KIND OF POLLSInternet PollsSamples obtained via subscribed respondents or advertisingBetter response ratesRelatively CheapAvoid interviewer biasEasy to allow for and code open-ended responses to questionsAre the representative? Considerable debateProbably the future of pollingHOW POLLING RELATES TO VOTINGPolling allows us to predict/explain election results wellElection outcomes are usually very similar to contemporaneous pollingSome common polling questions are particularly useful:Presidential Approval:“Do you approve or disapprove of the way Barack Obama is handling his job as President?”Strongly related to vote choicePresidents much below 50% rarely win re-electionPresidents about 50% rarely lose re-electionsCommonly asked on surveysEasy break down by stateEasy to study over timeThe generic Congressional Ballot:“If the election were held today, would you vote for the Republican or Democrat in your district/state?”Allows comparisons of support for members of congress by party across districts and statesVery helpful for forecasting/explaining aggregate Congressional election resultsWhen DEM or GOP (REP) are ahead by more than a few % points, this can indicate an impending wave election (The republicans/democrats are really going to do well this year based on who’s ahead)Head-to-Head Questions:Similar structure for asking about House & Senate racesAsks respondents to choose b/t competing candidates (For president, congress, etc)Also good at predicting resultsBUT voters must know candidates (which is not always the case with unknown challenges in House/Senate races) for them to be usefulHow the undecided vote can determine election outcomes, & sometimes these people are a sizably important groupCHALLENGES AND PROBLEMS IN POLLINGNonattitudesA specific survey response given when the respondent actually doesn’t have an opinionCan occur when respondents don’t want to say “don’t know” or “no opinion” b/c they don’t want to appear ignorantCan be a problem, especially if people are asked about very complicated policies they don’t understand or haven’t heard of, or candidates that they haven’t heard of before.Defining the CHOICE SetHow should you decide the choices that you offer to people to answer a question?Maybe using deliberation among survey designers, or in response to focus groupsBut more complicated than you might thinkISSUES TO THINK ABOUT:Do you want to offer people very different choices, that aren’t designed to be ranked or scaled, or those that are?What are the commonly held positions on this issues? (Not always obvious, think of abortion)Do we want to allow open ended responses?Looking at Sub-Groups and size of Sub-groupsMost polls have b/t 500 & 2,000 respondents, with an average around 1,000This provides sufficient sample size for inference for the target population as a wholeNot always the case if you want to look at a sub-group within the target populationEXAMPLE:You have a 1,000 person survey of all American adults with lots of questions about politics and want to examine the relationship b/t foreign policy views, and partisanship among Jewish AmericansAssuming you have a representative sample, you probably only have 20-25 jews (jews are 2.1 % of the American population)You cant do a very advance statistical analysis with only 20-25 jewsWord Choice and Survey PretestingHow you word surveys matter A LOT, even with small changesIt’s not always obvious how in advanceYou have to be careful not to bias your results with word choiceTo be sure of any uncertain working your survey, you should pre-test itUsing survey experiments or in a computer labARE THERE MEDIA EFFECTS ON VOTING?For a long time, there was a debate in POLS over whether the media mattered in influencing voting and election outcomesIt can be challenging to study media effects, so this debate wasn’t resolved for awhile:This is partially b/c the best tools for studying media effects are lab experiments and content analysis, but these tools weren’t common 20 years agoToday, they are commonLet me explain how media lab experiments (story of pens vs hand sanitizer) and content analysis work.So does media matter for voting and elections? YES!How does the media matter?PRIMES – the issues that effect vote choiceHere “Primes” meaning “puts these ideas at the top of peoples heads/on the tips of their tongues, so that they are thinking about & recall these”Sets the issue agenda of what the candidates compete overMay influence answers to “most important national problem” questionsAnswers to this question are often related to election outcomesFRAMES ISSUES, choosing how they are portrayedAffects what political facts people knowThe media also affects how candidates are portrayedEx: AL GORE was boring, George W. Bush was stupid, Obama was/is coolThis can be good/bad for candidatesIf you use the media poorly as a candidate , it can wreck your campaignEX: 2010, DE Senate Candidate Christine O’Donnell(REP), who poorly confronted accusations that she had once experimented with witchcraftMedia portrayals are not necessarily inaccurateBill Clinton actually did stop unannounced in fast food places and talked policy in ways regular people could understandThe Media AlsoShapes how voters view the economyAffects Policy mood on racial policyCreated linkage b/t race and welfare policies in public mindMay have some effect on voter preferences due to biased reportingYou hear lots of talk lamenting negative ads in mediaAre they really so bad? (Geers view: HE likes negative ads; Negativity allows challenge of opposing claims, Negative Ads create accountability, Public needs to know bad things about candidates)GEER (2006):Positive ads are difficult to challenge (EX -> Its hard to argue a candidate does not bring “New Leadership)Negative Ads usually need to be backed up with facts to be credibleAs a result, negative ads bring more info to voters than positive adsPublic hates negative ADSBUT while people often don’t like negative ads in general, usually what they think of when they say they don’t like them is negative character attacks, not negative issue adsThe % of negative character ads in presidential campaigns has not increased over timeThere are more negative issue ads now b/c the parties are more polarizedNegative issue ads are valuable b/c they help draw distinctions and give the public a choicePOLS206: 8.5 LECTURE 18KIND OF POLLS-Internet PollsoSamples obtained via