Glossary — GM State of The Sector

Base / valid base

The number of respondents included in a percentage or analysis. A valid base excludes people who did not answer the relevant question and may also exclude responses such as “don’t know” or “not applicable”.

Below the radar (BTR) organisations

Informal or unregistered VCSE organisations not captured on formal registers.

Benchmark

A substitute value used when a direct measure is unavailable, such as using the Real Living Wage as the hourly benchmark for replacement-cost valuation.

Cleaned mean

An arithmetic average calculated after applying the study’s data-cleaning rules, such as removing invalid entries, correcting clear formatting problems or excluding values that cannot be used.

Confidence interval

A range intended to show the uncertainty around an estimate under specified statistical assumptions.

Confidence level

The long-run level of confidence attached to a confidence interval, commonly 95%. It does not mean there is a 95% probability that a fixed population value lies within one particular interval.

Deduplication

Identifying and removing repeated records or multiple responses that represent the same organisation or survey submission.

Denominator

The number used as the basis for calculating a percentage. In the report, this is usually the valid base for the question.

Estimate type

The particular analytical rule used to produce a modelled total, such as a cleaned mean or a winsorised estimate.

Extrapolation

Scaling findings from responding organisations to the estimated whole sector. In this study, averages are calculated within organisation size groups and multiplied by the estimated number of organisations in those groups according to the sample frame.

Frequency table

A table showing how many respondents selected each answer and the corresponding percentage.

Headcount

The number of individual people employed, regardless of whether they work full time or part time. Headcount is different from FTE.

Lower Layer Super Output Areas (LSOAs)

LSOAs are small geographic units used for reporting local statistics, population sizes, and census data in England and Wales

Margin of error

An approximate indication of sampling precision, here expressed as plus or minus a number of percentage points at a stated confidence level. The conventional formula assumes random sampling, so it is a benchmark rather than a complete measure of error in this survey.

Mean

The arithmetic average: the total of all valid values divided by the number of valid values. Means can be strongly affected by very large organisations or other outliers.

Median

The middle value after valid values are ordered from smallest to largest. It is less affected by extreme values than the mean.

Missing data

Information that was not supplied, could not be matched or was excluded because it was invalid. Missing data cause valid bases to vary between questions.

Modelled estimate

An informed estimate of a sector-wide total produced from survey data and a population model. It should not be read as an exact count.

Multiple-response question

A question where respondents can select more than one answer. Percentages can therefore add to more than 100%.

N / valid N

These terms are used in frequency tables. ‘N’ is a count of responses or organisations. ‘Valid N’, which is the same as ‘valid base’, is the number included in the calculation after missing or excluded responses are removed.

Outlier

A value that is much higher or lower than most other observations. Genuine outliers can reflect the very large differences between VCSE organisations, but they can also distort averages and extrapolated totals.

Output per job

An economic benchmark from the Office for National Statistics estimating the average output generated per filled job within an industry. It is used with FTE employment to estimate paid-staff GVA.

Paid staff FTE Gross Value Added (GVA) estimate

The paid staff economic contribution estimate is an indicative GVA/output proxy. It multiplies the extrapolated paid staff FTE estimate by a VCSE sector-adjusted output-per-job benchmark. This benchmark is produced by mapping the survey main activity profile to SIC 2007 sections and matching those sections to ONS output-per-job data.

Percentage point

The arithmetic difference between two percentages. For example, a rise from 20% to 25% is an increase of 5 percentage points, not 5%.

Precision

How stable survey estimates are likely to be, based mainly on sample size.

Replacement cost

The volunteer replacement-cost estimate values volunteer time by estimating what it would cost to replace that time with paid labour at the Real Living Wage. It is calculated by multiplying the annualised estimate of volunteer hours by £13.45 per hour.

Representativeness

The extent to which responding organisations resemble the wider sector on characteristics that can be checked, such as locality, size or legal form.

Routed question / routed base

A follow-up question shown only to respondents who gave a particular earlier answer. Its results describe that subset and should not be interpreted as applying to all organisations.

Sample / Achieved sample

The organisations whose survey responses are included in the analysis.

Sample frame

The estimated number and profile of VCSE organisations in the area, including registered organisations and an allowance for below-the-radar groups. It provides the totals to which survey findings are scaled.

Sampling error

The difference that may arise because findings are based on a sample rather than information from every organisation. It is only one source of uncertainty; non-response, measurement and coverage can also affect findings.

Stratification

Dividing organisations into groups, such as locality and income-based size bands, so estimates can reflect major differences in organisational scale and profile.

Weighting

A statistical adjustment considered where the achieved sample differs from the sample frame on benchmark variables. Unweighted results are those in which each responding organisation contributes equally to the percentage or average. Weighted results are those which are adjusted so that responses more closely reflect known or estimated characteristics of the wider sector. The findings in this report are unweighted.

Winsorisation

A method for limiting the influence of extreme values by replacing values above or below a chosen cut-off with the cut-off value rather than deleting the observations. A 90% or 95% winsorised estimate uses the corresponding percentile as the upper cut-off under the study’s rules.