[su_tab title=”Projections” disabled=”no” anchor=”” url=”” target=”blank” class=””]
Following 2016, the PPD Election Projection Model added Slightly to the scale. The model maps three electorates for gauging potential outcomes. For each party, one electorate assumes a higher turnout among one party’s base juxtaposed to the other on a 4-cycle average, while the third assumes a normal base turnout.
Battleground: The truest of tossups. Neither party’s candidate has a clear edge and is generally seen as a 50/50 proposition. For our model, it means the three electorates we mapped out give each a victory if their base outperforms juxtaposed to the average, while the base model returns an even race.
Slightly: This year, we felt that there was a place for a tossup projection between Battleground and Leaning. This saw a race with a party’s candidate holding a very slight edge in the base model, albeit by an uncomfortably small margin.
Leaning: The traditional Leaning projection now means what we always meant it to mean, which is that the race is actively leaning toward one party’s candidate. High base turnout for either candidate can still mean their victory, but the base model has begun to show a clear advantage.
Likely: High base turnout for one candidate still does not spell their certain victory, while the base and high-base models obviously agree.
Solidly: While some would say something like, “there’s a 10% chance this will happen,” we say that Solidly is assigned when the model projects almost no scenario in which one party can prevail against the other.
Please note: While we will place the model-only projections separately, the mixed-model will adjust as actual voting data is collected.
[su_tab title=”Updates” disabled=”no” anchor=”” url=”” target=”blank” class=””]
[su_posts id=”73258,73240,67748″ posts_per_page=”2″]
[su_tab title=”Methodology” disabled=”no” anchor=”” url=”” target=”blank” class=””]
Projections are determined by PPD’s comprehensive model that includes several variables. We break them down into two separate categories. But first, a quick word on polling.
Polling is provided when available, and in-house polling is conducted by Big Data Poll. But polling in and of itself is not enough to determine a credible race rating. Media and public polling is weighted to value based on accuracy.
The model uses several variables that we consider to be statewide, including but not limited to polling, state demographics and political leanings, prior voting history and trends, candidate strength, voter ideology, fundraising, voter registration and affiliation (separate).
Candidate recruitment and strength also considers incumbency, experience, GOTV (get out the vote) apparatus and campaign organizations. But it also considers ideology juxtaposed to the state.
Voter sentiment includes the direction of the state, economic optimism, real GDP and unemployment.
Presidential approval rating factors into the state as well, but the national political environment, or voters’ sentiment toward each party — including the ever-important variables of the economy, both state and national — carries significant weight. As we saw in the Obama era, high approval ratings may not be enough to save or defeat an incumbent party if the national sentiment blows hard enough one way or the other.
[su_tab title=”History” disabled=”no” anchor=”” url=”” target=”blank” class=””]
Thank God for the Internet.
Read Post-2014: PPD’s Big Debut: Most Accurate Election Projection Model Of 2014, Hands Down
Read Post-2016: PPD the Most Accurate Election Projection Model for Second Straight Cycle