Talk:Results of the 2008 Democratic Party presidential primaries/Archive 11

Criteria for including/excluding events
While adding nominating events to the Overview of results table, I've run across many events that I'm not sure we should include because they have no impact on the allocation of delegates to candidates. It occurred to me that we may need to establish criteria by which we'll judge whether or not to include an event.

We could simply include all events regardless of their impact on candidate standings, but I think that this will clutter the table with too many uninformative rows. Those editors who feel that it's too complicated to include any events at all would probably find including all events to be particularly cumbersome. Some discussion about which events to include took place earlier (see here and here) and the consensus at that time was also to avoid including all events.

I think that the criteria for deciding which events to include or exclude should be based on the impact those events have on the final allocation of delegates to candidates. I find it useful to group events into 4 categories describing their degree of impact:


 * direct impact: These are events at which attendees decide some or all of the final numbers of pledged delegates who will be bound to each candidate. Most state primaries belong in this category since their results are the final word on how many pledged delegates will be bound to each candidate. Some, but not all, caucus events also belong in this category. For example, Iowa's April 26 district conventions belong in this category, because their results are the final word on allocating 29 of Iowa's 45 pledged delegates to candidates. Iowa's January 3 caucuses, however, do not belong in this category, because their results do not directly determine the final number of pledged delegates who will be bound to candidates. However, Iowa's January 3 caucuses do fit into either of the next two categories.
 * proxy impact: These are events at which attendees choose delegates who will act as their proxies at subsequent events in deciding the final delegate-to-candidate allocation. Iowa's January 3 caucuses belong in this category because some of the delegates that attendees choose at this event will go on to attend the April 26 district conventions where, as described above, the delegates make a final decision about the number of pledged delegates that will be bound to each candidate.
 * indicator impact: These are events at which a popular vote for candidates takes place, but the vote has no direct impact on the allocation of delegates to candidates. Washington's and Idaho's primaries belong in this category. The results of neither primary are used to allocate delegates to candidates, but they do provide an indicator of voters' presidential preferences, which could influence pledged delegate and superdelegate votes. Interestingly, Iowa's January 3 caucuses also belong in this category since a popular vote is what determines how many of the "proxy delegates" are allocated to each candidate. In fact, the first event in just about every caucus state falls into this category since they involve popular votes for candidates.
 * zero impact: These are event that have no impact whatsoever on the allocation of delegates to candidates. Many (but not all) state conventions belong in this category because they choose only the actual persons who get to attend the Democratic National Convention, not the number of delegates who are pledged to vote for each candidate. --Bryan H Bell (talk) 01:00, 26 March 2008 (UTC)


 * Modification to remove intial caucus events from "indicator impact" category:
 * direct impact: These are events at which attendees decide some or all of the final numbers of pledged delegates who will be bound to each candidate. Most state primaries belong in this category since their results are the final word on how many pledged delegates will be bound to each candidate. Some, but not all, caucus events also belong in this category. For example, Iowa's April 26 district conventions belong in this category, because their results are the final word on allocating 29 of Iowa's 45 pledged delegates to candidates. Iowa's January 3 caucuses, however, do not belong in this category, because their results do not directly determine the final number of pledged delegates who will be bound to candidates. However, Iowa's January 3 caucuses do fit into either of the next two categories the next category.
 * proxy impact: These are events at which attendees choose delegates who will act as their proxies at subsequent events in deciding the final delegate-to-candidate allocation. Iowa's January 3 caucuses belong in this category because some of the delegates that attendees choose at this event will go on to attend the April 26 district conventions where, as described above, the delegates make a final decision about the number of pledged delegates that will be bound to each candidate.
 * indicator impact: These are events at which a popular vote for candidates takes place, but the vote has no direct impact on the allocation of delegates to candidates. Washington's and Idaho's primaries belong in this category. The results of neither primary are used to allocate delegates to candidates, but they do provide an indicator of voters' presidential preferences, which could influence pledged delegate and superdelegate votes. Interestingly, Iowa's January 3 caucuses also belong in this category since a popular vote is what determines how many of the "proxy delegates" are allocated to each candidate. In fact, the first event in just about every caucus state falls into this category since they involve popular votes for candidates.
 * zero impact: These are event that have no impact whatsoever on the allocation of delegates to candidates. Many (but not all) state conventions belong in this category because they choose only the actual persons who get to attend the Democratic National Convention, not the number of delegates who are pledged to vote for each candidate. --Bryan H Bell (talk) 20:50, 26 March 2008 (UTC)

