Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
5. The Structuring of Beliefs. Are Many Opinions Random? (re Converse) GV917 Political Knowledge One way of examining if people reply to questions at random because they are illinformed is to ask them knowledge questions Political Knowledge questions indicate the extent to which people are actually ignorant about politics – and hence may have nonattitudes A battery of Knowledge questions were asked in the 2001 British Election Study Knowledge Questions in the 2001 British Election Study True False Don’t Know Polling Stations close at 10pm on Election Day 82 7 10 The Conservative Party policy is that Britain should never join the Euro 32 46 22 The Liberal Democrats favour a system of proportional representation for Westminster Elections 60 5 35 The minimum voting age in Britain is 16 10 85 5 Unemployment has fallen since Labour was elected in 1997 77 13 10 94 3 Only taxpayers are allowed to vote in general 3 elections Knowledge Questions in the 2001 BES On average 74 per cent of respondents gave the correct answer They were much more likely to give the correct answers about electoral arrangements (87%) than about party policies (61%) However, there is not much evidence of complete ignorance of politics in the data – random guessing would produce a Normal Distribution centred on 3 correct answers The Distribution of Political Knowledge in 2001 (Mean = 4.4) Political Knowledge, Age and Sex in 2001 Age Knowledge Scores Up to 24 3.9 25-34 4.3 35-44 4.6 45-54 4.7 55-64 4.4 Sex Male 4.7 Female 4.2 Political Knowledge and Occupational Status in 2001 Occupational Status Knowledge Scores Professionals and Managers 5.2 White Collar Middle Management 4.9 White Collar Clerical 4.6 Skilled Manual 4.3 Semi-Skilled Manual 4.0 Unskilled Manual 3.8 Results Political Knowledge is not that high – with a mean score of 4.4. On the other hand it is far from being random, since the distribution of scores is skewed heavily to the high end Some people undoubtedly guessed some of the answers – but only 14 per cent got a score of 3.0 – which is expected if people are guessing Another Way of Detecting Non-Attitudes – Asking Questions about Non-Existent Issues Patrick Sturgis and Patten Smith asked questions about two fictitious issues (See ‘Fictitious Issues Revisited: Political Interest, Knowledge and the Generation of non-attitudes’ Political Studies, 2010, vol 58: 66-84) In their research they wanted to investigate the extent to which individuals would express an opinion about two non-existent Bills before Parliament: the ‘Monetary Control Bill’ and the ‘Agricultural Trade Bill’. They were interested in finding out the extent to which people who were politically knowledgeable and politically interested would do this and also if there were any question order effects in surveys containing these questions. Responses to Two Fictitious Issues in a Survey (N=1,024) Responses Monetary Control Bill % Agricultural Trade % Strongly Support 0.8 0.4 Tend to Support 9.7 5.2 Neither Support nor Oppose 26.1 20.7 Tend to Oppose 3.6 4.1 Strongly Oppose 1.2 1.3 Never Heard of it 45.6 55.6 Don’t Know 13.0 12.8 Results Just over 15 per cent of respondents expressed an opinion on the Monetary Control Bill, and just over 11 per cent on the Agricultural Trade Bill (if we count ‘Neithers’, ‘Never Heard of it’ and Don’t know as non-opinions) The research showed that respondents who are politically knowledgeable and interested in politics are less likely to respond fictitiously There are question order effects too – respondents are less likely to express an interest in politics if they have just been asked about a fictitious issue – the latter appears to put them off. Are Attitudes Stable Over Time? Attitude Stability over time – the case of British Membership of the Euro in 2001 How Much Do Attitudes to the Euro Change over Time? This is tricky for most people because it is a complex issue and experts disagree This became an election issue in 2001 with the Conservatives campaigning on a platform of ‘Save the Pound’ The Liberal Democrats campaigned to join the Euro more or less right away Labour campaigned on a promise to join if the conditions were right Attitudes to Joining the Euro just before the General Election of 2001 – pre election survey aq23 Attitude Towards Joining Euro Currency Valid Missing Total Frequency 1 Definitely join 173 2 Join if conditions right 853 3 Out for at least 4/5 507 years 4 Rule out on principle 624 8 Don't know 141 Total 2297 9 Refused 5 2303 Percent 7.5 37.1 Valid Percent 7.5 37.1 Cumulative Percent 7.5 44.7 22.0 22.1 66.7 27.1 6.1 99.8 .2 100.0 27.1 6.1 100.0 93.9 100.0 Attitudes to Joining the Euro just after the General Election of 2001- post election survey bq35 Attitude Toward Joining Euro Currency Valid Missing Total Frequency 1 Definitely join 137 2 Join if conditions right 824 3 Out for at least 4/5 605 years 4 Rule out on principle 597 8 Don't know 138 Total 2301 9 Refused 2 2303 Percent 5.9 35.8 Valid Percent 5.9 35.8 Cumulative Percent 5.9 41.7 26.3 26.3 68.0 25.9 6.0 99.9 .1 100.0 26.0 6.0 100.0 94.0 100.0 Changes in Attitudes to the Euro during the election of 2001 There were clearly some changes in attitudes to British membership of the Euro which occurred during the campaign 45 per cent said join immediately or if the conditions are right before the election, and 42 per cent said this after the election But these are the net figures and hide a lot of gross turnover in opinion Stability of Attitudes over Time – The Row percentages in a Table Attitude Towards Britain Joining the Euro - Crosstabulation of Pre and Post Election Survey % within aq23 Attitude Towards Joining Euro Currency aq23 Attitude Towards Joining Euro Currency Total bq35 Attitude Toward Joining Euro Currency 2 Join if 3 Out for 1 Definitely conditions at least 4 Rule out join right 4/5 years on principle 8 Don't know 1 Definitely join 43.9% 39.3% 8.1% 3.5% 5.2% 2 Join if conditions right 4.7% 63.5% 22.5% 5.8% 3.5% 3 Out for at least 4/5 1.6% 25.2% 44.3% 25.6% 3.3% years 4 Rule out on principle .6% 7.7% 24.9% 61.4% 5.3% 8 Don't know 5.6% 26.8% 11.3% 21.1% 35.2% 5.9% 35.8% 26.2% 26.0% 6.1% Total 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% Findings? Only 44 per cent of the who supported joining before the election also supported this policy after the election 64 per cent were consistent in saying that Britain should join if the conditions are right 44 per cent were consistent in waiting for 4 to 5 years 61 per cent consistently ruled out membership in principle The Extent to which people were consistent over time – the total percentages in a table aq23 Attitude Towards Joining Euro Currency * bq35 Attitude Toward Joining Euro Currency Crosstabulation % of Total aq23 Attitude Towards Joining Euro Currency Total bq35 Attitude Toward Joining Euro Currency 2 Join if 3 Out for 1 Definitely conditions at least 4 Rule out join right 4/5 years on principle 8 Don't know 1 Definitely join 3.3% 3.0% .6% .3% .4% 2 Join if conditions right 1.7% 23.6% 8.4% 2.1% 1.3% 3 Out for at least 4/5 .3% 5.6% 9.8% 5.7% .7% years 4 Rule out on principle .2% 2.1% 6.7% 16.6% 1.4% 8 Don't know .3% 1.7% .7% 1.3% 2.2% 5.9% 35.8% 26.2% 26.0% 6.1% Total 7.5% 37.1% 22.1% 27.1% 6.2% 100.0% Findings? 55 per cent of respondents remained consistent between the two waves of the panel survey – 45 per cent were inconsistent We see this from the main diagonal of the table But ‘Definitely Join’ and ‘Join if the conditions are right’ are not that different. Are people really being inconsistent if they move from one of these categories to the other? How Can this be explained? One explanation is that there are a lot on non-attitudes and these are the 45 per cent of respondents who change their minds Another explanation is that there are some non-attitudes, but measurement error accounts for a lot of this. One example, of measurement error is people shifting one category between waves of the survey – they are not responding randomly, but rather probabilistically. A Recoded Version of the Table Europre * Europost Crosstabulation % of Total Europos t Europre Total 1.00 Join 2.00 Wait 3.00 Dont Join 1.00 Join 35.1% 6.6% 2.5% 44.2% 2.00 Wait 10.0% 10.9% 7.5% 28.4% 3.00 Dont Join 2.7% 6.3% 18.5% 27.4% Total 47.7% 23.8% 28.5% 100.0% Probabilistic Responses In the revised version of the table 65 per cent are consistent rather than 55 per cent The really inconsistent people are those who move from join to don’t join. These make up only just over 5 per cent of the sample Some may be answering at random, but others may have genuinely changed their minds in response to the campaign Conclusions Some people clearly do answer at random, since switching right across the spectrum is clearly not likely to be a rational decision except for the small number who changed their minds in the campaign However, switching between adjacent categories is not so unlikely. So when measurement error is taken into account the “randomizers” may not be that large a group Knowledge of politics is limited, but it is clearly a long way from being zero with people just guessing the answers In addition ‘useful’ knowledge such as when the polling station closes is quite high. Knowledge of party policies may be less useful because it is less important to people An Alternative - Collective Opinions (see Page and Shapiro) They argue that attitudes are neither perfectly formed or random, but are derived from some underlying values and beliefs New information will push the individual’s preferences back and forth over time – but this might be the product of uncertainty as much as randomness Each individual will have a true long-term preference which can be observed by sampling people over time Collective Opinions (see Page and Shapiro) If this is true then at any given moment the public as a whole will have real collective attitudes – but it is defined on average Random deviations will occur away from the average for any one individual – but collectively the public will have a ‘rational’ ie. Meaningful opinion. This averaging out process results from the law of large numbers Condorcet ‘Jury’ Theorem More than two centuries ago Condorcet demonstrated that if a number of individuals tried to answer a factual question and each had a better than 50 % chance of being correct, then a collective decision by majority vote had a much better chance of being correct than a decision by a single person Suppose an individual has a 60 % chance of being correct (and a 40 % chance of being wrong) Suppose we add two more individuals to the ‘jury’ What are their chances of being correct by majority voting? Condorcet ‘Jury’ Theorem Note that if at least two of them vote for the 0.60 branches, they will come to the right decision. So probability of a correct decision under majority rule = 0.216+0.144+0.144+0.144=0.648 – the probability of the jury being correct has increased by 0.048 when three people are on it compared with only one. 0.6 0.216 0.6 0.144 0.144 0.6 0.096 0.144 0.4 0.096 0.4 0.096 0.4 0.064