Abstract
BACKGROUND:
With the rapid rise in the elderly population and their labour force participation, quality of work life (QoWL) of elderly workers becomes an important concept. A valid instrument to measure elderly workers QoWL is a prerequisite to further in this direction.
OBJECTIVE:
To develop and validate the Quality of Work Life Scale-Elderly (QoWLS-E) for elderly workers 60 years and above in Sri Lanka.
METHODS:
The development and validation of 35 items in QoWLS-E was carried out in two stages. Using a literature search and expert opinion, the items were developed in English language and later translated to Sinhala language. The initial scale consisted of 38 items and a principle component analysis (PCA) was conducted among 275 elderly workers in selected administrative divisions of Colombo district. Then a confirmatory factor analysis (CFA) was conducted among a separate group of 250 elderly workers to confirm the factor structure of the developed scale.
RESULTS:
PCA identified 9 principle components accounting for a variance of 71%, which was later confirmed in the CFA (RMSEA-0.07, SRMR-1.0, NNFI-0.87, GFI-0.82, CFI-0.96). The final QoWLS-E with a structure of 9 domains namely; physical health, psychological, welfare facility, safety, job content, co-worker, supervisor, flexibility and autonomy having 35 items correlated satisfactorily with Cronbach’s alpha of 0.77 and test – retest reliability of 0.82.
CONCLUSION:
QoWLS-E is conceptually and culturally appropriate to assess Quality of Work Life Scale in elderly. It could be a useful tool to describe and monitor improvement of QOWL in elderly.
Introduction
The world’s elderly population is growing rapidly and in 2050 one in six persons in the world will be over 65 years and above [1]. According to the World Health Organization (WHO) from 2015 to 2050 the world population over 60 years will rise from 12% to 22% [2]. The elderly population in developing countries is also increasing at a rapid rate and the Sri Lankan population structure follows a similar trend. Sri Lanka is expected to have 6.3 million (27.4%) of elderly people aged more than 60 in 2050 [3].
The average regional total fertility rate of 5.5% in Southeast Asia in 1970 declined to 2.4% in 2015, representing an increased elderly population [4]. This transition marked a decrease in the traditional working age population, thus an increased old age share in the labour force. The projections show that the ratio of elderly people to the people of working age will rise from 1 in 4 in 2015 to 1 in 2 in 2050. In the United States of America, those over 55 are the fastest growing segment in the workforce. By 2030 it is expected that the workers 75 years and older to rise by 11.7% [5]. According to Chartered Institute of Personnel and Development [6], in 2015 over 30% of the United Kingdom workforce was aged 50 and above. Organization for Economic Co-operation and Development (OECD) projects that aging will be a major social and economic challenge. Further, they propose the decline in the labour force can be reduced if the older workers delay their effective age of retirement [7]. Also, the empowerment of older persons to fully and effectively participate in economic, political and social lives of their societies including income generation and voluntary work was set as a goal in the Madrid International plan [8]. Studies have shown engagement of elderly population in the workforce is a significant predictor of survival (mortality odds ratio-0.4) among the older workers compared to non-workers [9] and unemployed workers had a poor health outcome than the employed older workers (OR = 2.75, 2.46–3.07) [10].
Sri Lanka has already identified the importance of the elderly workforce. The National Charter for Senior Citizens of Sri Lanka [11] has stated that opportunities should be given to willing and capable elderly workers to engage in employment. At present, 25% of the elderly population are economically active [12] and provide a significant contribution to country’s economy.
Hence, it is important to attract and retain elderly workers, and for this, maintaining a good quality in their work life is vital. Knowledge on the level of Quality of Work Life (QoWL) of elderly workers would help to identify issues specific to elderly workers and to intervene for further improvement.
The QoWL is a multi-dimensional concept which was originated in 1930 [13]. It is the employee satisfaction on personal and working needs through participating in the work while achieving the goals of the organization [14]. This concept had been extensively studied by occupational psychologists, experts in human resource management, public health experts etc. among working age populations. Quality of Work Life has an objective component which includes work characteristics and subjective component as workers satisfaction [15]. Different theories and models had been used to explain this concept [16, 17] and some of the main dimensions of QoWL are safe and healthy work environment, organizational culture, relation and co-operation, health & well-being, training and development, compensation and rewards, facilities, job satisfaction, job security, autonomy of work, adequacy of resources, work life balance and time pressure [14, 17–19].
