Impact analysis of COVID-19 on Nigerian workers’ productivity using multiple correspondence analysis

Abstract

As the COVID-19 pandemic became a global health concern, many business activities have had to adjust to the protocols required to keep people safe, thereby altering the work structures of many professionals. With data gathered from 466 respondents in Nigeria, of which approximately 70% are from the South-West, this study shows how the factors associated with the health crisis have affected work productivity during this period. The snowball survey research design techniques with the two-way interaction model were employed. Multiple Correspondence Analysis was used to analyse and understand multiple and pairwise qualitative factors that influence productivity. The first part of the analysis identified boredom, remuneration, internet availability, fear of COVID-19 and depressing news of COVID-19 as the factors that had significant impacts on workers’ productivity. The second part of the analysis shows how the categories of the five significant factors were either associated or not with productivity. An analysis of each of these factors showed that fear of the disease was associated with slight productivity but access to internet facilities and remuneration were strongly associated with improved work productivity, while boredom and depressing news about COVID-19 were associated with non-productivity during this period. Further evidence also showed that training and new skills acquisition might improve workers’ productivity much more. We, therefore, recommend dynamic skills acquisition, training, and investment in tools and services that will enhance flexibility with the changing work structure that comes because of global crises.

Introduction

Over the past centuries, epidemics and pandemics have severely affected people across the globe [30]. The influenza pandemic was the earliest that occurred at three different times in one century. In 1918, the Spanish flu pandemic resulted in about 100 million deaths within two years, while the Asian flu that occurred in 1957, and the Hong Kong flu, both claimed about 4 million lives [22,53]. In December 2019, a novel coronavirus outbreak, causing severe acute respiratory syndrome, was reported in the Wuhan province of China [23,28]. As the unfolding dynamics of human-to-human transmission emerged, the disease epicentre witnessed outstanding hospitalisation rates within a very short period. Thus, COVID-19 became a public health concern of global scope with infections of over 762 million cases and over 6.8 million deaths [51].

The situation is particularly challenging for developing nations because of their fragile healthcare systems, high poverty, and low literacy rates. The extent of undetected cases in some developing countries in the early days of the outbreak was unknown due to limited access to testing procedures in some dense rural regions [1]. In those regions, many people lived and worked daily under poor conditions with poor hygiene1 . They simply ignore precautionary measures due to indifference and inadequate information about the pandemic. Many who lived in poor communities were unable to regularly wash their hands due to the unavailability of soap and running water. This may have led to a high number of COVID-19 cases and deaths [2]. However, as of 11^th April 2023, 266,665 cases and 3,155 deaths from COVID-19 have been confirmed in Nigeria, with a multi-sectoral national emergency operations centre (EOC) still activated at level 2 to coordinate the national response activities ([4]; Nigeria Centre for Disease [33]).

Apart from the COVID-19 mortality rates, there are several more devastating effects on relative terms. More developed countries struggled to maintain their economic status as businesses and cash flow were reduced, thereby affecting the economies of such countries. On the other hand, the economies of developing countries are the most hit as these depend on advanced economies for importation. The developing countries export lesser products, and their economies suffer slow growth of Gross Domestic Product (GDP), which could lead them to recession [35]. The interruption of international economic activities and businesses negatively affects tax revenue generation in these countries. Goods and services become expensive and unaffordable to low- and middle-class citizens [29]. Thus, there is a decline in profits derived from different sectors such as tourism and direct trade links among countries and continents across the globe [35].

Furthermore, to check the spread of COVID-19, many countries introduced travel bans, lockdowns, and shutdowns of businesses and commercial activities. Schools were forced to close. There was a general shutdown of major cities and towns. Consequently, this led to tapping the power of Information Technology, which allowed many workers and businesses to work from home and deliver their daily tasks/jobs via the Internet. Businesses sell their goods and services on online platforms, while schools started online and electronic learning platforms for teachers and learners [15].

