A fourteen-item scale covering both feeling and functioning, built for population and workplace monitoring. Requires a free licence.
| Full name | Warwick-Edinburgh Mental Wellbeing Scale. Short form SWEMWBS |
| What it measures | Positive mental wellbeing. Feeling and functioning together |
| Items and time | Fourteen items, two to three minutes. Short form seven items |
| Response scale | Five points, from none of the time (1) to all of the time (5) |
| Recall period | The last two weeks |
| Score range | 14 to 70. Short form 7 to 35 raw, then converted |
| Developed by | Universities of Warwick and Edinburgh for NHS Health Scotland, 2006 |
| Cost and licence | Copyright University of Warwick. Free for non-commercial use after registration. Commercial use is charged |
| Languages | About 30 versions listed by Warwick, many with the short form too |
| Ages | Validated 13 to 74. Short form from 11. Not recommended under 11 |
| Best used for | Population and workplace monitoring. Evaluating programmes at group level |
| Pair with | A capability or performance measure, and something specific to the setting |
The WEMWBS is fourteen positively worded statements about thoughts and feelings over the past two weeks. It was built to monitor the mental wellbeing of whole populations, and that is what it has done, in the Scottish Health Survey, the Health Survey for England and a long list of evaluations since. It reaches further than the WHO-5 because it asks about functioning as well as feeling, and it is the measure to reach for when you want one wellbeing scale that is short enough for a survey and wide enough to mean something.
The wording is copyright, so this describes the items rather than quoting them (see the licence note below). The fourteen statements cover feeling optimistic about the future, feeling useful, feeling relaxed, feeling interested in other people, having energy to spare, dealing with problems well, thinking clearly, feeling good about yourself, feeling close to other people, feeling confident, being able to make up your own mind about things, feeling loved, being interested in new things, and feeling cheerful. Each is answered on a five-point scale: none of the time (1), rarely (2), some of the time (3), often (4), all of the time (5).
About half the items are about how life feels and half are about how a person is working: clear thinking, confidence, making your own decisions, dealing with problems. That mix is the point of the scale, and it is why it correlates with both mood measures and psychological wellbeing measures.
NHS Health Scotland wanted a way to monitor positive mental health in the population and commissioned a team at Warwick and Edinburgh to build one in 2006. They started from an existing New Zealand scale, the Affectometer 2, ran an expert panel and focus groups to decide what positive mental health should include, and tested the result on 348 students and on a representative Scottish population sample of 1,749 adults drawn from two 2006 surveys. The validation paper by Tennant and colleagues came out in 2007.
Two years later, Stewart-Brown and colleagues used Rasch analysis to find a seven-item subset that behaves as a proper interval scale, which became the SWEMWBS.
Score each item 1 to 5 and add them. There is no reverse scoring, because every item is positive, and every item carries equal weight. The total runs from 14 to 70, and higher is better. Warwick’s guidance is not to score the scale if more than a small number of items are missing.
The short form is scored differently. Add the seven items to get a raw score from 7 to 35, then convert it using the table Warwick publishes, because the Rasch analysis showed the raw scores are not evenly spaced. The converted scores still run from 7 to 35, but the gaps between them vary, so a converted 20 and a converted 25 are the right distance apart in a way that raw scores are not. The Scottish surveys now use the short form with this conversion. If you use the short form, report the converted score and say so.
The mean in UK population samples is around 51 on the full scale, and in the original validation the median was 51, with the middle half of people between 45 and 56. The distribution is close to normal with a slight tail at the low end, which is unusual for a wellbeing measure and useful, because it means the scale has room at the top. Tennant’s team found no ceiling effect in either sample.
Warwick are firm that the scale was not built to put people into categories, and their FAQ discourages cut-points. If you need groups for reporting, they suggest either standard deviations around the mean or quintiles. You will see bands in the literature, with low wellbeing ending somewhere in the low forties and high wellbeing starting around sixty, and some studies have tried to line the low band up with probable depression. Treat those as conventions chosen by particular authors. They are not clinical thresholds, and the cut points differ from one report to the next.
For change, the useful figure is about three points. Maheswaran and colleagues worked out in 2012 that the scale’s measurement error is between two and a half and three points, and Warwick’s guidance is that a change of about three or more points can be considered meaningful. Smaller movements in a group are within the noise. The scale was always meant for group comparisons, before and after something changed, or against a national figure from the same country, rather than for watching one person.
In the 2007 validation, internal consistency was 0.89 in students and 0.91 in the population sample, which is high enough that the authors noted some redundancy among the items. Confirmatory factor analysis supported one underlying factor in both samples. One-week test-retest reliability in 124 students was 0.83.
