RECORD aims to always supply the best quality information and analysis, and the Manager is charged with ensuring this.

It is fully realised that data, for and of itself, is of little use unless it is verified and validated to as high a degree as is possible to ensure its accuracy within workable limits. Poor quality data will not result in improvements in decision making or even in recording itself. All of the data entered into the RECORD database will be validated as far as is possible given the constraints of time, people, and especially money, which we work under. What this means in practice is that we will endeavour to ensure that:

• All RDB and Nationally Notable species records will be double checked with the relevant county recorders before they are entered onto the database;
• Recorders will be encouraged to deposit voucher specimens (where applicable) of RDB and Nationally Notable species with the World Museum, Liverpool.
• County recorders and other experts (local, regional and national) will be asked to come on board the project to act as primary filters for data dealing with their particular subject. They will be much more likely to spot outliers and discrepancies than members of RECORD staff.
• All paper based records (e.g. notebooks, journals, etc.) are archived with Liverpool Museum along with annual printouts from the RECORD database, so that original records are traceable and are available for future research.

Quality Control Issues:

RECORD recognises the importance of holding and providing access to information which is as accurate as possible. To this end a number of methods of data validation and verification are currently in place and more are planned for the future. But, how do validation and verification differ ?

Data Validation:

Data validation is the checking of data, provided to RECORD, for inconsistencies and for errors imposed by incorrect formatting and typographical mistakes. Examples of these are as follows:

• incorrect dates:
• these can be from formatting errors in a spreadsheet where the field has been incorrectly described, this can lead to the field being interpreted incorrectly by the database
• from date entry errors where a day number has been entered which is greater than the number of days possible in the specified month
• a date in the future has been entered
• incorrect O.S. Grid Reference:
• reversed eastings and northings
• incorrect 100Km square identifier (e.g. SJ instead of SK)
• incorrect or odd number of placement characters (e.g. 7 or 9 fig ref)

Many of these potential problems can be caught (and some rectified) by computerised routines. Formatting errors can, in the main, be identified and fixed prior to the data getting into the database. Likewise, the incoming data can be compared to the stripped down sub-set of the database which we use for the online web searching and mapping facility, and this allows duplicate data to be removed before it gets into the main dataset.

Grid references can be compared against their textual names. RECORD holds in its hierarchical locations data, central grid references for most towns and parishes within Cheshire, as well as for most of the known wildlife sites (SSSIs, SBIs, SINCs, LNRs, Wildlife Trust Reserves, and just plain ‘good areas’). Using this data and a Pythagorean algorithm it is possible to check incoming grid references attached to records along with the textual site name; check that against our own central grid references and to work out the distance the supplied reference is to what we would consider the centre of the area. This enables us to catch grid references where a ‘6’ has been keyed as a ‘9’, or a ‘3’ as an’8’, due to the difficulties of reading poor or smudged hand-writing. Where errors are caught but the answer is not simple then the data has to be removed from the data-set to be imported to the database and then it needs to be inspected manually and where possible corrected. This may mean, if at all possible, going back and checking with the original recorder.

Data Verification:

Data verification is the development of an understanding of how likely the specified species is to be found in the site and habitat quoted within the record provided by the recorder or extracted from past literature. This area of data accuracy is one which requires an in-depth knowledge of the species group, the species within the group which are to be found within the Cheshire region, and the habitat requirements of the species quoted in the record, as well as the habitats available at the time of the record in the site in which it is reported.

RECORD staff are too few in number to be able to provide this degree of expertise as it would rely heavily on the interests of the staff members and there would never be a complete coverage of all orders and families of species. As a consequence, RECORDD has had to attempt to find other sources of record verification and corroboration as well as a means of simplifying and reducing the task.

Currently, two separate approaches are under way to attempt to resolve the verification problem. These are:

• statistical approach
• county, regional and national experts' approach

a) The ‘statistical approach’ is based around the premise that a species recorded once for the site or for the county is less likely to be correct than a species which has been recorded several times, over a long time-frame by a number of different recorders. Consequently, computerised routines to check through the database for singleton or low numbers of records for species are currently in development. The outputs from these statistical runs should provide lists of species and records which need further investigation by experts in their particular taxonomic fields.

b) The ‘experts' approach’ centres around using other people, external to RECORD, who are local, regional or national experts in particular taxonomic groups to verify the data via the application of their expertise and knowledge. Currently, as at 2006, RECORD is beginning to send out copies of the data, which relates to a local expert’s taxonomic expertise, for checking.

However, it is planned to expand RECORD’s capabilities in this area by the end of 2006. This expansion will see the capability of local, regional and national checking data pertinent to their expertise via the Internet. This reduces the risk of spurious data getting out into the world but more importantly allows the data in the RECORD database to be checked where we do not have a local expert to cover the specific taxonomic grouping (e.g. parasitic hymenoptera).

Cautionary Note:

A short cautionary note needs to be appended to the above which is that wildlife (animals and plants) do not tend to read our rules and follow our ideas. Whilst the ecological knowledge we possess may suggest that a species cannot possibly be found out of its currently understood distribution or habitat requirement(s) we can be proven wrong. Consequently, RECORD will, as a norm, not delete data which may appear to be incorrect but will flag it on the database as potentially in error.

This will mean that future work either on the species, which may show it has a wider distribution than previously thought, or on the specific record which may indicate that a voucher specimen is held in a museum, will allow the record to be resurrected without having to re-find and re-key it from the original paper copy. It also enables recognition of records which have been published and then their identity corrected in a later publication. These originally incorrect records will always come back to haunt record centres and there is a requirement to be able to understand which records are correct and which were in error.