Princeton Election Consortium

Innovations in democracy since 2004

Outcome: Biden 306 EV (D+1.2% from toss-up), Senate 50 D (D+1.0%)
Nov 3 polls: Biden 342 EV (D+5.3%), Senate 50-55 D (D+3.9%), House control D+4.6%
Moneyball states: President AZ NE-2 NV, Senate MT ME AK, Legislatures KS TX NC

Bustin’ out?

September 25th, 2012, 6:00am by Sam Wang

Governor Chris Christie (R-NJ) was partially correct when he said about Romney: “If the election were going to be held tomorrow that would be a problem.” Take a look at today’s Meta-Analysis update:

The EV estimator is nearing the edge of the November +/-2 sigma band (shown in yellow). Something fairly notable is happening.

In one respect, Governor Christie might be wrong: there has been a problem for Romney for months, not just recently. He was behind last week and he’s behind today. However, the kind of shift we are seeing is likely to be accompanied by an accompanying shift at the downticket level. Looking at the trend, a further surge for the Democrats is likely in the coming days. For Congressional Republicans, an election tomorrow would have the advantage of containing the damage at current levels.

At the Presidential level, despite the steepness of that curve, there are barriers to the EV estimator going higher. The gray confidence interval is narrowing, suggesting that there are not that many more electoral votes to be had with small opinion shifts. Indeed, the closest Romney states are Indiana, Missouri, Tennessee, and Arizona – with current leads of 5% or more.

Consistent with the idea of an Obama ceiling, today the great majority of probability, over 60%, is concentrated in a single peak, at 347 EV. It is very unusual for this distribution to be so sharply peaked. No states are in the knife-edge zone.

Yet the Popular Vote Meta-Margin is not at an extreme value. This reflects that fact that in many swing states, President Obama’s lead is only moderately-sized, for the time being.

Tags: 2012 Election · President

71 Comments so far ↓

  • Ralph Reinhold

    You may be right. There is some jiggling toward Obama in a few red states. If Obama’s trend holds, GA, TN, SD, IN, and MT could move to toss up or beyond. Current polls in a couple of them are 5%.

  • RadicalCentrist

    I’m having trouble seeing how a 4-5% margin in the popular vote translates to bringing new states into play. If you average the margin in swing states, it comes to around 3-4%. I realize thre are probably 15-20 deep red states in the South and Plains where the Romney margin is 20% or more, but aside from Texas none are in the top 10 in population. Those should be balanced by California, New York, Illinois, and Massachusetts, where Obama will win by 20% or more, not to mention New Jersey, where he should win by at least 10%. So it would seem to me that Obama would have to win the national vote by 10% or so to bring states like GA or TN into play.

  • MarkS

    It seems to me that sudden moves of the EV estimator are closely related to your sharp one-week time cutoff on the included polls. Is there really no useful information in an 8-day old poll? Hard to believe.

    • JamesInCA

      From the FAQ: “For the current snapshot, the rule for a given state is to use the last 3 polls, or 1 week’s worth of polls, whichever is greater. A poll’s date is defined as the middle date on which it took place. In some cases 4 polls are used if the oldest have the same date.”

      So, if polls are sparse, the most recent 3-4 polls will be used, even if more than a week old. If more than 3-4 polls are released in a week, we’re probably nearing the election (like now), and polls from longer ago probably don’t contribute much that isn’t accounted for by the last week’s polling.

    • AySz88

      According to the code (write_statistics() in, the one-week window is a week from the most recent poll, not one week from the current date. So the jumps happen only when the most recent poll changes, and not just arbitrarily as time passes. (I agree that it would be quite weird to have an estimator that can change when no polls have been released!)

      I do think an exponentially falling weight could be better, but it’s probably not a huge difference.

    • Brad Davis

      @AySz88 But how do you go about selecting a value for lambda that isn’t totally arbitrary? And if it’s arbitrary then it’s just adding more unsupported opinion into the data and hence noise? On the other hand, while Sam’s current method may be equally arbitrary, it doesn’t require estimating lambda properly and it’s far easier to understand.

    • Sam Wang

      As currently formulated, the Meta-Analysis limits itself to a latest-poll-minus-7-days scale, more or less, as AySz88 has determined from examining our code. That sets a natural time scale, so that you know how fresh the measurement is, more or less. Time-dependent weighting is an option. However, there is a risk of using data that came before some game-changing event. So I agree with Brad Davis’s concern.

