Text only more accessible viewing
WeD home page ESRC Research Group on Wellbeing in Developing Countries University of Bath
Research

- Research Statement
- Methods Toolbox
- Countries
- WeD Publications
- WeD Working Papers
- WeD Briefings

People
News
Links
Country Web Sites
Members Area


ESRC logo CDS logo
QOL toolbox

 

The Psychometric Development of WeD-QoL Measure (Phase 4)
In a nutshell. Full psychometric report to follow.

1. Ultimately the aim of this process is to get a battery of items that efficiently and reliably measures QoL in Developing Countries. It will consist of a set of easy to interpret factors that make sense, psychologically and contextually, and could potentially be replicated on further samples of people in developing countries.
2. We cannot determine the precise outcome until the analyses have been completed, but the resulting WeD-QoL measure will be usable as a battery in its own right, the score or scores from which will indicate level of subjective QoL, and can be used in further analyses, as a dependent or independent variable.
3. Prior to the psychometric analyses, we do not know what the content of this battery (or set of scales) will be. We do not know the best number of factors (as that depends on decisions we make with the statistical output), or what the items are which load on those factors (and therefore form part of the battery).
4. In Phase 1 we have gone through the process of administering a large pool of items to a large sample of the population for whom the test is to be designed (people in developing countries). Via Phase 2 and 3 we have included in this pool of items those which we believe (from earlier phases of the research) to measure some aspect of QoL. In Phase 4 we will now use the data to develop a simple and valid measure of WeD QoL.
5. The psychometric process will determine the 'best' scales from the pool of items. 'Best' means being the most interpretable, reliable (internally consistent), valid (in that it can be shown to measure what we intend it to measure), replicable (across further samples) and simple (in terms of factor structure).
6. Scale development involves a process of item and factor analysis, resulting in a range of several potential factor structures. These are then assessed (via exploratory and confirmatory factor analysis) to decide which is the 'best'. Each of the resultant factors becomes a sub-scale, with the items that load on each factor being the items which that subscale comprises. Then, reliability analysis for each subscale is conducted in order to get the best internal coherence. In this process some items may be deleted from the ultimate measure.
7. The traditional approach would be to then re-administer the scale to a new sample and subject it to further psychometric analyses. Due to the high costs of fieldwork in multi-cultural research, our planned alternative approach is to ‘virtually’ re-administer the scale on the existing databases. This is a procedure which has been applied successfully in previous research.
8. Iterations of the same analysis process will be performed on both the pooled 4 country data and the individual country datasets, starting with a general solution based on the 4 country data then a solution which builds in (as restrictions) the differences between countries.
9. The aim is to achieve a cross-cultural scale, which focuses on the common routes of the multiple cultures, and is valid and reliable for each country. This will enable comparability of QoL between the four countries.
10. A second aim is to be able to explore the uniqueness of QoL in each country, bearing in mind the country-specific items that were included in the item pool at Phase 3 alongside the main set of items which were common across all countries. To do this, exploratory then confirmatory factor analysis will also be run on each country dataset, with reliability analyses to achieve the best solutions. The specific country solutions will then be compared with a) the 4 countries solution and b) with the restricted (to that country) 4 countries solution mentioned in (8) above.
11. One possible outcome is that we achieve the same factor solution for all the countries separately as for the pooled data, with the country specific items loading on particular factors within that. These country specific items could then add to the strength of the measure for those countries, whereas others may be discarded.
12. Another possible outcome is that exactly the same set of items (from the common item pool) provides the best factor solution for all four of the countries.

Back to QOL Toolbox


Photo 1
Photo 2
Photo 3
 

| Research | People | News | Links | Country Web Sites | WeD Members |

Wellbeing in Developing Countries (WeD) ESRC Research Group, University of Bath, Bath BA2 7AY
This page maintained by j.french@bath.ac.uk. Last updated: 9 December, 2005

Copyright ©2002 University of Bath · Disclaimer · Privacy Statement