = 38
|
2
|
3
|
4
|
5
|
6
|
7
|
1. Global total confirmed cases of COVID-19
|
.94
|
.79
|
.80
|
.81
|
.55
|
.74
|
2. Total confirmed cases of COVID-19 in Taiwan
|
-
|
.92
|
.92
|
.92
|
.76
|
.90
|
3. Volume of the mentions of COVID-19 on social media
|
|
-
|
.91
|
.97
|
.91
|
.88
|
4. Volume of the mentions of face masks on social media
|
|
|
-
|
.93
|
.81
|
.95
|
5. Number of news reports on COVID-19
|
|
|
|
-
|
.84
|
.89
|
6. Google search query volume of COVID-19
|
|
|
|
|
-
|
.83
|
7. Google search query volume of face masks
|
|
|
|
|
|
-
|
Note: All values are significant at the .05 level.
Data collected between December 31 and February 6 were used.
|
Figure 1 shows that when Taiwan announced its first confirmed case on January 21, it aroused the first wave of news media reports and attention to COVID-19 on social media and prompted the public to search for COVID-19 and mask-related information on social media. After two more cases were added on January 24, the news media and the volume on social media climaxed again. The attention of the news media toward COVID-19 continued to increase, as did the volume of discussion on COVID-19 on social media. Thus, social media rapidly promoted discussion of the topics in news reports, and there was a high correlation between the number of news reports and the volume of the mentions of COVID-19 on social media (r = .97). Since the spread of COVID-19 to Taiwan, its Google search query volume peaked in less than five days (January 25), and the number of news reports and the volume of mentions on social media also peaked on January 28 when the first confirmed local case in Taiwan was identified as a household infection.
The new form of public epidemic awareness
Although the correlation between the Google search query volume of COVID-19 and face masks (r = 0.83) was quite high, the trend chart showed that the rising period and time of reaching a peak were not the same.
Figure 1 shows that the volume of the mentions of face masks on social media and Google search query volume of face masks increased on January 24 after the government implemented the export ban on face masks and reached peaks on January 28 and January 31, respectively. In this period, the government released 6 million epidemic prevention face masks for sale in convenience stores daily on January 28–30, with a purchase limit of three face masks per person at a price of NT $8 per face mask. The second local case of household transmission occurred on January 31, and the WHO declared the outbreak to be a public health emergency of international concern on the same day. Furthermore, in the week of January 24–31, among the top 25 keywords of rising searches on Google Trends, seven were related to face masks, especially “medical face masks,” “where to buy,” and “names of each face mask shop.” This implies that public epidemic awareness began to increase in just a few days, from the perception of the importance of COVID-19 (Google search for COVID-19) to the perception of its threat and the adoption of protective measures (Google search for face masks).
