Talk:Historical rankings of presidents of the United States

Update

This sentence "The bottom 10 often include James Buchanan, Warren G. Harding, Andrew Johnson, Franklin Pierce, Millard Fillmore, William Henry Harrison, John Tyler, Ulysses S. Grant, Zachary Taylor, and George W. Bush" is incorrect. Many, including polls on this page, say that Trump is the worst or one of the worst Presidents. Trump should replace Grant or Bush in this sentence. Pennsylvania2 (talk) 19:20, 14 January 2021 (UTC)

I'd say, wait until he is a past president. And there needs to be a discussion about which one would be replaced in the bottom 10 list. ~Anachronist (talk) 21:59, 14 January 2021 (UTC)

After rearranging the paragraphs in chronological order, it appears that Bush has been bumped up into the third quartile to make way for Trump in the bottom quartile. It would still be wrong to say the bottom 10 "often include" Trump, though, because only two surveys have included Trump so far. ~Anachronist (talk) 02:44, 16 January 2021 (UTC)

i don't think we need to include Trump yet because it's too early. wait till 2025 because results of presidents' actions only begin to show after 4-5 years out of office.84.54.77.22 (talk) 14:50, 12 February 2021 (UTC)

New at editing, but I believe the entire paragraph about the C-Span ranking should be updated to include their 2021 poll. 15:09, 30 June 2021 (UTC) — Preceding unsigned comment added by 156.111.111.78 (talk)

Ideological balance of experts surveyed

Ideological balance is mentioned, briefly, in only two places in the article: Two surveys conducted by the WSJ mentioned in "Notable scholar surveys", as well as a subsection that was deleted and I just restored about rankings based presidents' leadership about diversity and inclusion.

However, there are far more surveys represented in this article than just those, and we have no way of knowing the ideological balance.

As I have previously stated (archived now), if one sorts any of the tables by political party, a rather glaring pattern jumps out: The Democrat group is predominantly blue/green and the Republican group is predominantly red/orange.

Possible explanations:

1. Republican presidents are in fact generally worse than Democrat presidents.
2. The table is constructed from data based on survey responses from biased experts.
3. The surveys have a systemic sampling bias favoring one ideology.

Those aren't mutually exclusive explanations. Democrats may make better presidents and the surveys may have built-in bias.

It is an undeniable fact that Democrat presidents are generally viewed more favorably by historians, but what information do we have about the experts who were surveyed? Well, some work has been done to answer that. This source mentioned in a section above describes research results that "call into question such ratings insofar as they exist absent the political and ideological context of the reviewer". The article characterizes the differences in ranking are colored by the ideology of the experts who ranked the presidents, and says the population of those experts leans significantly toward Democrat.

Given those facts, a short sentence in the "Notable scholar surveys" about the possibility of bias in the survey results is needed. ~Anachronist (talk) 19:36, 31 May 2021 (UTC)

"Republican presidents are in fact generally worse than Democrat presidents." That does not really translate well in the rankings. Counting the Democrats and where they are typically ranked:

• Andrew Jackson. Ranked Blue. He is typically considered one of the most influential presidents, the founder of Jacksonian democracy.
• Martin Van Buren. Ranked Orange. His presidency is primarily associated with the Panic of 1837, and the costly Second Seminole War. His term is remembered as a disaster, and he was nicknamed "Martin Van Ruin".
• James K. Polk. Ranked Green. A highly successful expansionist, having annexed Texas, California, and vast Mexican areas. Typically ranked highly, but tensions over slavery in the annexed areas are considered one of the main causes of the American Civil War. He gets part of the blame.
• Franklin Pierce. Ranked Red. His support for the Kansas–Nebraska Act is considered one of the causes of the American Civil War. He was deeply unpopular, and is typically ranked among the worst presidents.
• James Buchanan. He tried to turn Kansas into a slave state, helped split the Democratic Party into two rival factions, and the American Civil War effectively opened in his term. He is actively condemned by most historians, and ranks among the worst presidents.
• Grover Cleveland. Ranked Green. Typically considered a successful leader, and is praised for his personal honesty. His failure to effectively face the Panic of 1893 is the main criticism against him.
• Woodrow Wilson. Ranked Blue. His principle of "ethnic self-determination" had global effects in the 20th century, and his international police is highly regarded. His support for racial segregation is the main criticism against him.
• Franklin D. Roosevelt. Ranked Blue. Typically regarded among the greatest president. His New Deal redefined liberalism for several decades, he is credited for ending the Great Depression, and he was an effective war leader in World War II.
• Harry S. Truman. Ranked Blue. Deeply unpopular during his term, reassessed as a folk hero a few decades later. Described as exceptionally hard-working, his "integrity and accountability" was contrasted favorably with dishonest presidents such as Richard Nixon.
• John F. Kennedy. Ranked Green. Well-regarded for his charisma, for his support for the civil rights movement and the Apollo space program, and his New Frontier policies. Blamed for the Bay of Pigs Invasion, the increased American involvement in the Vietnam War, and his scandalous personal life.
• Lyndon B. Johnson. Ranked Green. Praised for his civil rights policies, his creation of Medicare and Medicaid, and his war on poverty. Widely blamed for his perceived mishandling of the Vietnam War, and the high casualties of the war.
• Jimmy Carter. Ranked Orange. His term is largely associated with stagflation, with the Iran hostage crisis, with the 1979 oil crisis, with the Three Mile Island accident, and with a deterioration of relations with the Soviet Union. His presidency is considered a failure, with the Camp David Accords as his only long-term success.
• Bill Clinton. Ranked Green. Highly regarded for his effective economic policies, and a series of achievements in international policy. Blamed for his scandalous personal life, and perceived to be untrustworthy by a majority of Americans.
• Barack Obama. Ranked Green. Highly regarded for his effective economic policies, his significant changes to financial regulation, and attempt to overhaul health care in the United States. Whether his military policies were effective is still considered questionable.
• Joe Biden. Not ranked yet.