The quality of the job and the environment is very important for any age of a worker. However, the QoWL of elders have common dimensions with the young workers as well as compositional differences [20]. Elderly workers’ expectations regarding a job is likely to differ from that of a younger worker. The occupational features where the elderly workers usually value are; flexibility, personal meaningful, intellectually stimulating, sociable and offers adjustments for the disabilities and the health conditions. They also consider the employer support and the importance of their work for the organization. They will be looking for strong work place relationship and autonomy too [21].
Even though there are several tools to measure QoWL in general, they may not capture the true QoWL of elderly workers due to the unique nature of the needs of elderly. With the increasing number of elderly joining the workforce, their quality of work life becomes important. Hence, this study intended to develop and validate a tool to assess the QoWL of elderly workers.
Methods
The tool development was done in three main phases, namely: (i) identification of tentative domains and item generation, (ii) development of QoWLS – E and (iii) evaluation of QoWLS-E.
In the absence of a universally accepted definition for elderly workers and QoWL for elderly workers, the following definitions were developed through Delphi technique involving a panel of subject experts which included specialized professionals in Occupational Health, Geriatrics, Administration, Sociology, as well as selected elderly workers. Elderly worker was defined as, “A worker who is aged 60 years or above, who had been working for the last six months, for a minimum of twenty hours per week at the time of recruiting for the study. The worker may belong to any skill level and can be a paid regular employee or a self – employed person who works for a wage or salary in cash or in kind. Casual workers and seasonal workers were excluded”. Quality of Work Life of an elderly worker was defined as, “A state in which the elderly worker achieves optimal physical and psychological health preserving the elderly workers autonomy through suitable job characteristics while developing human capacities and work life balance to achieve optimal job satisfaction”.
Identification of tentative domains and item development
Both deductive and inductive methods were used for item generation. Based on deductive method items were generated from extensive literature search and assessment of available tools. The inductive method helped to extract the relevant items and domains were identified from the discussion with the panel of experts consisting of specialists in public health, managers, sociologists, psychologists and elderly workers [22]. Afterwards, possible items were identified, duplicates were removed and a common item list was prepared. The selected items were converted into unambiguous simple statements. The face and content validity were assessed by a panel of experts and a selected group of elderly workers. During this process the relevance of content, representativeness and technical quality was assessed by the experts individually using a predetermined rating scale. Consensus was reached between the experts using the modified Delphi technique. These statements were originally drafted in English and later translated into Sinhala language. Since the education levels of the elderly workers differ, the tool was planned to be an interviewer administered tool [23]. The responses for the QoWLS-E were recorded in five-point Likert scale.
Scale development
The scale was pretested with 20 elderly workers and the relevance of the questions to the concept of QoWL in elderly workers as well as whether the answers produce a valid measurement were assessed. The modified scale was used for the data collection for the exploratory factor analysis to identify the domain structure of the newly developed scale. Using multistage sampling [24] the study participants were identified from the elderly workers who fit for the working definition and residing more than one year in the study area, while excluding the workers who were acutely ill or could not be contacted after 3 visits. Multistage cluster sampling was used where a cluster was defined as a Gramasevaka Niladhari Division (GND) from which 10 elderly workers were enrolled with the support of the field health care workers for the study. The GND is the primary administrative unit of Sri Lanka and these GND’s were identified from the selected Divisional Secretariat Division (DSD) which is the secondary level of administrative units in a district. This DSD selection represented the urban and rural settings, all ethnicities, occupational groups and skill levels [25] of Colombo district.
The data collected from 275 elderly workers, in 29 GND’s of three DSD’s in Colombo district, were subjected to Exploratory Factor Analysis (EFA). A questionnaire on socio-demography of the participants and the developed QoWLS-E were the data collecting instruments. To obtain reliable results, sample size was calculated based on seven respondents per item in the questionnaire [26].
Trained sociology graduates interviewed the participants at their residences after obtaining informed written consent. The necessary administrative clearances were obtained from local administrative and health officials and ethical approval was granted by the university ethics review committee (P/81/02/2018).
All 38 items in the tool subjected to PCA using SPSS 21 and prior to the PCA any violations of assumptions such as normality, linearity, inter-item correlation, and multi-collinearity were assessed. The extraction of the factor structure was done by subjecting the drafted QoWLS-E to Principal Component Analysis (PCA) with varimax rotation. The retained items in each domain had an eigen value greater than 1. Following PCA the 38 items were reduced to 35 items organized into 9 domains.
