Validation of the Stress Overload Scale and Stress Overload Scale–Short Form among a Setswana-speaking community in South Africa

Abstract

Although there has been extensive research on the phenomenon of stress, there is still a lack of assessment tools, especially in the South African context, that have strong theoretical underpinnings, tapping into both internal depletion of resources and the excessive external demands from the environment in the measurement of stress. The aim of this study was to validate the Setswana version of the original 30-item long form of the Stress Overload Scale as well as the 10-item short form (Stress Overload Scale–Short Form), both evaluating experienced personal vulnerability and external event load. A sample of N = 376 adults living in a rural community in the Northern Cape Province of South Africa were randomly selected to partake in the study. Emerging model fit indices of confirmatory factor analysis testing the hypothesized two-factor structure of the original Stress Overload Scale were not convincingly good. However, we found a remarkable improvement in model fit indices in the case of the Stress Overload Scale–Short Form. Concurrent validity was shown for the Stress Overload Scale–Short Form in significant correlations with depression and emotional well-being. We conclude that the Setswana version of the Stress Overload Scale–Short Form is a psychometrically sound instrument for measuring stress in the present context; however, further validation of the original Stress Overload Scale in diverse samples is necessary to provide stronger support for the hypothesized two-factor structure.

Keywords

Northern Cape Province psychometric properties South Africa stress Stress Overload Scale

Although stress research is dated to decades ago, it still informs contemporary thinking (Wheaton & Montazer, 2010), and research has pointed to two major theoretical positions (physiological and psychological) informing the understanding of stress. From a physiological viewpoint, stress is understood as the ‘wear and tear’ of body systems, which is due to repeated exposure to heightened neural response from chronic stressors (Selye, 1974). From a psychological theoretical position, the transactional stress model postulates that stress occurs when an individual appraises the demands of the environment as exceeding existing personal resources (overload), with such evaluation being likely to lead to biological and psychological changes that could negatively influence well-being (Lazarus & Folkman, 1984).

Several self-report measures of stress have been created where the emphasis is either on demands (e.g., Cohen, Kamarck, & Mermelstein, 1983; Schlotz, Yim, Zoccola, Jansen, & Schulz, 2011) or resources (Windle, Bennett, & Noyes, 2011), but not considering overload as the result of depletion of resources and excessive demands, and therefore failing to adequately capture this construct (Amirkhan, 2012, 2016). It is in response to the shortcoming of previous stress scales and a need for a scale with strong theoretical underpinnings that the Stress Overload Scale (SOS) was created by Amirkhan (2012). He developed the SOS to capture overload, which was due to the depletion of personal resources (personal vulnerability) and the depletion of external resources to handle demands (event load). This tool has been found to predict illness (Amirkhan, 2012; Amirkhan, Urizar, & Clark, 2015) and cortisol responses (Amirkhan et al., 2015), as well as distinguish stressed and non-stressed populations. In addition, considering the burden of administering the 30-item scale in large-scale epidemiological studies, Amirkhan (2016) created the Stress Overload Scale–short Form (SOS-S) which comprises 10 items measuring both aspects of stress. In creating the SOS-S, consideration was given to items with strong factor loadings, and test scores with psychometric strength in terms of high reliability and validity. The authors also prioritized items that were comprehensible across diverse populations. Both the original SOS and the SOS-S have been validated in different community and college samples, but all of these were in the United States (Amirkhan, 2012, 2016; Amirkhan et al., 2015). These scales have not been validated in other contexts including rural African communities where many kinds of stressors and limited resources are rampant.

Despite the prevalence of studies on stress (cf., Pappin, Wouters, & Booysen, 2012; Turan et al., 2011), there is minimal literature on the psychometric properties of the stress measures being used in the South African context. One exception is the work of Marais, Mostert, and Rothmann (2009), who determined the psychometric properties of the Maslach Burnout Inventory (Schaufeli, Leiter, Maslach, & Jackson, 1996). Other scales such as the Stressful Life Events Scale (Kaminer, Grimsrud, Myer, Stein, & Williams, 2008) and the Perceived Stress Scale (Cohen et al., 1983) had been used but not validated in the South African context (Olley, Zeier, Seedat, & Stein, 2005).

