This guide introduces the main validation concepts and outlines some of the tools which are available in the DDb to help you check your data. We are not asking you to examine every last one of the records that sit on the system for your vice-county – this is unrealistic, largely unnecessary and may well induce mutiny.
Validation States in the DDb
The validation symbol for each record in the DDb is listed in the Record column on the left-hand side of a results table. The text in the record uses the colouring system below. Hold the cursor over the symbol to the left of the word “Record” in the column header to see the key:
| Icon | State | Description | Mapped? | Shown in queries by default? |
|---|---|---|---|---|
 |
Unchecked | Records are given this validation state when they enter the DDb by default |  |
|
 |
Confirmed | A record that has been checked subsequently in the DDb and is recognised as accurate is given the “confirmed” validation state. | |
|
|
Needs verfication | This state is applied when the record is not currently acceptable, but further investigation or correction of some inconsistent details might make it valid. |  |
|
 |
Doubtful | This state is used if you feel the record details might have some merit, but for whatever reason the record is thought to be incorrect. | |
|
 |
Rejected | The "rejected" state is applied if the record is obviously erroneous and fundamental flaws mean that the details aren't worth salvaging. If the strike through is diagonal, it means the record has been rejected outside of the DDb (e.g. often in MapMate) | |
|
 |
Duplicate | It is not necessary or recommended to mark records as “duplicate” as the DDb can detect and exclude most duplicates automatically. | |
|
Priorities for validation
Best practice is to view and validate records on a regular basis, as this allows you to keep on top of new records entering the system and stops a backlog building up. However, if you have yet to attempt to validate records using the database, you might already have hundreds of thousands of records for your vice-county (thank you!), so where to begin?
Well, we have developed summary reports that highlight subsets of your data that would be worth scrutinising more closely. These can be found by visiting your my county page and clicking the data validation tab. (If you haven’t visited the my county page yet, you can find the tab in the menu bar at the top of the screen after logging on to the DDb). The data validation page contains, from the top:
- A chart showing a breakdown of records for your county that meet one of the validation states. You’ll see that there are lots of “unchecked” records. Don’t worry – this is not the priority list and you do not have to check them all!
- A table listing the number of records that fall within or outside of the vice-county boundary. The vast majority are, hopefully, inside your vice-county boundary, but it’s worth checking the ones that aren’t. As the header for the table explains, records are hidden on public maps (and will be excluded from the Atlas) if a grid-reference is further than 2km from the specified county.
- A section titled Records to target for validation for Atlas 2020. These are your priority records for validation, split into three main categories: ‘Singletons’, ‘Rarities’ and ‘Possibly anomalous records’. By clicking on a blue highlighted number in the needs validation column you’ll be sent to a ‘search query’ page where you are able to view and start validating the records (see below).
How to apply a ‘validation state’ to a record?
Time is on your side, so break up the validation task into small manageable chunks. For example, you might first concentrate on nationally rare and scarce species. Focus on finding records that you feel are suspect, and mark them accordingly (see below). This will take them off the maps, and you’ll then be able to delve deeper into such records at your (relative) leisure. Conversely, if you consider a single unconfirmed hectad record of a rare or scarce species to be accurate, then confirmation is useful. It is rare that you will need to reject a record, but if you do, be cautious and only reject based on the details of the record, not the summary.
To validate a record:
- First click on a blue highlighted number within the priority species categories. This will bring up a search page. Press the display results tab to retrieve and view the list of records. The list can be resorted by clicking on the relevant header e.g. click on the taxon header to arrange species A-Z or Z-A. You can also sort by recorder, locality, grid ref, date, etc. in the same way.
- At the bottom of the page, you will see a number of tabs, including select all, selected records action and mark validation state as. As you look through the list, pick out doubtful records and tick the box in the first column associated with the record. It’s usually best to work on small sets of records at a time, rather than the whole search results.
- If you want the summary of a record, click on the validation symbol to the left of the pencil (edit) icon. If you find an error, try to understand how it happened – often you can see a pattern that can lead to other records with the same problem. For example, if you spot a site with a mistyped grid-reference (the cucaera website is helpful here), you’ll often find that a large set of records loaded at the same time that share the same mistake. Taxon mistakes may be the result of data entry errors – if you spot one such mistake then have a look to see if it was repeated more widely.
- Once you’ve finished examining the list, check the relevant validation state in the mark validation state as box, and click on apply change to selected records. You will be asked to give a brief explanation if you have marked records anything other than ‘confirmed’.
- 'Confirmed' implies that a record has been individually checked and that you are happy that it is likely to be valid. There is no need to confirm uncontroversial records and your priority should usually be to weed out mistakes instead.
- Later in the validation process, you may wish to view only records that you have assigned to one particular validation state e.g. ‘needs verification’. You’ll notice that the search query page mentioned above includes a row titled ‘validation state’, which is set at ‘unclassified’ as the default. You can select a different validation state attached to records from the associated dropdown menu.