Acquiescence Bias Or a Perceived Middle Ground MengLi
Acquiescence Bias? Or a Perceived Middle Ground? Meng-Li Yang Academia Sinica, Taiwan
Agree-Agree Answers to Opposite Assertions • Acquiescence: agree-agree answers to opposite assertions • Acquiescence is considered to be a bias, reducing validity of the measure. • It was estimated that about 10% of respondents were acquiescent (Krosnick & Presser 2010). • Factors related to acquiescence: the social norm to be polite and agreeable (assumed only) people of lower social class deferring to higher class **lower or not using cognitive ability—lower education, fatigue, satisficing personality traits of compliance or agreeableness?
Bias or Not? • Acquiescence was a continuing force among respondents with a college education. • Schuman and Presser (1981) called for a more complex model of acquiescence. • Acquiescence is often measured under disguise and is always assumed to be a bias without further exploring. • Voluntary remarks of several respondents to a CATI survey in 2018 suggested that they had a reason for giving agree answers to opposite assertions.
Perceived Middle Ground • Two assertions could be compatible under certain conditions or through some mechanism– a perceived middle ground. • Ex. Innocent people may be wrongly executed if retaining the death penalty. versus Criminals who killed innocent people might kill again if released from jail. So the death penalty should be retained in case of necessity. Possible PMGs—supportive systems, e. g. , better criminal justice system, life imprisonment without parole
This study • Asks respondents to explain for agree-agree answers • Explores three possible sources in agree-agree answers: acquiescence, error and a perceived middle ground • Explores the relationships of altruism and compliance (facets of the agreeableness personality factor) with acquiescence • Conducted a web-based survey. • Respondents were on-line survey panel managed by our survey research center. • The panel had 2079 members. 1252 returned a completed questionnaire (RR=60. 2%). (19 partially completed)
Procedures • Instruction: Many people have different opinions on the following issues. We would like to know your attitudes towards the opinions. • Respondents received 5 pairs of opposite assertions. Each is followed by an opposite argument. Answered on a 4 -point scale plus a DK option. • After all pairs were answered, agree-agree answers were probed. Response options included (1) I clicked on a wrong answer. (2) Both sounded sensible. (3) Other (please give your opinion). (1) Choose again (3) Write in a space
Questions Used Death 1. Innocent people may be wrongly executed if retaining the death penalty. 2. Some criminals killed innocent people. They might kill again if released from jail. So, the death penalty should be retained in case of necessity. Regulations 1. The government should make stricter laws requiring industries to protect the natural environment of Taiwan. 2. The government should loosen laws on environmental protection, so that industries may reduce production cost and the economy develops faster.
Questions Used Relationship 1. To protect the environment, Taiwan needs economic growth. 2. Economic growth necessarily does harm to the environment. Medication 1. Drugs are poison. Avoid them as possible. Taking drugs is the last choice. 2. If drugs can make one feel better, it would be foolish to suffer and not to take them. Technology 1. As technologies continually develop, they will solve all the problems that humans currently face. 2. As technologies continually develop, they will bring about more unsolvable problems.
