Previous Month | RSS/XML | Current | Next Month

September 28th, 2016 (Permalink)

Logic Checking the First Debate, Part II

In this part, we will look closely at the he-said/she-said part of the exchange between the candidates concerning the murder rate in New York City (NYC). To refresh your memory, here's the relevant part, though you can see the larger context in Part I, below.

Clinton: Well, itís also fair to say, if weíre going to talk about mayors, that under the current mayor, crime has continued to drop, including murders. So there is…

Trump: No, youíre wrong. Youíre wrong.

Clinton: No, Iím not.

Trump: Murders are up. All right. You check it.

Now, the thing that started this back-and-forth was Clinton's claim that the number of murders continued to decline under the current mayor of NYC, Bill de Blasio. Putting aside the factual question for the moment, why is this logically an important enough point for the debaters to wrangle about? Because, according to Trump in the earlier part of this debate―see Part I, below―the so-called stop-and-frisk policy was discontinued by de Blasio.

Because if the number of murders continued to decline after S&F ceased, that would be evidence that S&F had little if anything to do with the decline. This, of course, is why Trump vehemently denied it, claiming that the number of murders increased.

If Trump is right, the increased murder rate is additional evidence that S&F was indeed contributing to the decrease in the number of murders. This is an application of Mill's method of difference―see Source 4, below―that is, if S&F were causally responsible for at least some of the decline in the murder rate, then when S&F is ended we should expect that the murder rate will at least slow its decline, stop completely, or even begin to climb again.

This is why the issue of who is right about the murder rate in NYC is not a trivial dispute about numbers, but is important to the larger issue of whether the S&F policy is effective in reducing crime. Unfortunately, the factual question is not a simple one, but here are the facts―see Sources 1 & 2, below. The number of murders for the last year of S&F, 2013, was 335. For 2014, De Blasio's first year as mayor, the number was 333. For 2015, it was 352. For the year to date, it's 246. The year-to-date figure, of course, can't be compared directly to the statistics for the previous full years, but the NYPD also provide the comparable figure for this time last year, which was 257.

So, who's right: Clinton or Trump? Is Clinton right because the murder rate in NYC declined slightly from 2013 to 2014 and is on track to decline from 2015 to this year? What about the increase from '14 to '15? Do we just ignore that? Is Trump right because of the '14-'15 increase? But then, what about the decreases from '13-'14 and '15 to the year-to-date? Perhaps when Trump claimed that "murders are up" he was thinking of the substantial '14-'15 increase; maybe he had seen that statistic, but not the incomplete numbers for the current year. When Clinton claimed that "under the current mayor, crime has continued to drop, including murders", perhaps she was only considering the "continued" decrease from last year to now.

So, who's right? Neither? Both? I'll let you decide.


  1. "CompStat, Volume 23, Number 37", New York City Police Department, 9/12/2016-9/18/2016 (PDF)
  2. "Seven Major Felony Offenses", New York City Police Department, Accessed: 9/28/2016 (PDF)
  3. "Debate Fact-Check: How Donald Trump and Hillary Clinton Stack Up", ABC News, 9/27/2016
  4. Michael K. Green, "Mill's Methods: Method of Difference", Logic Tutor, 1998

September 27th, 2016 (Permalink)

Logic Checking the First Debate, Part I

I mentioned in the previous entry that many critics of past presidential debates have complained about a lack of "clash" between the candidates; well, there was plenty of clash in the first presidential debate of this election year. In fact, at times the moderator, Lester Holt, appeared to have very little control over the proceedings, which descended into childish bickering.

One point on which the candidates clashed was the effect of past policing policies on the murder rate in New York City (NYC):

Clinton: …[W]e do always have to make sure we keep people safe. There are the right ways of doing it, and then there are ways that are ineffective. Stop-and-frisk was found to be unconstitutional…in part, because it was ineffective. It did not do what it needed to do. …

Trump: …[W]hen it comes to stop-and-frisk…. [I]n New York City, stop-and-frisk, we had 2,200 murders, and stop-and-frisk brought it down to 500 murders. Five hundred murders is a lot of murders. Itís hard to believe, 500 is…supposed to be good? But we went from 2,200 to 500. And it was continued on by Mayor Bloomberg. And it was terminated by current mayor. But stop-and-frisk had a tremendous impact on the safety of New York City. Tremendous beyond belief. So when you say it has no impact, it really did. It had a very, very big impact.

