Two recent surveys on the eve of the South Carolina Republican Primary reveal significant flaws such, which invalidate the results.
The principal owner of the Wall Street Journal happens to be Rupurt Murdoch. Murdoch is the principal owner of FOX News. Murdoch seems to be against the candidacy of Donald Trump. Perhaps this is why his survey uses flawed methods.
As it is, already, in THE DES MOINES REGISTER AND BLOOMBERG COMMIT POLLING FRAUD ON DECEMBER 12, 2015, FOR THE UPCOMING GOP CAUCUS and WHY EVERY PRESIDENTIAL ELECTION POLL FOR 2016 YOU SHALL SEE OR HEAR WILL BE WRONG, I have shown you significant flaws in how polling has been done, especially for nationwide polls
Let's look at two wrong polls.
The Augusta Chronicle
Claimed Result: Trump 27, Rubio 24, Cruz 19, Bush 11, Kasich 7, Carson 8
Claimed Winner Margin: Trump +3
Status: Invalid survey results
Why?: The sample is not drawn at random from telephone numbers who then get qualified first as adults, then as likely voters. Assertion of definitely voting and exclusion based on lack of recall of polling station tainted sample.
Flawed Methodology: Likely Republican primary voters were selected at random from a list of registered voters. Only voters determined likely to attend the 2016 South Carolina Republican primary were included in the sample.
Two questions served as selection variables. The first question asked whether the respondent intends to attend the primary on February 20th; all responses other than “definitely” or “voted absentee” were excluded from the dataset. The
second question asked whether the respondent knew where their primary will be held (e.g. church, school); all responses of “I don’t know” (for those who did not vote absentee) were excluded from the dataset.
NBC/WSJ/Marist
Claimed Result: Trump 28, Rubio 15, Cruz 23, Bush 13, Kasich 9, Carson 9
Claimed Winner Margin: Trump +5
Status: Invalid survey results
Why?: The sample is not drawn from the first adult contacted regardless of sex. Instead, the survey targeted the youngest male first.
As well, the randomness of the survey became further eroded when matching against another sample with embedded error.
Also the survey is spread over a few days, which calls into question voting based on different news about candidates.
Flawed Methodology:
This survey of 2,567 adults was conducted February 15th through February 17th, 2016 by The Marist Poll sponsored and funded in partnership with NBC News and The Wall Street Journal. Adults 18 years of age and older residing in the state of South Carolina were interviewed in English by telephone using live interviewers.
Landline telephone numbers were randomly selected based upon a list of telephone exchanges from throughout the state of South Carolina from ASDE Survey Sampler, Inc. The exchanges were selected to ensure that each region was represented in proportion to its population. To increase coverage, this landline sample was supplemented by respondents reached through random dialing of cell phone numbers from Survey Sampling International.
Respondents in the household were randomly selected by first asking for the youngest male.
Both samples were matched by telephone number to a voter registration list for the state. Voter information was appended to each matched case. Assistance was provided by Luce Research for data collection and L2 for voter registration information.
After the interviews were completed, the two samples were combined and balanced to reflect the 2013 American Community Survey 5-year estimates for age, gender, income, and region except for race, which is from the 2010 census.
Results are statistically significant within ±1.9 percentage points. There are 2,282 registered voters. The results for this subset are statistically significant within ±2.1 percentage points. Among registered voters, 823 had a phone match in the voter file and 1,459 did not. The results for these subsets are statistically significant within ±3.4 percentage points and ±2.6 percentage points, respectively. There are 1,227 voters in the potential Republican primary electorate. The potential Republican primary electorate in South Carolina includes all voters who prefer to vote in the Republican presidential primary and those who identify as Republicans or Republican leaning independents without a primary preference.
There are 836 voters in the potential Democratic primary electorate. The potential Democratic primary electorate in South Carolina includes all voters who prefer to vote in the Democratic presidential primary and those who identify as Democrats or Democratic leaning independents without a primary preference. The results for these subsets are statistically significant within ±2.8 percentage points and ±3.4 percentage points, respectively.
There are 722 likely Republican primary voters and 425 likely Democratic primary voters defined by a probability turnout model which determines the likelihood respondents will participate in the 2016 South Carolina Republican/Democratic Presidential Primary based upon their chance of vote, interest in the election, and past election participation. The results for these subsets are statistically significant within ±3.6 percentage points and ±4.8 percentage points, respectively. The error margin was not adjusted for sample weights and increases for cross-tabulations.
POLLING DONE RIGHT
The SC House GOP has done polling right.
Claimed Result: Trump 34, Rubio 18, Cruz 19, Bush 12, Kasich 8, Carson 5
Claimed Winner Margin: Trump +15
Why?: Respondents were contacted on the same day. The questions were asked in the same way. The calls were placed at random. The first adult contacted was the accepted contact.
Solid Methodology:
Likely Republican presidential primary voters were selected at random from a list of registered voters based on party participation and registration date. Household ages and locations were extracted as well from this list. In excess of 180,000 households were called over the duration of the poll. Each household participant was asked to opt-in to the survey if they planned to vote in the 2016 SC presidential primary.
Using automated voice response technology allows each participant to hear the questions exactly the same way, from the same voice, spoken with the same tone and nuance. Calls are placed to randomly-selected phone numbers through a stratified process that insures approrpiate geographic representation. These calls were placed from 10:30am to 8:30pm local time on Thursday, February 18th.