Dimadick (talk) 18:02, 1 June 2021 (UTC)

Yes, you can find bad Democrats and good Republicans. What I suggested above is to group the list by party affiliation, then stand back from your screen and look at the dominant colors. The Democrat group is predominantly blue and green, while the Republican group is predominantly red and orange. My original point was to propose reasons why this is the case. ~Anachronist (talk) 05:10, 5 June 2021 (UTC)

Even with Trump, 3/4 of the top-ranked presidents are Republican, and 3/4 of the bottom-ranked are Democratic, and people are going to pay more attention to that than to party averages. I suspect part of the issue is that the Republican party used to stand for the "all men are created equal" American ideal, which historians tend to value, while the Democratic party was opposed to it, and now the situation is (broadly) reversed. — kwami (talk) 07:30, 3 July 2021 (UTC)

Dimadick, you are probably correct. Research (https://www.insidehighered.com/news/2016/10/03/voter-registration-data-show-democrats-outnumber-republicans-among-social-scientists) shows there to be 33.5 Democrat historians for every Republican. And like all of us, historians interpret the world from their own socio-political perspective. But this is a talk page: a forum for discussing how the article could be improved. And I don't see any way of removing the glaring bias in the statistics. Theeurocrat (talk) 19:16, 3 July 2021 (UTC)

Accessibility Issues in Article

Hey y'all,

There are several tables in the article that solely rely on the use of color to convey information. The sole use of color to convey information could create a barrier for people with disabilities to perceive the content in the article. Please review Wikipedia:Manual of Style/Accessibility and WCAG 2.0 Success Criterion 1.4.3 for additional information.

This can be resolved by additionally adding in a text-based method of identifying the information communicated in color; the colored backgrounds do not need to necessarily be removed altogether or changed (unless they violate color contrast requirements; normal sized text should have a color contrast ratio of 4.5:1 between the text color and its background content, and can be checked using the WebAIM Contrast Checker), but there does need to be text-based alternatives to the information being communicated through color alone.

Thank you!

SunnySydeRamsay (talk) 18:04, 10 June 2021 (UTC)

I strongly oppose this change. The color adds useful visual information and if colorblind people cannot distinguish them, they can still look at numbers. The text was already there; I see no valid reason to remove colors. Speculating on how the colors "could" create a barrier isn't a valid reason to remove them. Also, colorblind people can distinguish shading. The accessibility guideline also does not require that we remove them. The colors should be added back with different shadings and contrast between the color and the text.

While I applaud the WP:BOLD edit, I am reverting this change to the long-standing consensus version until an actual consensus can be achieved. At the moment, there is none. ~Anachronist (talk) 01:13, 20 June 2021 (UTC)

Testing the color version of the page with the colorblind filter at https://www.toptal.com/designers/colorfilter it appears legible for all forms of colorblindness. Until it can be demonstrated that the colors actually cause a problem, the colors should stay. If the colors are a problem, then Wikipedia:Manual of Style/Accessibility provides ample advice on how to improve the colors rather than remove them. ~Anachronist (talk) 01:23, 20 June 2021 (UTC)

It's not speculation. Please review Wikipedia:Manual of Style/Accessibility as you linked. The first bulletpoint under color clearly states to not communicate information with the sole use of color. It clearly defines how conveying information using color alone can be an accessibility barrier to people with disabilites. WCAG 2.0 Success Criterion 1.4.1 Use of Color also requires that color not be the sole method of communicating information. These are international standards that are in place specifically to maximize the accessibility for all people with disabilities. ~ SunnySydeRamsay (talk) 21:53, 20 June 2021 (UTC)

Information isn't being communicated solely through color. The numbers are all there, as you well know. The color enhances and clarifies the information already present.

Having worked for an online publisher of scholarly journals where this was an issue, I am quite familiar the various accessibility guidelines including MOS:COLOR (back in 2017 there were several conflicting guidelines also). No guideline advocates removing all colors; rather, they recommend using colors with appropriate contrasts and shades. Colors can coexist nicely with information accessible to disabled people.

The W3 guidelines do not prohibit the use of color to enhance the communication of information. There are guidelines about proper color usage for readers with disabilities, and those guidelines do not mean that the colors should be removed completely. Instead, colors and contrasts between text foreground and background should be improved.