Clinton: Well, itís also fair to say, if weíre going to talk about mayors, that under the current mayor, crime has continued to drop, including murders. So there is…

Trump: No, youíre wrong. Youíre wrong.

Clinton: No, Iím not.

Trump: Murders are up. All right. You check it.

So who is wrong? The question is more complicated and harder to answer than you might at first think. There are two issues here that need to be untangled:

  1. The causal issue of whether the stop-and-frisk (S&F) policy was effective in reducing crime in NYC, with Clinton claiming that it wasn't and Trump claiming that it was.
  2. There is a difference of opinion over the statistics of murders in NYC, with Trump claiming that the number of murders has increased and Clinton claiming that it has decreased.

In this first part, I will examine only the first, causal issue, leaving the issue of NYC murder statistics until tomorrow. But first, how are these two issues related? The only evidence given by the candidates to support their positions on S&F are the alleged decreases and increases in the murder rate. Clinton initially claimed that: "Stop-and-frisk was found to be…ineffective. It did not do what it needed to do." However, she did not say who found this out, so we just have to take her word for it.

Trump, in contrast, vehemently insisted that S&F was effective: "…[S]top-and-frisk had a tremendous impact on the safety of New York City. Tremendous beyond belief. So when you say it has no impact, it really did. It had a very, very big impact." He also used causal language in saying that S&F "brought down" the number of murders.

The evidence that Trump provides for this causal claim is that the number of murders had gone from 2,200 to 500. There's much in this statement that is unclear. Presumably, these are annual numbers of murders for NYC, but Trump does not say which years. To support his claim that S&F caused the decline, 2,200 would have to be the murder rate in the year before S&F began, and 500 in the year following its end, or something similar. Clinton didn't actually challenge this part of Trump's claim, so let's accept for the sake of argument that it's correct.

So, assuming that the annual number of murders in NYC really did decline during the period of S&F to about a quarter of what it had been, how strong of evidence is this that S&F brought about that decline? Not very strong. The national murder rate has been declining until very recently, so how much, if any, of the decline in the number of murders in NYC is the result of S&F, and how much is the result of a general decline in murder? It would take a careful statistical study of the differences between NYC, and perhaps other cities that have had S&F policies, compared to similar cities in the same period who lacked such policies, to tease out just how much of the decline is due to S&F. Has such a study been done? Maybe, maybe not, but we don't hear a thing about it from either candidate.

So far, neither candidate's conclusion is warranted by the evidence provided; maybe S&F reduces murder rates, but maybe it doesn't. However, we still need to look at the he-said/she-said on NYC murder rates, which we will do in Part II.

To Be Continued…


Fallacy: Cum Hoc Ergo Propter Hoc

September 25th, 2016 (Permalink)

Debate Preparation

Sitting on a sofa on a Sunday afternoon,
Going to the candidates' debate,
Laugh about it, shout about it,
When you've got to choose.
Every way you look at it you lose.
―Paul Simon, "Mrs. Robinson"

The first of this election's presidential debates is scheduled to occur tomorrow. As in past election years, I expect to comment on each debate here shortly after they happen, though I won't promise to do so since I may not have anything worth saying.

This debate will have a single moderator: Lester Holt, the NBC Nightly News anchorman. Holt will ask the candidates questions, each candidate will have a couple of minutes in which to answer, and then additional time to respond to each other. Holt is also responsible for selecting the questions that he will ask.

Many people have criticized these "debates" in past years on the grounds that they're really joint press conferences. What's often been missing is what's called "clash" in debate jargon, that is, the debaters directly responding to each others' claims and arguments. However, earlier presidential debates often had panels of journalists who asked questions, and the formats sometimes discouraged follow-up questions or direct responses. The format of tomorrow's "debate" at least leaves open the possibility of clash, but much will depend upon the moderator and the candidates themselves.