Working from home through remote interactions is, therefore, the new normal for many career professionals in many parts of the world [7]. Unfortunately, this sudden shift is a challenge for less developed countries that are characterized by low wages, poor housing infrastructure, poor access to internet services and other modern facilities that could make working from home bearable [44]. This has implications for higher operational costs borne by business organizations that are then unable to achieve their targets [39]. Amid these global losses experienced, the productivity loss in less developed countries might be more significant than that of advanced countries (Fuentes and Mies, 2012). As the pandemic rages, it is expected to have a direct effect on workers and their productivity levels. Morbidity resulting from the infection implies that work productivity drops significantly until after recovery. This results in sick leave or work absenteeism to seek medical attention. Hence, the productivity of individuals is impacted by these factors [21,52].

This paper examines how COVID-19’s impact on workers in Nigeria has affected their productivity. This will inspire policy formulation to enable workers to be focused, result-oriented, effective and more productive in challenging circumstances. This is vital because of the impact of labour productivity on economic growth, income and supply chains. We examined how this situation caused a change in variables relating to work productivity using the two-way interaction model. The study provides insights that have implications for matters of employment and career decision-making, implications for tasks that should be domiciled in the workplace or decentralised, and implications for required training to improve skills.

The remainder of the paper is structured as follows: Sections 2 and 3 review related studies of the subject of discussion while Section 4 describes the methodology which comprises the research design and methods of the study. Section 5 discusses the findings of the investigation carried out, and section 6 concludes the discussion.

Pandemics, Epidemics and their Effects

Studies conducted after disease outbreaks have linked productivity activities to human health and well-being, as well as work absenteeism, increased workload, quarantine and lockdown. Guimbeau et al. [19] investigate the outcomes of the 1918 Influenza Pandemic and how it affected productivity, among other issues. The study presents evidence that the pandemic had an impact that lasted over twenty years on health, education, and productivity. Since there are lessons to learn from past events of this kind [11], it is therefore important to revisit the ripple effects of past events on ‘well-being on productivity’ to understand the trajectories of newer events.

The studies of Kavet [25] and Gupta et al. [20] show that poor health and well-being can lead to productivity losses during a pandemic. Kavet examined the significance of the influenza pandemic of 1976-1977 and estimated productivity loss as an indirect cost of the pandemic that is linked to high morbidity and mortality due to the disease. Also, after the outbreak of SARS in 2003, Gupta et al. [20] found productivity loss to be an indirect cost associated with quarantines, illnesses, and mortalities during that period. The illnesses and morbidities suffered by individuals led to work absenteeism, which is another factor that leads to productivity loss during a pandemic. An increase in work absenteeism during the peak of winter resulted in reduced productivity and contemplated resignation from work altogether [16,43]. The high risk of exposure to infection and increased workload were the major factors responsible for this [43].

The work absenteeism observed in these studies inadvertently leads to an increase in workload on the few personnel or people remaining. These challenges were also observed in the HIV/AIDS and Ebola outbreaks in South Africa [6,31]. Most of those infected were the younger ones that could have made significant contributions to work productivity. However, absenteeism seems to be a major reason for low productivity during these times. Furthermore, the cost of social distancing to reduce the spread and effects of pandemics require quarantines, school closures, and lockdowns. This contributed more to the productivity loss during the influenza pandemic as it significantly increased absenteeism among workers [38]. Verikios et al. [49] assess the economic activity in response to the influenza pandemic with a quarterly Computable General Equilibrium (CGE) model. Their finding suggests that global economic activities are strongly affected by pandemics with high infection rates.

COVID-19 and Productivity Impacts

The upsurge of the COVID-19 pandemic as reported by the WHO on 11^th March 20202 has led to a new way of life of social/physical distancing, lockdowns, isolation, and loneliness [5]. As the outbreak raises concerns across the globe, researchers continue to evaluate its global impact [3,32,35]. Some have evaluated the productivity impact on academia [13,26,40], firms [7,52], households (S. R. [9]), daily life [48] and labour markets [45]. Productivity refers to the ability to convert inputs into output for profitability [14,47]. Thus, productivity can be said to lie in the ability to meet socioeconomic needs. In addition, productivity can be a function of output, capital stock, labour, and the technological conditions of production [18]. It can also consider a cluster of performance indicators such as the quality of output, input cost, innovation and technology, training, intensity of workplace collaboration, technological efficiency and job design. Thus, Green [18] considered the measure of performance indicators as a better approach to productivity measurement.

