Democratic Underground Latest Greatest Lobby Journals Search Options Help Login
Google

Non-linear optimization (Excel "Solver") proves rBr is Null and Void

Printer-friendly format Printer-friendly format
Printer-friendly format Email this thread to a friend
Printer-friendly format Bookmark this thread
This topic is archived.
Home » Discuss » Topic Forums » Election Reform Donate to DU
 
TruthIsAll Donating Member (1000+ posts) Send PM | Profile | Ignore Tue May-31-05 01:21 AM
Original message
Non-linear optimization (Excel "Solver") proves rBr is Null and Void
Edited on Tue May-31-05 01:41 AM by TruthIsAll
The problem:

Dertermine the precinct partisanship response rates required
to derive the absolute MINIMUM mean Kerry exit poll percentage
resulting from a 100,000 trial simulation.

The only constraints are 1) the weighted average response rate
equals 53% and 2) the following precinct partisanship
weightings:

       strong Bush >>>>>  strong Kerry
       40	415	540	165	90

Kerry%	10	30	50	70	90
Bush%	 90	70	50	30	10


The Excel "Solver" tool is a non-linear optimization
algorithm. Here is the description of Solver taken from
Excel’s online documentation:

"Solver is part of a suite of commands sometimes called
what-if analysis tools. With Solver, you can find an optimal
value for a formula in one cell — called the target cell — on
a worksheet. Solver works with a group of cells that are
related, either directly or indirectly, to the formula in the
target cell. Solver adjusts the values in the changing cells
you specify — called the adjustable cells — to produce the
result you specify from the target cell formula. You can apply
constraints to restrict the values Solver can use in the
model, and the constraints can refer to other cells that
affect the target cell formula. Use Solver to determine the
maximum or minimum value of one cell by changing other
cells."


Solver will derive the combination of response rates (subject
to the 53% weighted average response constraint) which will
MINIMIZE Kerry's total average simulation percentage. Rather
than "hunt" and "peck" and try hundreds of
combinations, we let the Solver algorithm derive the optimal
(minimal) solution.

Solver executes the 100,000 trial simulation, checks to see if
a minimum has been reached. If not, it iterates to try another
solution by changing the partisanship response rates and
re-executing the simulation. It performs this iteration
procedure until it finds a minimum, subject to the pre-set
iteration limit. 

The Solver algorithm went through its gyrations to derive the
optimal solution and returned the following result:

USING  PRECINCT PARTISANSHIP RESPONSE DATA SUPPLIED BY
MITOFSKY, THE SOLVER ALGORITHM COULD NOT FIND A  KERRY AVERAGE
SIMULATION PERCENTAGE  LOWER THAN 50.20% (50.24% MEDIAN). 

THERE IS NO FEASIBLE COMBINATION OF RESPONSE RATES CAN PRODUCE
THE RECORDED 50.73% BUSH VOTE. IN FACT, THE PRECINCT RESPONSE
RATES RETURNED BY SOLVER TO DERIVE KERRY’S 50.20% MINIMUM
PERCENTAGE ARE  IMPLAUSIBLE - WHICH MEANS KERRY DID FAR BETTER
THAN 50.20%.

Here are the final precinct category weightings calculated by
Solver to obtain the 50.20% Kerry minimum:

40 Partisan Bush (80-100) : 32.37%
415 Solid Bush (60-80)    : 70.33%
540 Non-partisan (40-60)  : 46.69%
165 Solid Kerry (60-80)   : 43.65%
90 Partisan Kerry (80-100): 18.93%

	Kerry Exit Poll simulation - 10,000 trials 
Using Solver algorithm to minimize Kerry percentage					
			K	B	Other	
		Mean	50.20%	48.80%	1%	
		Med	50.24%	48.76%	1%	
		Max	55.84%	43.16%	1%	
		Min	44.05%	54.95%	1%	
		Win	6951	3049		
						
	StdDev	0.15	0.05	0.05	0.1	0.15
	RespRate	32.37%	70.33%	46.69%	53.65%	18.93%

						
Printer Friendly | Permalink |  | Top
Bill Bored Donating Member (1000+ posts) Send PM | Profile | Ignore Tue May-31-05 01:29 AM
Response to Original message
1. Looks vaguely familiar. nt
Printer Friendly | Permalink |  | Top
 
autorank Donating Member (1000+ posts) Send PM | Profile | Ignore Tue May-31-05 02:02 AM
Response to Reply #1
2. Bill, how Delphic. Could you elaborate. I'm curious....and I need an
excuse to stay up even later than I'm staying up.

:hi:
Printer Friendly | Permalink |  | Top
 
Bill Bored Donating Member (1000+ posts) Send PM | Profile | Ignore Tue May-31-05 11:37 PM
Response to Reply #2
3. Hope you got some sleep.
I just meant that Febble, USCV and some others have been playing around with this stuff too. 1250 precincts distributed the way Mitofsky said back in Jan., and how to simulate the outcomes, slopes, fraud, etc.
Printer Friendly | Permalink |  | Top
 
autorank Donating Member (1000+ posts) Send PM | Profile | Ignore Wed Jun-01-05 12:33 AM
Response to Reply #3
4. No but I'm going to get some now. This is interesting since it's
the same problem but with a different outcome. In this case, "the work is shown."

