key: cord-0838378-yba84x03
authors: Akintande, Olalekan; Olubusoye, Olusanya
title: Datasets on how misinformation promotes Immune perception of COVID-19 pandemic in Africa
date: 2020-07-17
journal: Data Brief
DOI: 10.1016/j.dib.2020.106031
sha: 6bd5afed188ed38713858798c37d54e319249c08
doc_id: 838378
cord_uid: yba84x03

The dataset investigates the magnitude of the misinformation content influencing scepticisms about the novel COVID-19 pandemic in Africa. The data is collected via an electronic questionnaire method and twenty-one Africa countries randomly participated. Responses were received from all the five regions of Africa. The data is structured to identify some leading misinformation been propagated in the media. For data, in brief, we performed a descriptive analysis of the data and also examine the degree of each selected misinformation content on the immune perception of respondents using Confirmatory Factor Analysis. Another research can use the dataset to investigate how misinformation and religion misconception promote ignorance about disease or pandemic in Africa or the dataset could serve as supplementary material for further investigation of COVID-19 pandemic in Africa.

Decision Science

Specific subject area Investigating how misinformation contents about the pandemic promote scepticism. Table  Figure Analysed (misinformation content/items) Raw from the survey (all questions including demographic information)

Electronic survey shared via social media: Whatsapp and email, available at: Link to Questionnaire, raw data and analyzed data: https://data.mendeley.com/datasets/chrc52k4f5/draft?a=67edb0 27-ecb5-4278-9cac-97f0ef735900

Raw from the internet platform used for the collection. We attached the analysed and filtered as supplementary data.

The descriptive statistics examine the magnitude of misinformation and the statistical significance of each content. The confirmatory factor analysis examines misinformation content contribution to immune perception or scepticisms about the pandemic in Africa.

The data is collected via an electronic questionnaire. As a dataset limitation, we are unable to cover a very large population (n = 563 responses), however, the sample space of the respondents covers all the regions of Africa and we received responses from 21 Africa countries. 

 The dataset will enable readers/users to understand how misinformation promotes scepticisms about the pandemic (COVID-19) and towards adherence to preventive measures as well as causes of spread.

 The data method adopted helps identify the most conceived and promoted misinformation about COVID-19 pandemic in Africa [1] .

 Since misinformation misleads and could cause much havoc in time of the pandemic, it is important to world health bodies to understand the trend and danger it poses to plan for future occurrence.

 Another research can use the dataset to investigate how misinformation and religion misconception (this was not considered in this dataset but it is available in the raw data) promote ignorance about disease or pandemic in Africa [1] , [2] .

 The dataset could also serve as supplementary material or template for further investigation of COVID-19 pandemic and related issues in Africa [3] .

The data is collected via an electronic questionnaire. For the data used, we title it as "misinformation" in the repository. The data is an extraction from the raw dataset (named as Perception of COVID-19 disease in the repository. The perception data is the raw response received from the survey) which contains mainly questions (i.e., the ten-misinformation content/items are used) that address the misinformation about the COVID-19 pandemic in Africa. All the questions are coded on a five-point Likert scale. Table 1 presents the procedure or method used to score and classify each misinformation item/content as either promoting immune perception (score = 4) or not promoting immune perception (score = 0). Table 2 presents the descriptive result; the magnitude of misinformation items, which classifies each misinformation content/item as either promoting ignorance or not promoting ignorance. Table 3 presents the statistical significance of each misinformation content. That is, it presents individual strength or magnitude of each misinformation item in promoting ignorance of the COVID-19 pandemic in Africa. Table 4 presents the joint influence (correlation) of the misinformation content/item in promoting ignorance of COVID-19 in Africa. Figure 1 is the path diagram illustration of the six identified (most misleading) misinformation content with respect to the outcome in Table 2 . The figure is used to depict the item path or contribution to the immune perception of COVID-19. As a tradition, Figure 1 is necessary to arrive in Figure 2 . Hence, Figure 2 presents the statistical result (in magnitude) of individual item contribution to promoting ignorance of COVID-19 pandemic in Africa. Table 4 is a combination of all output or results of the confirmatory factor analysis conducted for the data. Table 4 (a), presents confirmatory factor analysis (CFA) model fitness. Table 4 (b) presents the individual contribution of the misinformation contents leading to immune perception against COVID-19 in Africa. Furthermore, Table 4 (c), presents the value estimate of 1 against COVID-19 to prove that the level of perception of the virus remains constant or unchanged as long as this misinformation continues. In Table 4 (d), the error correlation of the items is presented. Figure 2 summarizes the output of the path diagram in Figure 1 .