Yes, I still maintain that the removal of colors was based on speculation that the colors present a problem. They don't, as demonstrated by the test link I included above. No evidence has been presented so far showing where the page fails, but I'll do one better and include such evidence now: Color Contrast Accessibility Validator. Running this article through it, out of the 23 pairs of colors currently in use, only three were found to have insufficient contrast for accessibility — and one of them is from a system template, not this article. If we fix those two (basically using the acceptable orange/black as used elsewhere in the article, instead of orange/white), then there is no more accessibility problem, and the article complies with guidance. This is a trivial fix. ~Anachronist (talk) 15:32, 21 June 2021 (UTC)

Update: I have changed the white-on-orange combination to black-on-orange throughout the article (it was correct in the first table, but not the other tables) and now the only color pair that fails the guidelines is the light-blue-on-blue pair from the system template, which isn't a problem for this article's content. ~Anachronist (talk) 16:10, 21 June 2021 (UTC)

If the colors do not present new information, couldn't all users simply look at the numbers and make a quartile determination by themselves? By color segregating quartiles, you've performed a mathematical operation that communicates additional information to a user. If it didn't provide additional information, then there wouldn't be a purpose of adding the color, no? Again, my advocacy in totality isn't to remove the color, my advocacy is to include a secondary format to indicate this information. As it is, this is, by definition, a violation of WCAG 2 SC 1.4.1. ~ SunnySydeRamsay (talk) 17:36, 21 June 2021 (UTC)

As a reference to other contributors, WCAG 2.0 Success Criterion 1.4.1 is as follows: "Color is not used as the only visual means of conveying information, indicating an action, prompting a response, or distinguishing a visual element. (Level A)" ~ SunnySydeRamsay (talk) 18:11, 21 June 2021 (UTC)

Oh, and as a secondary point, my issue has never been color contrast; I checked color contrast and everything appeared to be 4.5:1, though it appears you've noticed something I missed in regard to color contrast, so kudos. ~ SunnySydeRamsay (talk) 20:34, 21 June 2021 (UTC)

The colors don't present new information, they serve as a visual aid, not only to those who can see the colors, but also to those who cannot (as they see different shadings). This becomes meaningful when sorting the table certain ways, as described in a section above, where one gets a visually striking grouping that can be seen from studying the numbers but is much clearer with colors. WCAG 2.1 SC 1.4.1 is violated only if color is the only means of conveying information. But it isn't. Therefore there is no violation. Honestly I cannot understand your objection; the coloring in this document are supported by every standard mentioned here so far.

This article already separated the numerical ranges into quartiles before I found it. One could also make a continuous gradient of colors, but quartiles is a fairly standard statistical representation that approximates a gradient. All the colors do is indicate an approximate quantity represented by the number. There is no information being solely conveyed by color, just relative differences. I'd much prefer replacing the quartile colors with a gradient. I'm trying to figure out how that could be done in a reasonably maintainable way, however, and I am at a loss. The quartile colors strike me as a compromise between desired presentation and complexity.

As to the secondary point, if accessibility matters (and it does) contrast is a far more important feature than hues, in my experience. In the previous job I mentioned, adhering to international standards of contrast mattered most. As the guy managing multiple projects for multiple online journal publishers, I was typically caught in the middle between our QA team's evaluation of page accessibility and the customer's branding requirements. Fortunately all my customers were reasonable and understanding, agreeable to departures from branding to accommodate accessibility. ~Anachronist (talk) 01:19, 22 June 2021 (UTC)

The new information is a summation of mathematical operations. It is not equitable or accessible to give color-sighted users quick access to such information while setting the expectation that people who are colorblind need to perform the same mathematical operation hundreds of times to obtain the same information. I think a solution that retains the color and adds an additional text-based indicator is a simple way to create an accessible and equitable experience. I'm certainly not opposed to having color in the table; it increases how visually engaging the table is which is also a key component of accessibility esp. for people with cognitive disabilities. Just the main goal is to ensure an accessible experience for all.

To clarify, when I said "my issue has never been color contrast" I meant I didn't notice any color contrast issues in the document. It is important to note, however, both color contrast and use of color criteria are both Level A requirements under WCAG. If a website is not compliant with all WCAG 2.0 A criteria, they are considered exceedingly difficult for people with disabilities to use. To be legally compliant in several countries and governmental entities such as the United States, Canada, Australia, and the E.U., websites must be WCAG 2.0 AA compliant. WP:ACCESS lists WCAG compliance as an objective of the Manual of Style. By the standards of WCAG, 1.4.1 Use of Color and 1.4.3 Contrast (Minimum) are equally important in that they are both Level A requirements.

Gradients would definitely be cool, just hard to implement without a programmatic solution probably. ~ SunnySydeRamsay (talk) 19:16, 23 June 2021 (UTC)

I just learned that gradients are implemented in {{Weather box}}. So maybe that can be adapted to a general purpose table of numbers. I'll look into it. I've been meaning to familiarize myself with Lua anyway. ~Anachronist (talk) 19:34, 23 June 2021 (UTC)

If you still want to do a gradient, though I'm not too experienced with Lua, I tried making a gradient for one of the ranking charts on this page on User:Jikat2/sandbox. Would this work for the page? Jikat2 (talk) 19:32, 1 July 2021 (UTC)

@Jikat2: I hadn't noticed that this conversation continued after my last comment a month ago. Your gradient is very cool. You managed to figure out what I could not. The colors aren't what I would use (I prefer a red-yellow-green-blue gradient) but I think that would be a great improvement to the tables here. Based on the discussion further down, though, this seems to have taken a different direction though and it's back to quartiles. ~Anachronist (talk) 21:32, 23 July 2021 (UTC)

So, I just added C-SPAN's newly released 2021 survey and tried my hand at making the colors easier per WP:BOLD. My initial color scheme went from blue to orange, but Jikat2 felt that wasn't intuitive enough for the reader, which was a good point. Just now I've added a classic green-to-red scale that I hope is much easier for readers to quickly interpret, as follows here:

•  Green  backgrounds indicate first quartile.
•  Yellow  backgrounds indicate second quartile.
•  Orange  backgrounds indicate third quartile.
•  Red  backgrounds indicate fourth quartile.

