Pundit Metrics

Ridings and Redistributions

Riding / Region Election Stats

Riding Election Metrics

Party Election Metrics

Financial Metrics

---

Ridings and Redistributions

Rep Order - Representation Order

Every 10 years or so, after a major census, the boundaries of federal electoral districts are adjusted according to population (but subject to a few other constraints). The number of seats will often be increased as well. The resulting final redrawing of all riding boundaries ("Redistribution") is legally called the "Representation Order", and it is identified by the first year in which it would take effect for a general election.

This database includes ridings from both the 1996 representation order (used for the 1997 and 2000 elections) and the 2003 representation order (used in 2004 and 2006, and will be used in the 40th General Election on Oct. 19, 2009 or earlier).

Often riding names can change as well, both during and after a redistribution. But even where the riding name stays the same after redistribution, it is still considered a different riding in law and in all the various data-sources (having a different Rep Order, a different FED code, a different census population, etc.).

For further information from Elections Canada:

• Redistribution of Electoral Districts
• Readjustment of Electoral Boundaries and Representation in the House of Commons
• Representation in the House of Commons of Canada

FED Code - Federal Electoral District code

A 5-digit code assigned by Elections Canada that uniquely identifies a riding within the life of a Representation Order. It is used in the National List of Electors to identify the riding where a voter is registered, by Statistics Canada to cross-reference census data with federal ridings, and it is a reliable unique numeric identifier for a riding.

OK, that was the tech-talk. In English, the FED code is the riding number you can use to look up information about that federal riding. Make sure you know which Rep Order the code comes from.

Tr - Transposition

Once a new Representation Order takes effect, Elections Canada announces how much of the census population in each of the old ridings can be found in each of the new ridings. From those figures, the number-crunchers get two sets of percentages (old-to-new, and new-from-old). For example

• 58% of the old Hamilton East riding went into the new Hamilton East--Stoney Creek riding (the rest went to Hamilton Centre), and
• 53% of the old Stoney Creek riding also went to the new Hamilton East--Stoney creek (the rest went to Niagara West--Glanbrook)

• 49% of the new Hamilton East--Stoney Creek riding came from the old riding of Hamilton East, and
• 51% of it came from the old riding of Stoney Creek.

Based on this Transposition, you can apply the second set of percentages to the results for the previous election, and see that the Liberals would have won the new riding of Hamilton East--Stoney Creek if the 2000 election had been fought on the new boundaries ... at least they *probably* would have, assuming everyone voted the same way regardless of which candidates were running. This is what the 2000 Tr resultset tells you, and why there are no candidates shown for it.

I probably don't need to spell out the rest of this cautionary tale for you ... namely that the Transposition can't tell you which incumbent Liberal candidate would have won it then (Sheila Copps from Hamilton East or Tony Valeri from Stoney Creek), which one had the "better" geographic claim to the new riding (it looks pretty even-steven), nor of course whether the new boundaries made the riding more winnable for the Liberals over the longer-term (they lost it to the NDP in 2006).

Two more examples demonstrate the kinds of caution to apply when interpreting transpositions:

• Saskatoon--Rosetown--Biggar was the riding most often reported by the media as the seat that would have had a different outcome with the new boundaries. The NDP's Dennis Gruending had won the 1999 By-election on the old riding boundaries by a comfortable 14 votes per poll. Less than a year later during the general election, he lost to Carol Skelton of the Canadian Alliance by just 68 votes, one-half a vote per poll. Had the 2000 election been fought on the 2004 boundaries, Gruending would have won that contest. However he lost the subsequent 2004 general election to Skelton by 14 votes per poll.


• In West Nova, Liberal candidate Robert Thibault was able to narrowly defeat the incumbent Progressive Conservative M.P., Mark Muise, by 4 votes per poll during the 2000 general elections. Had that election been fought on the 2004 boundaries, Muise would kept his seat. But by 2004, even after the merger of the PCs and the Canadian Alliance, Thibault was now the incumbent and he handily won the new West Nova riding by 19 votes per poll.

In other words, it's nice to know the party vote transposition numbers, but incumbency matters a lot too.

Riding / Region Election Stats

Elec - Electoral Event (GE-General Election, By-Byelection, Tr-Transposition)

Data are presented for all available types of electoral events.

