Worker job satisfaction has been linked to salient measures of performance such as productivity, absenteeism, and workforce turnover. As such it is a construct that economists care about. I’ve recently reviewed research on the determinants of job satisfaction in order to prepare for a study on pay-for-performance reforms in the health sector. And I’ve found a few surprises…
First and foremost I am surprised by how job satisfaction is measured in much of the economic literature. Usually an economic survey will ask a single question to the worker, something very much like:
“How satisfied are you with your job”?
And the response is typically recorded as a 5-level Likert item from 1 (not at all satisfied) to 5 (fully satisfied).
Such a simple open-ended question can be subject to various interpretations. After all there are various dimensions to a job. What if a worker is satisfied with some job aspects but not others? And what if different workers place differing levels of importance on these different dimensions? Can this still be a meaningful summary measure?
Well the question certainly appears to capture some meaningful content:
- One example of this measure is found a recent paper by Clark, Kristensen, and Westergard-Nielsen that finds job satisfaction among Danish workers to increase with the average wages of co-workers (the authors reason that higher wages provide a positive signal about the worker’s own future earnings).
- Perhaps unsurprisingly, job satisfaction has been found to be significantly related to both intrinsic factors (associated with the job itself, such as worker autonomy or the opportunity to learn new skills) and extrinsic factors (received for completing the job, such as salary) in a variety of populations.
- Relatively fixed personality traits such as optimism or neuroses partly determine job satisfaction responses but not entirely, suggesting that environmental factors in the work place also play an important role in determining job satisfaction.
- Job satisfaction is also positively associated with higher white blood cell counts and more robust immune systems, at least in a sample of female Japanese white-collar professionals.
So clearly there is information in the responses to this question.
However there are also obvious difficulties with this straight-forward single measure of satisfaction: job satisfaction can be interpreted in different ways by different subjects or even differently over time by the same subject. Hence the single item indicator exhibits low-levels of what psychologists term internal reliability, meaning that a subject may give inconsistent responses over short periods of time, even over a period as brief as an hour. Multi-item indicators that separately delineate various dimensions of the job tend to exhibit more internal reliability.
A related issue is that the collapse of all dimensions of job satisfaction into a single measure threatens what is termed “construct validity”. A scale deemed to have construct validity actually measures the psychological construct it purports to measure. And it is not clear from the condensed single question measure above whether the respondent’s notion of “satisfaction” matches that which is intended to be measured by the researcher.
So there are good reasons to measure job satisfaction in a more robust manner. However, if we leave the single question aside and seek multi-item satisfaction measures we quickly find a disquietingly large number of survey instruments purporting to measure job satisfaction. Fortunately there is a systematic review of such instruments from van Saane and colleagues that attempts to distinguish among the measures (unfortunately it is 10 years old – I haven’t found a more recent review).
The authors identify 29 job satisfaction instruments and review all of them for reliability and construct validity according to specified criteria. For reliability, a measure deemed satisfactory had to exhibit internal consistency and test-retest consistency to a certain high degree. In order to assess content validity, the authors specified 11 work domains, such as autonomy, growth/development, and financial reward, to represent the span of possible dimensions of job satisfaction. A measure was deemed to have sufficient content if it covered at least eight of these domains.
Of the 29 satisfaction measures surveyed, only seven were found to have sufficient reliability and construct validity. Most of these measures were designed to measure satisfaction among a specific work force such as emergency physicians or psychiatric nurses. A few, though, are meant for general applicability including the Jobs in General Scale.
All of these measures were developed to assess satisfaction among a developed country workforce in a formal employment setting. Adapting a scale to a developing country setting such as the health workforce in remote areas of Zambia will likely take a round of exploratory testing which should yield a better understanding of the specific empirical setting – as Markus says, context certainly matters.
About the Author:
Senior Economist, Development Research Group, World Bank
JED FRIEDMAN is a senior economist in the Poverty and Inequality Unit of the Development Research Group. His research interests include the measurement of poverty dynamics and the interactions between poverty and health. Jed holds a B.A. in Philosophy from Stanford University and a Ph.D. in Economics from the University of Michigan. Before joining the World Bank in 2003 as a Young Professional, Jed worked for one and half years at the RAND Corporation. Till date, he has published over 20 peer-reviewed articles and is currently the principal investigator for impact evaluations on: the effectiveness of malaria control programs in India, Nigeria, and Zambia; national health financing reforms in Kyrgyzstan, Zambia, and Zimbabwe; and conditional cash transfers in the Philippines.