Any thoughts on this scheme? As for Jikat2's gradient proposal, I like it, though red and blue feels risky. What would you think about a green-to-red gradient? --FlagFreak ^talk 21:36, 1 July 2021 (UTC)

Thanks for your input! As for your color scheme, I like the idea of a green-red scale, and think the green, orange, and red look good. However, the second quartile is supposed to be above-average, but in your scale, it is represented by yellow, which does not often have positive connotations. Potentially a greenish-yellow or light green could be used for the second quartile, with a different, darker green for the first quartile. I've also taken your advice about the red-blue gradient into account, and replaced it with a red-yellow-green gradient on my sandbox, though I am slightly concerned about how accessible it is for the colorblind.

However, although I do think the new color scheme works, I am not entirely sure what the need is, and I feel like the new color scheme could throw off people who visit the page often. Although I'm not completely opposed to the new colors and think they are a very good alternative that aren't worse than the old colors, and as such am not going to revert the change, I'd like to hear if someone other than us might prefer the original colors. Jikat2 (talk) 01:02, 2 July 2021 (UTC)

I strongly dislike the updated colors. Green and yellow are difficult to distinguish. Reducing saturation almost guarantees less differentiation between colors. If the issue is that colors shouldn’t be used for accessibility reasons, simply changing colors does nothing to address this. And using less readable colors for those without color-related disabilities does not make it somehow better.

Comment DO NOT Post a header in the main article to redo all the colors when there is no consensus that the page is inaccessible to colorblind people. In fact, there seems to be a strong consensus that the colors are accessible as is. Sleyece (talk) 23:45, 30 June 2021 (UTC)

Comment It is significantly harder to understand the information conveyed with the new gradient colors, unfortunately. Muttnick (talk) 13:49, 2 July 2021 (UTC)

Looks like we're keeping the dark color scheme (blue, green, orange, red) for now, since there's no agreement on a lighter color scheme one yet - which I do believe would be easier on the eyes for the most readers. In the meantime, it's a good idea to keep the overcolored template at the top of the article until there is consensus. To see the effect for yourself, I'd suggest taking screenshots of various color schemes and importing them into the Coblis color blindness tool. The current dark scheme works well for anomalous trichromacy, but is very poor for dichromacy; these users would not be able to clearly distinguish between dark red and dark green. We need to leave the tag up until this is fixed. I'd like to propose an even simpler green-yellow color scheme (1st quartile, 2nd quartile, 3rd quartile, 4th quartile), so we can simply fall back on brightness to quickly communicate a president's position. --FlagFreak ^talk 19:32, 2 July 2021 (UTC)

I think that may work! Muttnick (talk) 21:12, 2 July 2021 (UTC)

1st and 2nd quartile look pretty much the same to me. Three's a line break separating them on my screen, and they'd presumably be easier to tell apart if they were adjacent for me, but they'd be separated in the tables too. — kwami (talk) 07:36, 3 July 2021 (UTC)

Sorry if I stepped on anyone's toes. I saw the tag at the top of the article and went ahead and fixed it per ACCESSIBILITY before I saw this discussion. (I'm not color blind, but I still had trouble reading the tables.) I used the closest colors from the 9-color scheme at Paul Tol's Notes, as suggested by ACCESSIBILITY, but the red/pink was nearly identical to the color for the Republican party, which was confusing, so I shifted orange & red to yellow & orange/brown.

 Blue   Green   Yellow   Brown 

They're not pretty colors, but they do strike me as easier on the eye than either the original dark colors or the rainbow pastels above, both of which make my eyes swim a bit.

But if I stepped on your toes, my apologies. All 4 colors are now in #aabbcc; format, which should make it easy for you to find-replace them with whichever scheme people agree on.

I also moved all the scholarly surveys up together under the main heading for scholarly surveys, and added the survey results of African American scholars that were discussed in the 'alternative' section but not shown. Unfortunately, I can't access an edition of the book with an evaluation of Obama or Trump, but will keep looking. — kwami (talk) 07:08, 3 July 2021 (UTC)

I think this looks good! Muttnick (talk) 14:16, 3 July 2021 (UTC)

Sorry, this color scheme does very poorly on the Coblis colorblind simulator, in almost every type of colorblindness. I appreciate kwami's effort though, and I think it works far better than the previous darker color scheme. I'd like everyone to take a look at the 2 green-to-yellow gradients proposed by myself and Jikat2. --FlagFreak ^talk 19:28, 4 July 2021 (UTC)

Why would ACCESSIBILITY recommend a color scheme that isn't accessible? You mean that after 20 years of WP, we still haven't figured out how to pick colors for the colorblind?