Proposal 1: exclude only proxy and zero impact categories
Using the above categories [ without the modification --Bryan H Bell (talk) 20:50, 26 March 2008 (UTC) ] as a guide, here is the criteria I propose for deciding which events to include or exclude from the Overview of results table. I propose we exclude events that fall into the "zero impact" or "proxy impact" categories and include events that fall into the "direct impact" or "indicator impact" categories. To put it simply, I propose we include only events in which either delegates are bound to candidates or a popular vote takes place. This criteria would retain each state's initial event as well as the non-binding primaries in Washington and Idaho, but it would jettison many intermediate events that have little impact on the allocation of delegates to candidates. I think the results of events with "proxy impact" will be difficult to obtain. As an example of how my proposed criteria would affect the inclusion or exclusion of events, here is a run-down for one state: Washington. If my proposal is accepted, we will need to remove several events from the Overview of results table and should also add some explanatory text at the top of the section. --Bryan H Bell (talk) 01:00, 26 March 2008 (UTC)
 * Washington precinct caucuses (indicator impact, proxy impact) included
 * Washington primary (indicator impact) included
 * Washington legislative district caucuses (proxy impact) excluded
 * Washington county conventions (proxy impact) excluded
 * Washington congressional district caucuses (direct impact, proxy impact) included
 * Washington state convention (direct impact) included


 * I take it the March 15 Iowa County Conventions would also stay as an updated proxy impact? (And if I recall corectly, the upcoming Iowa Congressional Districts would stay as direct impact for district delegates and the Iowa State Convention as direct impact on the state wide delegates.) Jon (talk) 13:20, 26 March 2008 (UTC)


 * Oppose. I don't follow all the nuances of the multiple-event states, but it seems to me that any categorization that lumps the January 3 Iowa caucuses and the Washington primary in the same category (indicator impact) and includes them on that basis alone is flawed somehow. The initial caucus events should clearly be included, while I think there is a legitimate argument for the Washington primary being excluded. They should be considered under different categories. Using your categories, I guess I'm arguing to include "direct" and "proxy" events and to exclude "indicator" and "zero" events. My main reason for supporting a multiple events table was to track changes in estimated delegates in events like the March 15 Iowa County Conventions. It seems like you're proposing to exclude that event.Northwesterner1 (talk) 17:24, 26 March 2008 (UTC)


 * Good point that the "indicator impact" category might an inappropriate grouping of events and by extension a flimsy basis for including events. I don't know that it necessarily means the entire categorization scheme is flawed. Perhaps merely modifying the definition of "indicator impact" above to remove the word direct would suffice, limiting the category to popular vote events with no impact on delegates (i.e. Washington and Iowa primaries alone). With that modification, I would agree that including "direct impact" and "proxy impact" events while excluding "indicator impact" and "zero impact" events could make sense. Another option would be to exclude only "zero impact" events.
 * It wasn't my intention to exclude the obviously notable March 15 Iowa county conventions, but by the criteria I've proposed, it would be excluded since it represents a "proxy impact" only. Oops. My intention in excluding "proxy impact" events was that I suspected a great many of them will have little impact on the delegate allocations and thus no reliable sources will report their results. I've already encountered this issue with North Dakota's legislative district conventions. I also hoped to cut down on the events which would list their total delegate count as "0 (of x)". Finally, I thought that excluding "proxy impact" events would be a good way to cut down on the total number of events, keeping the table trim and simple. Ah, well. I tried. Let me make another attempt to make this entire scheme work by adding a modified categorization and two new proposals. --Bryan H Bell (talk) 20:30, 26 March 2008 (UTC)

