Journal Review on Passos et al.

’s (2020) study

Introduction

In a report released by the World Health Organization (WHO) in March 2022,

governments were called out to “step up” their assistance and support to protect the

mental health of their constituents. This came about from WHO’s scientific brief that

emphasized the 25% increase in anxiety and depression cases worldwide during the

first year of the COVID-19 pandemic.

Passos et al.’s (2020) study is indeed a timely and relevant research to assess the

impact of how the pandemic affected the mental health of selected adults from Brazil

and Portugal. The study results provided insights that could help the governments of the

two countries in formulating policies and interventions taking into account the socio-

demographic profile and other variables (e.g., life satisfaction, anxiety, and depression

status) of their target beneficiaries. Moreover, the study was published September 2020,

only few months after the global pandemic was announced.

1
 Methodology of the Study

Research Tool

The study sought to describe the mental health status of adult population between

Brazil and Portugal and analyze the differences between the results of each country. It

used three questionnaires: Satisfaction with Life Scale (SWLS) to assess life

satisfaction, Generalized Anxiety Disorder-7 (GAD-7) to assess anxiety levels, and

Patient Health Questionnaire (PHQ-9) to assess depression levels. Cronbach’s Alpha

was calculated to check the internal consistency of each questionnaire which was also

reported as acceptable.

Table 1. Summary of Cronbach Alpha for each questionnaire used.

Questionnaire Cronbach’s Alpha scores Satisfactory range (Taber, 2018)

SWLS 0.88
GAD-7 0.90 0.60 – 0.90
PHQ-2 0.83

However, the study only looked for internal consistency, not much on reliability. Though

the said questionnaires have been used in various reliable studies already (Glaesmer et

al., 2011), more updated questionnaires should also be considered. For instance, an

updated version of the SWLS – the Short Life Satisfaction Questionnaire for Lockdowns

(SLSQL) – was developed to capture respondents’ life satisfaction during crisis

situations like the pandemic.

2
Sampling Method

Because the goal of the research is to provide timely advice regarding mental health

initiatives that could be done by the governments of Brazil and Portugal, the researcher

perhaps used convenience sampling – snowball sampling – which noticeably affected

the profile of the respondents. Both authors are employed and are exposed in an

academe setting thus it can be assumed that they are surrounded with professionals

also working inside the academe. Thus, the educational level of the respondents

reported higher educational attainment (51.3% composes post-graduates, 36.9% has

university degrees, and only 11.8% represents those with high school degree and

below).

Using convenience sampling, by its name, is more convenient but tends to limit the

generalization of the data being scrutinized. It also has risks that the sample selected

does not truly represent the population. Nevertheless, the researcher was able to gather

550 responses, way beyond the required number of samples (n=385).

Data Analysis

In terms of the study methodology, the researchers employed multiple linear regression

analysis which seeks to examine the relationship between dependent and independent

variables. Such methodology also analyzes contributory factors and degree and

direction of their association.

3
When employing multiple regression analysis, the research must undergo four

assumptions, as emphasized by Osborne and Waters (2002):

1. Variables are normally distributed. Otherwise, the results will not provide reliable

conclusions regarding the relationship and significance of the independent

variables relative to the dependent variable.

2. There must be a linearity between the dependent and the independent variables.

If this relationship is disregarded, the results of the regression will underestimate

the true relationship between the variables.

3. The variables should not have correlation with each other to avoid

multicollinearity.

4. The variables must be homoscedastic or having equal variance.

Table 2. Assumptions in Multiple Regression Analysis – actual study methods and

recommendations

Multiple Regression What else could have been

What the Study Did
Analysis Assumptions done
Shapiro-Wilk and/or
Kolmogrov-Smirnov Test can
1. Variables are normally
further strengthen the
distributed. Visual analysis of QQ plot
assumption that the variables
and histograms;
follow a normal distribution.
Mann-Whitney Test for
Though visual analysis could
2. There is a linear variables that not normally
have helped, Analysis of
relationship between the distributed
Variance (ANOVA) can also
independent and dependent
help in confirming the linear
variables.
relationship of the variables.
Durbin-Watson test can be
used to check if there are no
3. Variables should not be No test was conducted for
correlations between the
autocorrelated. this.
identified independent
variables.