The studies of Cui et al. [13], Kim and Patterson [26] and Purwanto et al. [36] focus on the effects of the outbreak on the work productivity of teachers and academia. The studies estimated the cost of school closure on teachers’ productivity and wages during COVID-19, with an expected outcome that online learning would improve the value of labour. By switching to online teaching, teachers gain the advantage of work flexibility and reduction of cost and stress in commuting to work daily [36]. However, the disadvantage of reduced work motivation, health issues due to depression and fatigue, and electricity and internet costs remain, resulting in low productivity. In addition, reduced learning will lead to losses in labour productivity and marginal future earnings of about 15% of the gross domestic product. Also, an assessment of productivity during the COVID-19 outbreak shows a wide difference between the genders in academia [13,26]. Cui et al. and Kim and Patterson used the “difference-in-differences estimation” method to analyse surveys that showed a high gender gap between male and female academics. Their study shows that the females submitted fewer papers when compared to the males [26]. Also, the research productivity of female academics dropped by 13.9% relative to their male counterparts while working remotely, which was due to the imbalance of domestic responsibilities, among others [13].

Bartik et al. [12] reiterated that working remotely can affect the productivity of workers after analyzing surveys from both small and large businesses, and the authors further discovered that productivity loss by remote working can be minimized where workers are better educated and in higher-paid industries. Most of the sampled population engaged by Yang et al. [52] believed that remote working would be sustained in their company even after the pandemic is over, while the authors noted that it might lead to some health problems over time and then result in reduced work productivity. Productivity loss can also be due to certain factors – such as task significance, task interdependence and job insecurity – that make people go to the workplace when ill or factors that make people stay away from work [24]. Tan et al. [46] quantified the immediate psychological effects of returning to work after the easing of the lockdown because of COVID-19. They posited that post-traumatic stress disorder, anxiety, depression, stress and insomnia are factors that affect workers’ input after the resumption of work.

The effects of COVID-19 are also evident in the work productivity of health workers. These effects make them stay away from work or show up while ill because their roles are essential or for fear of losing their jobs [8,42]. For example, O'Kelly et al. [34] shed light on the direct effects of the pandemic on work scheduling and stress among pediatric urologists. Information gathered from the survey reveals 70% disruptions in operations, while altered work schedules disrupted the regular productivity of 90% of respondents. Bao et al. [10] and Ralph et al. [37] investigate COVID-19’s effects on IT professionals’ well-being and productivity while working from home. The quantitative analysis was based on a large dataset of IT developers and analyzed by non-parametric inferential statistics and structural equation modelling. The results suggest that COVID-19 hurts the well-being and productivity of software developers, with both positive and negative effects of working from home [10,37]. This depends on the nature of the projects each worker is handling. The study also reveals that a close relationship exists between productivity and well-being.

These studies suggest that the pandemic could leave a long-lasting impact on the value of future workers and their productivity [7,52]. Productivity in a less developed country could be based largely on determinants such as infrastructures, occupational stress, work-family balance, income, family size and energy resource capacities. How these scenarios could play out differently in a country like Nigeria where certain peculiarities may make them respond differently from the rest of the world remains to be seen. Thus, the focus of this study is to quantify the dependence of the above determinants on capacity development and work productivity, as occasioned by COVID-19, using the two-way interaction model.

Materials and Methods

This study uses the two-way interaction model with the data. It evaluates the factors that affect the work productivity of various professionals during the COVID-19 crisis and lockdown in Nigeria. The collected data have many impact factors that include fear of COVID-19, depressing news of COVID-19, work now boring, remuneration motivates work, satisfactory internet, overall mental health, loss of passion, palliative, regular electricity and alternative electricity. All the impact factors were categorised into the same five categories: Strongly Agree (SA), Agree (A), Neither Agree nor Disagree (NAD), Disagree (D) and Strongly Disagree (SD).

Study Design and Data Collection

The study employed a snowballing type of survey research design for data gathering. This offers a verifiable dataset, an investigation of important variables [41], and a better understanding of the variables that inhibit work productivity during COVID-19. The survey research design is also a valuable tool for assessing opinions and trends and is used because of its affordability and easy access to information. While it is challenging to completely expunge bias and attain 100% accuracy in a survey, surveys are still a powerful and accurate instrument of representative research if they are objective and well-designed. The research methods applied in this study allow the establishment of factors affecting the productivity of various professionals during COVID-19.