• General Elections (GE) have been held in Canada when Parliament is dissolved on the Prime Minister's request or after the fall of a government on a vote of non-confidence in the House of Commons, but in any event no less often than every 5 years. However, Parliament recently adopted legislation setting a fixed election date every 4 years, unless the government should fall in the meantime.
• Byelections (By) are called by the Prime Minister when a seat becomes vacant due to the resignation or death of its current sitting Member.
• Transpositions (Tr) are explained above.

Type - Result Type (Preliminary, Validated, Official)

Official results are those reported back to Parliament by the Chief Electoral Officer. Several steps take place in the counting process between the Preliminary results announced on election night, the Validated results released by the Returning Officer the following day, and the announcement of the Official results. This database contains Official results unless otherwise indicated (usually shortly after an electoral event has occurred).

Population

The census population number coming from Statistics Canada, reported by Elections Canada as part of the Official Voting Results, and used by them to calculate riding result transpositions and by the electoral boundaries commission to recommend redistributions.

Electors

This is the final number of eligible voters who are both registered on the National List of Electors and eligible to vote in the given Federal Electoral District (aka "riding"). It is the denominator in the calculation of Percent Turnout.

Voted

The final number of eligible voters who cast a ballot in the given Federal Electoral District (aka "riding") during a particular Electoral Event. It is the numerator in the calculation of Percent Turnout, and also the denominator in the calculation of both the Candidates' Percent of the Vote and the Percent Margin of Victory.

Seats

The number of seats in that region, province or local region for that particular election. Seats can increase between representation orders. It is the denominator in the calculation of Pct Seats.

Polls

The final number of polling stations (including urban, rural, special, advanced, and roving polls) for which results are reported in the given Federal Electoral District (aka "riding"). It is the denominator in the calculation of Votes Per Poll.

Riding Election Metrics

% TO - Percent Turnout = 100 * (Voted / Electors)

The percent of eligible electors who actually cast a ballot in the given Federal Electoral District (aka "riding") for the particular Electoral Event.

Turnout is often thought to be related to the income demographics of a riding, how tight the contest is in the riding or nationally, the weather, negative political advertising, whether there is a well-defined issue in the election, and whether the Electoral Event is a byelection or general election. Turnout levels, in general, have been declining in western liberal democracies over the past few decades, the reasons for which are a topic of considerable academic and practical interest.

Pundits might compare a riding's turnout to the regional or national turnout for a given election, and to the riding's turnout for other elections.

NV - Non-Voters = (Electors - Voted)

The number of eligible electors who did not cast a ballot in the given Federal Electoral District (aka "riding") for the particular Electoral Event.

Contest

A textual description of who the first and second place contenders were in the riding for that election. In other words, who was the race between.

Although it's not always the case, a party in 2nd place during one election may be in a better position to win during the next election, and is certainly able to attract better candidates, more donations, and better organizers than a party in third, fourth or worse place. First and second place parties have also had the right to supply enumerators and polling day workers, but this does not confer the advantage anymore that it once may have.

Margin - Raw Margin of Victory = Winner's votes - 2nd Place votes

The number of votes separating the first and second place finishers in a given Federal Electoral District (aka "riding") for the particular Electoral Event. In other words, how many votes did the winner win by.

Pct - Percent Margin of Victory = 100 * (Raw Margin / Voted)

The raw margin expressed as a percent of the number of electors who voted. In other words, how many percentage points did the winner win by.

PerPoll - Raw Margin as Votes Per Poll = Raw Margin / Polls

The raw margin divided by the number of polls. In other words, how many votes per poll did the winner win by.

This figure is often used by political organizers to assess whether the riding was winnable or losable on the basis of local voter-contact and get-out-the-vote organizing techniques. As a local organizer you never want to lose of course, but it's particularly humiliating to lose by a narrow margin ... and you'll go down in history as an organizer or a candidate for winning the squeakers.

Remember "Landslide Annie" McLellan? She had a cakewalk in Edmonton Centre (2.9 votes per poll in 2004) compared with Tony "too-close-for-comfort" Clement who won Parry Sound-Muskoka by just 28 votes, or 0.1 votes per poll, in 2006.