The scale correlated 0.77 with the WHO-5, 0.74 with Ryff’s psychological wellbeing scales, 0.73 with the Satisfaction With Life Scale and 0.71 with positive affect, and in the expected negative direction with negative affect and with the GHQ-12 (around minus 0.5). Correlations with the impression management scale of a social desirability measure were as low as or lower than for the comparison scales, so it is not unusually prone to people making themselves look good. It discriminated between population groups in the way the UK psychiatric morbidity surveys would predict.
What the 2007 paper could not show was sensitivity to change, and its authors said so. That came later. Maheswaran’s 2012 analysis of twelve intervention studies found standardised response means from minus 0.10 to 1.35, with nine of the twelve studies showing a response, and concluded that the scale is responsive to a wide range of mental health interventions at the group level. Measurement invariance across age and gender has since been examined in several countries, and translated versions have been validated in, among others, Spanish, Italian, Norwegian, Chinese and Brazilian Portuguese.
The full fourteen-item WEMWBS and the seven-item SWEMWBS. The short form is better for repeated use and for surveys where space is tight, and it is the one with the interval-level scoring. Warwick lists about thirty language versions, many in both lengths. The scales are validated for people aged 13 to 74, the short form has been used with children from 11, and Warwick does not recommend either under 11.
The scales are copyright the University of Warwick, with Edinburgh and NHS Health Scotland. Anyone using them registers first. Universities, public bodies, charities and other non-commercial users get a free licence that runs for twelve months and can be renewed. Commercial users pay. You may not change the wording, the order or the layout beyond text size, and translations have to go through Warwick. None of this is onerous, but it has to happen before the scale goes into a survey platform or onto a web page, and it is the reason this page describes the items rather than quoting them.
It covers feeling and functioning in fourteen items. A normal-ish distribution with no ceiling means it can show improvement in a population that is already doing reasonably well. National reference data in the UK means an organisation there can compare itself to the country. It has a strong validation record and a decade and a half of use in evaluation, and low vulnerability to social desirability, which matters anywhere people suspect their answers are seen by someone with power over them.
It was built in the UK, its focus groups were British, and spirituality was left out by design. That is fine in many settings and a real gap in others. Its behaviour in other cultures should be checked before scores are compared across them, and anyone working with communities whose understanding of wellbeing centres on family, community or spirit will find it thin.
It is a monitor. It will show you whether positive mental wellbeing in a group changed over months. It cannot tell you what changed it. If you want to say anything about an AI deployment, that needs a design around the measure, a reading before the system arrived, a comparison group, or staggered rollout across teams. Without one of those you have a before-and-after number and a story, and the story is doing all the work.
Three points is the smallest change worth noticing, and the scale is not very sensitive to short-term shifts, so it is a poor choice for watching a group week to week. Use the WHO-5 for that and the WEMWBS every few months.
And it is a self-report scale like the others. The one arrangement I would avoid is having an AI system that monitors people also administer the wellbeing survey. People stop answering candidly, and you will not be able to tell.
I would choose the WEMWBS when I wanted a single wellbeing scale for a workplace, a school or a service, I could measure every three to six months, and I cared about group-level change. It beats the WHO-5 when functioning matters as much as mood, and loses to it when I need to measure often or the licence is going to be a problem. Pair it with something that captures what the system is doing to people’s skills or judgement, because the WEMWBS cannot see that, and with an indicator chosen for the particular setting (psychological safety where there is monitoring, loneliness where there is a companion system, and so on).
Registration, the user guide, the SWEMWBS conversion table and the translations are all at the University of Warwick’s WEMWBS pages, warwick.ac.uk/services/innovations/wemwbs.
The validation paper is Tennant, R., Hiller, L., Fishwick, R., Platt, S., Joseph, S., Weich, S., Parkinson, J., Secker, J. and Stewart-Brown, S. (2007), The Warwick-Edinburgh Mental Well-being Scale (WEMWBS): development and UK validation, Health and Quality of Life Outcomes, 5, 63, doi.org/10.1186/1477-7525-5-63.
The short form is Stewart-Brown, S., Tennant, A., Tennant, R., Platt, S., Parkinson, J. and Weich, S. (2009), Internal construct validity of the Warwick-Edinburgh Mental Well-being Scale (WEMWBS): a Rasch analysis using data from the Scottish Health Education Population Survey, Health and Quality of Life Outcomes, 7, 15, doi.org/10.1186/1477-7525-7-15.
On responsiveness, Maheswaran, H., Weich, S., Powell, J. and Stewart-Brown, S. (2012), Evaluating the responsiveness of the Warwick Edinburgh Mental Well-Being Scale (WEMWBS): group and individual level analysis, Health and Quality of Life Outcomes, 10, 156, doi.org/10.1186/1477-7525-10-156. All three are open access.