INTRODUCTION

This chapter holds the background on which the study was found, the statement of the problem, the conceptual framework, the significance of the study to various people and fields, the scope and limitations of the study, and the definition of terms used.

Background of the Study

The first Covid-19 case in India was reported on January 30, 2021 from Kerela; ever since that day, the number of cases drastically increased each day. Consequently, the Indian government had taken several measures to contain the spread of the said virus. One of these is by having a lockdown. Lockdowns in India had compelled its people to stay at home and just rely on media for virtual gatherings, being updated from the news, and other forms of entertainment. For this reason, Indians have been more dependent on the media.

Media can be very useful and beneficial for people. Today’s generation is indeed reliant on different forms of media. There is no doubt that its impact on people is huge. With regards to its impact on the people, it is important to pay attention to its effects on the mental health of the people, especially the Indians, since they are the main participants of this study.

Mental health is a state of well-being in which the individual realizes his or her own abilities, can cope with the normal stresses of life, can work productively and fruitfully, and is able to make a contribution to his or her community. Mental health of the Indians amid the pandemic was observed to be negatively affected, since there was the fear of being infected11 Purohit (2021). But the mental well-being of the people is not just affected by fear. There are things that can influence it, such as the time a person spends on media.

For this study, the focus is on the impact of daily media hour usage to the mental health of the Indian Public during the initial phase of the global health pandemic crisis, as dictated by found dataset from the study of Mukherjee, Maity, and Chatterjee (2021), entitled Media use pattern as an indicator of mental health in the COVID-19 pandemic: Dataset from India.

During that said phase of the pandemic, there were fake news spreading regarding the Covid-19; this includes the rumored intensity of the virus and some exaggerated information. There were also some authenticated facts found about Covid19 that were circulating in the media. With this, the amplification of media usage is suspected by the researchers to be a factor for the mental anxiety and mental wellbeing of the Indians during the first month of the pandemic. Such assumption was grounded in the research findings22 Saxon et al. (2019) in the Ebola outbreak in Africa. This had shown the impact of media amplifications observed from the people of the USA, where uncertainties about the threat of Ebola led to stress and anxiety.

Statement of the Problem

This study aimed to determine the influence of media usage or engagement on the mental health of people in India during the initial phase of the global health pandemic crisis. Specifically, it sought to answer the following questions:

1. What are the respondents’ demographic profile according to their: a) media usage level, b) mental well-being level, and c) mental anxiety level?
2. What is the significant relationship between the respondents’ media usage and well-being score?
3. What is the significant relationship between the respondents’ media usage and anxiety score?

Hypotheses

1. \(H_0\): There is no significant relationship between the media usage and well-being score during the initial phase of pandemic.
2. \(H_0\): There is no significant relationship between the media usage and anxiety score during the initial phase of pandemic.

Conceptual Framework

⊕Figure 1. Conceptual Paradigm.

The system approach (Input-Process-Output system) was used in describing the conceptual framework of the study. As shown in Figure 1, the input consisted of the respondents’ media usage in hours per day, well-being score, and anxiety score. The process contained the means on how the study would determine the relationship between the media usage and well-being score of respondents as well as the relationship between the media usage and anxiety score. The well-being level was determined through the use of the Warwick-Edinburgh Mental Well-Being Scale, while the anxiety level was known through the use of Beck’s Anxiety Inventory Scale. Results based on the data that were gathered from the input were listed under the output area. The researchers also provided recommendations based on the findings of the study.

Significance of the Study

To the people of India. This study will benefit the general public of India as it will shed some light on the impact of their media usage on their mental health state, particularly during the initial phase of the global pandemic health crisis.

To the mental health professionals. This study is beneficial to the mental health advocates as it serves as a reference for the certain cases they’ll be handling, especially since the pandemic still holds up to this day.

To the policymakers. The provided data could help the policymakers of India in making necessary actions regarding the mental health conditions during the COVID-19 pandemic.

To the future researchers. This study will benefit the future researchers as it will serve as a guide or a reference data, and gives an overview of their topic. It can also be used as the background of the study or related literature for their studies relative to this research.