Given that the SOS and SOS-S cover both the objective and subjective aspects of stress and have shown good validity and reliability indices in the United States, the aim of this study is to report on the psychometric properties of both the SOS and SOS-S in a Setswana-speaking community in the Northern Cape Province of South Africa as part of a larger endeavour to profile this resource-poor community with the aim of implementing health promotive interventions (cf., Coetzee, 2011). This will be, as far as we are aware, the first study that validates the SOS and SOS-S in a non-American population. The findings would provide evidence on whether these theoretically well-grounded scales could serve as valid measures of stress in other contexts apart from America and specifically in this South African context.

Method

Participants

A random sampling technique was used to select a total of 400 households from the Valspan community in the Northern Cape Province of South Africa. One person per household was asked to complete a set of questionnaires, which included the SOS. A total of 376 participants completed the SOS. The community selected for this study has been described as having limited resources such as education (Statistics South Africa, 2011), housing, health services, infrastructure, and reliable transport (Coetzee, 2011).

Instruments

Stress Overload Scale (SOS)

The SOS (Amirkhan, 2012) is a 30-item scale (six of which are filler items) that has been designed to tap into the experience of stress overload. This scale comprises two subscales: Event Load and Personal Vulnerability. Each item was rated on a 5-point Likert scale ranging from 1 (not at all) to 5 (a lot). Greater scores indicated higher levels of stress with the exception of item 5, which was reverse scored. Reliability analysis showed that the SOS scores were internally consistent with a Cronbach’s alpha coefficient of .96 (Amirkhan, 2012). The two-factor structure of the scale was confirmed and scores on the subscales were found to be significantly correlated with scores on other stress indices (for the Personal Vulnerability subscale) and with major and minor life events (for the Event Load subscale; Amirkhan, 2012). Amirkhan et al. (2015) found that this scale had adequate criterion-related validity with significant correlations between scores on the instrument and illness, sick days, and workdays missed.

Stress Overload Scale-Short Form (SOS-S)

This 10-item version of the SOS (SOS-S) was created by extracting items from the original long form (Amirkhan, 2016). The 10-item scale also comprises the Event Load and Personal Vulnerability factors and is scored on the same 5-point Likert scale with no reversed-phrased items. Using a diverse sample, scores on this scale demonstrated internal reliability with Cronbach’s alpha values of .92, .94, and .95 for Personal Vulnerability, Event Load, and overall stress load, respectively (Amirkhan, 2016). Scores on the SOS-S were found to be strongly correlated with scores on the original SOS, perceived stress, and depression (Amirkhan, 2016).

Mental Health Continuum–Short Form

This 14-item scale measures three components of positive mental health, namely emotional, psychological, and social well-being (Keyes, 1998, 2006). On a scale from 0 (never) to 5 (every day), participants rate how often they have experienced statements related to the different dimensions of well-being. Scores on the Mental Health Continuum–Short Form (MHC-SF) have shown excellent internal consistency reliability (α > .80) and discriminant validity in adolescents (ages 12–18) and adults in the United States (Keyes, 2005) and South Africa (Keyes et al., 2008). In this study, we obtained Cronbach’s alphas of .74, .79, .81, and .83 for scores on emotional, social, and psychological well-being as well as the overall MHC-SF, respectively.

Patient Health Questionnaire

This 9-item scale is a screening, diagnostic, and monitoring tool measuring the severity of depression (Kroenke, Spitzer, & Williams, 2001). Each item in the Patient Health Questionnaire (PHQ-9) is scored on a range from 0 (not at all) to 3 (nearly every day), with higher scores indicating higher levels of depression. Scores on the scale were found to be internally consistent with a Cronbach’s alpha coefficient of .89 and scores on the scale displayed good test–retest reliability (Kroenke et al., 2001). Bhana, Rathod, Selohilwe, Kathree, and Petersen (2015) reported Cronbach’s alpha of .76 in a South African sample. Cronbach’s alpha for scores on the PHQ-9 in this study was .71.

Procedure

After initial advertisement in the community, a community research meeting was held in order to explain the aim of the study. Interviewer-administered questionnaires were used to conduct a survey in Setswana, the native language of the majority of the population in the region. The survey was carried out by trained fieldworkers who were local home-based care givers and social work volunteers from the community. Measures of psychosocial health, including the MHC-SF, the PHQ-9, and the SOS were administered as part of the survey. These questionnaires were translated to Setswana using a research committee approach (Brislin, 1973; Van de Vijver & Leung, 1997), where they were first translated to Setswana by one translator and then back-translated to English by another translator. A research committee of subject specialists compared the original and back-translated versions and adaptations were made to the translated version if discrepancies were detected. To ensure that the questionnaire items were well understood, the questionnaires were administered beforehand to 10 participants in a comparable community in Potchefstroom, South Africa.

