Resilience is a dynamic entity and is highly contextualized, with both internal and external factors influencing its conceptualization (20). Only identification of resilience traits does not serve the purpose alone, a detailed insight into the phenomenon is needed. As resilience measurement is not a stable trait so the main focus should be resilience development, tracking progress and focusing on resilience reintegration. Studies in health professional’s context describe different sets of related themes of resilience across different physician specialties (3, 7, 11, 17, 21). Hence, there is a dire need of a valid and reliable tool for resilience measurement in health professionals in order to measure, track, improve and report effectiveness of resilience interventions.
This study aimed to cross evaluate the validity and reliability of MeRS-37, which was originally developed and validated among Malaysian medical doctors, in Pakistan context. The findings of this study highlighted several important findings. First, MeRS-37 showed a good reliability in Pakistani health professionals’ population with Cronbach’s alpha value of 0.90, and thus confirming that MeRS-37 possesses high internal consistency and is a cross valid tool for reliability measurement. Second, the internal consistency of the four domains of MeRS ranges from 0.7 to 0.86 indicating acceptable to good reliability. Third MeRS possess a positive factorial structure, as 35 out of 37 items achieved standardized factor loading value of greater than 0.5, which indirectly supports its internal structure validity. Fourth, MeRS depicts high discriminant validity as its domains are exclusive and independent and showed a correlation value of less than 0.85. Lastly, CFA yielded that MeRS-37 was a poor fit model, Modified MeRS or MeRS-14 (with 4 domains and 14 items) is the best fit model for resilience measuring for health professionals with high goodness of fit indices.
This study provides evidence that MeRS is a valid and a reliable tool. Reliability is represented by internal consistency and stability, and it refers to the ability of a tool to reproduce similar results if repeated over time (22). The original MeRS-37 has scale reliability of 0.90 with domain reliability ranging from 0.76–0.86 (18). The modified MeRS-14 has a scale reliability of 0.87 while domain reliability ranges from 0.78–0.83. Both the scales show reliability in acceptable ranges (23) and have values close to each other. These values are in accordance with the recommendations by Briggs and Cheek (1989) for obtaining balance between amplitude of measurement and internal consistency (24). However, in terms of internal consistency, the original MeRS-37 is superior to MeRS-14. In comparison to CD-RISC (Connor- Davidson resilience scale) with Cronbach’s alpha value of 0.794 (25), BRS (Brief Resilience Scale) with 0.80 (26), ARS (Academic Resilience Scale) with 0.82 (27) and READ (Resilience Scale for Adolescents) with 0.79 (28), the Cronbach’s alpha value of both MeRS-37 and MeRS-14 are higher, and thus suggesting a higher reliability comparable to other scale for resilience measurement. Furthermore, this study was conducted at two different hospitals of Pakistan with two different cohorts of house officers and medical officers. The heterogeneity of the subjects strengthened its validity across Pakistani health professional’s context.
The CFA for MeRS-37 showed that 95% of the items achieved standardized factor loading of greater than 0.55, which indicates its good factorial structure (29). Higher factor loading indicates that items have high contribution towards the construct being measured. Yet it is a poor fit for goodness of fit indices. With step wise removal of item and establishment of corelations, the goodness of fit indices improved for MeRS-14 making it the best fit model, with 100% of items achieving standardized factor loading of greater than 0.64. In the first step, items with factor loading of less than 0.55 were removed, in the next step, items with factor loading less than 0.6 were removed and the absolute and incremental fit indices were improved. Furthermore, the items were examined individually for multiple overlapping corelations with other items and identification of problematic items was done. Subsequent removal of those items improved the indices, and thus yielded a 14 item best fit model. The poor fit indices of MeRS-37 may be attributable to number of variances in data sample as compared to the original sample it was made on. The contextual variance between Malaysian and Pakistani cultures, the differences in hierarchy of health professionals, working environment of medical doctors, population characteristics, personality differences, provision of formal or informal resilience training and nature of medical training provided at undergraduate levels are all grass root level variances, which could have contributed to the results. MeRS-14 represents the same constructs of resilience as MeRS-37 but with less number of items and improved goodness of fit indices. So, this study needs to be replicated at different levels of same population and also verified for different populations across the globe in order to yield a universal best fit model of MeRS.
The discriminant validity for both MeRS-37 and − 14 are high with correlation values of less than0.4. Discriminant validity refers to the ability of the domains of the tool to measure constructs independently with low corelation. A value of less than 0.85 between the factors indicates good discrimination (30). This indicates that items in the scale are well defined, non-redundant and measure the desired constructs efficiently.
Thus, in short MeRS is a valid and reliable tool for resilience measurement in health professionals. MeRS-37 has higher reliability and discriminant validity as compared to MeRS-14 but was a poor fit for goodness of fit indices. MeRS-14 has better factorial structure with best fit for goodness of fit indices. Both these tools provide a useful measurement criterion for benchmarking resilience levels.
This study is unique as it is the first study that cross validate MeRS in different geographical context. This study provides evidence to support validity and reliability of MeRS in Pakistani population. It incorporated participants from two different tertiary care hospitals in two different working categories thus making the sample more heterogenous. It further paves pay for the replication of study at different levels of medical and health personals and practitioners, namely medical students, clinicians, medical teachers, nursing staff and para medics. However, the study was conducted on a small scale at a snapshot with only two strata of doctors involved, which is its limitation. A study involving health professionals at population level and recruiting a more diverse and versatile sample will further elaborate on its psychometric properties. Secondly a lot of items were removed from the original MeRS during analysis. This needs further verification through data triangulation by replication of the study in different settings with variable data sets. Other sources of validity such as construct validity, consequence validity, response process validity and content validity should also be measured in order to ensure suitability & usefulness of this tool.