Again, I don't much care which colors are used, though I would appreciate something that is an intuitive scale. But your proposal distinguishes only 3 of the 4 quartiles. (Yes, I can see that you have 4 colors, but two of them are practically indistinguishable. The two greener/darker ones would need to be spaced out more in color space.) As for Jikat2's proposal, I'm not sure. I find such schemes really annoying on maps, because it's very difficult to match the color with the ranking. But since the ranking in these tables is given, that's not a problem. I think I just have an automatic dislike for such schemes that I need to reconsider. — kwami (talk) 21:49, 4 July 2021 (UTC)

First, I do think your color scheme looks good! Second, I couldn't find an edition of the book with an evaluation of Obama/Trump either, but I did find this paper, which on page 12 claims that the authors place Obama in the "Racially Neutral" category of their rankings in the 2012 edition of the book, though the actual book has a typo that puts Obama in the "Racist" category. I'm not entirely sure about Wikipedia's source policy, but a scholarly source referring to the book is good enough to place Obama in the ranking, right? Jikat2 (talk) 16:19, 3 July 2021 (UTC)

Good find! Yeah, that should be good enough. 2ary sources are preferred anyway. Esp. if there's a typo in the original that would throw us. — kwami (talk) 21:37, 3 July 2021 (UTC)

@Jikat2: According to Wallace, Obama was moved to 'Antiracist' "after a universal review of his record." Trump was added to 'White Supremacist'. We really should confirm the entire list with the 9th ed., in case others were reevaluated (though the old Obama evaluation had been before he completed his presidency). My lending circuit only has up to the 2017 edition. — kwami (talk) 18:08, 6 July 2021 (UTC)

Blue-green-yellow-brown is obviously not a series. (Go ahead; hand them to people and see what various series they make.) If red and blue are out for partisan reasons, and if white cells/areas would mess up grayscale, that leaves only green-yellow. Make the yellow light, so that darker=better, and make the text also signify quartiles by size or boldness, also bolstering the dark=good thing. 96.42.252.219 (talk) 17:35, 3 July 2021 (UTC)

Ok, so I did make a gradient from dark green to light yellow on User:Jikat2/sandbox. However, at this point, I feel that we're liable to just go on like this forever, with people proposing more color schemes, and more people disagreeing with these new color schemes, without ever reaching consensus. Perhaps we could hold a vote on the different schemes? Jikat2 (talk) 18:49, 3 July 2021 (UTC)

I really like this color scheme, possibly the best proposed so far. It avoids the red-blue confusion with the party labels. My only request would be to lighten the green enough so that all the numbers can be readable in black, so editors don't have to manually change cells to white text by hand. --FlagFreak ^talk 19:24, 4 July 2021 (UTC)

Disagree. You do not need for it to be a series (it never was to begin with). The colors clearly differentiate between categories. Muttnick (talk) 19:34, 3 July 2021 (UTC)

It's the ROYGBV series, reduced to the 4 colors in the middle (OYGB). According to Paul Tol's Notes that's an orange, not brown, but a dark orange is brown. So we could change 'brown' to 'orange' if that would make it more intuitive for people. At least, it seemed an intuitive order to me. — kwami (talk) 21:34, 3 July 2021 (UTC)

Support The color change is good; Same basic scheme, but with softer tones. -- Sleyece (talk) 05:32, 5 July 2021 (UTC)

I was partly wrong; you could also just drop the yellows and do light green to mid-green. Even better. The four quartiles obviously are a series (that's why they're labeled first second third fourth quartiles; you could arbitrarily do quintiles or deciles), but your colors are not serial in any visible way. Everyone knows about ROYGBIV, and ROYGBIV is not relevant. You really think people see RGB (for example) as a series? I don't know how you folks operate, but I suggest you vote. Good luck. — Preceding unsigned comment added by 68.117.48.13 (talk) 15:16, 7 July 2021 (UTC)

I liked this version of the colors best. I'd like to see a real gradient, however, with red-yellow-green-blue, or the current muted colors. ~Anachronist (talk) 21:32, 23 July 2021 (UTC)

(For clarity, the 'current muted colors' are orange-yellow-green-blue.) — kwami (talk) 23:45, 23 July 2021 (UTC)

WEIGHT problem with Alvin Felzenberg?

We have a section dedicated to Alvin Felzenberg (The Leaders We Deserved), the only historian we give individual attention to. All the other section are polls of historians or political scientists. Isn't that a WEIGHT problem, promoting a single individual out of all the many historians who are lumped together in the polls? Should we delete the section? It's under 'criticism', but it's not a criticism of presidential ranking so much as a disagreement with the results -- something that we could repeat for dozens of historians. — kwami (talk) 09:45, 5 July 2021 (UTC)

I think it's fair to include him, given there are multiple sources mentioning him that aren't his own book, making his opinion notable, and we do include opinions from single historians like Ron Walters or Julian E Zelizer. I do agree that he didn't deserve his own section, though. Jikat2 (talk) 16:27, 5 July 2021 (UTC)

It's not the section heading I think is UNDUE, but the three paragraphs we spend on his book. I'd be fine with a mention and moving the bulk of it to his article. But just removing the header doesn't change the weight we allocate to him, it only makes it covert. If half the paras in the criticism section are on a single topic, then I think we should set them off as a subsection. I just think it's UNDUE to spend half the criticism section on the opinions of a single scholar.

For the other scholars we mention, we aren't promoting their work. Donald reports something Kennedy said. For Zelizer, we just quote him as saying that rankings are a weak evaluation. Walters notes race as a neglected criterion for ranking, and compares two other multi-scholar polls -- we don't give his personal ranking.

The only scholar we single out for his personal ranking is Felzenberg. We say that he 'finds fault with conventional wisdom in certain areas and agrees with it in others' -- but that's true for every scholar. For example, one scholar I just came across was reported (in a 2ary source) as saying that he'd voted for Trump to be 43 in CSPAN-2021, but that others ranked him higher, and he laid out why he disagreed with them. We could mention dozens of cases like that, quibbling over every number.

