Talk:Opinion polling for the 2016 Philippine presidential election

Metro Manila-exclusive surveys[edit]

These should be on another section. –H T D 15:56, 26 November 2015 (UTC)[reply]

Are the graphs original research? They are certainly misleading. We should remove them.[edit]

Do the lines on the graphs here violate WP:No original research? They look like original research to me. The graphs look like someone took the results of surveys, plotted the results, and then drew lines through the graph, but significantly NOT connecting the dots. If the lines were connecting the dots, then the lines would be OK as they would be representing the actual results in a more understandable way. But the lines do not. The lines seem to be drawn by the author to show some sort of average or some sort of momentum. The lines provide therefore a narrative -- a conclusion -- an analysis -- and that violates both WP:No original research and WP:Neutral point of view unless the lines are from a reliable WP:Third-party source. Until the lines can be sourced, we should remove them from this and all other articles and delete the graphs from Wikimedia Commons. --Iloilo Wanderer (talk) 03:59, 25 January 2016 (UTC)[reply]

These is a graph that unites all surveys by all pollsters that's here. As they have different methodologies, it would be inaccurate to link every plot point as if they are related to each other. Instead of making separate graphs per pollster so you could link the plot points, I've added a trendline that uses polynomial regression to link all plot points together.

On the next update I'd be using a moving average as polynomial regression is quite limited on how many "ebbs" and "crests" it allows.

FWIW, this is the method used in other opinion polling articles. Go see them yourself. –H T D 07:49, 25 January 2016 (UTC)[reply]

Thanks for your (technical) explanation, Howard.

It's a close call but personally, I favor retaining them for the time being.

Any very topical article with new facts emerging almost daily has special problems with both sourcing and keeping a neutral point of view.

As the year gets older, trends may emerge that are less prone to being labelled 'original research'... BushelCandle (talk) 09:13, 25 January 2016 (UTC)[reply]

The graph may be fine. On the surface, they don't seem to have an inherent bias. Shhhhwwww!! (talk) 11:23, 25 January 2016 (UTC)[reply]

So, to summarize the above: the lines -- the trendlines -- were added to show a trend, to tell a story, to provide a narrative, to present a certain point of view of what is happening, what have you. This point of view's source is not a third-party source but an editor, in this case User:Howard the Duck. Trendlines are of course a tool of statistics, which is a technical branch of mathematics that is widely misunderstood and abused, therefore leading for the famous quote supposedly from Disraeli "There are three types of lies in this world: lies, damned lies and statistics." The use of trendlines is particularly tricky and debatable, and often used and abused in politics (for one notable debate about trendlines see article on the hockey-stick graph). How do we know that this editor's work is accurate or his use of polynomial regression is appropriate? It seems to be accurate to me and conforms to my reading of the polls, but I am not a reliable third party source. Here, the editor himself says he will be changing the method for his next calculation of the line. The Wikipedia community is usually especially careful when it comes to including statistics. I still do not see how the original statistical work by this editor is not in fact original research. If the same practices are followed elsewhere, then maybe they should be eliminated elsewhere. Let's start here and move onto other articles. I think that because of the wide abuse of trendlines in politics to attempt to show a trend, momentum, that is not in fact there, we should avoid adding trendlines to graphs and maybe this means avoiding graphs all together. Just let the numbers speak for themselves. --Iloilo Wanderer (talk) 06:48, 26 January 2016 (UTC)[reply]

If one looks at Real Clear Politics (see Democratic Presidential Nomination contest for an example) the avoid the problem of graphing different polls by creating a RCP Poll Average and graphing that. Then they provide a long list of poll result numbers (with links). They of course can come statistical work (RCP Poll Average) because they are not Wikipedia. They can be and are a source of original research. We can model our page on a RCP page without their graph or their poll average (which is a debatable but useful statistical tool). Note too that RCP does not have the nice smooth lines that seem to show the nice smooth trends that Wikipedia's graphs do. --Iloilo Wanderer (talk) 06:57, 26 January 2016 (UTC)[reply]

The appropriate response is WP:SKYISBLUE -- you don't need a source to say that 1+1=2; in this case, you'd just plot all of the data included in this article at Excel, make an XY graph, add a trendline, and choose "polynomial regression" as what was used in Taiwanese general election, 2016. You can't make up mathematical results, as what is being implied in saying that it is "made up". FWIW, the hockey stick graph must've just one source (there's only one "line", so it's not a "trendline" in the sense that it is being used here.

I'd agree looking at it now this format is not exactly suitable here at it is limited to the number of "ebbs" and "crests" which yours truly has total control of (full disclosure: the order for the graphs is "2"), that's why I'm switching to a more flexible moving average as what was used in Opinion polling for the next United Kingdom general election, Opinion polling in the Canadian federal election, 2015 and several others. –H T D 09:20, 26 January 2016 (UTC)[reply]

Ten green bottles[edit]

Is this edit correct in implying that, as of 17 February 2016^[update], there are only 5 remaining candidates accredited by COMELEC, please? BushelCandle (talk) 10:59, 17 February 2016 (UTC)[reply]

(When I accessed the official list at http://www.comelec.gov.ph/?r=Archives/RegularElections/2016NLE/Candidates/NationalCandidates just now, it was still showing 8 candidates, not 5:

BINAY, JOJO (UNA)
DEFENSOR SANTIAGO, MIRIAM (PRP)
DUTERTE, RODY (PDPLBN)
MENDOZA, MEL (PMP)
POE, GRACE (IND)
ROXAS, MAR DAANG MATUWID (LP)
SEÑERES, ROY (WPPPMM)
VALENCIA, DANTE (IND)

so what am I missing?) BushelCandle (talk) 11:42, 17 February 2016 (UTC)[reply]

One of them got disqualified, one died, one withdrew. Weird. But that's just normal in Philippine elections. The Revision history explains everything. Shhhhwwww!! (talk) 12:41, 17 February 2016 (UTC)[reply]

Thanks for your very prompt and helpful reply and I apologise for my ignorance, Shhhhwwww!!. My only excuse is that I am currently flying routes where I get very little news from the Philippines. If there are indeed only five bottles left "hanging on the wall", I now understand your point about "a leading candidate" being WP:PEACOCK words. It was entirely valid and I apologise for thinking otherwise. (It might be an idea to update our relevant articles by giving details of the three missing candidates' demise, rather than just deleting their names, though.) Thanks again! BushelCandle (talk) 13:21, 17 February 2016 (UTC)[reply]

Color key missing from latest graph[edit]

Viewing the latest polling graphs, it seems that the latest update of the image lost the key which explains which trend line belongs to which candidate... BushelCandle (talk) 08:52, 18 February 2016 (UTC)[reply]

Question regarding SWS and Bilang Pilipino[edit]

Is SWS-Bilang Pilipino unreliable or something of the source? The most recent survey by SWS listed in this page was the March 30 to April 2 survey released on Business World but a large portion of surveys conducted by SWS were released on Bilang Pilipino, is there any reason why these are omitted? 124.104.231.9 (talk) 10:51, 20 April 2016 (UTC)[reply]