Ethical considerations

The study has been approved by the Health Research Ethics Committee of the North-West University with approval number NWU-00002-07-A2.

Data analysis

The SPSS (version 23.0) and Mplus (version 7; Muthén & Muthén, 1998–2015) software packages were used for data analysis. Using the Shapiro–Wilk test, the psychometric evaluation of both the SOS and the SOS-S commenced with item analysis in order to determine whether the scores represented a normal distribution. Associations between items were analysed using Spearman’s rank order coefficient. Acceptable corrected item-total correlations (correlations between scores on the items and the total score of the remainder of the items in the subscale) were set at r > .30 (Nunnally & Bernstein, 1994).

A confirmatory factor analysis (CFA) using robust maximum likelihood (MLR) estimation was conducted to test the hypothesized two-factor structure. Missing data (missing at random) were handled by full information maximum likelihood estimation. The following model fit indices will be presented: chi-square (Hu & Bentler, 1999), comparative fit index (CFI; Bentler, 1990), Tucker–Lewis index (TLI; Schermelleh-Engel, Moosbrugger, & Müller, 2003), root mean square error of approximation (RMSEA; Steiger, 1990), and the standard root mean square residual (SRMR; Bentler, 1995). The criteria for an acceptable model fit for these goodness-of-fit indices were considered to be CFI > .95, TLI > .95, RMSEA < .08, and SRMR < .08 (Hu & Bentler, 1999). Given the lack of fit for the original SOS (see section on ‘Results’), an exploratory factor analysis (EFA), based on principal axis factor extraction and direct oblimin rotation, was also conducted to explore the underlying factor structure of this scale.

Reliability was evaluated using Cronbach’s alpha. External validity was determined by correlating scores on the SOS-S with scores on the PHQ-9 and the MHC-SF. All the analyses were done with the six filler items excluded.

Results

For the original SOS, scores on items from each of the subscales were significantly correlated with scores on the other items in the respective subscales with the exception of scores on item 5 (‘In the past week, have you felt confident’/‘Mo bekeng ee e fetileng, a o ile wa ikutlwa o itshepile’) that showed insignificant correlations with scores on a number of other items on the Personal Vulnerability subscale. Except for item 5 that had a corrected item-total correlation of .16, the corrected item-total correlations exceeded the critical value of .30 for all items in each subscale, and values ranged from .34 to .56 and from .40 to .64 for the Event Load and Personal Vulnerability subscales, respectively. All the item scores had a negatively skewed distribution with a significant deviation from a normal distribution (p < .01). Descriptive statistics for the items are depicted in Table 1.

Table 1.

Means, standard deviations, and factor loadings on two-factor structure for the 30-item Stress Overload Scale.

Item	Mean	SD	EL	PV
SOS2	2.94	1.19	.37
SOS3	2.2	1.34		.45
SOS4	2.55	1.29	.46
SOS5	2.2	1.2		.17
SOS7	2.52	1.23		.41
SOS8	2.9	1.22	.43
SOS9	2.93	1.2		.53
SOS10	2.9	1.16	.46
SOS12	2.58	1.2	.51
SOS13	2.62	1.19		.58
SOS14	3.1	1.15	.42
SOS15	3.06	1.15		.59
SOS17	2.77	1.19		.64
SOS18	2.91	1.26	.53
SOS19	2.53	1.27		.67
SOS20	2.75	1.28	.65
SOS22	3.12	1.25	.46
SOS23	3.01	1.24		.67
SOS24	2.17	1.1	.58
SOS25	2.51	1.22		.69
SOS27	2.55	1.24		.61
SOS28	2.89	1.27	.54
SOS29	2.67	1.34		.65
SOS30	2.76	1.18	.53

SD: standard deviation; EL: Event Load; PV: Personal Vulnerability; SOS: Stress Overload Scale.

The factor loadings from a two-factor confirmatory factor analysis using the robust maximum likelihood estimator are presented. Filler items are not presented in the table. Item 5 was reverse scored.