Jill Lepore had a worthwhile article in The New Yorker about a week ago on the history of presidential debates and the role that debate plays in self-government:

How to argue is something people are taught. You learn it by watching other people, at the breakfast table, or in school, or on TV, or, lately, online. Itís something you can get better at, with practice, or worse at, by imitating people who do it badly. More formal debate follows established rules and standards of evidence. For centuries, learning how to argue was the centerpiece of a liberal-arts education. … Debating, like voting, is a way for people to disagree without hitting one another or going to war: itís the key to every institution that makes civic life possible, from courts to legislatures. Without debate, there can be no self-government. The United States is the product of debate.
Source: Jill Lepore, "The State of the Presidential Debate", The New Yorker, 9/19/2016

The presidential debates this year, assuming they take place, may be different than those of previous years. Donald Trump is a famously unpredictable candidate, and both candidates bring a lot of historical baggage to the debates. There's been much speculation in the media that one or the other of the candidates might back out: Clinton, because of her health problems; or Trump, because, well, he's Trump. However, Trump's also complained that the debates, like everything else, are "rigged", and that the schedule for a couple of the debates is "Unacceptable!" because they compete with football games. Nonetheless, tomorrow's debate appears to be go for launch.

I prefer to let a little time pass before commenting, so the earliest that an entry on a given debate is likely to appear here is the day after the debate, and some may take longer. This is because, in contrast to television pundits, I spend some time thinking about what was said before I comment on it. There's plenty of knee-jerk commentary around, so if that's what you're interested in, I'm sure you can find it elsewhere. Much of that, of course, is spin coming from partisan pundits attempting to influence your opinion in favor of their candidate. I will endeavor to keep my commentary to logical and related issues that are within my field of competence, since I doubt that you're interested in my political opinions, nor should you be.

Perhaps it should go without saying, but I won't attempt to comment on every logical topic raised in a given debate, or identify every single fallacy committed, because this simply wouldn't be possible. Rather, what I choose to comment on will be decided by what I happen to notice and what interests me. However, I will strive to be even-handed and non-partisan, but I won't attempt a kind of artificial balance between the two candidates. If you think I've missed a logical point or fallacy example in any of the debates, please let me know about it.


Resource: Pre-Debate Warm-Up, 10/3/2012. In this entry, I discuss some of the most common logical problems to be on the lookout for in political debates.

September 3rd, 2016 (Permalink)


Pope Francis looking into female clergy members

No comment.

September 2nd, 2016 (Permalink)

Hey, MoE!

It's been a while since I did a "Poll Watch" entry, but with the general election only a couple of months away, this is a good time to do so. Here's how The Washington Post reported the results of a recent survey:

As of today…Americans' views of [Hillary Clinton] just hit a record low. A new Washington Post-ABC News poll shows 41 percent of Americans have a favorable impression of Clinton, while 56 percent have an unfavorable one. … Her previous low favorable rating this year was in July, when it was 42 percent…. Her previous high for unfavorable views was in June, when 55 percent disliked Clinton.

In other words, the "change" in Clinton's favorability rating is a shift of just one percentage point from "favorable" to "unfavorable". As far as I know, no national political poll has a margin of error (MoE) so low that a one percentage point difference would be statistically significant. In fact, the margin of error for this particular poll is 3.5 points, so that one point is well within it.

Oddly enough, some reporters appear to have an intermittent ability to understand the margin of error . For instance, in ABC News' account of the same poll, we read the following:

[Trump] scored particularly low with blacks, 84 percent of whom view him unfavorably. Given the sample sizes, thatís not a significant difference from the 91 percent of this group who responded similarly in early August….

The report doesn't go into the details but, presumably, the subsample of blacks who took the survey was so small that even seven percentage points is "not a significant difference". It's nice to see that the reporters actually understand the effects of sample size on the margin of error, and the fact that sometimes a difference in numbers is not necessarily a difference in reality. But if they understand that for Trump's unfavorability with blacks, why can't they understand it for Clinton's unfavorability with all Americans?


  1. Aaron Blake, "A record number of Americans now dislike Hillary Clinton", The Washington Post, 8/31/2016
  2. Gregory Holyk & Gary Langer, "Poll: Clinton Unpopularity at New High, on Par With Trump ", ABC News, 8/31/2016

Resource: How to Read a Poll

Previous Month | RSS/XML | Current | Next Month