Two new sections are now available on the Atlas - Gubernatorial Election Polls for 2006 and Senatorial Election Polls for 2006. The new polling sections include the ability to view individual polls in detail as well as summary pages comparing recent polls by state and by polling agency. The main page includes a compilation of the most recent three polls for each state and presents a map of the result showing the current status of who is ahead on a per-state basis as well as a table showing the current view of the pick-ups for each part. Also included is a trend graph illustrating the number of states ahead for each party starting in February of 2006.

The 2004 Presidential Predictions turned out to be a very popular Atlas site feature - with almost 2,000 unique entries (and over 12,000 total version entries). The compiled map showed the correct winner of every state - with the exception of Ohio - which missed being called by merely two Republican predictions (1,000 Democratic calls vs. 998 Republican).

For 2006, new prediction scripts and databases for the Gubernatorial and United States Senate races have been created. The predictions have been expanded to include independent candidates, show party “pick-ups”, and has more integration with the Atlas Forum.

Participate and Enjoy!

Total Ballots Cast are figures published by most election agencies that show the total number of people having cast ballots. Typically, a number of these ballots for any given contest (e.g. President) are invalid - either the voter decided not to choose a candidate for President (called a “blank” or “under-vote”) or the voter selected multiple candidates (called a “void” or “over-vote”).

Figures for Total Ballots Cast in the 2004 and 2000 general elections for President have been added to the database at the state, county, township, and precinct levels for all states in which this data is available. These data may be accessed by using the customize form at the bottom of any detailed data table (Atlas Membership required for county, township, and precinct-level). In addition, there are options to display the total number of invalid ballots and percent invalid for each in the table. For an example, take a look at the 2004 state-level data table with the new data included. Data for previous years are also in process of being added to the database.

Swing Map

Trend Map

Upon a suggestion from Alcon, I have created swing maps by county between the 2000 and 2004 elections. These maps describe the change in margin percentage (i.e. the difference between the Democratic vote percentage and the Republican vote percentage) between the 2000 election and the 2004 election. The maps may be found on the state summary pages (example). These maps show the change in support on a percentage basis for each county between the successive elections. In the New Jersey map to the right, the margin in every county swung towards Bush. The very light blue represents a change of < 5%, the light blue between 5% and 10%, the blue between 10% and 15%, and the dark blue 15% to 20% (see the Atlas Master Key for complete color and margin assignments).

(edit: 20051101) A second set of maps have been produced that show the county “trend” between the 2000 and 2004 elections. The trend map differs from the swing map in that the margin change is normalized to the nation-wide margin change (i.e. MarginCounty2004 - MarginNational2004 - (MarginCounty2000 - MarginNational2000) where each Margin is (VoteD - VoteR)/TotalVote). So, as an example, consider Hunterdon County, New Jersey:

To calculate the swing in Hunterdon County, subtract the margin percentage in 2000 from the margin percentage in 2004 -> an increase for the Republicans of 1.58%. Therefore the swing map shows Hunterdon County for Republican by < 5%. The trend is a normalization of the county margin change based on the national margin change. The trend is calculated by subracting the national margin swing from the county margin swing. Since the margin for Republican nationwide increased by 2.98% and the margin in Hunterdon County swung a smaller 1.58% towards the Republicans, the trend in Hunterdon County is 1.40% towards the Democrats. The map therefore shows Hunterdon County for Democratic by < 5%.

Feedback is welcome on all aspects of the feature (color selection, number of bins, whether it is useful or interesting, etc.).

The Atlas now has a Wiki! The community pages are intended as an area for contribution by interested parties in the area of Presidential Elections, History, US Politics, as well as repository for community interaction. I envision the wiki to include additional information related to elections - such as defining terms (examples include Fusion, Caucus, Primary, etc.), articles on historical election patterns, candidate biographies, political party information, interesting trivia, etc. Cross-linking between the main atlas database and the wiki helps to improve the educational aspect of the site. Please send me an email if you are interested in contributing to the Wiki.

The Atlas Electoral College Calculator has been a very popular feature on the site. One of the alternate uses of the tool has been to create maps for discussion on the forum (such as what-if scenarios, predictions, etc.). A common feature request for the calculator is to show popular vote percentages of the winning candidate as different shades of red and blue (and green) in the same manner as presented with the Atlas results maps. This functionality has now been added to the calculator. The user has an option to turn the feature on or off in the Map Options box (in the off-state, the map generator simply ignores the percentages shown in the boxes). The name has also been changed from the Electoral College Calculator to the Electoral College Calculator and Map Generator.

Yesterday saw the debut of the new mock election. A signficant improvment over the one that I hosted in 2000, the mock election tracks results via data, maps, and statistics in all US States as well as Canadian Provinces, European and other countries. Either the free or paid levels of site membership is required in order to participate in the mock election.

I have put a lot of work into trying to solve several shortcomings with the popular User Prediction feature. The problems have been the following:

• Lack of the ability to predict individual Congressional Districts in Nebraska and (especially) Maine.
• User annoyance for Members of managing multiple logins and passwords
• The presence of some degree of spamming by some users
• Due to ip limitations to control spamming, the inability of some legitimate users to create predictions

I have attempted to address all of these issues with today’s upgrade. The User Prediction section now has the ability to call and change results for individual congressional districts in those states that choose electors by congressional district (Maine and Nebraska). The compiled median results also support these individual electoral votes.

Members no longer have to choose username and password (essentially the ability to skip step 1). Once members have logged into the site, the software automatically recognizes their prediction. For those many site visitors and predictors, I have created a new Free version of site membership. This “User” level of membership allows for creating and updating user predictions, posting comments to polls, and participating in the soon-to-be-launched mock election. The one hassle, however (and I apologize in advance), is the need to register. This process also includes a email-verification step in an attempt to limit spamming.

Current predictions can be completely upgraded to the new format (all versions) - individual congressional districts are called for the same candidate as the currently-chosen statewide winner. The process is:
1. If you have not signed up, please do so here.
2. Then, upgrade your user prediction here

Please submit a bug if there are any problems.

The Electoral College Calculator has been well-received! Its even now featured on one of the support sites for Wesley Clark (http://www.us4clark.com/). Atlas visitors have also submitted several suggestions for improvements. Without losing the elegant simplicity of the current EV calculator, I’ve added a second, “deluxe” version that can be viewed via the link in the 2004 section. This version includes a dynamic bar graph showing the totals of each democratic, republican, and tossup electoral votes with the all-important 270 EV milestone. In addition, the page features the ability to generate a map from your state selections. Play around with it and let me know what you think. Note that the current state polygons of the generated map are only the first-draft. The rough-edges will be refined in due time. The same functionality will eventually be ported to the user prediction section to help alleviate the upload problems that have been plaguing PC users.