I think one para for Felzenberg (without a subheading) would be justified, but IMO, because it's in the 'criticism' section, it should describe his criticism of presidential rankings in general, not his quibbling over individual numbers. If we want to report his personal ranking, then IMO it should go in the summary Table 1 with all the rest. — kwami (talk) 23:43, 5 July 2021 (UTC)

Ok, I attempted to rewrite his section into one paragraph, specifically talking about who he was and what his specific criticism of the ranking process was. However, I still don't believe he deserves his own subheading, and so I have removed it. Jikat2 (talk) 00:05, 6 July 2021 (UTC)

I've read Alvin Felzenberg's 2008 book (The Leaders We Deserved) and am not impressed. It's lightweight popular history that is not aware of scholarship in recent decades. His main argument is to use multiple indicators, which in fact is used these days. He exaggerates the claim that rankings depend a lot on the ideology of the rankers. (I think that applies to Reagan but rarely to other presidents.) Michael Genovese in his published review says: " Felzenberg is upset—with some justification—at the liberal bias he sees as so prevalent in the ranking of U.S. presidents by historians and political scientists. To remedy this, he has provided a counter to the liberal bias with a conservative bias. In doing so, he commits all the sins of which he accuses liberals. This book is a mirror image of the work he finds so troubling....It is unscientific, impressionistic, and highly subjective" Bottom line in my opinion he is a minor player in the ratings game. Rjensen (talk) 00:44, 6 July 2021 (UTC)

So do you support removing the section, cutting it down even further, or leaving it as it is? I personally do agree that some of his criticisms seem redundant and that his rankings do show much of his own biases, but that doesn't really change the fact that it is a notable criticism, as his criticism is mentioned in multiple secondary sources, and so I feel that some discussion should be included in the criticism section. Jikat2 (talk) 00:59, 6 July 2021 (UTC)

I'm fine with no subheading when we have 3 paras for 3 critiques. It was just when half the 'criticism' section was devoted to Felzenberg's book that I thought we needed a subheading for it.

Should we include anything from Genovese's review? — kwami (talk) 02:44, 6 July 2021 (UTC)

Yes include Genovese quote--I looked and Genovese is the ONLY scholarly review of the book I could find. That is, editors treated it as not very notable. cite = Genovese, Michael A. "The Leaders We Deserved (And a Few We Didn't): Rethinking the Presidential Rating Game." Presidential Studies Quarterly 40.4 (2010): 799-800. Rjensen (talk) 05:01, 6 July 2021 (UTC)

Added. — kwami (talk) 06:01, 6 July 2021 (UTC)

2007 Rasmussen poll

I'm wondering who was polled for this. I doubt half of Americans even know we had a president Polk, much less know enough about him to have an opinion. I'm not sure half of Americans could distinguish the two Roosevelts, but two thirds have an opinion on Cleveland? When we quote Rasmussen as saying certain presidents were "rated favorably by at least 80% of Americans," we imply that this was a representative poll of the entire American population. I find that hard to believe. According to the memorability section, only 11% of respondents could name Cleveland as a president, and those were college students. Do we need to identify who the Rasmussen respondents were? — kwami (talk) 05:59, 6 July 2021 (UTC)

I agree -- Rasmussen is an outlier and gives too little info to be useful for this purpose. Probably drop it altogether. Rjensen (talk

Yeah, unless we can clarify what's going on, that would probably be best. Anyone else? — kwami (talk) 08:30, 6 July 2021 (UTC)

I looked at Rasmussen's website. they give out zero sampling info on that old 2007 poll (unless you join or pay them) --that is unacceptable for our rules re reliable source. Rjensen (talk) 08:50, 6 July 2021 (UTC)

Removed, if someone can justify keeping it, pls revert. — kwami (talk) 08:56, 6 July 2021 (UTC)

British polls

To give a bit more balance, or at least perspective, I've added a couple British polls. There's some Irish input in the 2nd. I don't know of any other country that has done polls like this. It would be interesting to see what scholars in Canada, Mexico, other NATO countries and the Pacific allies of the US think. — kwami (talk) 18:59, 23 July 2021 (UTC)

median ranking

In the summary table, some of the 4th quartile ranks had been colored as 3rd quartile when the median of an odd number of presidents was colored as 2nd quartile. (The biggest problem was in the Sienna poll, because they didn't adjust the ranking numbers for ties, so quite a few 3rd and 4th quartile rankings had gotten bumped.) I fixed those, but thought maybe we should have an intermediate green-yellow for the median rank of an odd number. If ppl don't like this, just revert the chartreuse #99CC00; back to green #44bb99;. (For consistency, we might also use intermediate colors for borderline 1-2 and 3-4 quartiles, but those aren't as consequential: they'd only appear in our two polls that rank 42 presidents, and we only cover those in summary.) — kwami (talk) 01:10, 30 July 2021 (UTC)

Sorting by most-frequent quartile gives screwy, and very misleading, results

I get the idea of adding this column to the summary table, but it's problematic, even disregarding the fact that history has changed its mind about e.g. Grant and Jackson since polling began. For one thing, you get weird results if you sort by that column. I did that once, and it put Grant at the very bottom of the list, with W at 2nd from bottom; tried again, and Trump was at the bottom, with W at 2nd. But the latter order was just because they're the most recent presidents to score in the 4th quartile, and the Grant-last order reflected the column I'd sorted by just before. So if we keep the table as it is, perhaps we would disable the sort function for the 'most-frequent quartile' column?