Scope and Delimitation of the Study

This study focused on the relationship between the media usage pattern and the mental health of people in India during the initial phase of the global health pandemic crisis. The researchers used the dataset from the study entitled Media use pattern as an indicator of mental health in the COVID-19 pandemic: Dataset from India33 Mukherjee, Maity, and Chatterjee (2021). They gathered the data on levels of media engagement, mental well-being level, and mental anxiety level among some of the Indian adults through the respondent-driven convenient sampling method. Since the target population was Indian nationals, who were adults, fluent in English language, and had access to social media, other sectors of the population were not considered in this study. The data was collected by Mukherjee, Maity, and Chatterjee (2021) through a web-based cross-sectional online survey that was conducted three weeks after the enforcement of a nationwide lockdown from April 16 to 22, 2020. The online survey was composed of questions for media usage and standardized questionnaires, such as Warwick-Edinburgh Mental Well-Being Scale and Beck’s Anxiety Inventory (BAI) Scale, which can provide more reliable results since their validity were checked by experts beforehand. Other demographic variables were not considered in this study.

Definition of Terms

COVID-19

A disease caused by a new strain of coronavirus where people with this disease experience moderate respiratory illness but can recover without needing any special treatment.

Lockdown

A restriction policy for people or communities to stay where they are, usually due to specific risks to themselves or to others if they can move and interact freely.

Media Use

The respondents’ engagement in the media in hours per day.

Mental Anxiety

An anxiety disorder that leads to extreme nervousness, fear, and worry which can negatively affect how a person behaves and processes emotions.

Mental Well-being

A state of health in which an individual realizes his or her own potential to withstand the normal stresses of life and can contribute to his or her community by working productively.

Pandemic

An outbreak of a disease that occurs over a wide geographic area (such as multiple countries or continents) and typically affects a significant proportion of the population

Social Media

A computer-based technology that facilitates the sharing of ideas, thoughts, and information through the building of virtual networks and communities.

Virtual Gatherings

Is a meeting that happens online rather than physically with all the participants in the same meeting room.

Virus

A microorganism, smaller than bacteria, that invades and grows in living cells as they use their chemical machinery to replicate and to keep themselves alive.

METHODOLOGY

This chapter contains the research design, the sample and sampling technique, the research instrument, the data gathering procedure, and the statistical treatment used by the researchers.

Research Design

This study used a descriptive-correlational method as the research design to be able to arise with sufficient interpretation of findings about the significant relationship of the media usage pattern and the mental health of people in India during the global health pandemic crisis. Descriptive-correlational method was used because it was designed to determine the existence and degree of a relationship between two or more variables, in this case, the relationship between the respondents’ media usage and well-being score, as well as the relationship between the respondents’ media usage and anxiety score.

Research Respondents

The data gathered were from adult resident citizens of India, without regard to the state (province) of residence, caste, creed, religion, and sex; the researchers did not consciously exclude any social group in the data collection process.

The profile of the sample population is characterized by certain demographics, specifically age, gender, habitat and educational qualification. The respondents were categorized according to their age (young, middle-aged, or old) as this demographic characteristic determines a certain portion of the active social media users in India. They were also categorized according to their gender (female, male, or intersex), habitat (urban, urban-municipal, or rural) and education qualification (undergraduate, postgraduate, or above postgraduate). Such demographic-based categorization was conducted as these factors were influential in their psychosocial reflection, maturity, and responses, respectively.

The sample has 426 respondents, who were in the ages of 18 years and above. Such range of age was set as one of the characteristics of the concerned sample since out of the 300 million active social media users in India, a figure found according to the survey conducted by UN Volunteer India and Sports (2019), 93% of it are concluded to be adults based on the survey report of NapoleonCat (2019).

The ages of 18 years and above were also mindfully chosen by Mukherjee, Maity, and Chatterjee (2021) as a characteristic of the sample, as it considers the effective English knowledge of the respondents, which greatly plays a role in the psychosocial perception of the COVID-19 information found in media.

The age-distribution of the total adult social media users in India are 80% young, 17% middle aged, and 3% old aged, according to reports of NapoleonCat (2019). Majority of them are men (78%) and so the female social media users are the remaining 22%. As per other demographic characteristics such as habitat, 29% of total social media users in India are from rural areas, 38% are from urban-municipal, and 33% are from metropolitan, according to Internet and India (2014). With respect to the education qualification based on the average enrollment ratio in higher education, 80% of the 300 million Indian social media users are undergraduates, while 11% of them are postgraduates and 1% are above-postgraduates, according to Ministry of Human Resource Development (2019). These percentage-distributions of the adult social media users in India are the ideal percentage-distribution of the sample of respondents, for Mukherjee, Maity, and Chatterjee (2021), that could participate in the web-based cross-sectional online survey

However, due to the imposed lockdown, the data collection method of Mukherjee, Maity, and Chatterjee (2021) is limited. The ideal percentage-wise distribution mentioned earlier could not be attained in the demographic profiling of the respondents. There was an overrepresentation in the female, middle-aged, urban municipal, postgraduates, and above postgraduate categories, as depicted in Table 1 below.