Supplementary data and materials include the misinformation content responses before score titled "misinformation Raw data"; raw data from the survey, titled "perception of COVID-19 disease"; R script for the confirmatory factor analysis (CFA; M 2); and questionnaire for the survey in pdf form. All there are also available in the Mendeley repository.

We developed ten items (misinformation contents) about COVID-19 which have formed major debates on African media space. These questions were diligently followed from the onset of the pandemic from all media platforms (between January 20 and March 25, 2020) as most circulated and conceived among Africans. We conducted opinion polls (oral engagements) on Kenyatta University (KU) Campus in Nairobi (between February 1 and February 25, 2020) as first case contact to access the general belief about the misinformation about COVID-19 pandemic spreading on social media space. KU like other African higher institutions has a large population of international students from all across Africa (making up to 90% of foreign students) and other foreign nationals including Europe, Asia, and America. Essentially, these questions were coined from the misinformation shared and conceived by a larger percentage of African social media users. Apart from our initial poll on the KU campus, we also follow posts on social media space (Facebook, Whatsapp, Tweeter, YouTube, and E-news) between January 20 and March 30, 2020, and make records of several post followers, likes, retweets/shares, seen and read. By March 30, 2020, we made a compilation of the top ten misinformation contents about COVID-19 and sent out an e-questionnaire through Whatsapp and Email. We also shared the questionnaire with individuals on various international Whatsapp Groups and appeal with them to also share with their countrymen and women to gain wider coverage. The questionnaire is restricted to only persons of Africa descent. We also did a sample of all the misinformation on different fact check websites (see, e.g., https://www.hsph.harvard.edu/india-center/myths-vsfacts/) to see how relevant the questions are to the study.

We are unable to cover a very large population (n = 561 responses), however, the sample space of the respondents covers all the regions of Africa and we received responses from 21 Africa countries. 

All responses were coded on a five-point Likert scale. Response options for all questions were "strongly agree", "agree", "neutral", "disagree" and "strongly disagree". The data aims to examine how misinformation contents promote people's poor perception of COVID-19 in Africa. Thus, strongly agree or agree implies "being influence by" and strongly disagree or disagree implies "Not being influence by" the misinformation contents. We assume being "neutral" as undecided and could go either way. We score each response on a scale of 4, 4, 1, 0, and 0 respectively. That is, the score of '4" implies that the item promotes ignorance about the pandemic, score of "0" implies the item does not promote ignorance about the pandemic, and the score of "1" implies that the item neither. The range is computed by multiplying the score by the total number of items (m =10). See equation (1) for the magnitude score formula. In essence, the aim is to score individual response per item and measure the magnitude of influence of the item on ignorance about the pandemic on the whole sample. Hence, each participant is scored as in Table 1 . Similarly, for Table 2 , the range is between "0 and 40". Score "below 10" is considered as misinformation content "Not promoting" ignorance about COVID-19 and "above 10" is considered as misinformation content "Promoting" ignorance about COVID-19.

Thus, the magnitude score ( ) is computed as follows:

where n -is the total number of respondents, -is the j th item score of the i th respondent, m -number of items, and -is the total number of items; counts against j th item at a time. This means that we compute (as in 1) for the i th score of the respondents per j th item. Test of the significant (presented in Table 3 ) influence of the misinformation on the immune perception against COVID-19 in Africa, we conduct a Chi-square test statistics on each of the items. The reference point is the P-value. The Chi-square value is just mathematical output which is used in obtaining the P-value by the computer. Thus, it has nothing to do with the magnitude score in Table 2 . Therefore, if P-value is less than α (= 5%), the null hypothesis (H 0 ) is rejected. That means that the misinformation content/item promotes ignorance about the pandemic. See Table 3 for the P-value of each item.

