forecasted outcomes
Each day, we use our models to forecast the outcomes in all upcoming state and federal general elections with filed candidates. This table includes a running log of the outcomes predicted by those forecasts.

Schema

Field
Description
timestamp
The date and time that a forecast was generated.
campaign id
Unique ID maintained by Deck representing a specific campaign. We define a campaign as an instance of a specific candidate running for a specific office or party nomination. If the same candidate ran in a primary and general election in both 2018 and 2020, they would have four campaign entries across those contests.
election state
The state (in the form of a two-letter abbreviation) that a district’s jurisdiction is associated with.
election stage
One of four values (“primary”, “primary run-off”, “general”, or “general run-off”) indicating which stage a given election’s contests were at.
election date
The final date (or only date) of voting for a given election.
office name
The name of the elected office associated with a given district.
district name
The name of a given district. Georgia State House District 2 will be named “2,” for example.
candidate party
The candidate’s stated party affiliation. Note: if a candidate is on the ballot in affiliation with multiple parties (such as the Working Families Party, which is a common second ballot line for New York progressives), we only include one party here: the one with the most electoral success.
candidate name last
A given candidate’s primary last name -- typically their full, legal last name.
predicted vote share
The vote share we expect this campaign to receive in its contest. These values are presented at two significant digits to avoid the impression of false precision.
margin of error
The margin of error (in percentage points) for our predicted vote share at a 95% confidence level. These values are presented at two significant digits to avoid the impression of false precision.
probability of winning
The probability that a given campaign will win in a contest. These values are presented at two significant digits to avoid the impression of false precision.
explanation
A JSON string with the aggregate SHAP values for a given category of features in our forecasting model.

Example entry

[{
"timestamp": "2022-07-28T10:49:24Z",
"campaign_id": "559cf193-5b42-4421-a7fa-44bfcdf63ba7",
"election_state": "AL",
"election_stage": "general",
"election_date": "2022-11-08",
"office_name": "STATE HOUSE",
"district_name": "74",
"candidate_party": "DEMOCRATIC",
"candidate_name_last": "ENSLER",
"projected_vote_share": "0.51",
"margin_of_error": "0.06",
"probability_of_winning": "0.62",
"explanation": [
[
{
"category": "candidate: history",
"impact": -0.017
},
{
"category": "candidate: issue stances",
"impact": -0.003
},
{
"category": "candidate: party",
"impact": 0.009
},
{
"category": "contest: number of candidates",
"impact": 0.001
},
{
"category": "data quality indicators",
"impact": -0.001
},
{
"category": "fundraising: contributor traits",
"impact": -0.001
},
{
"category": "fundraising: in-district",
"impact": 0.003
},
{
"category": "fundraising: overall",
"impact": -0.004
},
{
"category": "media: overall share",
"impact": -0.011
},
{
"category": "media: sentiment",
"impact": 0.001
},
{
"category": "national mood",
"impact": 0.004
},
{
"category": "previous cycle results",
"impact": 0.02
},
{
"category": "voter traits",
"impact": 0.034
}
]
],
}]
Export as PDF
Copy link
On this page
Schema
Example entry