Abstract
Randomised controlled trials investigating homelessness have
been utilised more frequently in the last few years to help evaluate, understand, and ultimately reduce the number of homeless households in the UK.
During the designing stage of any trial, power calculations are used to help
determine the required sample size and the minimum detectable effect of
interest. These calculations, however, require a number of assumptions to be
made regarding the structure and size of the data to be used. In order to
support researchers looking to conduct trials in this field, we estimate intracluster-correlation-rate values that are essential in the design of such trials. We estimated ICCs for the incidence of households threatened/experiencing
homelessness through time, categorised by region and by support needs.
Intracluster correlation estimates ranged between 0.1 and 0.2 categorised at
region level, reaching values up to 0.5 when further subcategorised by support
needs. The existence of clustering of households threatened/experiencing
homelessness emphasises the need for the provision of ICC values to
researchers in order to facilitate the successful implementation of future
randomised controlled trials
been utilised more frequently in the last few years to help evaluate, understand, and ultimately reduce the number of homeless households in the UK.
During the designing stage of any trial, power calculations are used to help
determine the required sample size and the minimum detectable effect of
interest. These calculations, however, require a number of assumptions to be
made regarding the structure and size of the data to be used. In order to
support researchers looking to conduct trials in this field, we estimate intracluster-correlation-rate values that are essential in the design of such trials. We estimated ICCs for the incidence of households threatened/experiencing
homelessness through time, categorised by region and by support needs.
Intracluster correlation estimates ranged between 0.1 and 0.2 categorised at
region level, reaching values up to 0.5 when further subcategorised by support
needs. The existence of clustering of households threatened/experiencing
homelessness emphasises the need for the provision of ICC values to
researchers in order to facilitate the successful implementation of future
randomised controlled trials
| Original language | English |
|---|---|
| Journal | European Journal of Homelessness |
| Volume | 17 |
| Issue number | 3 |
| Publication status | Published - 13 Oct 2023 |
Keywords
- power calculations
- ICC
- household homelessness