It shows that new local cases, the government’s emergency release of face masks, and lax purchase restrictions on face masks have aroused a public sense of insecurity and anxiety created by the increased chance of potential infection and the public quickly buying face masks. The phenomenon of panic buying of face masks was reported by news media during that period.33
The implement of name-based rationing system for face masks
Public complaints also prompted the government to increase the allocation of resources for mask-related epidemic prevention materials. The government took emergency measures on January 31, including the daily collection of 4 million face masks from manufacturers, of which 2.6 million were for the public and sold through convenience stores, drugstores, and related chain stores, with a purchase limit of three face masks per person per day at a unified price of NT $6 per face mask on February 1.34 Subsequently, the government launched a name-based rationing system for face masks on February 6. Based on the principles of prioritizing healthcare workers and ensuring equal purchase opportunities for all, the government adopted the method of unified collection, allocation, and price, and stipulated a purchase limit. Each person is allowed to buy two masks per National Health Insurance card (for identification) at a price of NT$5 per face mask in a week.35 The response measure is to ensure that the Taiwanese are protected from other people’s hoarding of goods and to avoid the more significant social burden caused by those who intend to drive up prices. After the policy was put in place, the government and the public worked together to create open data applications such as Face Mask Maps to enable the public to quickly find sales locations and inventory, thereby improving the efficiency of buying face masks, reducing the negative impact of long queues and inability to buy face masks, and making proper use of information technology to achieve epidemic prevention.36
After the implementation of the new mask rationing system, the government also successively put forward some face mask countermeasures such as increasing collection, extending the export ban, adding production lines, and increasing the purchasing limit of face masks to three face masks per person per week (TCDC, 2020 March 2),37,38 which were expected to enable everyone to have face masks and protect themselves. Such positive actions have reduced public anxiety about not being able to buy face masks; accordingly, the volume of the mentions and search query of face masks have dropped significantly and become less affected by new cases of infection. More recently, an “name-based rationing system for face masks new version 2.0” was implemented in which the online ordering mechanism was added on March 12.39 The purpose of this new mechanism is to ensure even distribution better and make it more convenient to obtain face masks for people such as office workers and students who lack time to go to pharmacies and public health centers. From the perspective of crisis management, these face mask countermeasures not only achieve a cogent allocation of medical resources and reduce the possibility of infection but also take into account the panic of the public so that the people have sufficient control over the purchase of face masks, thus preventing the epidemic of fear.
The period of highest Google search query volume of face masks and the loss of control
According to the results of the ANOVA (Table 2), the types of fear had a significant effect on Google search query volume of “face masks” [F(3,358) = 5.67, p < .05]. Moreover, the results from post hoc comparisons using the Scheffé test with bootstrap method (Table 3) indicated that only the mean score for the loss of control group (M = 42.65, SD = 26.12) was significantly different from the other groups (M = 26.73, SD = 14.61, for mistrust; M = 24.65, SD = 18.38, for severity; M = 22.08, SD = 21.20, for uncertainty; M = 26.12, SD = 12.10, for susceptibility; M = 25.77, SD = 8.67, for without fear).
Table 2. The result of one-way analysis of variance
|
|
|
|
Sum of squares
|
df
|
Mean square
|
F
|
p
|
Between groups
|
6163.83
|
5
|
1232.77
|
5.67
|
<.05
|
Within groups
|
76726.75
|
353
|
217.36
|
|
|
Total
|
82890.58
|
358
|
|
|
|
Table 3. Post hoc Scheffé test with the bootstrap method
|
|
Descriptive statistics
|
Multiple comparison a
|
N
|
Mean
|
SD
|
2b
|
3b
|
4b
|
5b
|
6b
|
1. Mistrust
|
48
|
26.73
|
14.61
|
2.08 (3.31)
|
-15.92* (3.92)
|
4.65 (3.18)
|
0.62 (2.49)
|
0.96 (2.65)
|
2. Severity
|
34
|
24.65
|
18.38
|
-
|
-18.00* (4.16)
|
2.57 (3.46)
|
-1.47 (2.84)
|
-1.12 (2.98)
|
3. Loss of control
|
20
|
42.65
|
26.12
|
|
-
|
20.57* (4.06)
|
16.54* (3.54)
|
16.88* (3.66)
|
4. Uncertainty
|
39
|
22.08
|
21.20
|
|
|
-
|
-4.04 (2.69)
|
-3.69 (2.84)
|
5. Susceptibility
|
131
|
26.12
|
12.10
|
|
|
|
-
|
0.34 (2.04)
|
6. Without fear
|
87
|
25.77
|
8.67
|
|
|
|
|
-
|
Note. Values in parentheses are bootstrap standard error. a Post hoc Scheffé test with the bootstrap method was used.
b The reference group for the multiple comparisons of mean * The mean difference in bold is significant at the .05 level.
|
Changes in the number of different fear events at different periods
A Pearson chi-square test was performed to examine the change in the proportion of events in different types of fear across various time periods. The proportion differs by the time periods [χ2 (10) = 41.16, p < .05]. A cross-table (Table 4) showed the counts and expected counts of events in different categories.