⊕Table 1. Descriptive statistics of demographic characteristics (N = 426) and percentage-based representation of target population.

Sampling Technique

Mukherjee, Maity, and Chatterjee (2021) used a respondent-driven convenient sampling method for the determination of sample. The approximate target population is 2.5 million, considering the percentage of the adult Indian social media users concluded from certain survey reports mentioned earlier. The portion of this percentage that has effective knowledge in English Language was considered as well to arise with the estimate of the target population (2.5 million). Because the target population is too large for Mukherjee, Maity, and Chatterjee (2021) to conduct a simple random sampling, a respondent-driven convenient sampling was applied instead in determination of the sample. Respondent-driven convenient sampling employs a non-probability sampling technique, where samples that were conveniently accessible to the Mukherjee, Maity, and Chatterjee (2021) were selected. The design followed in gathering the samples was mainly grounded on the respondent’s action of forwarding the survey link to their contacts. This is great since the sample would then have a wide range. The design was also beneficial as it reduced the sampling bias in the data collection method and had also lessened the dependence on the initial convenience sample.

Research Instrument

⊕Access the survey here on on https://data.mendeley.com/datasets/sktz4xv4vh/4

A web-based cross-sectional online survey was administered to determine the mental well-being level and mental anxiety level among the Indian resident adults and their engagement, in terms of the amount of time spent, with the media.

The questionnaire was divided into four parts. The first part was composed of the demographic information of the respondents. In the second part, the respondents were asked on the amount of time spent with the media during the lockdown period.

The third part included the Warwick-Edinburgh Mental Well-Being Scale to assess the mental well-being level of the respondent. Fourteen questions were answered based on a five-point Likert type scale, with scores ranging from “none of the time”= 1 to “all the time”= 5, for items such as “I’ve been feeling confident.” The total score obtained was used as the mental well-being score. The mental well-being levels were categorized into high (score, >56), medium (score, 47–56), and low (score, <47).

The last part was the Beck’s Anxiety Inventory Scale, which consisted of 21 self-reported items (four-point scale) that were used to assess the mental anxiety level of the respondents during the pandemic. The summated score of each respondent was used as the mental anxiety score. The mental anxiety levels were also categorized in the BAI scale itself as high (score, >35), medium (score, 22–35), and low (score, <22).

Data Collection Procedure

The web-based cross-sectional online survey concerned with the daily media hour usage, the mental well-being state, and the mental anxiety state among the Indian sample was conducted three weeks after the implementation of the nationwide lockdown. The said date of data collection was preferred for the procedure since it was the first phase of Covid-19 transmission in India. In the first phase, enormous information about the said viral infection’s intensity, both legitimate and manufactured, were circulating in media that could impact the mental health of the general public during the pandemic44 Mukherjee, Maity, and Chatterjee (2021).

The survey was administered as a Google form questionnaire. The distribution of this Google form’s link was done on social media channels and was only handed out to the Indian people with WhatsApp or Facebook accounts, and with enough knowledge about the English language. All the survey’s participants had provided some necessary information and informed consent before the start of the survey. As for the actual survey, the respondents filled in the four parts mentioned earlier in the Research Instrument section. The participants were also requested to forward the survey’s link to some of their contacts, since the chosen research sampling design is convenient-sampling technique.

Statistical Treatment

Having this study as descriptive-correlational, such that the researchers are interested in discovering the impact of the media usage in the mental health of the people in India during the initial phase of the global pandemic health crisis, the use of Simple Linear Regression Analysis was then preferred as the appropriate statistical treatment. Through Simple Linear Regression Analysis, the (a.) significant causal relationship between the media usage (hours/day) and the well-being score of the respondents during the initial phase of pandemic could be determined, as well as (b.) the significant causal relationship between the media usage (hours/day) and the anxiety score of the respondents. In this statistical tool, the media usage in terms of hours per day will be treated as the independent variable (x), while the well-being score and the anxiety score of the respondents will be treated as the dependent variables (y)’s.

The Linear Regression Analysis on these variables was performed in R-Studio, specifically with the “lm” function, in order to maximize and take advantage of the stats package of the software. It is also best and smart to use R-Studio in performing the statistical test, since the researchers were dealing with hundreds of respondents’ data.