Proposal 2: exclude only indicator and zero impact categories
Using the above modified categories as a guide, propose we exclude events that fall into the "indicator impact" or "zero impact" categories and include events that fall into the "direct impact" or "proxy impact" categories. This criteria would retain each state's initial event, but it would jettison the non-binding primaries in Washington and Idaho. As an example of how the proposed criteria would affect the inclusion or exclusion of events, here is a run-down for one state: Washington. If this proposal is accepted, we will need to remove several events from the current Overview of results table and should also add some explanatory text at the top of the section. --Bryan H Bell (talk) 21:03, 26 March 2008 (UTC)
 * Washington precinct caucuses (proxy impact) included
 * Washington primary (indicator impact) excluded
 * Washington legislative district caucuses (proxy impact) included
 * Washington county conventions (proxy impact) included
 * Washington congressional district caucuses (direct impact, proxy impact) included
 * Washington state convention (direct impact) included


 * Support. Thanks for modifying the categories and clarifying the proposal. I like this option, excluding the non-binding primaries, as I see them as roughly akin to straw polls. However, I'm also okay with proposal 3 below. Northwesterner1 (talk) 21:09, 26 March 2008 (UTC)


 * Support — I would rather be inclusive when it comes to these events. – Zntrip 00:52, 27 March 2008 (UTC)


 * Support Jon (talk) 13:12, 27 March 2008 (UTC)

Proposal 3: exclude only zero impact categories
Using the above modified categories as a guide, propose we exclude events that fall into the "zero impact" categories and include events that fall into the "direct impact", "proxy impact", or "indicator impact" categories. As an example of how the proposed criteria would affect the inclusion or exclusion of events, here is a run-down for one state: Washington. If this proposal is accepted, no events will need to be removed from the current Overview of results table, but we should add some explanatory text at the top of the section as to why the "zero impact" events haven't been included. --Bryan H Bell (talk) 21:03, 26 March 2008 (UTC)
 * Washington precinct caucuses (proxy impact) included
 * Washington primary (indicator impact) included
 * Washington legislative district caucuses (proxy impact) included
 * Washington county conventions (proxy impact) included
 * Washington congressional district caucuses (direct impact, proxy impact) included
 * Washington state convention (direct impact) included

Consensus
It looks like consensus favors Proposal 2 above. I have therefore removed the following events from the Overview of results table: None of the other zero or indicator impact events had been included yet. I also made an attempt to explain all of this in the introductory paragraph of the Overview of results section. --Bryan H Bell (talk) 22:55, 29 March 2008 (UTC)
 * Kansas district conventions: zero impact
 * Democrats Abroad regional caucuses: zero impact
 * Democrats Abroad global convention: zero impact
 * Washington primary: indicator impact
 * Idaho primary: indicator impact


 * Thanks for all the great work, Bryan. I re-edited the intro a little, trying to make it a little clearer. I think it can be improved further still. Should we mention WA primary in the intro? Northwesterner1 (talk) 23:51, 29 March 2008 (UTC)


 * I was rather hoping you'd do a little editing to the intro. Your edits are an improvement. Thanks. The intro still seems a little long and complex to me, but I'm as yet uncertain how to shorten or simplify it. Explaining the absence of the WA primary would be good if we can do so without adding much to the intro's size and complexity. I think looking for ways to further streamline the table will also help in this effort. I hope to present a few ideas for this soon. --Bryan H Bell (talk) 04:00, 30 March 2008 (UTC)

Gravel
He's just joined the Libertarian Party. Can we now FINALLY remove him? — Nightstallion 11:45, 26 March 2008 (UTC)


 * Is just stating your now a member of the Libertarian Party sufficent for the DNC to no longer consider the person a party member in good standing or does he also have to file paperwork with them to be removed? Jon (talk) 13:23, 26 March 2008 (UTC)


