Open AccessMeasurement properties and implications of the Brief Resilience Scale in healthy workers
Author(s) -
Soer Remko,
Six Dijkstra Marianne W. M. C.,
Bieleman Hendrik J.,
Stewart Roy E.,
Reneman Michiel F.,
Oosterveld Frits G. J.,
Schreurs Karlein M. G.
Publication year - 2019
Publication title -
journal of occupational health
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.664
H-Index - 59
ISSN - 1348-9585
DOI - 10.1002/1348-9585.12041
Subject(s) - cronbach's alpha , confirmatory factor analysis , exploratory factor analysis , construct validity , reliability (semiconductor) , scale (ratio) , construct (python library) , psychology , structural equation modeling , psychometrics , clinical psychology , statistics , mathematics , computer science , power (physics) , physics , quantum mechanics , programming language
Objectives The aim of this study was to study measurement properties of the Dutch Language Version of the Brief Resilience Scale (BRS‐DLV) in blue and white collar workers employed at multiple companies and to compare the validity and factor structure to other language versions. Methods Workers (n = 1023) were assessed during a cross‐sectional health surveillance. Construct validity was tested with exploratory and confirmatory factor analyses (EFA and CFA) and hypothesis testing. Reliability was tested with Cronbach's alpha. Results A two‐factor structure of the BRS‐DLV had good model fit in both EFA and CFA, which could be explained by difficulties of workers with reversed order items. After excluding these inconsistent answering patterns, a one‐factor structure showed good model fit resembling the original BRS (χ 2 = 16.5; CFI & TLI = 0.99; SRMR = 0.02;RMSEA = 0.04). Internal consistency is sufficient (Cronbach's α = 0.78). All five hypotheses were confirmed, suggesting construct validity. Conclusions Reliability of the BRS‐DLV is sufficient and there is evidence of construct validity. Inconsistent answering, however, caused problems in interpretation and factor structure of the BRS‐DLV. This can be easily detected and handled because item 2, 4 and 6 are in reversed order. Other language versions differ in factor structure, most likely because systematic errors are not corrected for. To collect valid data, it is advised to be aware of inconsistent answering of respondents.