A Breusch-Pagan Test was used to check if heteroscedasticity is present in errors of the regression model. The homoscedascity of the errors of a regression model is one of the standards to determine whether the said model is adequate or not. Additionally, Pearson correlation coefficient was used in this research to determine the strength of the correlation and thus linear relationship between the media usage of the respondent & their well-being, and the daily media hour usage & anxiety level of the respondents, respectively, during the initial phase of the pandemic pandemic. With the use of Pearson correlation coefficient, there would be 5 degrees of correlation which includes:

• Perfect: The value is near ±1.
• High Degree: The value lies between ±0.50 and ±1.
• Moderate Degree: The value lies between ±0.30 and ±0.49.
• Low Degree: The value lies below ±0.29.
• No correlation: The value is 0.

RESULTS AND DISCUSSION

This chapter presents all the gathered data for examination, tabulation, and analysis. It also explains the specific answers to the objectives in a detailed manner, such as using pie charts and tables. For a thorough understanding of the study’s findings and results, the interpretations and methods of analysis were cautiously laid out.

---

Simple Linear Regression Analysis for Well-Being Score Data of Indian Public during the Initial Phase of the Global Pandemic Health Crisis

⊕Table 2. Regressional statistics on media hour usage and well-being scores of respondents.

Table 2 summarizes some valuable estimates of parameters for the simple linear regression analysis of the well-being score data. The data were tested to know if there is a significant relationship between the respondents’ well-being scores and their media use in hours per day. Here is the scatter diagram of y (well-being scores) versus x (media use in hours per day).

⊕Table 3. Coefficients of the intercept and the regressor variable (x): media usage in hours

	Coefficients for the intercept and the regressor
Coeffcient	Estimates
(Intercept)	56.61
MediaUse	-0.63
Observations	426

A. Fitted Simple Linear Regression Model

The fitted simple linear regression model for the Well-Being Score data is: \(\hat{y} = 56.61 + -0.63x\).

Practical Interpretation: This means that best fit simple linear regression model for the well-being scores of the respondents (y) for any and their media use in hours/day (x) is \(\hat{y} = 56.61 + -0.63x\).

B. Test of Significance for Regression for Well-Being Score and Media Use

To know if there is a significant linear relationship between the the well-being score and the media use in hours per day, the researchers test the significance of regression model for the well-being score data,

\(\hat{y} = 56.61 + -0.63x\).

at a significance level of \(\alpha = 0.05\).

The hypotheses are:

• \(H_0: \beta_1 = 0\)

• \(H_1: \beta_1 \neq 0\)

The test statistic is:

\(\lvert t_0 \rvert = \frac{ {\hat{\beta}_1}}{\sqrt{se({\hat{\beta}_1})} }\)

Rejection Criteria

Using P-value:

• Reject \(H_0: \beta_1 = 0\) if the P-value is less than 0.05.

Using fixed significance level:

#critical value
abs(qt(p = 0.025, df = 424))

[1] 1.965575

• Using \(\alpha = 0.05\) and a fixed significance level test, we would reject \(H_0: \beta_1 = 0\) if \(\lvert t_0 \rvert > t_{0.025,424} > 1.965575\).
• Failing to reject the hypothesis implies that there is no linear relationship between the Well-Being Score (y) of the Respondents and their Media Usage in hours per day (x), during the initial phase of the global pandemic health crisis.

Computation of Test Statistic \(|t_0|\)

⊕Table 4. Test statistic score and p-value of the regressor variable (x): media use in hours.

	Test statistic score and p-value of the regressor variable (x)
Predictors	Estimates	T-statistic \((t_0)\)	P-Value of \(|t_0|\)	df
MediaUse	-0.63	-3.46	0.001	424.00
Observations	426

• From the Table 4, the test statistic is \(\lvert t_0 \rvert = 3.46.\)

• The p value is \(P = 0.001\).

Conclusion:

Using fixed significance level test, the researchers reject \(H_0: \beta_1 = 0\) at the 0.05 level of significance since \(\lvert t_0 \rvert = 3.46 > 1.965575\).

Using the p-value, since \(0.001\) is less than \(0.05\), the researchers reject \(H_0: \beta_1 = 0\) at the 0.05 level of significance, and adopt \(H_1: \beta_1 \neq 0\)

Practical Interpretation: Since \(\beta_1 \neq 0\), there is a linear relationship between well-being score and media use in hours per day. Since \(\beta_1 = -0.6271x\), it can be concluded that media hour usage has a negative relationship with the well-being score during the initial phase of the Covid19 pandemic. This implies that a high media hour usage per day by the respondent during the said phase may indicate a low well-being level of the individual.

C. Adequacy of the Regression Model for the Well-Being Score (y) and Daily Media Hour Usage (x) During the Initial Phase of Pandemic

1. Mean Squared Error

The mean squared error (\(\sigma^2\)) of the Well-Being Score data could be determined through the statistical tool of the analysis of variance.

