Wikidata:WikiProject British Politicians/Data model

WikiProject British Politicians
Related datasets

This page outlines the data model being used for British politicians. Note that due to the complexity and scope of the project, a slightly flexible data model is used which may not be as detailed for older items as it is for modern ones - this is still a work in progress.

The data model we use is consistent with that developed for the EveryPolitican project. This allows us to represent all the core information needed to describe someone's parliamentary career - when they were elected, who by, for what party, how long they served, when any of that information changed, and why they left.

Service in Parliament is recorded using position held (P39) statements. Qualifiers on the statement show the start and end dates that it applies to, as well as the constituency, party affiliation, and so forth. Each discrete term in office is given a different statement, with a new one assigned whenever there is a definable break in service; for example, if -

  • Parliament is dissolved
  • an MP resigns their seat to stand for reelection
  • an MP is suspended from the House for a period of time
  • an MP changes party affiliation, including the whip being suspended to sit as an independent.

This means that an MP who served over, for example, the four Parliamentary terms 1979 to 1997 would have at least four position held (P39) statements, even if they had continuous service for the same constituency.

In a more complicated case, such as if they were one of the members who went from Labour to the SDP to the Liberal Democrats in that period, they might have six statements - two each in the Parliaments when they changed affiliations.

Standard qualifiersEdit

The basic data model for MPs is that all items should have statements along the lines of:

  1. start time (P580)
  2. electoral district (P768)
  3. parliamentary group (P4100)
  4. elected in (P2715) - the general election or by-election at which they were elected/re-elected - if there is no item for a specific by-election, use UK Parliamentary by-election (Q7864918)
  5. end time (P582)
  6. end cause (P1534) - the reason for ending this period, eg dissolution of parliament (Q741182), leaving party (Q30580660), death in office (Q5247364)

The qualifiers set out the details for that specific period in office. Items have at least one statement per Parliament in which they served, and two or more statements if they changed party during the period.

These qualifiers should all be present if relevant, but some may not be needed, depending on circumstances. Most obviously, end time and end cause cannot be defined for sitting MPs (and in fact, the easiest method to identify sitting MPs is to look for entries with no end time), but also not all terms will begin with an election.

Defining datesEdit

Terms are defined as beginning on:

  • the day of polling for their election (even if not declared until later);
  • the day of changing party affiliation;
  • the day they were declared elected by an election court, etc

This is inconsistent with some other approaches, which may use any or all of: the date of the returning officer's declaration (particularly for 1918/45, which were delayed by a few weeks); the date of the official return; or the date the member took their seat. The approach chosen here has the virtue of being straightforward, makes sense to most lay users, and builds on easily available information. It also avoids unintentionally missing members who died before the poll was declared, or who never took their seat for whatever reason.

Where a general election took place over several days (before 1918), members are treated as having been elected on the first day of polling. In some cases, Historic Hansard quotes a day during the election period as the start of the term, or it has been possible to identify the actual day of polling, and where this happens we have provisionally used that date. But the majority use the first day of polling.

Terms are defined as ending on:

  • the day of dissolution
  • the day they changed party affiliation;
  • the day of their death (even if not known until later);
  • the day they inherited a peerage (before Lords reform);
  • the day of the letters patent creating their peerage;
  • the day their election was voided by a court following an election petition;
  • the day of their resignation through accepting an office of profit; or
  • the day of their appointment to another incompatible office (using the Gazette date if possible);
  • and failing all else, the date Parliament declared they were no longer a member

There is some ambiguity in these dates; in some cases, they appear to have left Parliament in two different ways - eg inheriting a seat and then resigning a couple of days later. In these cases, we tend to go with the earliest reason for departure and omit the second.

Where a term is split in two by a member changing party affiliation, the two parts are shown as starting and ending on the same day. In some cases there may be a slight ambiguity about the exact day on which a member ceased to be affiliated to a party and so these dates may not always be reliable to the exact day, but they should always be the same, regardless.

One consequence of starting terms at the election and ending at dissolution is that for the period between, there are no MPs. This is constitutionally correct, but there is a general convention that MPs who are re-elected are retrospectively treated as having had their career continue throughout, and so various sources will sometimes show MPs who retire or are defeated at an election as holding office right up until the day of polling.