H 0 : Misinformation content/items do not promote ignorance about the pandemic H 1 : Misinformation content/items promote ignorance about the pandemic α = 5%. Lastly for this section, we conduct a correlation analysis to examine the joint influence of items (misinformation contents) on immune perception against the novel COVID-19. This is presented in Table 4 . 

For the second phase of the data analysis, a more advanced technique is considered to assess the effects of the misinformation items on the COVID-19 immune perception. Hence, we adopt the confirmatory factor analysis (CFA) techniques. CFA is a common structural equation model technique in which one specifies how observed variables (misinformation contents) relate to assumed latent (COVID-19 immune perception) variable(s) [4] , [5] . That is, it allows us to evaluate the structure or extent of the latent variable base on all available evidence. Following the classification in the exploratory section (see Table 2 ), we have the "six items one factor" CFA model. Although all the items significantly promote ignorance of the pandemic, the choice of the six items are based on the misinformation magnitude scale (in Table 2 ) and correlation result (in Table 4 ). This helps to avoid violation of some classical/statistical assumptions (such as multicollinearity etc.).

We adopt the variance standardization method. This method fixes the variance of each factor (items) to 1 but freely estimates all loadings and it is more robust for CFA of items greater than 3. In matrix notation, our CFA model is:

Where; ∑( ) -is the observed covariance matrix, -is the factor loading matrix, -is the variance-covariance matrix of the latent factors, and -is the variance-covariance matrix of the residuals.

The path diagram gives a visual representation of our CFA model because it is a symbolic oneto-one visualization of the measurement model and the model-implied covariance. For general understanding, the circle represents the latent variable -COVID-19 immune perception, squares represent observed items, the one-way arrow represents paths and two-way arrows represent either variances or covariances. Following our matrix notation above, the CFA path diagram is presented below and we adopt R programming for the CFA analysis. Figure 2 : Path diagram output Source: R output from Authors" computation Key: Item2 -Africa Weather and humid system prevent the spread of COVID19, Item3 -Africans are naturally resilient and resistant to most diseases, Item5 -Older people are most prone to the danger of COVID19, younger people are less prone, Item7 -Drinking hot water can prevent/resist/kill COVID19, Item9 -Hightemperature cure COVID19, Item10 -China made the novel Coronavirus (COVID19) to become the global power.

The data does not violate privacy, involve experiment, or human sample. The process of data collection or participation in the survey was voluntary. Thus, all respondents willingly participated in the data. We provided a section in the questionnaire where respondents either choose to be anonymous or provide their name and email to receive the published work from the data.

Can "fake news" impact the stock market

The spread of true and false news online

Systematic Literature Review on the Spread of Health-related Misinformation on Social Media

Exploratory and Confirmatory Factor Analysis: Understanding Concepts and Applications

Structural Equation Model

We sincerely appreciate the international students" community of Kenyatta University (KU) for their supports in participating in this survey and sharing the questionnaire among their countrymen and women to gain wider coverage.We sincerely wish to appreciate our colleagues at Old Ruwenzori; our Ghanaian, Rwandan, Burundian, Ugandan, Cameroonian and Nigerian friends. Essentially, we thank Mr Teh Exodus Akwa, Dr Olusegun Oke (in SA), Mr Emmanuel Owusu, Mr John Zagabe (KU International students Congressman) and Mr Mark Odhiambo for their supports and commitment in sharing our research questionnaire towards achieving the goal of this research. We also thank friends and relatives back home who are committed to our calls at all times. UI-LISA team, and all our respondents, you are most appreciated. 

The authors declare that they have no known competing financial interests or personal relationships which have, or could be perceived to have, influenced the work reported in this article.