⊕Table 5. ANOVA Table for Well Being Score Data.

–	Df	Sum Sq	Mean Sq	F-value	Pr(>F)
MediaUse	1	925	925.10	11.978	0.0005931
Residuals	424	32748	77.24

From the R computation, the mean squared error (\(\sigma^2\)) is equal to \(77.24\).

Practical Interpretation This value implies that the average squared difference of the actual observed values (personal response of the respondents), and the fitted values acquired using the regression model for the well-being score and daily media hour usage during the initial phase of pandemic is 77.24.

2. Assumptions about the errors

• Density Plot

We can see that the density plot roughly follows a bell shape, although it is slightly skewed to the left. Such bell shape in figure 4 means that the residuals are normally distributed.

• Residual vs fitted Plot

Based on the residual plot depicted by the figure 5, the residuals or errors seemed to follow a particular trend, which means that their variance is not constant. To verify this assumption, the researchers performed Breusch-Pagan Test to the data set, to check for the Homoscedasticity of the residuals.

Breusch-Pagan Test

The hypotheses are:

• \(H_0\): Homoscedasticity is present (the residuals are distributed with equal variance)

• \(H_1\) :Heteroscedasticity is present (the residuals are not distributed with equal variance)

Rejection Criteria

Using P-value:

• Reject \(H_0\) if the P-value is less than 0.05.

R Computation

    studentized Breusch-Pagan test

data:  lmWellBeingScore
BP = 12.699, df = 1, p-value = 0.0003658

Since the p-value of the Breusch-Pagan test score is 0.0003658, which is less than the significance level of \(\alpha=0.05\), the researchers reject the null hypothesis that claims the Homoscedasticity of the residuals. Instead, the researchers adopt the alternative hypothesis which claims for the heteroscedasticity of the errors.

Practical Interpretation: Although the errors follow a normal distribution in the density plot as depicted in figure 3, the errors do not have a constant variance upon performing the Breusch-Pagan Test. This means that the model for the daily media hour usage and well-being scores during the initial phase of the global pandemic health crisis, can’t be used confidently in practice.

3. Coefficient of Determination \(R^2\)

The coefficient of Determination is the amount of variability in the data accounted by the regression model55 Montgomery and Runger (2010).

⊕Table 6. Regressional statistics on media hour usage and well-being scores of respondents.

In Table 6, the \(R^2\) value is equal to 0.027. This practically that means that the model accounts only for 2.7% variability of the data. This value is expected based on the findings in the Breusch-Pagan Test.

Overall Practical Interpretation in the Adequacy of the Regression Model of the Daily Media Hour Usage (x) and Well-Being Scores (y) of Indian Public during the Initial Phase of the Global Pandemic Health Crisis

Although the errors have a normal distribution, they don’t have a constant variance. The coefficient of determination is also relatively low. Thus, the regression model for the daily media hour usage and the well-being score, is inadequate to be confidently used in practice for other datasets and predicting new observations.

D. Pearson Correlation Coefficient

Since the regression model is inadequate, the researchers opted to measure the degree of the correlation or linear relationship found in the test for significance of regression, which was a negative linear relationship between the daily media hour usage and the well-being score of an individual during the first phase of the Covid19 pandemic.

Test the Significance of Correlation Coefficient

cor(data1$MediaUse , data1$WellBeingScore ,method ="pearson") 

[1] -0.1657495

The computed Pearson correlation coefficient between the well-being scores and daily media hour usage is -0.1657495.

Test the Significance of Correlation Coefficient

The P value of the computed Pearson correlation coefficient (r) is 0.00092, at a significance level of \(\alpha=0.05\) and degrees of freedom of 424. Since the p-value is less than the significance level, then the correlation is statistically significant.

Since the computed correlation coefficient is significant, (\(r=-0.16\)), the researchers could then conclude that the well-being scores and daily media hour usage have a low negative correlation, and thus a weak negative linear relationship.

---

Simple Linear Regression Analysis for Anxiety Score Data of Indian Public during the Initial Phase of the Global Pandemic Health Crisis

⊕Table 7. Regressional statistics on media hour usage and anxiety scores of respondents.

Table 7 summarizes some valuable estimates of parameters for the simple linear regression analysis of the Anxiety score data. The data were tested to know if there is a significant relationship between the respondents’ anxiety scores and their media use in hours per day. Here is the scatter diagram of y (anxiety scores) versus x (media use in hours per day).