Another option would be to give each president an average score. We could still color them into quartiles of 11 presidents, but then when the reader sorts by that column they would get meaningful results. (Or at least as meaningful as the polls themselves.) I would exclude any polls conducted from before a president left office, as they did not evaluate the entire presidency. Those polls were news when they came out, but IMO have been superseded. (Indeed, the USPC poll gave Obama a provisional ranking, but did not let that affect the rankings of the other presidents, as they thought it premature to say where he'd end up.)

Straightforward enough to calculate in a spreadsheet, if something of a pain. — kwami (talk) 03:45, 6 August 2021 (UTC)

Okay, here are the results with provisional rankings (italic figures in our table) removed, and the others resorted to compensate. There are only a few cases where tenths affect the order, and only Pierce vs Johnson where hundredths matter, so I'll round off to the nearest tenth in the summary table in the article.

Overall ranking of US presidents, disregarding polls conducted while still in office

No.	President	Average ranking*	Quartile
1	George Washington	2.87	1
2	John Adams	14.70	2
3	Thomas Jefferson	5.46	1
4	James Madison	14.88	2
5	James Monroe	15.46	2
6	John Quincy Adams	20.16	2
7	Andrew Jackson	12.41	1
8	Martin Van Buren	26.47	3
9	William Henry Harrison	37.66	4
10	John Tyler	37.22	4
11	James K. Polk	14.41	2
12	Zachary Taylor	33.78	4
13	Millard Fillmore	37.88	4
14	Franklin Pierce	40.74	4
15	James Buchanan	42.83	4
16	Abraham Lincoln	1.78	1
17	Andrew Johnson	40.75	4
18	Ulysses S. Grant	33.16	4
19	Rutherford B. Hayes	28.11	3
20	James A. Garfield	30.24	3
21	Chester A. Arthur	30.62	3
22	Grover Cleveland	19.72	2
23	Benjamin Harrison	31.87	3
25	William McKinley	19.05	2
26	Theodore Roosevelt	5.20	1
27	William Howard Taft	23.35	3
28	Woodrow Wilson	8.53	1
29	Warren G. Harding	42.28	4
30	Calvin Coolidge	31.42	3
31	Herbert Hoover	32.47	3
32	Franklin D. Roosevelt	2.44	1
33	Harry S. Truman	7.87	1
34	Dwight D. Eisenhower	10.53	1
35	John F. Kennedy	13.06	1
36	Lyndon B. Johnson	14.12	2
37	Richard Nixon	31.81	3
38	Gerald Ford	27.99	3
39	Jimmy Carter	27.67	3
40	Ronald Reagan	14.62	2
41	George H. W. Bush	22.63	2
42	Bill Clinton	16.82	2
43	George W. Bush	34.55	4
44	Barack Obama	11.07	1
45	Donald Trump	41	4

*Average of (numerical position divided by number surveyed), × 44

Note that if you sort by quartile, the presidents are not ordered according to how well they're rated.

I've placed the average scores into the last column in the article. Revert or modify of course if this is not acceptable, but the previous version really was screwy.

Is there a good way to post the whole spreadsheet, with the embedded formulas to produce the averages, so that you can check my work and so that when new polls come out, someone else can add them to the table without having to recreate the spreadsheet? — kwami (talk) 05:02, 6 August 2021 (UTC)

We've actually been over this many times. Most Frequent Quartile is the consensus because it is the only aggregate method that does not generate new data. All other forms, no matter how math heavy are Synthesis, and a violation of Wikipedia Policy. Your well meaning, but most of what you're asking is "how can we best violate Wikipedia Policy w/ the most complex math possible".. That said, your tie breaking system seems very useful. -- Sleyece (talk) 11:35, 6 August 2021 (UTC)

Simple arithmetic (averages) is hardly "the most complex math possible". I tried making that column unsortable, but it breaks the header. We could at least get it to sort correctly, rather than implying that Grant is ranked the worst president. — kwami (talk) 11:56, 6 August 2021 (UTC)

No aggregate, no averages, nothing but counting. I tried to do what you're doing already, and I got an Indefinite block for it. -- Sleyece (talk) 12:10, 6 August 2021 (UTC)

I don't know what that means. Could you give me an example? If evaluations changed over time, so a president was ranked 5x Q4, 4x Q2 and 4x Q1, how should we rank him?

Also, we cannot in good conscience make the last column sortable. Removing sortability is not SYNTH -- I don't know how you get that. Claiming that Grant is ranked dead last, or that W is 2nd last, is not accurate. — kwami (talk) 23:58, 6 August 2021 (UTC)

The real problem here is that the only kind of quartile sort that is meaningful is a multi-key sort: First sort by quartile, then sort by a secondary key (like ranking). I don't recall HTML tables ever having the ability to do that. This isn't a spreadsheet after all.

