Living descriptive study · DM-MCS-004
Completeness of UK MCS Installer Data: A Field-Level Assessment
Abstract
This study measures field completeness in Data Menu’s UK MCS installer dataset. It counts non-empty values by column, reports each field’s coverage and summarises the distribution of coverage across the file. The assessment distinguishes populated values from independently verified information. It treats zero and false as values, while explaining that placeholders can also count as populated. The results identify fields that may warrant closer inspection for a specific use, without claiming to validate contact details, certifications or the real-world attributes represented by the cells.
Research questionWhich source fields are populated, where are empty cells concentrated, and what does this tell us about analytical readiness?
1. Background and scope
A dataset can contain many columns without every field being populated for every record. Field-level coverage is therefore a useful first step when choosing variables for analysis or planning data preparation. However, a single overall score can conceal the difference between a mandatory identifier and an optional attribute.
The Government Data Quality Framework distinguishes completeness from other dimensions of quality [3]. This report follows that distinction: it measures presence, not correctness. A filled telephone field does not demonstrate that a number works, and a filled certification field does not demonstrate that a status is current.
2. Findings and visualisations
The results below are calculated from the latest available source analysis. The methodology and discussion describe how to interpret them; numerical claims appear only after the results have loaded.
Loading research data
Preparing the latest available aggregate findings. Larger files may take a few minutes.
3. Methods
- Exclude completely empty rows. For each remaining row, convert supported cell values to text and trim surrounding whitespace. Count a cell as populated when the resulting text is non-empty.
- Numeric zero and Boolean false are retained as values. Text placeholders such as N/A or unknown are also non-empty under this measure. Spreadsheet formula cells use a stored result when available; formulas are not recalculated.
- For each column j, compute coverage as 100 multiplied by populated cells in j divided by all analysed rows. Report empty cells as the row total minus populated cells.
- Compute overall cell occupancy from the sum of populated cells divided by rows multiplied by fields. This gives every column equal weight and is not a weighted business-quality score.
- Summarise fields as fully populated, at least 95% but not full, at least 80% but below 95%, or below 80%. Use the unrounded populated-to-total ratio for assigning bands. These bands are descriptive choices, not an external quality standard.
4. Interpretation
The full field table supports assessment against a specific use. A field with low coverage may still be useful for a subset analysis, while a highly complete field may need substantial validation. Which fields matter most depends on the intended decision.
Empty cells do not necessarily indicate a collection error. Some variables are only relevant to certain records, and the file-wide denominator does not adjust for that applicability. Conversely, a placeholder can make apparent completeness high without providing useful information.
Overall occupancy is a compact description of the spreadsheet. It should not be advertised as dataset accuracy, contact deliverability or compliance. Adding a consistently filled indicator column can raise occupancy without improving the reliability of other fields.
5. Limitations
- Rows are the unit of analysis. The calculation does not deduplicate legal entities, establish whether a business is currently trading, or independently re-check its certification.
- The source is a commercial dataset rather than a probability sample. No sampling weights, population estimates, confidence intervals or causal claims are produced.
- This is a living analysis of the latest available source file. Later refreshes may change the figures. Comparisons require retained, dated snapshots and consistent definitions.
- The assessment does not validate email syntax or delivery, telephone reachability, URL availability, address accuracy or certification status.
- Completeness is measured by source column, not by semantic concept. Duplicate headings remain separate columns, and alternate columns containing the same information are not merged.
- No applicability adjustment, mandatory-field rule or placeholder dictionary is applied. Comparisons between fields or source versions require attention to changes in schema and collection practice.
6. Implications and further research
The assessment makes the file’s populated and empty cells visible at field level. It is a useful diagnostic for planning validation and explaining why an analysis used particular columns.
The next quality checks should be chosen for the intended use: validating formats, reconciling duplicate entities, checking freshness or auditing a documented sample against an appropriate source. None of these checks is replaced by a high completeness percentage.
7. Data access and disclosure
Data Menu produces this analysis and sells access to the underlying commercial dataset. This commercial relationship should be considered when using the study. The aggregate downloads are available with the findings; individual records and the source workbook are not published here.
These studies examine the same source dataset and are not independent replications. They are descriptive analyses, have not undergone external peer review, and are not official MCS statistics. Data Menu is independent of MCS and does not claim its endorsement.
View the source dataset →8. References
9. How to cite this study
Data Menu Research. Completeness of UK MCS Installer Data: A Field-Level Assessment. DM-MCS-004, methodology 2.0. Data Menu Limited. Living study, n.d. https://www.datamenu.co.uk/research/mcs-installer-data-completeness.
When citing numerical findings, use the snapshot citation supplied with the loaded results. Retain the source update, analysis time, analysis version and aggregate download so readers can identify the figures you used. A live page is not a permanent archive of a historical result set.