For the SOS-S, corrected item-total correlations ranged from .36 to .47 for Event Load and from .40 to .59 for Personal Vulnerability, all exceeding .30, and the items in each subscale were significantly correlated. All item scores had a negatively skewed distribution. Descriptive statistics for the items are depicted in Table 2.

Table 2.

Means, standard deviations, and factor loadings on two-factor structure for the Stress Overload Scale–Short Form.

Item	Description	Mean	SD	EL	PV
	In the past week have you felt/Mo bekeng ee e fetileng, a o ile wa ikutlwa?
SOS1	Inadequate/inyatsa?	2.21	1.34		.40
SOS2	Swamped by your responsibilities/o faraferwe ke maikarabelo a o nang le one?	2.93	1.23	.47
SOS3	That the odds were against you/o sa tsamaelwe ke dilo sentle?	2.94	1.20		.54
SOS4	That there wasn’t enough time to get to everything/o sena nako e e lekaneng ya go dira sotlhe se o tshwanetseng go se dira?	2.89	1.17	.51
SOS5	Like nothing was going right/jaaka e kete go ne go se sepe se se tsamayang sentle?	2.60	1.20		.58
SOS6	Like you were rushed/jaaka e kete o a itlhaganedisiwa?	3.05	1.16	.50
SOS7	Like there was no escape/jaaka e kete go ne go sena kgoro ya botso?	2.76	1.28		.60
SOS8	Like things kept piling up/jaaka e kete dilo di ne di koelegana?	2.54	1.25	.61
SOS9	Like just giving up/jaaka e kete o ka latlhela seditse?	2.91	1.27		.66
SOS10	Like you were carrying a heavy load/jaaka e kete o ne o rwele morwalo o o boima?	2.69	1.34	.53
Cronbach’s alpha				.66	.73

SD: standard deviation; EL: Event Load; PV: Personal Vulnerability; SOS: Stress Overload Scale.

The factor loadings from a two-factor confirmatory factor analysis using the robust maximum likelihood estimator are presented.

For the original SOS, a CFA of the two-factor model yielded inconclusive findings, where the fit was adequate according to the RMSEA and SRMR fit indices, but insufficient considering the CFI and TLI (χ² (251) = 647.98, p < .001, χ²/df = 2.58; CFI = .820; TLI = .802; RMSEA = .065; 90% confidence interval (CI) = [.059, .071]; SRMR = .059, see Table 1 for the factor loadings). As a result, we performed an EFA to explore the factor structure underlying the data. Kaiser’s criterion suggested the extraction of four factors with eigenvalues greater than 1, which also explained 48.46% of the variance. However, the grouping of items under the four factors was not theoretically interpretable. We therefore performed another EFA fixing the factors to two in line with the theoretical model. Only 37.57% of the variance was explained by this solution and the items did not follow the pattern of the hypothesized factor structure. The pattern that emerged was not theoretically supported as the pattern matrix showed that items 2, 3, 4, and 8 loaded onto one factor. All the other items, except for item 5, had factor loadings greater than .3 on the other factor. Item 5 did not load substantially on either of the factors.

We went further to explore a three- and five-factor structure and the variance explained was 43.31% and 53.02%, respectively. However, once again the grouping was not theoretically substantiated. The one-factor solution explained 31.18% of the variance.

For interest sake, we explored the fit of a four-factor and single-factor structure using CFA. Even though the same data that generated the factors in EFA were now used to conduct the CFA, the four-factor structure still demonstrated a poor model fit (χ² (250) = 766.05, p < .001, χ²/df = 3.06; CFI = .763; TLI = .739; RMSEA = .074; 90% CI = [.068, .080]; SRMR = .149). The one-factor structure obtained the following fit indexes: CFI = .684; TLI = .661; RMSEA = .085; 90% CI = [.079, .091]; SRMR = .152. Considering all the findings together, the intended two-factor solution displayed the best fit of all options that were investigated, but the fit was not convincingly good.

For the SOS-S, the CFA for the two-factor structure provided preliminary support for the validity of this version of the instrument among the present sample. The following fit indices emerged: χ² (34) = 53.82, p < .001; CFI = .966; TLI = .954; RMSEA = .039; 90% CI = [.017, .059]; SRMR = .038. The factor loadings are shown in Table 2. The two-factor solution explained 50.18% of the variance.