An online Google Forms questionnaire was the primary instrument for data collection. Online forms became more desirable because of the government's restricted movement of people. The study was conducted between May and June 2020. The questionnaire asked questions on the level of productivity of each participant to understand the factors that affected their productivity during the COVID-19 lockdown. The questionnaire was divided into four sections. The first section contains socio-demographic factors while the second focuses on the productivity level and work environment. The third section addresses how family and COVID-19 news have influenced work and productivity. The final section seeks to understand the effects of economic status on productivity levels. A five-point Likert scale ranging from Strongly Agree to Strongly Disagree was used. The responses were anonymous, ensuring the confidentiality and reliability of the data. The URL for the survey is: https://bit.ly/3fUmQoL

The study population comprised residents in the six geo-political zones of Nigeria. The survey was collected from a cross-section of workers in academic and non-academic organizations. They were reached by first sending the questionnaire to colleagues, social network groups, and contacts with emails and phones while requesting them to also forward the survey to their contacts. In essence, the snowball sampling technique was used in the gathering of data. This technique works like a chain referral, allowing the leveraging of people-network to reach potential participants that are not within reach or access. The rationale for choosing to snowball is that it is most useful during COVID-19 when potential participants are hard to locate in person for the survey study [17]. The acquired data, from the online survey, was completed by 466 respondents. This sample size was obtained from a sample size calculator, using a moderate margin of error.

Data Analysis

Data was collected on the individual productivity perception that has four categories namely very productive, productive, slightly productive and not productive. The data were analyzed using the Statistical Analysis System (SAS). SAS is used to plot the four categories of the response variables (individual productivity perception) and the five categories of each of the five significant impact factors (see Table 6). The Correspondence Analysis (CA) routine in SAS was used to explore relationships between categorical data, as well as to find a low-dimensional graphical representation of the rows and columns of a cross-tabulation or contingency table. This is vital to make predictions with CA. Each row and column is represented by a point in a plot determined from the cell frequencies.

Table 6.

Analysis of productivity measure of the significant variables

Variable	p-value	Comment	Representation
			SA	A	NAD	D	SD
Work now boring	< 0.0001	Highly significant	wsa	wag	wad	wdg	wsd
Ongoing Remuneration	< 0.0001	Highly significant	rsa	rag	rad	rdg	rsd
Depressing news of COVID-19	= 0.0061	Highly significant	nsa	nag	nad	ndg	nsd
Internet (boost productivity)	= 0.0148	Significant	isa	iag	iad	idg	isd
Fear of COVID-19	= 0.0342	Significant	fsa	fag	fad	fdg	fsd

To make the plot manageable, the name of each categorical variable is abbreviated. Understanding the dataset requires evaluating whether there is a significant dependency between the rows and columns. This can usually be achieved with the chi-square statistical criterion to examine the association between the row and column variables. For the response variable (productivity perception), its categories are abbreviated as follows: pr0 for “not productive”, pr1 for “slightly productive”, pr2 for “productive”, and pr3 for “very productive”. Thus, the set {pr0, pr1, pr2, pr3} represents the levels of productivity. Similarly, the five categories of each of the five significant impact factors are abbreviated. Starting from “Work now boring”, wsd is used for “strongly disagree”, wdg for “disagree”, wad for “neither agree nor disagree”, wag for “agree”, and wsa for “strongly agree”. Thus, the set {wsd, wdg, wad, wag, wsa} represents the five categories of “Work now boring”. Since the categories in all five impact variables are the same, the first letter for each of the remaining four factors is simply replaced. For “Ongoing remuneration”, we have {rsd, rdg, rad, rag, rsa}; for “Depressing news of COVID-19”, we have {nsd, ndg, nad, nag, nsa}; for “Internet (boost productivity)”, we have {isd, idg, iad, iag, isa}; and finally, for “Fear of COVID-19” we have {fsd, fdg, fad, fag, fsa}. These representations are shown in Tables 5 and 6.

Table 5.