 * I don't quite understand why editors get so bunched up about removing Gravel. Of course we won't remove him. Any candidate who has run a national Democratic Party campaign should be listed on the table. However, it is true that Gravel has announced on his own web site that he is leaving the Democratic Party and joining the Libertarian Party. This amounts to a public announcement of withdrawal from the Democratic Party presidential race. I've therefore shaded as "Withdrawn" the remainder of the rows in Gravel's column and moved the column over. --Bryan H Bell (talk) 21:38, 26 March 2008 (UTC)


 * I think switching to another party validates listing him as withdrawn. See his campaign Web site. – Zntrip 00:49, 27 March 2008 (UTC)

Democrats Abroad Events
It looks like an editor has added the democrats abroad cacus delegates to the democrats abroad primary and deleted the line for the cacus instead of using it and deleted the comment on the primary about it only having part of the delegation. Jon (talk) 13:19, 31 March 2008 (UTC)


 * It looks like the Democratic Abroad subarticle has the same problem of now comingling primary & cacus results. Prior to the results coming in from their cacus these articles were all claiming something like 1 1/2 delegates would be betermined at the convention intead of the primary. Jon (talk) 13:23, 31 March 2008 (UTC)


 * I'm the editor that made the changes to both articles. A closer reading of the recently revised and official Democrats Abroad Delegate Selection Plan reveals (in section I.A.4) that the results of the primary determine the allocation of all 7 delegate votes to each candidate. Subsequent caucus events determine only who will serve as delegates, not how many delegates are bound to each candidate. I've added notes to this effect (with appropriate citations) in the Democrats Abroad subarticle. --Bryan H Bell (talk) 20:00, 31 March 2008 (UTC)

Proposal: Calculate percentages as a total of delegates TO DATE, not ALL delegates
 Several days ago, this edit on the main primary article changed the practice for reporting delegate percentages in the summary table on that article. Whereas the pledged delegate count used to read 53-47% (based on the percentage of all delegates projected to date), it was changed to 43-39% (based on the percentage of all delegates available, including states that haven't voted yet). There was no edit summary to explain the change, but I assume the editor changed it to match the results article. which has an extra column for "uncommitted delegates" and includes a percentage for that column (e.g. 43-39-18%)

I would like to revert to the original practice on the main primary article, reporting a percentage of all delegates projected to date. In addition, I would like to change the current practice on this results article. Candidate percentages should be calculated as a percentage of all delegates awarded or projected thus far, excluding future contests. Percentages should be removed from the uncommitted column. Percentages should also be removed from the "Uncommitted" superdelegates and the "Uncommitted" grand total, although they may be maintained in state rows where appropriate.

Reasons: Northwesterner1 (talk) 08:36, 2 April 2008 (UTC)
 * As the main article does not have an uncommitted column, it's very confusing when these numbers don't add up to 100.
 * The current practice is at odds with accepted statistical practice just about everywhere in the real world. When the numbers are rolling in for State X, CNN says "with 50% of precincts reporting, Candidate A has 60% of the vote, and Candidate B has 40% of the vote." They don't say, "Candidate A has 30% of the vote, Candidate B has 20% of the vote, and 50% is undetermined."
 * The current practice introduces a POV problem, as it artificially narrows the gap between the two candidates on all measures.


 * Mostly neutral. I feel ambivalent toward most of this proposal. Either way is fine. The one point I disagree with is that the percentages should be removed from the uncommitted column on the results article. As its footnote indicates, that column doesn't (and shouldn't) indicate the number of potential delegates in upcoming contests. It measures the pseudo-candidate "uncommitted" which is a valid candidate choice in most contests. We show percentages for all the other candidates, so we should show percentages for the "uncommitted" candidate as well. Part of the problem here is that we currently fudge this meaning somewhat in the superdelegate estimate row. I suppose the (weak) argument is that the superdelegates that have not endorsed a candidate have so far "picked" uncommitted. If you don't buy that argument, then perhaps a different way to fix the uncommitted issue you see would be to mark the delegate count in the uncommitted column of the superdelegates row as "0". However, that wouldn't prevent the issue arising again later if a particular contest again yielded some true uncommitted delegates (or if Michigan's delegates somehow got seated as is). Another possible solution would be to simply let the main and results articles show different percentages, each with the percentages appropriate to the particular way they display their numbers. --Bryan H Bell (talk) 09:32, 2 April 2008 (UTC)