Nite
Printer Friendly | Permalink |  | Top
 
TruthIsAll Donating Member (1000+ posts) Send PM | Profile | Ignore Wed Jun-01-05 01:26 AM
Response to Reply #4
5. Optimizations and response rates
Edited on Wed Jun-01-05 01:32 AM by TruthIsAll
Kerry wins 7 of 8 optimizations


1250	Precincts					
	High Bush		High Kerry	
	40	415	540	165	90	
	20%	40%	50%	60%	80%	
						
Optimize    Goal    Kerry Bush	Constraints (53% avg
response)		
1	Max Bush	50.30%	49.70%			
2	Max Kerry	50.72%	49.28%			
3	Min Bush	71.61%	28.39%			
4	Min Kerry	50.30%	49.70%			
5	Max Kerry	54.97%	45.03%<  Response [30%,70%]		
6	Max Bush	50.35%	49.65%<  Response [30%,70%]		
						
						

	10%	30%	50%	70%	90%	

      	  Goal	Kerry	Bush	Constraints 
7	Max Kerry	54.19%	45.81%<  Response [30%,70%]		
8	Max Bush	48.08%	51.92%<  Response [30%,70%]		
						
OPT 1						
Kerry	50.30%					
Bush	49.70%	<maximize				

		HighBush		High Kerry
	1250	40	415	540	165	90
	Kerry%	20%	40%	50%	60%	80%

0.53	Resp	89.70%	57.35%	50.88%	46.57%	41.17%
1.06	alpha	1.79	1.15	1.02	0.93	0.82
662	Resp	36	238	275	77	37
329	K Resp 	13	109	140	43	24
333	B Resp	23	129	135	34	13
300	K Ref	1	71	133	53	42
288	B Ref 	3	106	133	35	11
						
629	TotalK 	14	180	272	96	67
621	TotalB 	26	235	268	69	23
49.70%B Pct%	65.76%	56.63%	49.55%	41.92%	25.81%
49.52%B Exit%	64.12%	54.12%	49.12%	44.12%	34.12%
0.18%	WPE	1.64%	2.51%	0.43%	-2.20%	-8.31%
						
						

OPT 2						
Kerry	50.72%	<maximize				
Bush	49.28%					
		HighBush		High Kerry
	1250	40	415	540	165	90
	Kerry%	20%	40%	50%	60%	80%
0.53	Resp	50.30%	53.11%	54.04%	51.24%	50.67%
1.06	alpha	1.01	1.06	1.08	1.02	1.01
663	Resp	20	220	292	85	46
344	K Resp 	4	94	158	52	37
318	B Resp	16	127	134	33	9
290	K Ref	4	78	124	48	36
298	B Ref	16	117	124	32	9
						
634	TotalK 8	171	282	100	72
616	TotalB 32	244	258	65	18
49.28%B Pct%	79.9%	58.7%	47.8%	39.2%	19.5%
47.95%B Exit%	79.9%	57.5%	46.0%	38.5%	18.9%
1.33%	WPE	0.1%	1.2%	1.9%	0.7%	0.5%


OPT 3						
Kerry	71.61%					
Bush	28.39%	<minimize				
		HighBush		High Kerry
	1250	40	415	540	165	90
	Kerry%	20%	40%	50%	60%	80%
0.53	Resp	100.0%	6.3%	100.0%	0.0%	62.5%
1.06	alpha	2	0.127	2	0	1.25000001
663	Resp	40	26	540	0	56
614	K Resp 	16	1	540	0	56
49	B Resp	24	25	0	0	0
281	K Ref	0	156	0	99	27
306	B Ref	0	233	0	66	7
						
895	TotalK 16	157	540	99	83
355	TotalB 24	258	0	66	7
28.39%B Pct%	60.00%	62.21%	0.00%	40.00%	7.50%
	B Exit%	60.00%	94.94%	0.00%	#DIV/0!	0.00%
	WPE	0.00%	-32.73%	0.00%	#DIV/0!	7.50%
						
OPT 4						
Kerry	50.30%	<minimize				
Bush	49.70%					
		HighBush		High Kerry
	1250	40	415	540	165	90
	Kerry%	0.2	0.4	0.5	0.6	0.8
0.53	Resp	89.7%	57.3%	50.9%	46.6%	41.2%
1.06	alpha	1.79	1.15	1.02	0.93	0.82
662	Resp	36	238	275	77	37
329	K Resp 	13	109	140	43	24
333	B Resp	23	129	135	34	13
300	K Ref	1	71	133	53	42
288	B Ref	3	106	133	35	11
						