% Mrg 1-3 - Percent Margin between 1st and 3rd = 100 * ((Winner's votes - 3rd Place votes) / Voted)

The raw margin between 1st and 3rd place, expressed as a percent of the number of electors who voted. In other words, how many percentage points did the winner place ahead of the 3rd place finisher. A "Close 3-way Race" is defined here as one having a Margin Percent 1-to-3 of <= 15%. This threshold has not been statistically verified; it was merely selected based on eyeballing the data and seeing where there was a significant enough break (that was also a round number).

% Mrg 1-4 - Percent Margin between 1st and 4th = 100 * ((Winner's votes - 4th Place votes) / Voted)

The raw margin between 1st and 4th place, expressed as a percent of the number of electors who voted. In other words, how many percentage points did the winner place ahead of the 4th place finisher. A "Close 4-way Race" is defined here as one having a Margin Percent 1-to-4 of <= 25%. Again, the threshold would have to be reviewed academically, but seemed on inspection to capture seats with a four-way dynamic over one or more elections.

Party Election Metrics

Party Vote

The number of votes cast for a party in that riding, local region, province or region during a given election. It is the numerator in the calculation of Vote Pct and Votes Per Seat.

Vote Pct = 100 * (Party Vote / Voted)

The percentage of all votes cast during a given election which were cast for a given party (aka "Vote Share"). It can be calculated for a riding, local region, province, region or for an entire electoral event.

% Elec = 100 * (Party Vote / Electors)

The percentage of all electors during a given election (aka "Percent of the Electorate") who either cast a ballot for a given party, or in the case of NV (Non-Voters), who did not vote. It can be calculated for a riding, local region, province, region or for an entire electoral event.

When comparing a party's vote-share across time, it can be preferable to compare its percent of the electorate rather than its vote-share, in order to see whether its supporters switched to another party or just "stayed home".

For example, Karl Rove in the United States is credited with developing a strategy whereby certain electors, who did not typically vote, could be drawn to vote for one party based an issue they felt strongly about, while other issues could be employed to make opposition supporters discouraged, and perhaps therefore less likely to vote and more likely to "stay home".

Seats

The number of seats won by a party in a local region, province or region during a given election.

2nds

The number of second place finishes achieved by a party in a local region, province or region during a given election.

Strong 3s

The number of strong third place finishes achieved by a party in a local region, province or region during a given election. A third place is defined as "strong" where the Percent Margin 1-to-3 is less than or equal to 15% points. In other words, a third place finish in a three-way race. The threshold of 15% between first and third place was selected for this project on the basis of eyeballing the existing data and seeing where it seemed to make a significant difference (and also to keep it a round number). Statistical analysis would have to be undertaken to determine whether this is the optimal threshold.

Pct Seats = 100 * (Party Seats / Region Seats)

The percent of all seats in a local region, province or region won by a party during a given election.

Those who are supporters of some form of proportional representation may be interested to compare a party's percentage of seats with its percentage of the vote; always remembering that individuals' voting behaviour and parties' electoral strategies may not be the same, or yield the same results, under a different electoral system.

Nom - Number of candidates nominated

The number of candidates nominated by a party in a given local region, province or region during a given election. Compare it to the number of seats in that grouping to determine whether the party fielded a full slate of candidates. It is the denominator in the calculation of Pct Women.

Wom - Number of women candidates nominated

The number of women candidates nominated by a party in a given local region, province or region during a given election. It is the numerator in the calculation of Pct Women.

Rebate

The number of seats in which a party's candidate obtained a sufficient percentage of the vote in a local region, province or region during a given election, in order to qualify for a rebate of election expenses. The threshold used to be 15%, but was changed to 10% as of the 2004 General Election. In the old days we used to call missing that threshold "losing your deposit", and even though the rules about the return of a deposit have changed you will still hear old-timers use that expression to mean not obtaining a high enough percent of the vote to get the rebate.

Vote % ↑

The number of ridings in which a party's percent of the vote increased (or at least did not decrease; I used the comparison operation ">=") since the last electoral event in that riding. In most cases this would have been the last general election, but not in the cases where a by-election had occurred in the meantime.

Things do get a little complicated with the older data, however, for a couple of reasons. First of all, it's very difficult to write a good generic query that assesses how the Conservative vote might have gone up (or down) from either the Canadian Alliance and/or the Progressive Conservatives in each riding between 2000 and 2004, so I haven't even tried to this point. It may make more sense to run that as a separate analysis.