⊕Table 8. Coefficients of the intercept and the regressor variable (x): media use in hours

	Coefficients for The intercept and the regressor
Coeffcient	Estimates
(Intercept)	4.39
MediaUse	0.93
Observations	426

A. Fitted Simple Linear Regression Model

The fitted simple linear regression model for the Well-Being Score data is: \(\hat{y} = 4.39 + 0.93x\).

Practical Interpretation: This means that best fit linear regression model for the anxiety scores (y) of the respondents and their daily media hour usage (x) during the initial phase of the pandemic is given by the equation: \(\hat{y} = 4.39 + 0.93x\).

B. Test of Significance of Regression for Anxiety Score and Media Use

To know if there is a significant linear relationship between the anxiety score and the media use in hours per day, the researchers test the significance of regression model for the anxiety score data,

\(\hat{y} = 4.39 + 0.93x\).

at a significance level of \(\alpha = 0.05\).

The hypotheses are:

• \(H_0: \beta_1 = 0\)

• \(H_1: \beta_1 \neq 0\)

The test statistic is:

\(\lvert t_0 \rvert = \frac{ {\hat{\beta}_1}}{\sqrt{se({\hat{\beta}_1})} }\)

Rejection Criteria

Using P-value:

• Reject \(H_0: \beta_1 = 0\) if the P-value is less than 0.05.

Using fixed significance level:

#critical value
abs(qt(p = 0.025, df = 424))

[1] 1.965575

• Using \(\alpha = 0.05\) and a fixed significance level test, we would reject \(H_0: \beta_1 = 0\) if \(\lvert t_0 \rvert > t_{0.025,424} > 1.965575\).
• Failing to reject the hypothesis implies that there is no linear relationship between the Anxiety Score (y) of the Respondents and their Media Usage in hours per day (x), during the pandemic.

Computation of Test Statistic \(|t_0|\)

⊕Table 9. Test statistic score and p-value of the regressor variable (x): media use in hours per day.

	Test statistic score and p-value of the regressor variable (x)
Predictors	Estimates	T-statistic \((t_0)\)	P-Value of \(|t_0|\)	df
MediaUse	0.93	4.17	<0.001	424.00
Observations	426

• From the Table 9, the test statistic is \(\lvert t_0 \rvert = 4.17\)

• The p value is \(P < 0.001\).

Conclusion:

Using fixed significance level test, we reject \(H_0: \beta_1 = 0\) at the 0.05 level of significance since \(\lvert t_0 \rvert = 4.17 > 1.965575\).

Using the p-value, since \(P<0.001\), thus, it’s also less than the significance level, \(\alpha=0.05\). With this, we reject \(H_0: \beta_1 = 0\) and adopt \(H_1: \beta_1 \neq 0\)

Practical Interpretation: Since \(\beta_1 \neq 0\), there’s a linear relationship between anxiety score and media use in hours per day. Since \(\beta_1 = 0.93x\), it can be concluded that media usage has a positive relationship with the anxiety score during the initial phase of the pandemic. A high number of hour usage of medias during the said phase may indicate a high anxiety level of an individual.

C. Adequacy of the Regression Model

1. Mean Squared Error

The mean squared error (\(\sigma^2\)) of the Anxiety Score data could be determined through the statistical tool of the Analysis of variance (ANOVA).

⊕Table 10. ANOVA table for anxiety score data.

–	Df	Sum Sq	Mean Sq	F-value	Pr(>F)
MediaUse	1	2041	2041.40	17.392	3.69e-0.5
Residuals	424	49768	117.38

From the R computation, the mean squared error (\(\sigma^2\)) is equal to \(117.38\).

Practical Interpretation This value implies that the average squared difference of the actual observed values (personal response of the respondents), and the fitted values acquired using the regression model for the daily media hour usage and the anxiety score during the first phase of global pandemic is 117.38

2. Assumptions about the errors

• Density Plot

Figure 8 demonstrates a bell-shaped density plot for the residuals, although it is slightly skewed to the right. Such bell shape in figure 8 means that the residuals are normally distributed.

• Residual vs fitted Plot

Based on the residual plot depicted by the figure 9, the residuals or errors seemed to follow a particular trend, which means that their variance is not constant. To verify this assumption, the researchers performed Breusch-Pagan Test to the data set to check for the Homoscedasticity of the residuals.

Breusch-Pagan Test

The hypotheses are:

• \(H_0\): Homoscedasticity is present (the residuals are distributed with equal variance)

• \(H_1\) :Heteroscedasticity is present (the residuals are not distributed with equal variance)

Rejection Criteria

Using P-value:

• Reject \(H_0\) if the P-value is less than 0.05.