Scale evaluation
The Confirmatory Factor Analysis (CFA) was conducted to verify the dimensionality of the newly developed tool. The field study included 250 elderly workers from randomly selected 27 GND in the same DSD’s. The GND’s used for the EFA study were excluded, but the same sampling method was followed when collecting data for the CFA. The newly developed QoWL-S and questionnaire on the socio demographic were the data collection instruments and were administered by using trained data collectors at the participants resident. The data was collected from November 2018 to December 2018.
The CFA of the 35 items in the QoWLS-E were analyzed using Liseral 8.8 and the extent to which the data fitted the nine domains hypothesized in the PCA was confirmed. During the CFA, Robust Maximum Likelihood estimation method was used, and model fitness was assessed by three categories of fitness indexes; Absolute fit, Incremental fit and Parsimonious fit. Cronbach’s alpha coefficient of the Test re-test reliability assessed the internal consistency of the tool and values for Cronbach’s alpha were obtained for each domain and for the total tool.
Development of cut-off values
Since there is no ‘gold standard’ measure for QoWL in elderly workers, the cut-off values were developed considering the given highest and the lowest scores for the relevant domains by a panel of experts. Each answer in the QoWLS-E was given a score out of 5. Selected statements (2,3,9,20,21 and 33) were scored inversely as they were correlated with poor QoWL. Cut-off points for the overall tool as well as to each domain were decided by the same panel through modified Delphi technique.
Results
A total of 43 items were selected for the scale from the item pool after removing 5 duplicating items. The drafted QoWLS-E which had 38 items were used for the scale development phase and the socio demographic details of the participants selected for scale development through PCA are shown in Table 1.
Distribution of the study sample of PCA by their socio-demographic characteristics (N = 275)
Distribution of the study sample of PCA by their socio-demographic characteristics (N = 275)
GCE O/L* – General Certificate of Education Ordinary Level. GCE A/L**– General Certificate of Education Advanced Level.
Calculation of skewness and kurtosis values revealed the non-normality of the data set. A linear relationship was observed in all the bi-variate scatter plots. The adequacy of sample size was evaluated by KMO values (0.7) and Chi square value for Bartlett’s test of sphericity was 6452.35 with df of 703 and the p value was < 0.001. All the standardized scores were within the mean+/–3.29 value range. Following the PCA of the data set, 35 items remained and 12 domains were identified which were later combined into 9 meaningful domains based on the similarity of the concepts measured (Table 2). These domains were named accordingly.
Rotated solution with factor loading
aRotation converged in 16 iterations.
The 9-factor model, along with two other theoretically-plausible models–an eight-domain model and a ten-domain model, underwent scale validation using confirmatory factor analysis in a separate sample. The socio demographic details of the study participants of scale validation step is presented in Table 3. The models with 8, 9 and 10 domains were assessed for model fitness using CFA (Table 4). A satisfactory level of goodness of fit was obtained for the model with 9 domains. The Nine – Domain model has a chi – square value of 1182.48 and the Root Mean Square Estimate of Approximation value of 0.07. The Comparative Fit Index, Goodness of Fit Index and Parsimony Normed Fit Indexes were respectively 0.96, 0.51 and 0.58. The path diagram of the final 9-domain model is given in Fig. 1 and the final structure of the tool is shown in Table 5.
Socio – demographic characteristics of the study population of CFA (N = 250)
GCE O/L * – General Certificate of Education Ordinary Level. GCE A/L**– General Certificate of Education Advanced Level.
Model fit statistics of the ten domain, eight domain, and nine domain models
Path diagram of nine domain model.
Developed QoWLS-E: Final structure of the tool with items and domains
The mean domain scores and Cronbach’s α coefficients are presented in Table 6. The measure of internal consistency ranged from 0.63 to 0.89 and the internal consistency reliability of the entire instrument was 0.768. The test re-test reliability was assessed with a sub sample of 50 participants and the total correlation coefficient was 0.82.
The cut–off marks developed for the scale are shown in Table 7. There were 9 domains and arithmetic sum of each domain score gave the cut off mark for the total scale. The possible scores in the scale ranged from 35–175. The cut-off mark for the scale was decided as 126 and scores above this represents the higher QoWL.
Reliability and internal consistency of the domain
Cut off mark development for the scale
Principle component analysis enabled the identification of the relationship between the items and the domains of the tool. These relationships were further confirmed in a separate sample through confirmatory factor analysis. Identified domains were well aligned with the concept of QoWL of elderly workers. The identified domains in this tool are namely; physical health, psychological health, facility, safety, job content, co-worker, supervisor, flexibility and autonomy.