Table 4. Crossable for Sample characteristics with Chi-square analysis
|
|
Time periods
|
χ2 (df)
|
First stage (N = 47)
|
Second stage (N = 105)
|
Third stage (N = 207)
|
Types of fear
|
Untrustworthy (N = 48)
|
n
|
6
|
17
|
25
|
41.16* (10)
|
Expected n
|
6.3
|
14.0
|
27.7
|
Dread (N = 34)
|
n
|
8a
|
11
|
15
|
Expected n
|
4.5
|
9.9
|
19.6
|
Loss of control (N = 20)
|
n
|
2
|
10a
|
8
|
Expected n
|
2.6
|
5.8
|
11.5
|
Uncertainty (N = 39)
|
n
|
15a
|
10
|
14b
|
Expected n
|
5.1
|
11.4
|
22.5
|
Vulnerability (N = 131)
|
n
|
11b
|
34
|
86
|
Expected n
|
17.2
|
38.3
|
75.5
|
Other (N = 87)
|
n
|
5b
|
23
|
59
|
Expected n
|
11.4
|
25.4
|
50.2
|
Note. a The counts are 1.5 times greater than expected counts were in bold.
b The counts are 1.5 times less than expected counts were in Italic.
* The Pearson chi-square value is significant at the .05 level
|
To further ascertain which types of events in terms of fear have relatively large ratios compared to different periods, multinomial logistical regression was used (Table 5). We used the second stage as a reference group since it had the highest Google search query volume of face masks. The odds ratio for uncertainty (6.90) in the First stage is significant, indicating that the probability of uncertainty event happening in the First stage is 6.90 times higher than that in the Second stage, relative to the without fear group. On the other hand, the probability of the loss of control event happening in the Second stage is 3.23 (1/0.31) times significantly higher than that in the Third stage, relative to the without fear group.
Table 5. Results of the multinomial logistic model
|
Time
periods a
|
|
B
|
SD
|
Wald
|
df
|
Odds
ratio
|
95% Confidence Interval for odds ration
|
Lower
Bound
|
Upper
Bound
|
First
stage
|
Intercept
|
-1.53
|
0.49
|
9.56
|
1
|
.00
|
|
|
Mistrust
|
0.48
|
0.68
|
0.50
|
1
|
.48
|
1.62
|
.42
|
Severity
|
1.21
|
0.68
|
3.17
|
1
|
.07
|
3.35
|
.89
|
Loss of control
|
-0.08
|
0.92
|
0.01
|
1
|
.93
|
.92
|
.15
|
Uncertainty
|
1.93*
|
0.64
|
9.10
|
1
|
.00
|
6.90
|
1.97
|
Vulnerability
|
0.40
|
0.60
|
0.43
|
1
|
.51
|
1.49
|
.46
|
Without fear b
|
-
|
-
|
-
|
-
|
-
|
-
|
-
|
Third
stage
|
Intercept
|
0.94
|
0.25
|
14.69
|
1
|
.00
|
|
|
Mistrust
|
-0.56
|
0.40
|
1.94
|
1
|
.16
|
.57
|
.26
|
Severity
|
-0.63
|
0.47
|
1.83
|
1
|
.18
|
.53
|
.21
|
Loss of control
|
-1.17*
|
0.53
|
4.76
|
1
|
.03
|
.31
|
.11
|
Uncertainty
|
-0.61
|
0.48
|
1.58
|
1
|
.21
|
.55
|
.21
|
Vulnerability
|
-0.01
|
0.32
|
0.00
|
1
|
.96
|
.99
|
.53
|
Without fear b
|
-
|
-
|
-
|
-
|
-
|
-
|
-
|
a,b The reference categories are Second stage for time periods and Other for types of risk characteristic.
* The parameter estimate is significant at the .05 level.
|