Text Coding • A coder coded if there is a perceived middle ground. • I looked at the codes and made corrections. • Inter-coder reliability—Cohen’s Kappa values at least higher than 0. 40 (Fleiss 1981). Kappa for this study ranged from 0. 50 to 0. 93. Examples of No perceived middle ground: DK, paraphrasing, elaborations, both important in real life, philosophical considerations
Perceived Middle Grounds • Death (25 /216, 19. 8%; wrong answer (error)=14) life imprisonment without parole, better criminal justice system, better correction programs • Regulations (6/51, 11. 5%; error = 9) Make regulations that are environment-friendly but also reasonable for industries to. Tighten up regulations as industries grow. (gold medium) • Relationship (61/272, 25%; error = 20) Develop alternative industries that do minimum harm to the environment. Minimize harm to the environment while developing the economy. Use gains or new technology from development to amend the harm. Only when economy is developed to a certain extent will the people value, or have the ability to protect, their environment. (alternative, gold medium, mutually dependent)
Perceived Middle Grounds • Medication (71/247, 28. 7%; error = 10) Dependent on the seriousness of the illness or whether the symptoms were bearable; the life quality of people suffering from illnesses; side effects and addictiveness of the medication. • Technology (6/179, 3. 4%; error = 12) Both true but people would face fewer and fewer problems. Effects depend on humans’ attitudes or relevant regulations. Observations: 1. Few people claimed error for AA. 2. PMG possibility depends on issue difficulty
Results Death Penalty Regulations Relationship Medication Technology Non-substantive answers 70 35 257 a 108 123 Analytical cases 1182 1217 995 1144 1129 126 (10. 7%) b 52 (4. 3% ) b 272 (27. 3%) b 247 (21. 6%) 179 b (15. 9%) b Clicked a wrong answer 14 (11. 1%)c 9 (17. 3%) c 20 (7. 4%) c 10 (4. 1%) c 12 (6. 7%) c Both sounded reasonable 73 (59. 4%) c 32 (61. 5%) c 138 (50. 7%) c 146 (59. 1%) 110 (61. 5%) c Texts indicating acquiescence 14 (11. 1%) c 5 (9. 6%) c 46 (16. 9%) c 20 (8. 1%) c 51 (28. 5%) c A perceived middle ground 25 (19. 8%) c 6 (11. 5%) c 68 (25%) c 71 (28. 7%) c 6 (3. 4%) c Agree-agree
Distribution of Agree-Agree (AA) Answers AA Sum of real acquiescent answers (purged) 0 0 637 (50. 9) 1 0 2 0 3 0 4 0 5 0 1 129 (10. 3) 284 (22. 7) 0 0 0 2 18 (1. 4) 45 (3. 6) 89 (7. 1) 0 3 2 (0. 2) 6 (0. 5) 11 (0. 9) 4 0 0 5 0 786 (62. 8) Total Sum of PMG 637 (50. 9) 0 637 (50. 9) 1 0 2 0 0 413 (33. 0) 311 (24. 8) 102 (8. 2) 0 0 0 152 (12. 1) 105 (8. 4) 36 (2. 9) 11 (0. 9) 24 (1. 9) 0 0 43 (3. 4) 31 (2. 5) 8 (0. 6) 4 (0. 3) 0 2 (0. 2) 3 (0. 2) 0 5 (0. 4) 0 0 0 1 (0. 1) 2 (0. 2) 0 0 335 (26. 8) 100 (8. 0) 26 (2. 1) 4 (0. 3) 1 (0. 1) 1252 1091 (87. 1) 146 (11. 7) 15 (1. 2)
Personality items (7 -point scale) 1. Some people think that I am stingy and calculating. (altruism, Generous) 2. I often agree with others’ opinions simply to please them. (compliance, Pleasing) 3. I help others even if that takes extra efforts or time. (altruism, Helpful) 4. My family and friends often get into argument with me for all kinds of things. (compliance, Non-arguing) 5. I try my best to be thoughtful and polite. (compliance, Polite) 6. I easily feel compassion for others. (altruism, Compassionate) 7. Once I have decided on something, it is very hard for others to change my mind. (compliance, Cooperative) 8. If I feel offended by people, I will let them know. (compliance, Tolerant) 9. I am willing to have others tell me what to do. (compliance, Submissive)
Results • Higher scores indicate higher compliance or altruism. • Items on Compliance have low alpha (=0. 29); • Items on Altruism alpha= 0. 49. • Used individual items predict each issue (logistic regression), sum of AA, & sum of real acquiescence (purged) (ordinal regression). Controls: gender, education, age, marital status. • Trait Item is consistent in the sign when significant. • Pleasing and Non-arguing are most predictive (4/5), but have opposite signs. • Items on altruism are not predictive.
Significance of Coefficients from Regressions Death Regulations Relationship Medication Technology Sum Purge AA Purge AA Pleasing ns ns s+ s+ m+ ns s+ s+ s+ Non-arguing m- s- s- ns ns s- s- Polite ns s- ns ns ns Cooperative ns ns ns Tolerant ns ns s- s- Submissive m+ m+ ns ns ns s+ m+ s+ s+ s+ Generous ns m- s- s- ns ns m- ns ns ns Helpful ns ns ns
Conclusion • Schuman and Presser (1981): Broad generalizations (e. g. , Medication) are in reality impossible questions, and are susceptible to agree-agree answers. Indeed, many PMGs were observed. • Even some concrete issues (e. g. , the death penalty) do have a middle ground. • Real acquiescence (purged) is around 70% to 90% of AA answers. • The relationships of personality traits with AA are close to those with real acquiescence (purged).
- Slides: 17