Scottish House Condition Survey: Local Authority Analysis 2017-2019

Using Local Authority Data: Key Information

The release supplements the SHCS 2019 Key Findings report which was published in December 2020 and presents the latest national data for key measures of energy efficiency, fuel poverty, energy perceptions and housing quality. The local authority tables provide key indicators at local authority level relating to households and dwelling types. However they lag the main national data because three years are combined to mitigate the smaller sample sizes involved when analysing sub-national geographies. In this case, survey data from the period 2017-2019 are averaged. Consequently, the national rates presented here, and in the Excel tables, will not match those found in the main Key Findings report. Furthermore, the tables are a snapshot in time, and comparisons over time should only be made between releases with no overlapping years, e.g. comparing 2014-2016 to 2017-2019.

All stated estimates lie at the midpoint of a confidence interval which primarily depends on sample size. Over the three year period, the largest local authority sample sizes were for Glasgow (644 households) Edinburgh (613 households) and Fife with 443 households. The smallest sample is for Renfrewshire, with 190 households. Comparisons between all estimates should take account of the confidence limits, and caution should be taken if simply comparing the stated midpoints.

For example, the prevalence of damp in Angus was estimated to lie in the range 3-9%, while in West Lothian, in the range 1-5%. Despite the midpoint in Angus being double West Lothian (6% versus 3%), the extent of overlap between the two ranges means the survey has not detected a statistically significant difference between them. For this reason, and for clarity, this summary focuses only on statistically significant differences between local authority and national rates in the 2017-2019 period. National rates use the full sample (for most tables, 8,963 households) and therefore have smaller uncertainties, meaning observed differences are more likely to be real.

Confidence intervals are visualised in the accompanying plots as error bars, and are calculated at the 95% level, where there is a one in twenty chance the true value will lie outside these ranges. A statistical tool provided with the published local authority tables helps users determine if differences between any two estimates are significant at the 95% confidence level or not. This allows users to reproduce any of the analysis in this summary as required.