Data quality

Return of Quarterly Monitoring Returns

In 2024-25, all 32 local authorities returned complete or partial data to us, via the Quarterly Monitoring Return.

Validation process

Each year, data validation work is conducted on the data local authorities submit, before the data is included in the FPNC statistics.

To begin the 2024-25 validation process, each local authority’s data was checked, comparing against previous years’ data, to identify any large increases or decreases. Any large changes or unusual patterns in the data were noted, and summarised into queries to send to the local authorities. In 2024-25, we noted queries for 29 out of the 32 local authorities.

A set of validation template spreadsheets were then created, using the data submitted in the Quarterly Monitoring Returns. These validation templates summarised the data each local authority had sent us for Q1 to Q4 of 2024-25. The validation templates were then emailed back to each of the local authorities, along with any queries we had noted.  

Local authorities were asked to respond to any queries we had on their data, plus to confirm that the data submitted was correct. They could also send us updated numbers if the data had changed since they first returned the Quarterly Monitoring Return. Many local authorities have dynamic social care systems, so the numbers for quarters may change after local authorities submit their returns, due to updated data becoming available.

We also asked for clarification on whether missing values represented zeros, or instead were because data was unavailable.

Revision of earlier values

During the data validation process, some local authorities identified errors or changes to their previously submitted values for 2023-24, and / or earlier years. This has led to previously published values being revised for some tables. In 2024-25, the following three local authorities submitted revised values for historic years: Moray, Perth and Kinross, and West Dunbartonshire. It should be noted that the revisions made at local authority level generally had a comparatively small effect on the total Scotland values.

All revised values are marked up in the accompanying tables.

Process for estimation of missing values

As in previous years, not all local authorities were able to supply a value for all variables requested in the Quarterly Monitoring Return. Where possible, these missing values have been estimated.

For the 2024-25 publication we made a change to how we estimate some mising values. In line with previous publications since 2021-22, where we had previously received data and there was then a gap in the data series (and no later data), we extrapolated the missing data, using data from other local authorities to estimate the likely change in the missing data. The process used for this was as follows:

1. The sum of the numbers from all local authorities who had submitted data was calculated for the quarter with the missing data point(s).
2. The sum of the numbers from the same local authorities was calculated for the quarter before the quarter with the missing data.
3. If any local authority had not submitted data in one of the quarters, then their number was also excluded from the total for the other quarter. 
4. The quarter-on-quarter percentage change between the two totals was calculated.
5. That percentage change was applied to the latest available data point for the local authority whose data was being estimated.

This extrapolation process was repeated over the time series being estimated, and for all missing quarters. (Note this differs slightly to the method used in 2020-21 and previous years, when the year-on-year percentage change (considering Q4 only) was used.)

For 2024-25, we made a change to how we made some estimates for missing data. Where there was a gap in the data series for a local authority, but we had received data on either side of the gap, we used the data on either side of the gap to interpolate the missing values. The process used for this was as follows:

1. The difference between the last available data point before the gap and the first available data point after the gap was calculated.
2. That value was then divided by the number of missing data points plus one, to calculate the change between each data point over the course of the data gap.
3. Each missing value was then estimated by adding the calculated ‘change per data point’ from step 2 to the previous data point in the series (either known, or estimated), until the data gap was filled.

Vales for some local authorities have been estimated over multiple years and quarters. Estimates in longer time series (longer gaps) will be less reliable than estimates in shorter time series.

The following values from local authorities were estimated for 2024-25:

Scottish Borders:

• number of Care at Home clients receiving personal care services (all ages), all quarters of 2024-25. (Table 2c of the Quarterly Monitoring Return, table 6b in the accompanying tables).
• number of hours of personal care provided to Care at Home clients, age 65 and over, all quarters of 2024-25. (Table 2e of the Quarterly Monitoring Return; tables 7a and 7b in the accompanying tables).

South Ayrshire:

• number of Care at Home clients (both age groups), Q1 and Q2 of 2024-25. (Table 2a of the Quarterly Monitoring Return, tables 5a and 5b in the accompanying tables).
• number of Care at Home clients (all ages) receiving personal care services, Q1 and Q2 of 2024-25. (Table 2c of the Quarterly Monitoring Return, tables 6a and 6b in the accompanying tables).
• number of hours of personal care services provided to Care at Home clients (all ages), all quarters of 2024-25. (Table 2e of the Quarterly Monitoring Return, tables 7a and 7b in the accompanying tables).

For 2024-25, we reviewed estimates that had been made for previous publications since 2021-22. Where estimates for missing data had previously been made using the first method mentioned above (extrapolation) but it was now possible to estimate them using the second method (interpolation), we revised the estimate.

All estimated values, including those that are revised estimates for the 2024-25 publication, are marked up in the accompanying tables.

Missing values (non-returns)

Data points which were not provided by local authorities and which couldn’t be estimated are marked as non-returns, denoted by [NR], in the accompanying tables. Where one or more local authority value was a non-return (i.e. was missing), the Scotland total is ‘across all areas returned’. A list of the values which were missing in 2024-25 is provided in the Data Quality section of the main FPNC 2024-25 publication.

Obtaining annual estimates

In the Quarterly Monitoring Return local authorities supply four quarterly values. For ease of comparison between years, the written publication - and tables numbered ending with ‘a’ in the accompanying tables - report a single value for each year.

In order to convert the four quarterly values into a single annual value:

• numbers of people and numbers of hours of care provided both use the Quarter 4 value as a ‘snap shot’ that is considered to be representative of the annual value
• expenditure values are scaled up from weekly to quarterly, by taking the expenditure in the final week of each quarter as representative of average weekly expenditure for that quarter. The four quarterly expenditures are then added calculated and finally added together to give an estimated annual expenditure for each financial year.

Values for all four quarters are available for each variable in the accompanying tables.

Rounding

All values reported in the publication text and accompanying tables have been rounded for disclosure control. This is to remove the risk of readers being able to indentify individual people referred to by the data. The following rounding has been applied:

• numbers of people have been rounded to the nearest 10
• numbers of hours of care provided have been rounded to the nearest 100
• expenditure values have been rounded to the nearest £1,000

All values in the publication and accompanying tables - including totals and percentages - have been calculated using un‑rounded data. Rounding is the final step applied before values are reported. Percentages are calculated from the original un-rounded values.

Because of this, the Scotland-level totals in the main FPNC publication and in the accompanying tables may not be exactly equal to the sum of the rounded values presented in the tables for each local authority. Similarly, percentages presented in the publication may be different from what would be calculated from the rounded values presented in the report or accompanying tables.

Suppression of small values

Some values within the accompanying tables have been replaced with [c], to indicate the underlying data has been suppressed for disclosure control. All values that rounded to zero are suppressed in this way, in other words:

• numbers of people below 5
• numbers of hours of care below 50
• costs below £500

In some cases, larger values have also been suppressed in this way, if it would be possible to determine a person count of less than 5 from this value (e.g. expenditure values). Therefore, a [c] does not necessarily mean the true value lies within the ranges above.

Many of the data points at local authority level, for people aged 18 to 64, are small and therefore round to zero. Where there are large numbers of data points needing to be suppressed, tables for residents aged 18 to 64 are reported at Scotland-level only in the accompanying tables.