The Employment Precariousness Scale (EPRES), developed by researchers from the Health Inequalities Research Group - Employment Conditions Network (GREDS-EMCONET),1,2 was first validated in Spain with 2004-05 data, demonstrating good psychometric properties and construct validity,3 with subsequent applications in epidemiologic studies further reinforcing its construct validity.4

A revised version of the EPRES (hereafter EPRES-2010 for clarity), with changes aimed at overcoming limitations identified in the previous validation study,3 was included in the 2010 edition of the Psychosocial Work Environment Survey, a population-based survey conducted by the Spanish Union Institute of Work, Environment and Health (ISTAS).5

This revised version of the questionnaire (EPRES-2010) was applied in 2010, in the context of the highest unemployment rate (20%) seen in Spain since 1997, after which unemployment continued rising steeply, reaching 27% by the first trimester of 2012 and exceeding the highest unemployment rates observed since 1975.6 The importance of this lies in the fact that precarious employment is understood to be tightly linked to the unemployment rate: high unemployment rates reduce workers’ bargaining power and capacity to refuse poor employment and working conditions, thus increasing the precariousness of employment overall.

Thus, this study serves the double purpose of assessing the psychometric properties of the revised version of the EPRES (EPRES-2010) and, by doing so in a context of high unemployment, assessing its suitability for use under adverse labour market circumstances.

Study subjects (n=4,750) are in greater proportion male (56.2%), aged 30 to 49 years (56.9%), have complete primary (30.6%) or secondary education (24.5%), work in commerce (16.8%) and manufacturing (12.2%), hold permanent contracts (75.9%), and report tenures greater than 5 years (48.2%).

All item response categories were used. The proportion of missing items was low, although somewhat higher for temporary than permanent workers. Means for ‘disempowerment’, ‘vulnerability’ and ‘exercise rights’ were roughly around 1, but lower for ‘rights’ and for ‘temporariness’ among permanent workers, and higher for ‘wages’ and for ‘temporariness’ among temporary workers. Standard deviations were roughly around 1, with few exceptions.

With the exception of ‘temporariness’, subscale score ranges coincided with the theoretical 0-4 range. Percentages of ceiling effects were low, but floor effects were high for ‘rights’ and ‘disempowerment’, and for ‘temporariness’ among permanent workers. Subscale reliability coefficients were all at or above 0.7. The global score ranged from 0 to 3.3, had neither floor nor ceiling effects, and a reliability coefficient of 0.82 (Table 1).

Table 2 shows all items presented the highest loading within their theoretical subscale; loadings were all above 0.35. The only exception was item 1 in ‘vulnerability’, which loaded highest in ‘exercise rights’ (0.31). The rotated factor model explained 51% of cumulative variance (and 53% if item 1 in vulnerability is removed).

Correlations between EPRES-2010 subscales were all lower than their reliability coefficients, indicating unique reliable variance is being measured by each subscale.7 All inter-scale correlations were positive and low (≈0.3); correlations between the global score and each subscale were substantial (0.47-0.64). Although not a central objective of the study, we additionally performed all psychometric analyses for women (n=2,079) and men (n=2,671) separately, obtaining highly similar results.

The revised Employment Precariousness Scale (EPRES-2010) demonstrated good psychometric properties (acceptability, score distributions, reliability and structure) overall and for both women and men, with improvements in those aspects that were modified following previous recommendations.3

Regarding subscales subject to significant changes, we observed several expected improvements: a notable reduction of floor effects (and increased mean) in ‘temporariness’ among permanent workers, and a notable reduction of overall floor effects in ‘vulnerability’. The latter is possibly due both to the rewording of items and to changes in the socioeconomic context that increase workers’ labour market insecurity and vulnerability. In fact, unmodified ‘vulnerability’ items also exhibited increased means in 2010 (data not shown). Consistent with this, lower floor effects were also observed in ‘exercise rights’, which underwent minor changes, although the magnitude of this change was comparatively small.

In contrast, floor effects increased for ‘disempowerment’ for temporary and permanent workers. While this could be due to the broader range of response categories, we observed similar changes in both direction and magnitude for ‘rights’, which underwent no significant changes. Jointly, these increased floor effects suggest higher worker awareness regarding their terms of employment in 2010. In fact, the endorsement of ‘I don’t know’ responses fell on all comparable items in both subscales. This may be partly due to the greater proportion of long-tenured workers in this sample, who are better informed of their employment conditions.

Factor analysis showed that the reworded item ‘ability to demand better working conditions’, did not load into its corresponding subscale (vulnerability), possibly due to the positive wording now used, closer to that of ‘exercise rights’ items. In future applications of EPRES this item should be reworded in order to fit its corresponding subscale.

The study is not without limitations. Differences between this and the 2004-05 study3 must be interpreted with caution given important sample differences. The 2010 sample includes a smaller proportion of women, young workers, temporary workers, and short tenured workers (most precarious), and a larger proportion of workers aged 50 or more, with secondary education, and with permanent contracts (least precarious). Being a sample with less labour market disadvantage may explain the higher floor effects of ‘disempowerment’ and ‘rights’, whereas the reduction in floor effects in ‘temporariness’ and ‘vulnerability’ are most likely due to improvements in the questionnaire and the deterioration of the labour market situation.

The EPRES-2010 preserved the scale's good metric properties while improving its ability to capture employment instability among permanent workers and its capacity to measure ‘vulnerability’. Results also indicate that EPRES-2010 is appropriate for use in high-unemployment contexts. Furthermore, the multidimensionality of the scale appears most suitable in these circumstances, demonstrating increases in the employment precariousness of workers in some dimensions (vulnerability, exercise rights), despite falls in other more contractual dimensions (disempowerment, wages, rights) which reflect the disproportionate out-selection, from the labour market, of workers with temporary jobs.

This study was partially supported by CONICYT/FONDECYT Iniciación 11121429, by Pontificia Universidad Católica VRI/DID/104/2011, the European Community‘s Seventh Framework Programme (FP7/2007-2013) under grant agreement n° 278173 (SOPHIE project), Programa Estatal de Fomento de la Investigación Científica y Técnica de Excelencia del Ministerio de Economía y competitividad n° CS02013-45528-P (CriSol), and Fundación para la Prevención de Riesgos Laborales and the Generalitat de Catalunya.

A. Vives and F. González designed the analysis. A. Vives drafted the manuscript. F. González performed the analysis. A. Vives, J. Benach, C. Llorens and S. Moncada worked in the adaptation of the questionnaire. S. Moncada and C. Llorens provided all the data for this study. All authors contributed to the interpretation of results, critically revised the work and approved the final version to be published.

Carlos Álvarez-Dardet.

The corresponding author on behalf of the other authors guarantee the accuracy, transparency and honesty of the data and information contained in the study, that no relevant information has been omitted and that all discrepancies between authors have been adequately resolved and described

None.