Categorical rating of individual productivity

Category	Representation	Frequency	%
Very Productive	pr3	64	13.73
Productive	pr2	168	36.05
Slightly Productive	pr1	165	35.41
Not Productive	pr0	69	14.81

Results and Discussion

Analysis

The data obtained from the questionnaire were analyzed and the results are presented in Tables 1 – 8 and Figures 1 – 2 . The Tables show the productivity rating and measure of workers. Table 1 shows a cross-section of the professions that were considered in the survey. Some categories were collapsed as “Others” due to low sampling frequency to avoid having a long reporting table.

Table 1.

A cross-section of professions surveyed

Occupation	Freq	%
Lecturing/Research	98	21.03
Student (Tertiary)	74	15.88
Entrepreneur	59	12.66
Teaching	38	8.15
Civil Servant	37	7.94
IT Professional (Programmer, Designer, SW Engineer, etc)	28	6.01
Engineer	22	4.72
Finance (Banking, Broker, Insurance etc)	22	4.72
Medical Profession (Doctors, Nurses, Dentist, etc)	17	3.65
Lawyer	11	2.36
Agriculture	10	2.15
Others	50	10.69
TOTAL	466	100

Figure 1.

Chart of variables that motivate productivity level

Figure 2.

Multiple CA of individual productivity perception and significant variables measure

Productivity Measurement

Table 2 shows the productivity assessment of selected workers in Nigeria during the COVID-19 lockdown. The Table shows that 27.4% of respondents are essential workers. During the COVID-19 lockdown, these workers may have been overburdened, forming part of those whose productivity has been affected and job satisfaction diminished. With 60.3% agreeing that there will be more catch-up work to do after the lockdown, it suggests that their productivity did not improve. Though, other factors may then play roles to enhance productivity as shown in Figure 1.

Table 2.

Productivity assessment of selected workers in Nigeria during the COVID-19 lockdown

Variables	Yes		No		Maybe		Not Applicable
	Freq	%	Freq	%	Freq	%	Freq	%
Are you an essential services worker?	126	27.4	334	72.6	-	-	-	-
Do you work from home during the lockdown?	281	60.3	185	39.7	-	-	-	-
If you work from home, is the home environment conducive for your work productivity?	136	29.2	108	23.2	75	16.1	147	31.5
If you still go to work, do you feel safe going to the office?	94	20.2	90	19.3	61	13.1	221	47.4
Has this lockdown improved your productivity compared to before?	94	20.2	279	59.9	93	20	-	-
Did you get job satisfaction as before	128	27.5	265	56.9	73	15.7	-	-
Would you still be able to work if the COVID-19 lockdown continues for the next 6 months?	223	47.9	119	25.5	124	26.6	-	-
Do you think there will be more catch-up work (than usual) to do after lock-down/COVID saga?	281	60.3	94	20.2	91	19.5	-	-

Relationship of Remuneration with Productivity

Table 3 shows the relationship between compensation and reward with productivity. Receiving palliatives did not improve or reduce productivity. On the other hand, receiving remuneration and access to facilities like electricity and the Internet contributed significantly to improving productivity. Also, acquiring skills through training and courses contributed to improving productivity. This infers that the effect of compensation on productivity is significant, as seen in Figure 1 and Table 6.

Table 3.

Effect of compensation on productivity

Variables	Strongly Agree		Agree		Neither Agree nor Disagree		Disagree		Strongly Disagree
	Freq	%	Freq	%	Freq	%	Freq	%	Freq	%
My employer pays me remuneration or salary during this period	136	29.18	127	27.25	75	16.09	74	15.88	54	11.59
My employer offers me full remuneration or salary during this period	118	25.32	111	23.82	87	18.67	81	17.38	69	14.81
My remuneration has motivated my ability to work	88	18.88	105	22.53	110	23.61	98	21.03	65	13.95
I have received palliative during this period	18	3.86	55	11.80	86	18.45	114	24.46	193	41.42
The palliative has/could have motivated my career drive, if given	48	10.30	89	19.10	116	24.89	95	20.39	118	25.32
I have access to regular national electricity supply in my home	47	10.09	100	21.46	96	20.60	117	25.11	106	22.75
I have access to additional/ alternate source(s) of electricity	103	22.10	148	31.76	80	17.17	67	14.38	68	14.59
The network service/internet service coverage in my area is satisfactory	67	14.38	168	36.05	91	19.53	91	19.53	49	10.52
Access to Internet facility has improved my work productivity	74	15.88	173	37.12	101	21.67	76	16.31	42	9.01
Access to other facilities (laboratory, good furniture etc.) has improved my work productivity	41	8.80	94	20.17	113	24.25	120	25.75	98	21.03
I was able to acquire necessary skills needed to help my productivity (e.g., training, courses, more qualification etc.)	84	18.03	162	34.76	88	18.88	77	16.52	55	11.80