Given that the hypothesized two-factor structure of the original SOS scale was not supported by the CFA, internal reliability values and external validity were determined only for the SOS-S (see Tables 2 and 3). As a result of the evidence of the relationship between stress and mental illness (Siemer, Mauss, & Gross, 2007), we hypothesized that stress overload would be significantly positively related to depression as measured by the PHQ-9. We also hypothesized that the scores on the SOS-S would be negatively related to positive mental health as measured by the MHC-SF. We found that there was a significant relationship between scores on the PHQ-9 and scores on the Event Load (r = .42, p < .001) and Personal Vulnerability (r = .47, p < .001) subscales. On mental health, we found that scores on Emotional Well-being were significantly negatively correlated with scores on Event Load (r = −.15, p < .01) and Personal Vulnerability (r = −.21, p < .001; see Table 3).

Table 3.

Intercorrelations of the subscales Stress Overload Scale–Short Form with mental health and depression.

	EL	PV	EWB	SWB	PWB	Dep
EL	1
PV	70**	1
EWB	−.15*	−.21***	1
SWB	−.00	−.03	.38***	1
PWB	.01	−.06	.22***	.33***	1
Dep	.42***	.47***	−.12*	.07	−.12**	1

EL: event load; PV: personal vulnerability; EWB: emotional well-being; SWB: social well-being; PWB: psychological well-being; Dep: depression.

p < .05; **p < .01; ***p < .001.

Discussion

Using data from a Setswana-speaking community in the Northern Cape Province of South Africa, this study aimed to determine the factor structure and reliability of a Setswana version of the SOS (both the original version and the short form) in an African context. One of the rationale underlying the creation of this stress measurement instrument is that previous stress scales have focused exclusively on measuring either objective (demands) or subjective (depletion of resources) aspects of stress without much theoretical bearing on the concept of stress (Derogatis & Fleming, 1997), while the SOS attempts to capture both facets (Amirkhan, 2012, 2016). Given the high prevalence of stressors within the South African community, the dearth of validated measures for use in an African context, and the fact that this is the first paper to assess the factorial validity of both versions of the SOS in a non-US sample, this study is deemed important to fill these gaps. Confirmatory and exploratory factor analyses for the original version showed that the hypothesized two-factor structure displayed the best fit of all models tested, but the fit was still not convincingly good. Better fit indices were found when CFA was performed on the SOS-S. Scores on the subscales of both the original version and the short form displayed sufficient internal consistency reliability. Scores on the SOS-S (composite and subscale scores) had strong negative correlations with scores on a measure of depression and small to medium positive correlations with scores on a subscale of emotional well-being. Correlations with scores on subscales of social and psychological well-being were negligible. These findings will now be further discussed.

Reasons should be sought for the fact that the hypothesized and theoretically supported two-factor structure of the original version of the SOS did not fit the data well and that no other better factor structure emerged when EFA was performed on all the items in the SOS. First, the findings could pose questions about the nature and manifestation of stress in this cultural context. It is possible that stress is experienced or expressed in different ways across different cultures and that the indicators that will be able to tap into the experience of stress may differ from context to context. For instance, Niu et al. (2016) in their validation of the HIV/AIDS stress scale in a Chinese sample noted that cultural factors influenced the pattern of factor loadings that emerged. They noted that certain typical reactions to HIV-related stress such as the misuse of drugs, which was common in Western samples, did not have strong factor loadings in their study. Although this finding relates primarily to HIV/AIDS-related stress, the proposition that stressors and the appraisals thereof do take place in the context of culture can provide some insights into our current findings on the original SOS, but then the question arises as to why the short form proves to be valid. Qualitative research on the manifestation of stress within this Setswana-speaking rural context may help to develop an understanding of what stress is and how it operates within this and other non-Western samples. Such studies may lead to theoretical advances in the understanding and conceptualization of stress in diverse contexts.

Another possible explanation for the findings that emerged from the CFA performed on the original SOS could be translation. Although the scale was piloted among 10 laypersons prior to administering the instrument in this study, a post hoc analysis of the item content of the SOS with two Setswana-speaking laypersons suggested that certain Setswana item translations, especially in the original long form and not included in the short form, might have been too ‘academic’ to be able to communicate effectively with rural Setswana speakers. Another factor that could have played a role is the challenge of different Setswana dialects being spoken across different parts of South Africa. As a result, we suggest a revision of the original Setswana version with further inputs from laypeople from both urban and rural areas. We suggest that there is also a need for further validation of the English version of the original SOS in multi-cultural contexts to determine whether the fit indices that emerged in our study was due to possible translation issues of items contained in the long form but not retained in the SOS-S.