R Computation

    studentized Breusch-Pagan test

data:  lmAnxietyScore
BP = 9.7322, df = 1, p-value = 0.001811

Since the p-value of the Breusch-Pagan test score is 0.001811, which is less than the significance level of \(\alpha=0.05\), the researchers reject the null hypothesis that claims the Homoscedasticity of the residuals. Instead, the researchers adopt the alternative hypothesis which claims for the heteroscedasticity of the errors.

Practical Interpretation: Although the errors follow a normal distribution in the density plot as depicted in figure 8, the errors do not have a constant variance upon performing the Breusch-Pagan Test. This means that the regression model can’t be used confidently in practice.

3. Coefficient of Determination \(R^2\)

The coefficient of Determination is the amount of variability in the data accounted by the regression model66 Montgomery and Runger (2010).

⊕Table 11. Regressional statistics on media hour usage and anxiety scores of respondents.

In Table 11, the \(R^2\) value is equal to 0.039. This practically that means that the model accounts only for 3.9% variability of the data. This value is expected based on the findings in the Breusch-Pagan Test.

Overall Practical Interpretation in the Adequacy of the Regression Model for the Daily Media Hour Usage (x) and Anxiety Score (y) of Indian Public during the Initial Phase of the Global Pandemic Health Crisis

Although the errors have a normal distribution, they don’t have a constant variance. The coefficient of determination is also relatively low. Thus, the regression model for the daily media hour usage and the anxiety scores during the said phase is inadequate to be confidently used in practice for other datasets and predicting new observations.

D. Pearson Correlation Coefficient

Since the regression model is inadequate, the researchers opted to measure the degree of the correlation or linear relationship found in the test for significance of regression, which was a positive linear relationship between the daily media hour usage and anxiety score during the initial phase of the COVID-19 pandemic.

cor(data2$MediaUse , data2$AnxietyScore ,method ="pearson") 

[1] 0.198499

The computed Pearson correlation coefficient between the anxiety score and daily media hour usage is 0.198499

Test the Significance of Correlation Coefficient

The P value of the computed Pearson correlation coefficient (r) is 0.000037, at a significance level of \(\alpha=0.05\) and degrees of freedom of 424. Since the p-value is less than the significance level, then the correlation is statistically significant.

Since the computed correlation coefficient is significant, (r=0.000037), we could then conclude that the anxiety scores and daily media hour usage during the initial phase of Global Pandemic Health Crisis, have a low positive correlation and thus weak positive linear relationship.

---

The respondents’ demographic profile according to their: a) media usage Level, b) mental well-being level, and c) mental anxiety level

Respondent’s demographic profile according to their Media Usage Level

Figure 10 indicates the frequency and distribution of respondents according to their media usage. Based on the figure, 30.28% of the respondents have high media usage, 33.8% of the respondents have an average amount of media usage, and 35.92% of the respondents only have a low amount of media usage. The figure also shows that the respondents are evenly distributed when it comes to their media consumption.

Respondent’s demographic profile according to their Well-Being Level

Figure 11 presents the frequency and distribution of respondents according to their well-being level. In this category, 28.4% of the respondents fall on the high level of well-being, 45.78% comprise the middle level of well-being, and 25.82% have a low level of well-being. This data shows that the majority of the respondents fall on the middle level of well-being during the pandemic.

Respondent’s demographic profile according to their Anxiety Level

Figure 12 shows the frequency and distribution of respondents according to their anxiety level. The figure shows that only 6.57% of the respondents experienced high levels of anxiety during the pandemic, 11.5% of the respondents showed middle levels of anxiety, and 81.93% of the total number of respondents showed low levels of anxiety. In this category, the majority of the respondents only experienced mild anxiety compared to the other levels.

Summary Table ⊕Table 12. Summary table for the respondents’ demographic profile according to their: a) media usage Level, b) mental well-being level, and c) mental anxiety level

Category	Low Level	Middle Level	High Level
Media Usage	35.92%	33.8%	30.28%
Well-Being Score	25.82%	45.78%	28.4%
Anxiety Score	81.93%	11.5%	6.57%

The percentage-wise distribution of the respondents based on the three demographics is influenced not only by the linear negative and positive relationship found between the daily media hour usage & well-being scores of the respondents, and daily media hour usage & anxiety scores of the respondents, respectively. These were also influenced by the low coefficient of determination and low degree of correlation coefficient. Hence, the respondents were not nearly evenly distributed for the well-being score and anxiety score categories, unlike the media usage category.