Second, 2004 results have to be compared backwards to the 2000 Transposition. However Elections Canada only calculated the transposition for the first four parties in each riding, and lumped everyone else into the "other" category. Unfortunately this means that by the 2004 General Election, when the Green Party ran a full slate of 308 candidates for the first time and also increased their vote in many of the ridings they had contested before, it is not possible to actually measure that increase for them between 2000 and 2004, since they typically placed 5th or lower in 2000, when both the Canadian Alliance and Progressive Conservative parties were running candidates.

Finally of course, this metric does not take into account the impact of changes in turn-out, so eventually when I add the number of ridings in which a party's raw vote went up (and down), the two measures (raw and percent) should be read together.

Vote % ↓

The number of ridings in which a party's percent of the vote actually decreased since the last electoral event in that riding. All the same cautions apply to the interpretation of this data as discussed above for Vote % up.

Pct Wom - Percent Women = 100 * (Wom / Nom)

The percent of women candidates nominated by a party in a given local region, province or region during a given election.

Compare a party's performance on this metric with other parties' performance, and against any commitments made by that party.

Make sure to read it together with an assessment of how competitive ("winnable") the ridings are in which the party has nominated those women. One way to assess a party's commitment to electing women is to examine the gender of candidates nominated to replace retiring incumbents, or in seats where the party last placed a strong second.

Votes Per Seat = Party Vote / Party Seats

The number of votes cast for a party for every seat it won in a local region, province or region during a given election. The metric is undefined in the case where a party has won no seats (can't divide by zero).

This can be a measure of a party's vote "efficiency", in the sense that the more locally concentrated a party's vote, the fewer votes per seat it will register in the larger region. At the same time, it is also a measure of vote distortion in the first-past-the-post system, given that smaller parties with broad national support (think the NDP) will exhibit a larger number of votes per seat, than will smaller parties with concentrated regional support (think the Bloc Quebecois), while parties with overwhelming support in a certain region (the classic example is conservative parties in Alberta) will also demonstrate large votes per seat in that region.

Votes Per Seats and 2nds = Party Vote / (Party Seats + 2nds)

The number of votes cast for a party for every seat or 2nd place finish it secured in a local region, province or region during a given election. The metric is undefined in the case where a party has won no seats or 2nds (can't divide by zero).

This metric should be read across elections, to see whether a party's support is growing and if so, how effectively (or is shrinking, and if so, how deleteriously) ... remembering that the only real measure of electoral success is the actual number of seats a party is able to win. On the other hand, when you're in second-place you're at least still in the Contest. And, since changes to the federal Election Financing law came into effect, votes for a registered political party also dictate its subsequent financial subsidy.

Financial Metrics

Funds Raised Locally

Includes campaign contributions from individuals, and from corporations/businesses, trade unions, governments, and other groups when they were permitted (i.e., prior to the 2003 legislation changes), along with all the funds transferred from the local riding association. This figure gives an indication of how well-organized and self-sufficient the local party riding association (or independent candidate) is.

Num Ind - Number of individual contributors

The number of individual contributors during an election campaign. This number gives another indication of the organizational capacity of the local riding and/or candidate. It is the denominator used to calculate the average individual contribution (not yet shown).

Num Corp - Number of corporate/business contributors

The number of corporate or business contributors during an election campaign. This number gives an indication of the relative reliance of the local riding and/or candidate on such contributions prior to the 2003 legislative changes. It is the denominator used to calculate the average corporate contribution (not yet shown).

From Party

The amount of money transferred to the local campaign by the Party Head office/Central Campaign. This number can suggest that a riding was a party priority, or that it required support, or that its candidate was successful in commanding a greater share of central resources, for example.

Spent

The amount of money spent by the riding campaign on Election Expenses. It is the numerator in the calculation of Pct Limit, Spent Per Voter and Spent Per Vote.

Pct Limit = 100 * (Spent / Spending Limit)

The percent of the riding's Spending Limit for that Election spent by the riding campaign on Election Expenses.

Per Voter = Spent / Electors

The amount of money per voter spent by the riding campaign on Election Expenses.

Per Vote = Spent / Party Vote

The amount of money per vote obtained spent by the riding campaign on Election Expenses.