      In 2008, I experimented with identifying moments of likely opinion change, then take all data from that point in time forward. That did pretty well in extremely close states. Read about it here.

    • AySz88

      Brad – This is a bit nitpicky, but I think one can select a lambda that would be equivalent for a 7-day window for a particular simple case – for example, a constant linear trend.

      For example, if one poll were released every day showing a one-point increase per day, and n were the latest poll result, then:

      A 7-day window would give you n-3.
      An exponential weighted average would give you n – (x/(x-1)) where x is lambda.

      If you wanted those numbers to match, then x = 0.75 . So an exponential weighting with “lambda” = 0.75 would, in one sense, produce results equivalent to a 7-day window. (And interestingly, it looks like it would be much more sensitive to recent trends of higher order than linear.)

      Just for completeness, I should probably also mention the idea that to get rid of “jumps” (of the old-poll-expiring type), the weighting doesn’t have to be “exponential”, precisely – it just has to have the property that today, all past polls get hit by some penalty factor. But that penalty can change with each new day. (Simple example: poll weights decay at 0.9x/day until September 1, and then all poll weights start decaying at 0.75x/day rate for each day after that.)

      That gives you a “game-changing-event” knob if you want it (just increase the decay rate temporarily), but that obviously makes things even more complicated… How exactly would one use the knob? Maybe try to use Wang’s variance minimization for various stretches of the campaign (manually? automated with some sort of RANSAC?)? I think it’s interesting, but yes, not simple.

    • Brad Davis

      @AySz88 I don’t think it’s nit-picky at all. For one thing, the current method doesn’t use a 7 day window, even though you continue to refer to it in that fashion. It uses the last 3 polls OR 1 weeks worth of polls, whichever is great. So if those 3 polls are spread out over 3 weeks, then that falls outside of the 7 day margin. Even if you don’t use precisely an exponential function, you just weight polls according to their age, you still need to figure out the right way to give weights to each of those polls. Plus, that kind of a method would implicitly give added weight to polls done more frequently, because at any given point in time they’re likely to be the most recent poll. And it adds all sorts of weird and difficult to interpret side effects, with minimal benefit. Sam has been using this basic method very successfully since 2004. If it ain’t broke, why fix it? This seems like a solution in search of a problem.

    • AySz88

      :/ I think I should have specified that my ‘nitpicky’ comment was not directed at you, but at myself (for taking your question literally and defining a family of methods to “select a value for lambda that isn’t totally arbitrary”). :p

      The focus on the 7-day window is because that’s what the OP focused on. Though you make a good point that the “last three” polls part doesn’t exactly have a great analogue with exponential weighting. (It could be that a dearth of polling is itself a signal of stability.)

      But the main benefit I see is that exploring these parameters has the outside shot at bringing deeper insight. “This algorithm seems to work with these parameters” is awesome in its own way, but it’s not as satisfying to me as figuring out why that algorithm works with those parameters. Whether you think this is worth it is a matter of taste.

  • Partha Neogy

    “Yet the Popular Vote Meta-Margin is not at an extreme value. This reflects that fact that in many swing states, President Obama’s lead is only moderately-sized, for the time being.”

    And because in the very red states he is lagging worse than in 2008?

  • Bill N

    I live in TN and find it very hard to believe that Obama could ever carry TN. But, in all candor I have not seen any polls recently, so my skepticism is a gut hunch or at best a SWAG. I have seen some Obama TV ads, which has surprised me, since I would think it a waste of money if TN is out of play.

    In terms of the EV estimator and meta-margin, the question is, is this trend a lasting change moving to a new, stable level? Or will it revert back towards the “true median,” and if so, what is that true median? Any thoughts on this? It will be very interesting to watch.

    • Anbruch

      Are there close congressional races in TN? If so, Obama may be on the air in TN as much in support of those races as anything else—that and the fact that he’s got to put the money somewhere.