629	TotalK 14	180	272	96	67
621	TotalB 26	235	268	69	23
	B Pct%	65.8%	56.6%	49.6%	41.9%	25.8%
	B Exit%	64.1%	54.1%	49.1%	44.1%	34.1%
	WPE	1.6%	2.5%	0.4%	-2.2%	-8.3%

						
						

OPT 5						
Kerry	54.97%	< maximize; Response rates [30%,70%]				
Bush	45.03%					
		High Bush		High Kerry
	1250	40	415	540	165	90
	Kerry%	20%	40%	50%	60%	80%
0.53	Resp	30.0%	30.0%	70.0%	55.6%	62.5%
1.06	alpha	0.60	0.60	1.40	1.11	1.25
663	Resp	12	125	378	92	56
413	K Resp 	1	30	265	61	56
249	B Resp	11	95	113	31	0
274	K Ref	6	116	81	44	27
314	B Ref	22	174	81	29	7
						
687	TotalK 7	146	346	105	83
563	TotalB 33	269	194	60	7
	B Pct%	82.4%	64.8%	36.0%	36.3%	7.5%
	B Exit%	88.0%	76.0%	30.0%	33.3%	0.0%
	WPE	-5.6%	-11.2%	6.0%	3.0%	7.5%


OPT 6						
Kerry	50.35%					
Bush	49.65%	< maximize; Response rates [30%,70%]				
		High Bush		High Kerry
	1250	40	415	540	165	90
	Kerry%	20%	40%	50%	60%	80%
0.53	Resp	70.0%	58.1%	51.5%	47.1%	41.6%
1.06	alpha	1.40	1.16	1.03	0.94	0.83
663	Resp	28	241	278	78	37
332	K Resp 	8	112	143	44	25
330	B Resp	20	129	135	34	13
297	K Ref	2	69	131	52	42
290	B Ref	10	104	131	35	11
						
629	TotalK 10	182	274	96	67
621	TotalB 30	233	266	69	23
	B Pct%	74.4%	56.2%	49.2%	41.6%	25.6%
	B Exit%	72.0%	53.5%	48.5%	43.5%	33.5%
	WPE	2.4%	2.7%	0.7%	-1.8%	-7.9%

OPT 7						
Kerry	54.19%	< maximize; Response rates [30%,70%]			
Bush	45.81%					
		High Bush		High Kerry
	1250	40	415	540	165	90
	Kerry%	10%	30%	50%	70%	90%
0.53	Resp	30.0%	30.0%	66.8%	70.0%	55.6%
1.06	alpha	0.60	0.60	1.34	1.40	1.11
662	Resp	12	125	360	116	50
427	K Resp 	1	22	241	113	50
236	B Resp	11	102	120	2	0
250	K Ref	3	87	90	35	36
337	B Ref	25	203	90	15	4
						
677	TotalK 4	110	330	148	86
573	TotalB 36	305	210	17	4
	B Pct%	91.2%	73.6%	38.8%	10.4%	4.4%
	B Exit%	94.0%	82.0%	33.2%	2.0%	0.0%
	WPE	-2.8%	-8.4%	5.6%	8.4%	4.4%

						
						
OPT 8		Requires high Bush response in Kerry precincts				
Kerry	48.08%					
Bush	51.92%	< maximize; Response rates [30%,70%]				
		High Bush		High Kerry
	1250	40	415	540	165	90
	Kerry%	10%	30%	50%	70%	90%
0.53	Resp	70.0%	64.5%	48.7%	41.9%	38.2%
1.06	alpha	1.40	1.29	0.97	0.84	0.76
663	Resp	28	268	263	69	34
300	K Resp 	4	104	128	41	24
362	B Resp	24	164	135	29	11
301	K Ref	1	44	138	67	50
287	B Ref	11	103	138	29	6
						
601	TotalK 5	148	267	108	74
649	TotalB 35	267	273	57	16
	B Pct%	87.2%	64.4%	50.6%	34.7%	18.1%
	B Exit%	86.0%	61.3%	51.3%	41.3%	31.3%
	WPE	1.2%	3.1%	-0.7%	-6.5%	-13.2%

Printer Friendly | Permalink |  | Top
 
DU AdBot (1000+ posts) Click to send private message to this author Click to view 
this author's profile Click to add 
this author to your buddy list Click to add 
this author to your Ignore list Mon May 06th 2024, 01:59 AM
Response to Original message
Advertisements [?]
 Top

Home » Discuss » Topic Forums » Election Reform Donate to DU

Powered by DCForum+ Version 1.1 Copyright 1997-2002 DCScripts.com
Software has been extensively modified by the DU administrators


Important Notices: By participating on this discussion board, visitors agree to abide by the rules outlined on our Rules page. Messages posted on the Democratic Underground Discussion Forums are the opinions of the individuals who post them, and do not necessarily represent the opinions of Democratic Underground, LLC.

Home  |  Discussion Forums  |  Journals |  Store  |  Donate

About DU  |  Contact Us  |  Privacy Policy

Got a message for Democratic Underground? Click here to send us a message.

© 2001 - 2011 Democratic Underground, LLC