Regarding the findings on the factorial validity of the SOS-S, it is possible that the short form contained items that were more culturally sensitive to testing stress in the rural South African context, or that it just contained items without translational problems. The most parsimonious explanation may be that items in the SOS-S are just better indicators of depletion of resources and excessive external demands resulting in stress, as they were selected by Amirkhan (2016) because they had the highest loadings on the relevant factors and applicability in diverse cultural contexts. The fact that the SOS-S performed better than the long format is a bonus as shorter scales are usually preferred in large epidemiological and other health-related studies.

The pattern of positive correlations between scores on an instrument of depression and the SOS-S provided support for the scale’s external validity. Previous validation studies on perceived stress scales have shown that stress was positively related to depression (Karam et al., 2012; Morgan, Umberson, & Hertzog, 2014). Results of the original validation of the SOS scale also showed that depression was related to the experience of stress (Amirkhan, 2012, 2016). A likely explanation for the observed relationship is that increased levels of stress could have resulted in negative emotional reactions, which could over time result in depression. This explanation also resonates with the finding of a significant negative relationship between the emotional well-being subscale of mental health and stress. This finding is supported by Shiffrin and Nelson (2010), who have found that there was an inverse relationship between stress and happiness among a sample of students in the United States. Interestingly, the relationship between stress and the more eudaimonic (functioning well) components of positive mental health (social and psychological well-being) as well as overall positive mental health as operationalized by the MHC-SF were negligible. Although stress appears to be taxing on a person’s hedonic (feeling good) well-being, the association with living a meaningful, self-expressive, and relationally well life (indicative of eudaimonic well-being) is not significant.

This study makes an important contribution, but it was not without limitations. The study was conducted only in a specific region within South Africa. The sample included participants who were illiterate and therefore the instrument was administered in interview format. The potential impact of the specific sample and the method of data administration cannot be ignored and findings can therefore not be generalized. We acknowledge that collecting data in an interview format has the potential of influencing the standardization of the process among illiterate samples. In order to ensure a maximum level of standardization, all fieldworkers were taken through a thorough training process. Similar interview-format questionnaire administrations in rural areas with illiterate participants had proved to worked well (cf. Maré, Wissing, Watson, & Ellis, 2011; Vorster et al., 2000; Wissing, Wissing, Du Toit, & Temane, 2008). It is important to indicate that the rural and often illiterate community that participated in this study represents an important group for study and intervention who would be unethically excluded if only paper-and-pencil administration formats are used for data collection on matters that may be of importance to their well-being.

Further validation across multiple sites and using both paper-and-pencil and interview formats to administer the scale is necessary to provide conclusive evidence on the validity of the instrument. Also, the SOS-S was not administered as such, and conclusions regarding the short form were merely based on extracting the relevant items from the SOS for analysis. This could have influenced the findings and it would be necessary to administer only the short form items in a next validation study. As with most cross-sectional studies, conclusions regarding causal relationships between stress, mental health, and depression cannot be drawn.

Conclusion

The original SOS does not hold sufficient factorial validity to be used in the present context. However, the short form has proven to have sufficient psychometric properties that make this scale suitable for measuring stress in the present sample of South African adults. Given the limited evidence on validation of stress scales in African languages and the rural context of the study, and the evidence of extreme stress being present in this context, our findings represent a timely and important contribution to stress measurement in this context and open up the possibility of the use of the SOS-S in future health and well-being-related studies in this context. This is the first study to support the validity of Amirkhan’s SOS-S scale in a non-USA sample and may thus render support for the implied theoretical model of internal and external resource depletion and consequent stress overload (event load and personal vulnerability) on which Amirkhan’s scale is based. The remarkable improvement in the psychometric properties of the SOS-S when compared to the original SOS may indicate that certain items are more sensitive in tapping into stress in the South African context, but this needs to be explored further.

Footnotes

Funding

The financial assistance of the National Research Foundation (NRF) towards this research is hereby acknowledged. Opinions expressed are those of the authors and should not necessarily be ascribed to the NRF. Opinions expressed and conclusions arrived at are those of the authors and are not necessarily attributed to the NRF.