    • Donna WO

      Bill N, I also live in TN. You might have been right in believing that Obama could not carry TN, UNTIL all of the GOP talk over the Social Security/Medicare voucher system that they want to convert over to. I am a volunteer for the Obama campaign here in Nashville and I talk to more and more people every day, that are going to vote for Obama just because of those issues. There are life long Republicans, that are 62+year old, that are flipping only because they fear that they will not have the system. I spoke with a sweet widow today that explained, “The Social Security and Medicare that we have now might not be what I would like it to be but I certainly do not want to take the chance on losing it to some type of voucher system that I know nothing about nor has Romney explained any true details about. It is just not worth the risk!” So Bill, you never know, TN might be changing.

    • David D

      I can’t imagine that TN is in play. Could you be in a media market that spills over into NC or VA?

  • Xtalographer

    TN clearly not in play. Not even close. I would accept Indiana and Montana. Next would be Arizona and Missouri.

  • PatM

    Perhaps you’ve answered this question previously, but I’m curious to know how your analyses account for voter suppression (. If 700,000 voters are prevented from voting in Pennsylvania, for example, that could be decisive.

  • Donna WO

    TN, in which I live, is a half and half state. In the larger cities ie; Nashville, Memphis, Obama has the upper hand. Most of our celebrity population is behind Obama. Many of our over 62 population are changing their minds now that they fear the loss of the Social Security/Medicare system, thanks to the Romney/Ryan voucher platform. I honestly believe, the more of the 99% hear of the GOP plan and the fantastic screw up speeches by Romney/Ryan, the more of them will actually GO to vote and will vote Democratic.

    • wheelers cat

      What you are describing Donna is the Distributed Jesusland model. Remember the Jesusland map from 2008? Distributed Jesusland is that map minus the majority minority cities.
      Eventually we might see a kind of oil spot model where the urban population cents join up and overwhelm the heartland. In say 2050 when hispanic becomes the most populous demographic in the US.

  • bks

    If you click Andrew S. Tanenbaum’s “No Rasmussen Data” button:

    the result, strangely enough, is 347:191.


  • Spiny Norman

    “If the election were going to be held tomorrow that would be a problem.”

    Thanks for your help at the Convention, Governor Christie!

  • Ms. Jay Sheckley

    Mark S, it seems possible that using no polls older than 7 days gives the snapshot here some of its superior present-day focus. For example, North Carolina is now a near-sure blue, where it remains pink on , a site with many virtues, by the way, which I learned of here.

  • Bill N

    Thanks to all of you who live in TN for your comments and your efforts on behalf of the Obama campaign.

  • Ken

    I just spent 5 days in Indianapolis visiting family. My family is hard core GOP and particularly anti-Obama. They hardly discussed or even mentioned politics or the election. Even odder was the near complete lack of campaign signs or bumper stickers anywhere. During my entire time there I saw a single corner with about a half dozen signs, about half for Mourdock and the other half for Donnelly. You wouldn’t even know it’s an election year there – very defeated kind of atmosphere and very different from that here in DE where I live.

    • Dave F

      That’s similar to what I’ve noticed on Facebook. I have a number of very conservative “friends” who had been posting anti-Obama things for months. They gleefully called him “Failure in Chief,” “the food stamp President,” etc., repeated all of Rush’s best lines. But since the conventions, they have basically all shut up. They’re not going to vote for Obama, but the wind has been taken out of their sails. It’s no longer fun for them. And now my liberal friends are frequently posting anti-Romney stuff. You can viscerally see the enthusiasm measure swing in the last month.

  • steve in Colorado

    And in 2014 we will see another 2010 year with massive Republican gains as a backlash to the gains of this year… Incumbents beware! We are definitely lucky that the Republican candidates were so incredibly weak this year. If Obama was running against a Paul Ryan (much smarter than Santorum) who had a consistent conservative narrative, then he could have played as a ‘compassionate conservative’ instead of continually pandering to the base.
    I think it might have changed the whole dynamics of this election.

    • RadicalCentrist

      Compassionate conservative-where have I heard that before?

      Look, I think Republicans are fooling themselves that it’s the messenger, rather than the message. It’s both. The Ryan budget is incredibly unpopular (and deservedly so). So, why would anyone think that he would do better at the top of the ticket?

      As fo 2014, my crystal ball is cloudy. Typically mid-terms for a re-elected President dn’t swing too much either way.

    • Sam Wang

      Perhaps, but the number of restrictions you place generate a very low N (2006, 1998, 1984…). Also, in 2006, President G.W. Bush’s party lost 31 seats in the House and lost control of the chamber.