A secondary effect is that any count based on "number of days in office" will return lower results on Wikidata (which does not count dissolution periods) than when counted by another method; over a long career, this can amount to several months.

In some cases (particularly before about 1900) it can be difficult to determine the exact date of an event, particularly for resignations or party changes. In these cases, they can be set to month-precision (or even year-precision if necessary). These will show up in the query service as occurring on the first of the month but it is rare this will cause any significant confusion. A list of these is being maintained for future reference.

Levels of coverage and reliabilityEdit

Full details are at Wikidata:WikiProject British Politicians/Coverage

Broadly speaking, the model above applies to all "modern" MPs. As it goes back in time, coverage becomes a little less extensive.

  • Before 1970, mid-term changes and parties are not manually checked; party affiliation may miss any changes
  • Before 1923, many items are missing party data
  • Before 1885, start/end dates and by-elections may be inaccurate
  • Before 1820, a lot of per-term data for MPs is missing, and data on who sat in a given Parliament is no longer comprehensive

Prior to 1820, the majority of items have per-term items with no qualifiers, or in some cases simply the generic Member of Parliament of Great Britain (Q18015642) / Member of Parliament in the Parliament of England (Q18018860) items. In the long term, these will be phased out and replaced with more detailed values.

All members are covered from 1386 onwards, save for most of those who sat in 1422-1504, and 1640-1660, and those whose names are so lost to history as to not be recorded by the History of Parliament (a real problem before the mid-1500s). For members covered in the earliest four History of Parliament volumes (1386-1421, 1509-1629) all terms include constituencies but no start/end dates.

Special casesEdit

There are a number of special cases.

Removal of the whip or changes of partyEdit

This should end the current P39 entry (using end cause (P1534):leaving party (Q30580660) or suspension from a political party (Q35855188) (more rarely exclusion from a political party (Q30580630), but most leave before they are expelled) and start a new one with parliamentary group (P4100):independent politician (Q327591). If they change to a new party, then crossing the floor (Q5188683) and parliamentary group (P4100):new party.

If the removal of the whip (or suspension from the party) is intentionally temporary and lasts only a very short period, such as a few days, it may not be recorded as the sources are not always clear on this point. On the end of a suspension period, use end cause (P1534):rejoining party (Q35867887)

Independents and the SpeakerEdit

"Parliamentary group" is used rather than party membership; an MP who resigns the whip but remains a member of a party is treated as independent (and an MP who joins a party without taking their whip, likewise). Labour Co-Operative MPs are classed as such, rather than Labour. Independents are all assigned as independent politician (Q327591) even if they describe themselves as, for example, "Independent Conservative", but members of organised "independent groups" like Change UK (Q61751194) are treated as parties.

The Speaker is always classed as parliamentary group (P4100):Speaker of the House of Commons (Q464103). In the event the Speaker stood down but did not immediately leave Parliament, they are classed as independents after that point, unless known otherwise. Deputy speakers are impartial, but they are conventionally counted with their own party, so use the normal party affiliation here. (They can be listed as deputy speakers with a separate P39 item specifically for that role, similar to how a ministerial appointment is modelled)

Suspensions from the HouseEdit

Suspensions from the house are modelled with continuous service on the position held (P39) statement, but with a corresponding significant event (P793) statement indicating the period of suspension.

Delayed electionsEdit

It occasionally happens that polling is delayed in a specific seat at a general election (eg due to the death of a candidate). In this case, the start date is given as the day of the actual polling, but with elected in (P2715) set as the relevant general election. A significant event (P793):postponed election (Q35647389) qualifier can be added - this will help indicate what happened.

Electoral petitionsEdit

If an MP's election is voided by an election petition (either to deem someone else elected or to order a by-election), use end cause (P1534):election petition (Q13634247) to terminate the P39 entry, ending it on the day of the ruling. If someone else is deemed elected, begin their term with elected in (P2715):declared elected after an election petition (Q56123417) rather than the original election (otherwise this will lead to double-counting), again starting on the date of the ruling.

Members who do not take their seatsEdit

If an MP deliberately did not take their seat (eg contemporary Sinn Fein MPs or some historic MPs known never to have served), use significant event (P793) with a suitable qualifier: for modern abstentionists this might be abstained from taking their seat (Q111372257), while for historic members where a reason is not known, use returned but did not attend Parliament (Q60618465) or returned but may not have attended parliament (Q61677239) if it is unclear.