 * You're right. I misread the uncommitted column. I'm withdrawing that part of the proposal.Northwesterner1 (talk) 09:37, 2 April 2008 (UTC)


 * Thank you. That changes my vote to just plain neutral. --Bryan H Bell (talk) 19:51, 2 April 2008 (UTC)


 * Support On the main primaries article, I also found it confusing seeing the percentages not adding up to 100%. Andareed (talk) 09:52, 2 April 2008 (UTC)


 * Done. Well, it's not exactly overwhelming consensus here with two votes in support, but I'm going to make this change, as there is no opposition. Northwesterner1 (talk) 19:25, 5 April 2008 (UTC)


 * By the way, what's up with the Texas caucus, which shows 50% uncommited 22% Clinton 28% Obama? By the same logic, shouldn't that be 44% Clinton 56% Obama, with 41% of precincts reporting? The state article says there were no "uncommitted" delegates, only "undetermined" delegates.Northwesterner1 (talk) 19:25, 5 April 2008 (UTC)

Question about updating pledged delegate totals
I apologize in advance if I'm breaking any kind of convention.

I've been hearing various news sources updating delegate totals. However, these changes don't seem to be propagating to all the various news organizations (eg NY Times ). For example, there are these two articles I have heard/read (Chicago Public Radio and MSNBC ). Both of these articles are from what I would call reputable news sources and they correct previously published delegate totals. However, I have tried to find further references about these particular issues but have come up empty handed.

My question is basically: given how complex this particular race has been, what is the criteria for selecting sources for delegate totals for this article? DavidForero (talk) 17:50, 3 April 2008 (UTC)


 * All lines in this table (and the "main" democratic article) use the linked state pages for the pledged delegates and DemConWatch for the superdelegates. So when DemConWatch changes, we'll change the superdelegates to match. For a pledged delegate change, we fix the state page first and then the tables. (On the state primary pages, the prefered source for vote count is certified state results, but that's often not advaible for a couple of weeks, pending which we use a major news site. Also the states don't tend to keep track of delegates so a major news site is used for that was well.)Jon (talk) 18:00, 3 April 2008 (UTC)


 * For more context, see the section Reporting delegate totals in the main article Democratic Party (United States) presidential primaries, 2008. Northwesterner1 (talk) 20:56, 3 April 2008 (UTC)

Proposal: split Overview of results section
I propose we split the Overview of results section into two sections: National summary and Local contests. The first section would contain the first four "totals" rows of the current section's table. The second section would contain the remaining "events" rows. The split would provide the following advantages: Please add your support for, opposition to, or comments on this proposal below. Listing any advantages or disadvantages I've missed would be helpful. As much as possible, let's try to keep the focus on whether or not we should split the section and avoid getting too caught up in the details of how we'd format or wordsmith the resulting two sections. We can work out those details later, either with successive edits or new discussion sections on this talk page. --Bryan H Bell (talk) 06:58, 4 April 2008 (UTC)
 * Allows expansion or reformatting of information in the "totals" rows without trying to shoe-horn it into the existing table format. For example, we would be free to split the information into two totals tables: an "agressive" table that totals estimated pledged delegates with estimated superdelegates and a "conservative" table that totals only bound pledged delegates with no superdelegates.
 * Provides the ability to break up the current section's long explanatory text into smaller chunks.
 * Provides more flexibility in organizing the "events" rows. For example, we could make the events rows sortable by date or location, letting readers see a nation-focused cross-state progression of events or a state-focused local progression of events.