    • Matt McIrvin

      But Obama *is* running against Paul Ryan. Ryan’s selection, actually, brought Romney as close as he’s been to parity for a couple of weeks. Ryan’s convention appearance didn’t do the ticket any favors, though, and we haven’t heard a lot from him lately.

    • steve in Colorado

      My point is that he is STILL making the case that he is a true blue conservative. At this point, he should have used his etcha-sketch and become a moderate, ‘compassionate conservative’ in order to woo the moderate voters. Worked with GW Bush.
      But even with the selection of Paul Ryan as VP he doesn’t feel safe enough with his base to do that. McCain had the same problem, though it was less obvious due to the massive financial crash right before the election.

  • steve in Colorado

    Washington Post says the GOP majority in the House is safe…

    • Sam Wang

      Lots of prognosticators are saying that. In 2010, nearly all advance estimates were too low on the size of the Republican sweep. And that was with the help of district polls, which we currently lack.

      Go mouse over individual races at, especially those “leaning R,” think about how fresh many of those polls are and where things are headed. I do not know if you would be surprised, but evidently the Washington Post would be.

    • Greg

      A lot of the races listed as “strong” (R or D) are up for grabs. Look at MN — the 2nd and 6th are both in play, but both are listed as Strong Republican.

  • Matt McIrvin

    As I said a week or two ago, I suspect Obama actually has a pretty hard ceiling right about where he is now. Missouri and Arizona and Indiana are very tough nuts to crack at this point. I’m not even convinced about North Carolina.

    Curiously, in Missouri Rasmussen seems to be more Democratic-leaning than PPP.

    • Brad Davis

      I agree, it seems that Obama has hit a ceiling- but if the voters continue to be disenfranchised with Romney, how likely is it that some of those other states would come into play? A question to Sam might be, of the next tier states (AZ, GA, TN, SD, IN, and MT), how much would increased Democratic voter turn out need to be to move those states into play (since I doubt that we’ll see a lot of hard core GOP voters switching to DNC)? How do those numbers compare against historic standards?

      A little off topic, but can anyone recommend a site similar to this one but for economics?

    • Jack Tenold

      However, the booing of Ryan at the AARP convention might indicate that Romney is losing his best polling group, the seniors. If so, the ceiling may be considerably higher than we thought.

    • michael

      For Brad Davis: An excellent site for economics that many here would find interesting would be

  • Dave Kliman

    I noticed that there are very few polls of TX. As a matter of a fact mostly (R) yougov and (R++) rasmussen… and even from those only a small lead for romney, and no really recent polls… it would be interesting if there were more modern style cellphone/human interview polls done there…

    • Froggy

      Dave, Pollster ( lists a couple of Texas polls from this month, both with double digit Romney leads.

      It’s too early for the Obama campaign to coast, but what next? One possibility is campaigning in states with close Senate races, where Obama is running ahead of the Democratic candidate, as Prof. Wang suggested the other day. This would mean sending Obama to Massachusetts, Connecticut, Wisconsin, Virginia, and/or Nevada. (Two new polls out today show the Nevada senate race close, with Obama running 5 points ahead of Shelley Berkley in both.)

      Is there something to be said for campaigning in states where a Democratic candidate is outperforming Obama, but Obama is still reasonably close? I’m thinking about Montana, Missouri, or Indiana. Is it a given that an Obama appearance or ad buy would necessarily hurt the local candidate? (I see that Tester is touting his defiance of Obama, so maybe any Obama effort would undermine him.)

    • Sam Wang

      Yes, local races are where the game is now. They were before, as I’ve been saying for many weeks. Now the jig is up – everyone knows.

      Even more important than the Senate, I think House districts are where the real knife-edge situation is. Think what a big deal an Obama or Romney visit would be to a narrow race involving fewer than 500,000 voters. Speaking of which, soon I plan to comment on the substantial pushback I am getting from The Monkey Cage.

  • HJM

    This is an astonishing shift. The American people are unlikely to elect anyone president – regardless of party – who says half the country is mooching and can be written off. But if that is driving this change, the magnitude of the reaction is really something. But will it hold?

  • Olav Grinde

    There has been a huge turnaround in recent weeks! The despondency on the Republican side is palpable, while the Democrats are all fired up. That’s going to have a huge effect on turnout.

