[Previous] Globalism | Home | [Next] The Greeks Knew We All Make Mistakes

Bad Scholarship on VDare

From VDare: 93% Of Democrats Think It's Important That Fewer Whites Be Elected.

The article leads with a chart, which says that 75.1% of Democrats believe that "Whites are favored". The chart repeats later in the article. Problems:

  1. The chart has no source.

  2. The author falsely claimed that he always gives sources. Actually he had to be asked the source on Twitter because he hadn't given it. The source is these CNN exit polls.

  3. To get the chart, the author did math on the CNN exit poll data. He did not show his work.

  4. I checked what math he did by asking him on Twitter, since he didn't document it. His math was wrong. Where he got 75.1%, the correct answer was 75%. He added an extra significant figure to exaggerate how good his data was. He admitted I was right, but thought the matter deserved the comment "lol" rather than saying e.g. "My mistake, I will fix it."

The chart was brought to my attention by khaaan on Twitter, who questioned its sourcing.

Elliot Temple on November 8, 2018

Comments (9)


1. and 2. Yes it does. It's provided in the body of the post. In the interest of my readers' time, I don't link to the same source multiple times in the course of a single post. I've been doing this for a long time and have a well-deserved reputation for being meticulous with my source data.

3. Almost nobody wants to see the figures presented worked out. It's a blog post, not a research paper (though most people who read research papers also skip over the parts where authors show their work.

4. No, you are incorrect. Obviously Edison/CNN asked questions, got absolute numbers in their responses, and then put rounded percentages next to them. The absolute numbers are not difficult to work out. Having obtained them, I then created a graph with those absolute numbers of responses and rounded to the nearest tenth of a percentage.

Literally every single complaint you made, publicly, to try and grab a little attention for yourself was invalid. You're not doing VDare or our movement any favors with ignorant concern trolling.

I'm a lot better at this than you are--that's fine, I've been doing it for a long time--so if you have concerns you should attempt to have them addressed privately via DM. There are organizations like the SPLC and the ADL who love having useful idiots like yourself blow smoke for them. They use it to create the vague perception that VDare cannot be trusted. It can be.

Audacious Epigone at 11:19 AM on November 10, 2018 | #11354 | reply | quote

If a book has a chart on page 54 with no source in that chapter, but on page 33 there is an unrelated comment with a source which happens to be the same source from the page 54 chart ... is that reasonable? Is the chart sourced? No. There is nothing saying what the source for the chart is.

Re 3, if people don't want to see something, that is what footnotes are for. ez.

Re 4, You simply can't work out the exact absolute numbers based on rounded percentages. There are a bunch of different absolute numbers of responses that would result in the same rounded percentage being published. This is obvious. You are being careless because you're rationalizing instead of thinking.

By being so sloppy and then refusing to correct your errors, you are helping SPLC and ADL. You are making the lefty "fact checkers" partially correct when they attack sites like VDare.

Dagny at 11:32 AM on November 10, 2018 | #11355 | reply | quote

AE, regarding point 3, you could use a footnote. Doesn't have to be in the body.

It's important to show your original work in order to get mistakes corrected, which is something your *best* readers might help with.

Regarding SPLC etc, it is helpful to them if their opposition offers low quality scholarship that is easily refutable. The way to avoid this is by using good methods like being rigorous, showing your work and properly documenting sources. You don't seem to be a big fan of this approach, and are thus yourself (unwittingly) the left's handmaiden.

Finally, having a strong spirit of frank public criticism is good. Why should we limit our criticisms of scholarship to DM like fragile snowflakes? Doesn't VDare like spirited exchanges of ideas like at the Mencken conference or whatever? The left is gonna be vicious regardless - whispering our disagreements won't appease them.

I'm not curi, and I like VDare in general, but can't say I'm impressed with your response here pal.

Anonymous at 11:35 AM on November 10, 2018 | #11356 | reply | quote

CNN: 18778 respondants, 41% say whites are favored, 87% of those are dems.

How many of those respondents are dems? We can't know exactly but we can calculate a range. The *minimum* is: ceil(ceil(18778 * 0.405) * 0.865). Why? The lowest they'd round up to 41% is 40.5%. And they'd round 86.5% or more up to 87%.

So here's the math with ruby:

>> ((18778 * 0.405).ceil * 0.865).ceil

=> 6580

Let's sanity check that. What does the naive method get?

>> 18778 * 0.41 * 0.87

=> 6698.1126

Similar. Sanity check passed.

That naive method was actually used by Audacious Epigone, see https://twitter.com/AudaciousEpigon/status/1060703381176627201

Ok and what about the top end of the range?

>> ((18778 * 0.415).floor * 0.875).floor

=> 6818

So there is a range of 239 different numbers of individual respondents that CNN would have reported as it did. Audacious Epigone is factually and mathematically wrong.

I leave it as an exercise for the reader to correctly calculate the possible total numbers of dems who responded to the poll and to then figure out maximum and minimum percentages of dems who think whites are favored, and then see if the answers are all in the range 75.05% to 75.15% that could be correctly represented as 75.1%, or not. (Based on estimating it, I will be happy to bet money on *not*).

PS I know this depends on rounding rules. I assumed the naive rule of rounding .5 or more up and .49999 down, except also I fudged it and let .5 round down when convenient (but .50001 would round up, so I doubt that mattered).

curi at 12:02 PM on November 10, 2018 | #11357 | reply | quote

Even if all 18778 were dems (they aren't), then 239 variation is over 1% variance. So Audacious Epigone is incorrect to give an answer to a tenth of a percent.

curi at 12:12 PM on November 10, 2018 | #11358 | reply | quote

Even giving the answer as 75% would be questionable. It's a reasonable estimate that comes from math which makes sense (unlike the 75.1% which is just an error), but I don't think it's exact.

curi at 12:19 PM on November 10, 2018 | #11359 | reply | quote


Then in comments, he claims:

> I'm a lot better at this than you are--that's fine, I've been doing it for a long time

It's easy to discover that I have a blog post from Jan 2003 on this site. And I started before I had a blog. So it seems I've been doing it longer. He should be more careful about making false factual assumptions (he appears to assume he's being doing this longer than I have).

curi at 12:26 PM on November 10, 2018 | #11360 | reply | quote


>In 2008, Virginia’s seventh congressional district was 79 percent white. Today, it is 65 percent white.


>What do all three have in common besides a shockingly similar demographic change in the last decade? All three had been considered solidly Republican for decades, and the last week, all three Republican incumbents lost to Democrats.

the Virginia district was REDRAWN in a way which hurt Republican Dave Brat


vdare guy is trying to make some point about immigration and demographic change leading to GOP losses, leaves this detail out...

Anonymous at 3:28 PM on November 11, 2018 | #11364 | reply | quote

Today, Ann Coulter RT'd the VDARE post, saying:

> But they're NATURAL REPUBLICANS! From 80 to 60 Percent White: How to Lose a Congressional District

Alisa at 11:50 AM on November 12, 2018 | #11365 | reply | quote

What do you think?

(This is a free speech zone!)