SUMMARY, CONCLUSIONS, AND RECOMMENDATIONS

Summary

The objective of the study is to determine the influence of media usage or engagement on the mental health of people in India during the global health pandemic crisis. The data was collected by Mukherjee, Maity, and Chatterjee (2021) three weeks after the nationwide lockdown from April 16 to 22, 2020. The method of research used in this study was descriptive-correlational. Through the web-based online survey of Mukherjee, Maity, and Chatterjee (2021) , the respondents’ media usage, well-being score, and anxiety score have been known. The respondents were chosen by Mukherjee, Maity, and Chatterjee (2021) using the convenient sampling technique.

After analyzing the data, the findings were the following. Majority of the respondents only experience low levels of anxiety and have an intermediate well-being score. However the respondents are evenly distributed between the low, middle, and high levels when it comes to media consumption.

The data from media usage in hours per day had been tested for its relation on well-being score as well as anxiety score using the Simple Linear Regression Analysis and Pearson Correlation. With the use of simple linear regression analysis, the regression models for the daily media hour usage (\(x\)) & the well-being score (\(y\)), and the daily media hour usage (\(x\)) & the anxiety score (\(y\)) were found significant. The two models depicted a negative and positive linear relationship, respectively. However, the adequacy of the two regression models was low since their errors do not have constant variance, based on the Breusch-Pagan Test and the plots of the Residuals Vs. Fitted values.

Because of the low adequacy in the two regression models, the researchers opted to measure the degree of the found linear relationship between the daily media hour usage (\(x\)) & the well-being score (\(y\)), and the daily media hour usage (\(x\)) & the anxiety score, respectively, through the Pearson Correlation Statistical Test.

With the use of Pearson Correlation Test, the measured coefficient correlation for the daily media hour usage (\(x\)) and well-being score (\(y\)) was found to be statistically significant. This led the researchers to conclude that the computed coefficient could be used to determine the degree of correlation and thus the linear relationship between the aforementioned variables. Since \(r=-0.16\), the degree of correlation/negative linear relationship between the daily media hour usage (\(x\)) and well-being score (\(y\)) was low (weak).

As for the daily media hour usage (\(x\)) and the anxiety score (\(y\)), the measured coefficient correlation between the two variables is also found to be statistically significant, and could then be used for interpreting the two variables. Since \(r=0.20\), the degree of correlation/positive linear relationship between the daily media hour usage (\(x\)) and anxiety score (\(y\)), was also low (weak).

Conclusions

The explanation and discussion of answers from the following statement of the problems are presented below.

1. What are the respondents’ demographic profile according to their: a) media usage pattern, b) mental well-being level, and c) mental anxiety level?

• In terms of media consumption, the respondents are evenly distributed between the low, middle, and high levels. On the other hand, most of the respondents experience low levels of anxiety and have an intermediate well-being score.

2. What is the significant relationship between the respondents’ media usage and well-being score?

• With the use of simple linear regression analysis and Pearson correlation test, it can be concluded that there is a weak negative relationship between the respondents’ media usage and well-being score.

3. What is the significant relationship between the respondents’ media usage and anxiety score?

• Based on the data that were analyzed, it can be concluded that there is a weak positive relationship between the respondents’ media usage and anxiety score.

Recommendations

The researchers devised numerous recommendations after careful analysis of the data gathered in the study.

The people of India, as much as possible, should limit their exposure to media since it has been found out that they are negatively affected by it. They are highly encouraged to divert their time, especially when it comes to entertainment purposes only, on spending it with their family at home.

The government must provide the people with free and accessible consultations with mental health professionals. With this, mental health professionals must be well compensated and be provided with all the necessary equipment and knowledge. The government should also provide the people a safe place, where there will be no infections that could take place, for leisure. Thus, allowing the citizens to enjoy quality time while being away from the media.

Mental health professionals must learn to develop a new way of treating mentally ill people with a minimum influence of media.

Future researchers should focus more on knowing which type of media has the greatest impact on people’s mental well-being. This way, the media—as a whole—won’t be labeled as harmful for the people.

Furthermore, since this study is limited on the impact of the media usage of people to their mental well-being and anxiety during the initial phase of the pandemic, based on the dataset of Mukherjee, Maity, and Chatterjee (2021), future researchers are encouraged to conduct this kind of study during the current phase of the pandemic, and even during the post pandemic.

Future researchers are also encouraged to gather a different sample for this kind of study, so the conclusions made about the impact of the media engagement to the mental health indicators could be extended more as a general public, and not only on the Indian population.

---