 * Support I agree we could write clearer and more concise explanatory text by concentrating on the two parts separately. I also still like the idea of making event rows sortable by state. Northwesterner1 (talk) 07:48, 4 April 2008 (UTC)
 * Support Same as above. ChPr (talk) 09:29, 4 April 2008 (UTC)

✅ Though there haven't been many votes cast on this over the past week, there has been no opposition, so I've gone ahead and split the section. Now we can begin making refinements to each separate section. --Bryan H Bell (talk) 18:37, 10 April 2008 (UTC)


 * Looks good, Bryan! Thanks for doing all that work. The sortable tables aren't working quite right, though. They seem to be sorting numerical values alphabetically, e.g. 150 comes after 10 but before 20. I'll try to take a look at the syntax later unless you know right offhand how to fix it.Northwesterner1 (talk) 02:01, 11 April 2008 (UTC)


 * Thanks. The brackets might be causing the software to interpret some columns as text rather than numbers (see Help:Sorting). When I get a chance, I'll see if I can fix it (unless you fix it first). --Bryan H Bell (talk) 04:43, 11 April 2008 (UTC)


 * I played around with the instructions at Help:Sorting and couldn't get it to work. I decided just to make the first two columns sortable for now, as that seems the most important... Northwesterner1 (talk) 04:03, 13 April 2008 (UTC)


 * I tried messing around with this as well and couldn't get a satisfactory result. We could force the correct sorting by adding hidden text for each cell, but that seems like a clumsy, tedious, and error-prone way to handle it. I think your solution works best for now. --Bryan H Bell (talk) 21:20, 14 April 2008 (UTC)

Superdelegates
In an effort to standardize the superdelegate numbers across the article series, I have begun a conversation at Talk:List of Democratic Party (United States) superdelegates, 2008 that may interest some editors of this article. Northwesterner1 (talk) 22:03, 6 April 2008 (UTC)


 * Speaking of superdelegates, I think the blog post is behind on counting them. According to FirstRead, Clinton holds a 256-225 superdelegate lead over Obama. In the overall count, Obama leads by 133 (1,641 to 1,508). He has a 164 pledged-delegate lead (1,416 to 1,252). The pledged delegates are almost exactly the same, but the gap is with the count of supers. ChPr (talk) 09:01, 9 April 2008 (UTC)


 * I can't quickly reconcile the difference between First Read and the current DCW watch based on the count alone with First Read giving Clinton more delegates than DCW and yet Obama fewer than DCW. (If First Read had higher counts for both or lower counts for both it would make more sense.) DCW does however list every single delegate. Unless First Read does the same it's going to be impossible to reconcile them. Jon (talk) 13:19, 9 April 2008 (UTC)


 * FirstRead: Obama 1416 + 230 = 1646, Clinton 1252 + 260 = 1512
 * DemConWatch: Obama 1416 + 222 = 1638, Clinton 1253 + 246 = 1499
 * The difference between FirstRead's and DCW's pledged delegates, is that FirstRead has one remaining delegate to allocate from the Democrat's Abroad, which DCW gives to Clinton. FirstRead's superdelegate count is higher for both, however. ChPr (talk) 00:33, 11 April 2008 (UTC)

Popular Vote discussion
The discussion of counting popular vote is factually flawed with regard to one of the examples. Specifically, the numbers for Washington that are said to be of caucus participants are nothing of the sort. The reported numbers are of delegates elected by the caucus precincts, to be sent to the county & congressional district conventions. The number of these precinct delegates given each precinct was assigned based on the turnout there in the previous presidential election, and have no direct relationship to the actual participation in the precinct during the caucus. Obama did not receive 11,700 more votes, but rather, an unknown number of caucus voters sent 11,700 more delegates to the next level than did an unknown number of Clinton voters (a ballpark estimate would be at about 10 times as many participants as delegates). Not knowing exactly which numbers were included in the table, I cannot provide an alternative discussion of the problem, but a fix is in order. Agricolae (talk) 06:20, 9 April 2008 (UTC)


 * Good catch. I've returned it to the old example (Hawaii + Rhode Island), as we have an actual popular vote count for the Hawaii caucus.Northwesterner1 (talk) 07:27, 9 April 2008 (UTC)