References

Amirkhan

J. H.

(2012). Stress overload: A new approach to the assessment of stress. American Journal of Community Psychology, 49, 55–71. doi:10.1007/s10464-011-9438-x

Amirkhan

J. H.

(2016). A brief stress diagnostic tool: The short Stress Overload Scale. Assessment. Advance online publication. doi:10.1177/1073191116673173

Amirkhan

J. H.

Urizar

G. G.

Jr. Clark

(2015). Criterion validation of a stress measure: The Stress Overload Scale. Psychological Assessment, 27, 985–996. doi:10.1037/pas0000081

Bentler

P. M.

(1990). Comparative fit indexes in structural models. Psychological Bulletin, 107, 238–246. Retrieved from http://psycnet.apa.org/journals/bul/107/2/238.pdf

Bentler

P. M.

(1995). EQS structural equations program manual. Encino, CA: Multivariate Software.

Bhana

Rathod

S. D.

Selohilwe

Kathree

Petersen

(2015). The validity of the Patient Health Questionnaire for screening depression in chronic care patients in primary health care in South Africa. BMC Psychiatry, 15, Article 118. doi:10.1186/s12888-015-0503-0

Brislin

R. W.

(1973). Cross-cultural research methods. New York, NY: John Wiley.

Coetzee

(2011). Needs assessment conducted in the Vaalharts region, North West and Northern Cape Provinces, South Africa (A North-West Living Labs Baseline Projects) (Unpublished). Potchefstroom, South Africa: Research Logistics.

Cohen

Kamarck

Mermelstein

(1983). A global measure of perceived stress. Journal of Health and Social Behavior, 24, 385–396.

10.

Derogatis

Fleming

(1997). The Derogatis Stress Profile (DSP). In Zalaquett

Wood

(Eds.), Evaluating stress (pp. 113–140). Lanham, MD: Scarecrow.

11.

Bentler

P. M.

(1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling: A Multidisciplinary Journal, 6, 1–55. doi:10.1080/10705519909540118

12.

Kaminer

Grimsrud

Myer

Stein

D. J.

Williams

D. R.

(2008). Risk for post-traumatic stress disorder associated with different forms of interpersonal violence in South Africa. Social Science & Medicine, 67, 1589–1595. doi:10.1016/j.socscimed.2008.07.023

13.

Karam

Bérard

Sheehy

Huneau

M.-C.

Briggs

Chambers

, . . . OTIS Research Committee. (2012). Reliability and validity of the 4-item perceived stress scale among pregnant women: Results from the OTIS antidepressants study. Research in Nursing & Health, 35, 363–375. doi:10.1002/nur.21482

14.

Keyes

C. L. M.

(1998). Social well-being. Social Psychology Quarterly, 61, 121–140. doi:10.2307/2787065

15.

Keyes

C. L. M.

(2005). Mental illness and/or mental health? Investigating axioms of the complete state model of health. Journal of Consulting and Clinical Psychology, 73, 539–548. doi:10.1037/0022-006X.73.3.539

16.

Keyes

C. L. M.

(2006). Mental health in adolescence: Is America’s youth flourishing? American Journal of Orthopsychiatry, 76, 395–402. doi:10.1037/0002-9432.76.3.395

17.

Keyes

C. L. M.

Wissing

Potgieter

J. P.

Temane

Kruger

Van Rooy

(2008). Evaluation of the mental health continuum-short form (MHC–SF) in Setswana-speaking South Africans. Clinical Psychology & Psychotherapy, 15, 181–192. doi:10.1002/cpp.572

18.

Kroenke

Spitzer

R. L.

Williams

J. B.

(2001). The PHQ-9: Validity of a brief depression severity measure. Journal of General Internal Medicine, 16, 606–613.

19.

Lazarus

R. S.

Folkman

(1984). Stress, appraisal, and coping. New York, NY: Springer.

20.

Marais

Mostert

Rothmann

(2009). The psychometrical properties of translated versions of the Maslach Burnout Inventory–General Survey. SA Journal of Industrial Psychology, 35, 175–182.

21.

Maré

Wissing

M. P.

Watson

Ellis

S. M.

(2011). Psycho-social health of an African group before awareness of HIV status: A comparative study. Journal of Psychology in Africa, 21, 7–14.

22.