Relationship between COVID-19 and the Lockdown with Productivity

Tables 4 and 5 show the relationship between COVID-19 and the lockdown with productivity. The analysis consists of impact factors and individual productivity perception as the response factor with four categories, as shown in Table 5. Five of the nineteen variables fitted to the main-effect logistic model turned out to have significant relationships with productivity as shown in Table 6 .

Table 4.

Effect of COVID-19 and the Lockdown on Productivity

Variables	Strongly Agree		Agree		Neither Agree nor Disagree		Disagree		Strongly Disagree
	Freq	%	Freq	%	Freq	%	Freq	%	Freq	%
COVID-19 news has affected my overall mental health	46	9.87	105	22.53	155	33.26	103	22.10	57	12.23
I am usually fearful of contacting the COVID-19 virus	78	16.74	154	33.05	114	24.46	85	18.24	35	7.51
COVID-19 news makes me depressed and unproductive	35	7.51	87	18.67	174	37.34	110	23.61	60	12.88
COVID-19 news has reduced my passion and career enthusiasm	38	8.15	88	18.88	138	29.61	122	26.18	80	17.17
My career work feels boring to me during this period	54	11.59	119	25.54	129	27.68	112	24.03	52	11.16
The internet service coverage in my area is satisfactory	67	14.38	168	36.05	91	19.53	91	19.53	49	10.52
Access to internet has improved my work productivity	74	15.88	173	37.12	101	21.67	76	16.31	42	9.01

Multiple Correspondence Analysis

Multiple Correspondence Analysis (MCA) was conducted to study the association among the categorical variables. This is a method of analysis that is used to detect and represent underlying structures in a data set by representing data as points in a low-dimensional Euclidean space. Since the interpretation of the input from a categorical variable requires a base (reference) level among its categories to compare other categories with (that is, how the logistic model fitting is done), it may even quickly become more challenging when the factors and the response variable are both categorical with multiple categories. Thus, the same five significant impact factors already identified by the logistic regression model in Table 6 are used. All the results are obtained by assuming the “Strongly disagree” category as the reference level.

The PROC CORRESP multiple analysis reduces the data on the six groups to the two-dimensional plot of Figure 2, using a chi-square criterion to put them into groups. With the centre (0,0) as the reference point, clusters and elements in an observed cluster are assumed to be associated. A group is well defined if it lies on one side of the vertical axis (Dimension 2 axis) or one side of the horizontal axis (Dimension 1 axis). Members on opposite sides of the other axis have opposite signs. The distance of a group from the centre also indicates the strength of the group. The closer to the centre a group is, the weaker the strength of that group.

Group 1: {wad, rad, nad, iad, fad} is all to the left of Dimension 1 axis. They will have the same interpretation regarding productivity since they are all again on the lower side of Dimension 2 axis. Productivity category is missing. That is, none of the responses for all the five factors here is associated with any of the categories of productivity. This makes sense since the responses in this group are “neither agree nor disagree” on any of the five factors.

Group 2: {wsd, nsd, fsd} is located to the extreme right of Dimension 2 axis. They all have the same interpretation regarding productivity since they also lie to the right of Dimension 2 axis. With this group's extreme distance from the centre, this is a strong group in terms of strongly disagreeing on the three factors, namely, “Work now boring”, “Depressing news about COVID-19”, and “Fear of COVID-19”.