Construct of the QoWL had been measured using a variety of tools in different contexts [16]. However, with the specific physical, psychological and social needs of the elderly persons, it is unlikely that the general QoWL measures would effectively capture the situation among elderly workers. In a rapidly aging world, more and more elderly persons are expected to join the labour force. In this background, adapting the concept of QoWL to elderly workers is important and this study aimed to develop and validate a new tool to assess the QoWL of elderly workers in Sri Lanka.
Valid operational definitions for ‘elderly worker’ and the ‘QoWL of elderly workers’ for the Sri Lankan context were achieved by scientifically incorporating the inputs of a multi–disciplinary panel of experts. A widely used non–statistical method of expert rating was used to refine the pool of generated items [27]. Due to the diverse socio-economic levels, applied literacy and physical disabilities in the elderly workers, QoWLS-E was developed as an interviewer administered questionnaire (IAQ) to minimize the non-response bias. However, IAQ’s carry the inherent limitation of response bias where the respondent could provide inaccurate information to be more socially acceptable [28]. Therefore, training the data collectors, and asking clear questions were employed to minimize bias. The scale was developed in simple language which is more suitable to collect data when the study population is having varying levels of education [29].
QoWL in the elderly being a novel concept without a reference measure, the validity and the reliability of the QoWLS-E were assessed using triangulation of several methods. Since QoWLS-E was a new tool, PCA was carried out as recommended by Redding and colleagues [27], which provides the best option in identifying the underlying constructs [30]. Construct validity of the tool is further ensured by conducting CFA in a separate sample, which confirmed the validity of the factor structure identified through PCA.
The identified domains in this tool namely; physical health, psychological environment, welfare facilities, safety, job content, co-worker relationship, supervisor relationship, flexibility and autonomy, were all consistent with the existing theoretical concepts of QoWL [16, 17]. Though most of the available QoWL tools have similar domains or items [14, 31] this new QoWL-E was clearly more specific on geriatric work-related issues. Physical health emerged as a separate domain; the elderly workers’ physical health related issues were predominantly assessed through this tool. In addition, welfare facilities, specifically receiving health facilities from work place also emerged as an important area for QoWL. Flexibility in the job with regard to working times and days and autonomy of the elderly worker in making decisions, were also important domains that showed a greater value to the elderly workers. These unique features make the tool suitable to identify geriatric specific work-related issues.
Acceptable internal consistency reliability requires Cronbach’s α values of 0.7 [32]. The high internal consistency and test-retest reliability shown in the current study were similar to the other comparable multidimensional measures such as QoWL scale developed by Van Larr and team and Swamy & Nanjundeswarawamy’s QoWL tool. High internal consistency across the 35 items suggests that all the items together measure the same construct of QoWL.
Both Van Laar and team’s QoWL scale [31] and Swamy & Nanjundeswarawamy’s QoWL tool [16] were developed based on occupational settings but the present tool was developed based on the community setting. Since the current tool aimed to cover a diverse range of work settings, the number of domains extracted in the present scale was high compared to the scale developed by Van Laar and team. This permits the use of QoWLS-E to assess the QoWL of organizational based workers as well as self – employed workers, both in work settings as well as community settings. The study was based in a capital district of Sri Lanka which has a wide diversity of occupations hence, QoWL-E is capable of assessing the QoWL covering a wide range of the workforce.
The items included in QoWLS-E were largely specific for the QoWL in the elders and the number of items was not high (n = 35). Also, the administration time of the tool was 20 minutes on average. All these factors enhanced the applicability and the compliance of QoWLS-E for the elderly workers and it was proven by the high response rate.
Conclusions
A 35-item multidimensional scale (QoWLS-E), consisting 9 domains has shown good construct validity and reliability. QoWLS-E therefore, shows much promise as a tool to describe QoWL in elderly workers as well as an outcome measure of future interventions to enhance it. This will support generating evidence to retain much needed elderly workers in the labour force while improving their work-related wellbeing.
Ethical approval
Ethical approval was obtained from the Ethics Review Committee of the Faculty of Medicine, University of Kealniya, Sri Lanka (P/81/02/2018). Consecutive approval was obtained from the same Ethics Review Committee to continue the data collection.
Informed consent
Written informed consent was obtained from each participant prior to data collection.
Footnotes
Acknowledgments
Not applicable.
Funding
Not applicable.