Do not use this if they intended to take their seat but died before they could take the oath. returned but died before attending Parliament (Q108305146) may be appropriate in that case.

Ministerial by-electionsEdit

Until the 1920s, it was common for Ministers to have to stand for re-election on taking ministerial office (as this was held to constitute an office of profit under the Crown). The vast majority of these were purely notional, and so are not modelled. However, any ministerial election which resulted in a change - where the minister was defeated or did not contest the seat - is modelled as a normal by-election, with end cause (P1534) for the previous term set to ministerial appointment (Q51212178)

Members returned for two seatsEdit

Where a member is returned for more than one seat at a general election, they were required to select one to sit for and relinquish the others. They should have a separate statement for each seat, with the "retained" one presented as normal and the others ending with returned for another seat and chose to sit there (Q50393121) on the day they made the declaration.

Double returnsEdit

The modelling for this situation - where a returning officer declares two winners elected as a tie, and leaves it to Parliament or the courts to work out - is still being worked out.

External identifiersEdit

Further notes at Wikidata:WikiProject British Politicians/identifiers

Biographical items in Wikidata are, where possible, systematically crosslinked to external resources. In addition to the more generic identifers such as library authority control records, the following are of particular relevance for MPs:

Work is under way to crossmatch to other Parliamentary data systems.

For contemporary MPs, or those who held office in the last few years, there are a range of external services with useful content -

as well as a variety of standard properties with links to social media identities, etc.

Issues arising from the general Wikidata data modelEdit

Single names for itemsEdit

The general data model for Wikidata gives one name per item per language (the "label"). This is generally the name they are most commonly known by ("Tony Blair", not Anthony) and honorifics (Sir, Hon, etc) are not normally included. However, baronets and peers usually have a formal title (eg "Sir X, nth Baronet") and for holders of peerages this is usually their final title. Things are a bit mixed for modern figures who later recieve peerages; usually the very prominent ones keep their short form but a lot of others become "X, Baron Y" or similar. We use courtesy titles in a few cases where people are most commonly known by these - eg Lord Randolph Churchill. And if someone has changed their name (on marriage, on inheritance, etc) then we tend to use the most recent form unless they are universally known under the old one.

One inevitable effect of this is that names are often anachronistic - the name an MP is shown as having may be the title they inherited or acquired many years later. In some cases the name shown may bear very little relationship to the one they served in Parliament under.

It is also worth noting that labels do not have to be unique (there are 17x MPs exactly matching "John Smith"), but the combination of labels and descriptions (the short summary note at the top) should be unique.

At the moment, we do not have a good way of handling this problem - name (P2561) could potentially be used with date qualifiers, and there is also the related birth name (P1477) and married name (P2562). However, tying this into queries in a reliable fashion is challenging - in all cases it is much simpler to just use labels and accept the anachronisms.

Julian and Gregorian datesEdit

Wikidata uses a "proleptic Gregorian" calendar model, where dates are all stored internally on a Gregorian calendar but have a modifier saying if they are Julian or Gregorian and displayed accordingly. Dates before 1582 are automatically set to Julian when entered through the web service, but sometimes entered as Gregorian when added through the API, which can be confusing.

Ideally, all dates on items for British MPs should be Julian up to 1752 and Gregorian thereafter, but there will of course be a few anomalies for people active on the Continent. Qualifier dates on the position held (P39) MP statements, however, should all be entered as Julian before September 14 1752, and as Gregorian thereafter.

However, the real problem is in the query service. This gives counterintuitive results when displaying Julian dates: it converts them to the corresponding proleptic Gregorian figure. This means that a correct Julian date will display as a Gregorian date some days later, but very counterintuitively, it will also do this for Julian dates before 1582. As such, all 1582-1752 dates in the database will be reported as the corresponding Gregorian dates, and all pre-1582 dates will be reported as a somewhat confusing proleptic date that does not fit with the original or with any other sources.

There is no fix for this for the moment - the query service simply is not able to display a Julian version of a date even when it knows a Julian calendar model was used - but it is a known bug. Rounding to year precision may solve most of the problems, or if necessary adapting the query to offset dates on a case-by-case basis.