Group 3: {isa, fdg, wdg, ndg, pr3} is located to the right of Dimension 2 axis. They all fall below Dimension 1 axis, meaning they have the same type of association with productivity. The isa means that responses here strongly agree on the factor “Internet (boosting productivity)” as associated with “very productive”, pr3. The fdg means that responses here disagree on the factor “Fear of COVID-19” as associated with “very productive”, pr3. Similarly, the wdg means that responses here disagree on the factor “Work now boring” as associated with “very productive”, pr3. The ndg means that responses here disagree on the factor “Depressing news about COVID-19” as associated with “very productive”, pr3.

Group 4: {iag, rag, rsa, pr2} is to the right of Dimension 2 axis and very close to the centre. They all fall below Dimension 1 axis, meaning they have the same type of association with productivity. The iag means that responses here agree on the factor “Internet (boosting productivity)” as associated with “productive”, pr2. The rag means that responses here agree on the factor “Ongoing remuneration” as associated with “productive”, pr2. The rsa means that responses here strongly agree on the factor “Ongoing remuneration” as associated with “productive”, pr2.

Group 5: {fag, idg, rsd, pr1} does not clear any of the two axes as we have seen with the previous groups. This is a pointer that things may be muddled up here. Also, another weakness of this group is its closeness to the centre. The fag means that responses here agree on the factor “Fear of COVID-19” as associated with “slightly productive”, pr1. The idg means that responses here disagree on the factor “Internet (boosting productivity)” as associated with “slightly productive”, pr1. The rsd means that responses here strongly disagree on the factor “Ongoing remuneration” as associated with “slightly productive”, pr1. Association of this group to productivity is weak or slight.

Group 6: {wsa, fsa, nag, wag, isd, rdg, pr0} stays away from the two axes. Thus, this group has an association in the same direction regarding productivity. They have nothing to offer to enhance productivity. The wsa means that responses here strongly agree on the factor “Work now boring” as associated with “not productive”, pr0. The fsa means that responses here also strongly agree on the factor “Fear of COVID-19” as associated with “not productive”, pr0. The nag means that responses here agree on the factor “Depressing news about COVID-19” as associated with “not productive”, pr0. The wag means that responses here agree on the factor “Work now boring” as associated with “not productive”, pr0. The isd means that responses here strongly disagree on the factor “Internet (boosting productivity)” as associated with “not productive”, pr0. The rdg means that responses here disagree on the factor “Ongoing remuneration” as associated with “not productive”, pr0.

Further Analysis

Since computed correlations are not given in correspondence plots, and interpreting the plot is only descriptive, we need something that is as objective as the results from the logistic model. The logistic model can also give a computed probability for each category of productivity based on each possible combination of the categories of the five significant factors. Then, look for the combination of categories that gives the maximum probability for each category of productivity. The results of the investigation are shown in Table 7 , with the interpretation following. It makes sense to compute the minimum probability of ‘not productive’.

Table 7.

Computed probability for productivity category based on a possible combination of factors’ categories

Group	Category of Productivity	Maximum Computed Probability	Corresponding Combination of Factors’ Categories
A	Very Productive (pr3)	0.56184	{fdg, nsd, wsd, rsa, isa}
B	Productive (pr2)	0.50967	{fsd, nsd, wsd, rag, iag}
C	Slightly Productive (pr1)	0.47091	{fad, nag, wag, rsa, idg}
D	Not Productive (pr0)	0.46758 (minimum)	{fad, nag, wsa, rsd, idg}

Group A – {fdg, nsd, wsd, rsa, isa}: This category is associated with high productivity. Disagreeing on fear of COVID-19, strongly disagreeing on depressing news of COVID-19, and strongly disagreeing on work now boring, all regarding productivity, will enhance the chance of high productivity. On the other hand, strongly agreeing on ongoing remuneration and strongly agreeing on internet access to boost productivity enhances the chances of high productivity, according to the data. One can conclude that {fdg, nsd, wsd}, (all disagreeing), are strongly and negatively associated with high productivity, while {rsa, isa}, (both agreeing), are strongly and positively associated with high productivity. Hence, the opposing interpretations. Also, the two subgroups are on opposite sides of the horizontal axis. The categories in this group all lie to the right side of Dimension 1. Linking Group A to the plot, it seems Group 2 from the preceding sub-section is a subset of Group A.