    I would not be surprised if that results in Obama also winning North Carolina, Arizona, Indiana and Missouri — if the votes are fairly counted, mind you.

  • steve in Colorado

    Early voting appears that it will be even more widespread this year than in ’08. Here, we could conceivably get our ballots by Oct 16th (the entire state can choose to do mail in ballots). Conventional wisdom seems to say that Obama will disproportionately benefit from early voting. Do you have any thoughts on this, Prof. Wang?

    • Sam Wang

      Early voting is baked into the polls; pollsters ask this question. However, President Obama is leading by his largest margin yet this campaign season. For that reason alone, early voting helps because it locks in present gains. Kind of like selling a stock while it is high.

  • Brad Davis

    I recognize this question is a bit off topic, so I will understand completely if you choose to delete it Sam.

    I’ve been thinking a bit about the GOP’s economic plans, which to one degree or another all involve cutting taxes on the highest tax brackets as a way to increase spending and then increasing taxes or decreasing benefits for the poor and working class.

    It made me wonder what is the average residence time or half life of a dollar as a function of someone’s average income or net worth? My guess is that it would be strongly asymptotic (x/(x+1)). If that’s true (and it seems like it has to be true, as the less money you have, the greater it’s utility, and the less ability you have to save it for later), then reducing taxes on the rich will not produce as much economic growth as reducing the taxes by the same amount (in absolute terms) on the poor. Because they have to spend their money.

  • Olav Grinde

    Well, you could also try some Outcome Research. For instance examine what millionaires and billionaires actually did with their Bush tax cuts.

    Did Mr Romney, for example, invest in American jobs? No? Oh, they’re in a Cayman Island bank account? Well…uhm…we could always try another tax cut to create jobs.

  • Amitabh Lath

    Looking at the Monkey Cage as well as some other House predictions, I get the feeling they come up with a number they would like to hit first, and then create a model (some polls, some econ data, some gut feeling…) and a pinch of this and a teaspoon of that and voila, the model validates what they had thought all along.

    Let’s assume House races are two-component systems: a local term and a national term. The local-term coefficient probably does not move much (unless the candidate says/does something stupid).
    The national-term coefficient is probably skyrocketing up as the Median EV does.

    So given that polls have moved +D, it’s not unreasonable that the House median is also +D.

    What is really interesting is that all the polls are moving in a correlated way (NC, WI, MN, OH…).
    This is what is driving down the uncertainty of the
    median. This is where Sam’s 74% prob comes from. Not only median moving +D, but uncertainty shrinking.

    By the way, 74% is interesting (more for the small uncertainty than the +D median) but fluctuations happen. If it is real it should get more significant as D-day approaches.

  • Olav Grinde

    Wow, the numbers on today’s CBS/NYT/Quinnipac polls from Ohio, Florida and Pennsylvania can’t be fun reading for Team Romney. And Bloomberg har Obama +6 nationally.

    I wish we had more House polls. It would be encouraging to see polling confirmation of Sam’s calculation.

    Amitabh Lath: I wonder if it would help for the Democrats to air a whole bunch of local commercials in the last week or two before the election, where Obama implores voters to vote for their local Blue representative. That they’re needed in Washington.

    I’m surprised Obama and the Democrats haven’t talked more about Republican obstructionism. Of appointments, bills, compromises…

  • Amitabh Lath

    Olav, I really don’t know what moves this population who are undecided at this point.

    Numerically, they aren’t a big cohort. It’s what, a million or so people? Who are these guys? Do you know anyone still not decided? Maybe I live in a blue bubble.

    I would not have the first idea of what would move these guys. It’s a small enough number that free pizza for all of them would be cheaper than ads.

    I believe Axelrod et al are experts at this. I remember him from when he wrote for the Chicago Tribune. His articles (and Mike Royko’s) were must reads, even for 11 yr. olds. I have been very impressed with how effective and timely their ads have beeen, compared Romney’s.

  • MAT

    It’s a seven point race, outside of the uncertainty bands, in the Rand poll this morning.

  • Olav Grinde

    Undecided voters?

    Amitabh, Bill Maher has his own take on “the undecided voter”. It’s hilarious, but rather irreverent and I hope you don’t mind the occasional well-placed expletive…

    • Slovo

      Thanks for the link Olaf, but
      unfortunately the video is not there anymore. Instead there is this notice:
      “Bill Maher Slams ‘Dip…”
      This video is no longer available because the YouTube account associated with this video has been terminated due to multiple third-party notifications of copyright infringement.
      Sorry about that.