Morgan

E. S.

Umberson

Hertzog

(2014). Construct validation of self-reported stress scales. Psychological Assessment, 26, 90–99. doi:10.1037/a0034714

23.

Muthén

(1998–2015). Mplus user’s guide (3rd ed.). Los Angeles, CA: Author.

24.

Niu

Qiu

Luo

Chen

Wang

Pakenham

K. I.

. . . Xiao

(2016). Cross-culture validation of the HIV/AIDS stress scale: The development of a revised Chinese version. PLoS ONE, 11(4), e0152990. doi:10.1371/journal.pone.0152990

25.

Nunnally

Bernstein

(1994). Psychometric theory (3rd ed.). New York, NY: McGraw-Hill.

26.

Olley

B. O.

Zeier

M. D.

Seedat

Stein

D. J.

(2005). Post-traumatic stress disorder among recently diagnosed patients with HIV/AIDS in South Africa. AIDS Care, 17, 550–557. doi:10.1080/09540120412331319741

27.

Pappin

Wouters

Booysen

L. R. F.

(2012). Anxiety and depression amongst patients enrolled in a public sector antiretroviral treatment programme in South Africa: A cross-sectional study. BMC Public Health, 12, Article 244.

28.

Schaufeli

Leiter

Maslach

Jackson

(1996). The MBI-general survey. In Maslach

Jackson

Leiter

(Eds.), Maslach burnout inventory manual (3rd ed., pp. 19–26). Palo Alto, CA: Consulting Psychologists Press.

29.

Schermelleh-Engel

Moosbrugger

Müller

(2003). Evaluating the fit of structural equation models: Tests of significance and descriptive goodness-of-fit measures. Methods of Psychological Research Online, 8(2), 23–74.

30.

Schlotz

Yim

I. S.

Zoccola

P. M.

Jansen

Schulz

(2011). The Perceived Stress Reactivity Scale: Measurement invariance, stability, and validity in three countries. Psychological Assessment, 23, 80–94. doi:10.1037/a0021148

31.

Selye

(1974). Stress without distress. Philadelphia, PA: J. B. Lippincott.

32.

Shiffrin

Nelson

(2010). Stressed and happy? Investigating the relationship between happiness and perceived stress. Journal of Happiness Studies, 11, 33–39. doi:10.1007/s10902-008-9104-7

33.

Siemer

Mauss

Gross

J. J.

(2007). Same situation – Different emotions: How appraisals shape our emotions. Emotion, 7, 592–600. doi:10.1037/1528-3542.7.3.592

34.

Statistics South Africa. (2011). Statiscal release (Revised) (No. P0301.4). Pretoria, South Africa. Retrieved from http://www.statssa.gov.za/publications/P03014/P030142011.pdf

35.

Steiger

J. H.

(1990). Structural model evaluation and modification: An interval estimation approach. Multivariate Behavioral Research, 25, 173–180. doi:10.1207/s15327906mbr2502_4

36.

Turan

J. M.

Bukusi

E. A.

Onono

Holzemer

W. L.

Miller

Cohen

C. R.

(2011). HIV/AIDS stigma and refusal of HIV testing among pregnant women in rural Kenya: Results from the MAMAS Study. AIDS and Behavior, 15, 1111–1120. doi:10.1007/s10461-010-9798-5

37.

Van de Vijver

F. J. R.

Leung

(1997). Methods and data analysis for cross-cultural research. Thousand Oaks, CA: SAGE.

38.

Vorster

H. H.

Wissing

M. P.

Venter

C. S.

Kruger

H. S.

Kruger

Malan

N. T.

. . . MacIntyre

(2000). The impact of urbanization on physical and mental health of Africans in the Northwest Province of South Africa: The THUSA study. South African Journal of Science, 96, 505–514.

39.

Wheaton

Montazer

(2010). Stressors, illness, and distress. In Scheid

Brown

(Eds.), A handbook for the study of mental health (pp. 171–199). New York, NY: Cambridge University Press.

40.

Windle

Bennett

Noyes

(2011). A methodological review of resilience measurement scales. Health and Quality of Life Outcomes, 9, Article 8. doi:10.1186/1477-7525-9-8

41.

Wissing

J. A. B.

Wissing

M. P.

Du Toit

M. M.

Temane

Q. M.

(2008). Psychometric properties of various scales measuring psychological well-being in a South African context: The FORT 1 project. Journal of Psychology in Africa, 18, 511–520.