Group B – {fsd, nsd, wsd, rag, iag}: This category is associated with productivity. Strongly disagreeing on fear of COVID-19, strongly disagreeing on depressing news of COVID-19, and strongly disagreeing on work now boring, all regarding productivity will enhance a chance of high productivity. On the other hand, agreeing on ongoing remuneration and agreeing on internet (boost productivity) enhances a chance of productivity, according to the data. One can conclude that {fdg, nsd, wsd}, (all disagreeing), are strongly and negatively associated with productivity while {rag, iag}, (both agreeing), are positively associated with productivity. The categories in this group all lie on the right side of Dimension 1. Also, the two subgroups are on opposite sides of the other axis. Hence, the opposing interpretations. Linking Group B to the plot, there is no obvious inconsistency between Group B and Group A.

Group C – {fad, nag, wag, rsa, idg}: This category is associated with slight productivity. This is a weak combination in enhancing productivity. Neither agreeing nor disagreeing on fear of COVID-19, agreeing on depressing news of COVID-19, agreeing on work now boring, and disagreeing on internet (boost productivity), regarding productivity, do not enhance the chance of productivity. It is only rsa, strongly agreeing on ongoing remuneration, that has an effect on productivity. Thus, slight (or weak) productivity is the outcome. Regarding the plot, four of the five categories are on the left side of Dimension 1 while only rsa is on the right side. The subgroups {nag, wag, idg}and {fad, rsa} are on the opposite sides of the other axis, meaning they have opposing interpretations, as done above.

Group D – {fad, nag, wsa, rsd, idg}: This category is associated with no productivity. This is a very weak combination in enhancing productivity. Neither agreeing nor disagreeing on fear of COVID-19, agreeing on depressing news of COVID-19, strongly agreeing on work now boring, strongly disagreeing on ongoing remuneration, and disagreeing on internet (boost productivity), all regarding productivity, do not enhance the chance of productivity at all. Thus, there is no productivity enhancement in all of these. Regarding the plot, all five categories are on the left side of Dimension 1, supporting the same direction all of them have been interpreted on.

Reflection

This analysis is in two parts. The first part identifies factors that have a significant impact on productivity using p-value probability and odd ratios. Multiple logistic regression is applied to achieve this identification, and five impact factors are identified as significant with p-value < 0.05.

The second part of this analysis investigates how the categories of the five significant factors are related or not related to the categories of the response variable, productivity. The investigation of association is done via Multiple CA whose output is in Figure 2, discussed above. The 25 categories of the five significant factors and the 4 categories of the response variable are reduced to a 2-dimensional plot, using a chi-square criterion to put them into groups. We can roughly identify SIX groups as discussed. Note the careful use of the word “associated” in the interpretation of each group. This graphical method is only descriptive.

The two dimensions and the centre of the plot are applied to form a group with one of the dimensions (most likely the one with a higher percentage), to check the type of association with the second dimension. Also, the distance from the centre of the plot is applied to judge the strength of the group. The farther away a group is from the centre, the stronger it is in terms of association.

The implication, therefore, is that the five impact factors discussed – boredom, remuneration, internet availability, fear of COVID-19 and depressing news of COVID-19 – have significant impacts on workers’ productivity. Also, their health – physically and mentally – could also be impacted substantially.

Conclusion

This study assessed workers’ productivity in Nigeria during the pandemic with data collected from 466 respondents across the country. Multiple correspondence analysis was used to analyse and interpret the results in six groups. Three of the six groups have an association with a good level of productivity, while the other three do not. From these groups, it is observed that the five impact factors –boredom, remuneration, internet availability, fear of COVID-19 and depressing news of COVID-19 – have both positive and negative significant impacts on workers’ productivity. However, access to internet facilities is most strongly associated with work productivity during the COVID-19 pandemic. Further analysis used the logistic model to compute the probability and identify the maximum probability for each category of productivity based on each possible combination of the categories of the five significant factors. Evidence suggests that the COVID-19 pandemic is a call to professionals to be dynamic in skills acquisition, as well as training that will make workers more productive, flexible and adaptable to the changing work structure in times of crisis. In addition, motivation and further investment in work tools and services will enhance work in times like this.

Uncited References

[27,50]

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Editor: DR B Gyampoh