    • Sam Wang

      I think that is a very long article – at least page 1 was. Then I reached my RDA of Jay Cost.

    • Brad Davis

      My favorite line in the story is ‘I am not in favor of partisan weighing, per se, although some polls like the Rasmussen poll do it in a sensible and nuanced way.’. I don’t see very many blogs or columns which describe Rasmussen’s polling or analytical methodologies as being ‘sensible’, or ‘nuanced’. Right off the bat, that sounds to me like someone who is trying to preach to the choir, e.g. GOP supporters.

      He also likes to include historical precedents. That’s fine for Sunday Night Football- it serves as a talking point, but even then I don’t think those comparisons mean very much (does it really matter how many win’s the Chicago Bears have over Green Bay in games played in the snow in December???)

      My next favorite line in that article is ‘All told, we see a statistically significant relationship between Obama’s margin and the Democratic advantage in partisan identification. ‘ Really? Care to give a p-value or identify the method that you used to come up with this statement? And I don’t think it’s because he’s worried it’s too complicated for his readership, he posted a link to the Wikipedia entry for “bimodal distribution” (although personally, I would’ve liked to Wolfram).

      In short, tt seems to me that Jay Cost started with a conclusion (i.e. Obama isn’t nearly as far ahead of Romney as the polls suggest), ‘examined’ the data, and then came up with a ‘just-so’ story to establish the validity of his hypothesis.

      On the other hand, perhaps my analysis of his article is equally as biased.

    • RadicalCentrist

      Most of the polls do show a higher proportion of Democrats and a lower proportion of independents than the actual registration data. But that isn’t because the pollster are screwing with the #s or gremlins in the phone lines are routing the calls preferentially to Dems. It’s because people who are going to vote for Obama will tend to report themselves as Democrats. And the pollsters have no way to check how someone is actually registered, they simply record the affiliation stated.

    • steve in Colorado

      I read most of that- when he got the non-partisan guy from the National Journal on, he seemed at least as interested in discrediting him as in arguing his thesis- IE that they are oversampling Democrats. That interview could have been done in 5 minutes:
      HH: When you show that self identifying Democrats are outnumbering self identifying Republicans by 7%, it flies in the face of the 2010 stats of self identifying R’s outnumbering self identifying D’s by 1%. The swing shouldn’t be that high.
      SS: Well, the attitudes are what we are measuring, and self identifying with one of the parties is part of that, a main part that we don’t want to fiddle with. We can fiddle with age, gender, race etc. but that one (party) is part of the answer that we are looking for.
      HH: OK! Well, I disagree but thanks for your time.

  • Olav Grinde

    @MAT: With regards to the Rand Poll, I was struck by another statistic; respondents indicate that Democrats are 5.5 % more likely to vote that Republicans.

    In other words, we are seeing an enthusiasm gap that is nothing short of astonishing!

    If this comes true on Election Day, it will have a huge effect on downticket races — and may well (true to Dr Wang’s predictions) give the Democratic Party control of the House, and perhaps even increased representation in the Senate.

    • Matt McIrvin

      It’s the other way around, isn’t it? The Romney supporters are the ones more likely to vote. That’s how it’s usually been in US elections.

  • Brad Davis

    By the way, I love the picture of Chris Christie you selected. It makes me laugh every time.

  • Amitabh Lath

    I managed to see the Bill Mahr snippet on undecided voters. Very funny, and politically incorrect, as usual. But the glimpse of Salman Rushdie at the start of the segment was nice.

    Maybe undecided voters have just prioritized differently.

    Frankly, will your life change in any appreciable way if Romney wins? Mine probably won’t.
    Overwhelming sadness? yes. Life changing? no.

    I am motivated not by any personal gain/loss calculation, but the fear that if science-hating, natural-selection denying, CO2 ignoring luddites gain power, we will move closer to the next dark age.

    But maybe that’s not a huge deal for these people.

  • Ms. Jay Sheckley

    You’ve said watching polls is “like watching paint dry”. Here’s an example: After all the anxious map watching, today’s State by State Presidential Map is eerily like your 2008 Result Map

Leave a Comment