4
 4. Variance of errors is the Breusch-Pagan test checks for
Plotting residuals vs fitted
same across all levels the heteroscedasticity of the
predictive values
(Homoscedastic) model.
Should any of these assumptions are not met, the analysis could result to either Type I

or Type II errors. Type I error is also called as “false positive” or rejecting a null

hypothesis that is actually true i.e., tests show there is significant difference when there

is actually no difference. Meanwhile, Type II error, called as “false negative,” happens

when the null hypothesis was not rejected when in fact the alternative hypothesis is the

true statement. This means that it was declared that there is no significant difference

when there is a significant difference (University of California, n.d.).

Moreover, though the study used the coefficient of determination (R 2) to measure the

goodness-of-fit of the model generated, it could have also used ANOVA to further

validate if there are significant differences across the dependent variables between the

two countries.

Other Possible Methodology

Should the study be repeated, it could consider employing logistic regression analysis

instead because of the following reasons:

1. Whether or not a respondent has anxiety and/or depression was determined by:

In interpreting GAD-7, score of 5+ means there is anxiety.

In interpreting PHQ-2, score of 3+ means there is depression.

5
 Thus, the dependent variables for “anxiety” and “depression” are binary which

indicates that a logistic regression analysis is more fitting.

2. Unlike multiple linear regression analysis, logistic regression would provide a

better prediction of the probability that the independent variables would cause the

outcome being analyzed.

Similar impact, cross-sectional survey studies conducted by Baumann et al. (2021)

analyzing the COVID-19 effect on the mental health of students in France, and Alonso

et al. (2021) looking into the impact of the pandemic on Spanish healthcare workers

have used logistic regression analysis instead of multiple regression analysis. This is

primarily because they aim to determine the significant variables that will predict the

mental health status of specific groups.

Conclusion

In conclusion, the study provided relevant and timely insights that would assist in the

planning and formulation of initiatives to protect the mental health status of the adult

population in Portugal and Brazil. However, conducting additional validation measures

to test the assumptions of using a multiple linear regression analysis could have

provided more accurate results.

6
REFERENCES:

Alonso, J., Vilagut, G., Mortier, P., Ferrer, M., Alayo, I., Aragón-Peña, A., Aragonès, E.,
Campos, M., Cura-González, I., Emparanza, J., Espuga, M., Forjaz, M.,
González-Pinto, A., Haro, J., Fresneña, L., Martínez de Salázar, A., Molina, J.,
Ortí-Lucas, R., Parellada, M., Pelayo-Terán, J.M., Pérez-Zapata, A., Pijoan, J.,
Plana, N., Puig, M., Rius, C., Rodríguez-Blázquez, C., Sanz, F., Serra, C.,
Kessler, R., Bruffaerts, R., Vieta, E., and Pérez-Solà, V. (2021). Mental health
impact of the first wave of COVID-19 pandemic on Spanish healthcare workers:
A large cross-sectional survey. In the Revista de Psiquiatría y Salud Mental, 14
(2), pp 90-105. DOI: [Link]

Baumann, C., Rousseau, H, Tarquinio, C., Batt, M., Tarquinio, P., Lebreuilly, R.,
Sorsana, C., Legrand, K., Guillemin, F. and Bourion‐Bédès, S. (2021). Effect of
the COVID-19 outbreak and lockdown on mental health among post-secondary
students in the Grand Est region of France: results of the PIMS-CoV19 study. In
the Health and Quality of Life Outcomes. DOI: 10.1186/s12955-021-01903-9

Glaesmer, H., Grande, G., Braehler, E., and Roth, M. (2011). The German Version of
the Satisfaction with Life Scale (SWLS): Psychometric Properties, Validity, and
Population-Based Norms. In the European Journal of Psychological
Assessment 2011; Vol. 27(2), pp 127–132. DOI: 10.1027/1015-5759/a000058

Osborne, J. and Waters, E. (2002). Four assumptions of multiple regression that

researchers should always test. In the Practical Assessment, Research, and
Evaluation: Vol. 8 , Article 2. DOI: [Link]

Taber, K. (2018). The Use of Cronbach’s Alpha When Developing and Reporting
Research Instruments in Science Education. Res Sci Educ 48, 1273–1296
(2018). DOI: [Link]

University of California. (n.d.) Multiple Hypothesis Testing and False Discovery Rate.
Retrieved from Berkeley
[Link]

World Health Organization. (2022). COVID-19 pandemic triggers 25% increase in

prevalence of anxiety and depression worldwide. Retrieved from
[Link]
increase-in-prevalence-of-anxiety-and-depression-worldwide