Comments on: A Gravis Fraud

By: jimbo255

jimbo255 — Thu, 08 Nov 2012 08:17:30 +0000

The poll was NOT A FRAUD, they simply reported the incorrect sample size!

Take the following numbers:

___19______0_____16___________35

___36_____90_____66__________192

___77_____87_____24__________188

____9______1______0___________10

__547____359____196_________1102

__688____537____302_________1527

Divide by *1527*, not 914, and -- VIOLA! -- you get the reported percentages!

__1.24___0.00___1.05_______2.29

__2.36___5.89___4.32______12.57

__5.04___5.70___1.57______12.31

__0.59___0.07___0.00_______0.65

_35.82__23.51__12.84______72.17

_45.06__35.17__19.78_____100.00

By: Logarchism » Gravis Marketing: A Deeper Analysis

Logarchism » Gravis Marketing: A Deeper Analysis — Thu, 01 Nov 2012 10:02:52 +0000

[…] the October 14, I examined a poll of Gravis Marketing and found some dubious numbers in the crosstabs. Doug Kaplan, CEO of Gravis Marketing, has […]

By: Logarchism » A Gravis Response

Logarchism » A Gravis Response — Mon, 22 Oct 2012 19:01:05 +0000

[…] A Gravis Fraud […]

By: dcpetterson

dcpetterson — Mon, 22 Oct 2012 14:11:00 +0000

@logarchismsandbox-a424ed4bd3a7d6aea720b86d4a360f75:disqus
If you recall your question, you can try asking it again.

By: marbleman

marbleman — Mon, 22 Oct 2012 13:40:00 +0000

Thanks for the answer.

Unfortunately for me, I kind of need an answer in a timely manner (long story), but oh well.

By: mclever

mclever — Mon, 22 Oct 2012 12:40:00 +0000

@marbleman
DISQUS is turned on because of the debate tonight. Recent comments aren’t loaded into the DISQUS system, but should return when DISQUS is turned back off tomorrow.

By: marbleman

marbleman — Mon, 22 Oct 2012 12:16:00 +0000

Where did the comments go?

I had a question in my comment that I was really hoping for an answer to.

By: marbleman

marbleman — Sun, 21 Oct 2012 06:53:46 +0000

dawolf - while I am inclined to believe that Gravis may be a very suspect operation for many different reasons, something occurred to me about the most compelling part of your analysis - the idea that, say, a racial variable couldn't be re-weighted differently for different questions.

But what if a different number of people answered a specific question/refused to answer a specific question from question to question? Wouldn't your overall sample size and composition change in that case, and therefore, wouldn't you have to re-weight differently for different questions?

I am invested in believing this man is shady because he has been involved in bad things in his past - but the math has to work for me to claim his polling operations are suspect. Can you answer to this question I've posed? Am I missing something obvious here?

By: Michael Weiss

Michael Weiss — Sat, 20 Oct 2012 03:52:50 +0000

Working on it. I hope to have it written up this weekend. I apologize for the delay, but we're really late in the election cycle, so there's an amazing amount of stuff going on. And that's not counting my personal life.

By: ftspbr

ftspbr — Sat, 20 Oct 2012 03:44:46 +0000

Where's the interview you said you conducted? Been waiting.