This problem tends to come up whenever one has a data set with a large number of identical values (as we do with quartiles). It is not a problem when every item to be sorted has a different value. In a single-key sort on quartiles, the ordering within any quartile value is going to be an artifact of the order in which the list started, and the sorting logic applied to it. ~Anachronist (talk) 02:40, 7 August 2021 (UTC)

Any sorting method or calculation is fine as long as it avoids aggregates or personal opinions about results.. (i.e. Kwamikagami repeatedly changing the sort for Grant because they believe he was a better President (he was, for the record) than sorted without getting consensus.) -- Sleyece (talk) 02:47, 7 August 2021 (UTC)

Again, I'm not sure I understand. AFAICT, "aggregate" in this sense would be what I did in the table above. Is that what you mean? Because if I'm counting the quartiles, and you're counting the quartiles, then I fail to see the difference. (It would be nice if you answered the specific question above, rather than objecting to generalities, so I could be sure what you were talking about.)

As for your example, you appear to be accusing me of pushing a political bias. But it has nothing to do with what I think of Grant. I gave him as an example above, not because I care much about Grant but because he was the most egregious example -- can we agree that Grant is *not* ranked 44th overall among the president? It's not just Grant: I also object to showing W as the 2nd-lowest-ranked president and Jefferson as the 2nd-higest -- or any case where the summary table does not accurately summarize the data it contains.

I ask you again, since you are not responding to the actual questions, please explain how not sorting a column violates SYNTH. If sorting the last column gives poor results, then one obvious solution is to not sort it.

If we can't come to a common-sense way of sorting the general poll results that is neither SYNTH nor misrepresentative, and you insist that they must be sortable despite that producing misleading results, then I think the only solution would be to remove the last column as unencyclopedic. What does it add, exactly? Again, if you could answer my question about how we'd treat someone in the top half but with a mode in Q4, so I could ask the SYNTH people how to address it, that would help. — kwami (talk) 06:32, 7 August 2021 (UTC)

I've tagged the section as POV, as as it stands it gives a seriously distorted summary of what the polls actually are. — kwami (talk) 06:32, 7 August 2021 (UTC)

Tag whatever you like. Get consensus, you aren't God. -- Sleyece (talk) 08:20, 7 August 2021 (UTC)

Tags don't require consensus. Edits do, unless they clearly violate one of our policies. This last column arguably does, as we are ranking presidents ourselves despite our results conflicting with sources, so I might be able to get away with just deleting it. I'd rather have a constructive discussion with you, if you're willing to discuss it. But I've asked you some basic questions, and you refuse to answer. If you're not willing to engage in good faith, then you don't get any say in what happens. — kwami (talk) 05:58, 10 August 2021 (UTC)

Stay off my personal talk page please. I have no response to this nonsense. -- Sleyece (talk) 11:49, 10 August 2021 (UTC)

UK polls

@Rjensen, Jikat2, Anachronist, FlagFreak, Muttnick, SunnySydeRamsay, Theeurocrat, Dimadick, and Pennsylvania2:

Since we're at an unfortunate impasse, where we can't improve the article without getting consensus first, do you agree it's a good idea to broaden the perspective of this article by including non-US polls? A couple weeks ago I announced I'd be adding 2-3 UK polls to help balance the article (I don't know of polls conducted in any other country, or I'd add them too), and I added the first 2 without a problem. I've now signed up to the Times to access the 3rd. But now Sleyece and I are in a dispute over SYNTH, so he's blanket reverting any changes I make, and deleted the Times poll without giving any reason. Is there any reason not to add the 2008 Times poll to the summary chart?

I got the info from the online edition, but since then the folks at WP:Resource Request found a paper version in the Gale archives: MacIntyre, Ben (1 November 2008). "The big question: who is the greatest president of all time?". The Times. London. p. 42. — kwami (talk) 18:57, 10 August 2021 (UTC)

I agree that the UK polls are useful. The article has been horribly US-centric for several years. Wikipedia articles need to provide a global perspective. Dimadick (talk) 19:02, 10 August 2021 (UTC)

I also agree the UK polls are useful. I have no objection to their inclusion. Consensus must come first, though. -- Sleyece (talk) 19:06, 10 August 2021 (UTC)

Did you get consensus for all the polls included in this article? — kwami (talk) 23:55, 10 August 2021 (UTC)

Not useful....no RS compares US and UK polls and readers keen on UK will not come to this article. Rjensen (talk) 20:05, 10 August 2021 (UTC)

Actually, the British polls themselves make that comparison. It's a challenge to find a RS that compares even US polls with each other, and certainly no RS compares the polls we include here. And we're not talking about readers keen on the UK, we're talking about not being insular when presenting info to readers keen on the US. — kwami (talk) 23:55, 10 August 2021 (UTC)

Good point, British Prime Ministers aren't judged by the opinions of Americans. This "Global Perspective" thing seems like an "America is Bad" opinion which is WP:SYNTH. I've changed my mind, not useful. -- Sleyece (talk) 20:16, 10 August 2021 (UTC)

How could that possibly be anti-American? Have you read any of them? None of the British specialists in US politics are anti-American. There's not even much of a difference with them. I could see that objection if it were a state-sponsored Russian poll, but the opinions of US allies are relevant. And if UK PM's aren't judged by Americans, they should be. Insularity is one of the problems of WP. — kwami (talk) 23:51, 10 August 2021 (UTC)

Support adding any international polls given the polls are of same methodological value, i.e. academics with domain specific expertise in history. The purpose is to give as much of an objective ranking to Presidents as possible, so feasibly data from individuals with expertise regardless of their whereabouts in the world should be applicable here since the purpose is sufficiently broad enough.SunnySydeRamsay (talk) 03:18, 11 August 2021 (UTC)