Political ignorance & policy preference Eric Crampton University of Canterbury
What do we know? Know US House majority party 69% know it Can name their member of Congress 46% know it Know term of House is years 30% know it Can name one of their state senators 8% know it Source: Dye and Zeigler
Why should we care? Ignorance doesn t matter if: Errors are unbiased Effective cueing mechanisms exist (Lupia & McCubbins) Voter can correctly perceive a third party to have common interests and necessary knowledge OR Can use external factors to substitute for lack of knowledge about third party Democracy is a pathetic belief in the collective wisdom of individual ignorance. H.L. Mencken
Is ignorance unbiased? If political ignorance has predictive power after controlling for demographics, it s not unbiased. If the effect of ignorance isn t attenuated by cue availability, Lupia s solution doesn t help.
Is ignorance unbiased? 7 ' '' * 7 6 ' ''' * 6 5 ' '' * 5 4 4 3 ' 3 ε γ θ θ α ε θ θ α ε θ θ α ε γ θ α ε θ α ε θ α ε α = = = = = = = Z X Econ Cue Ig Ig Vote Z X Cue Ig Ig Econ Z X Cue Ig Ig Pol Z X Econ Ig Vote Z X Ig Econ Z X Ig Pol Z X Ig
Method NZES provides five broad measures of ignorance: Left-right party positioning MMP and consistency checks Quiz questions on Parliament Who was in last government Knowing own MP and MP s party
NZ: What do we know? basic facts about Parliament Parliament 3 year term 83% Enrolling to vote is compulsory 68% Non-citizens can vote 8% Can name own MP and Party 56% Party vote most important? 56% Representation threshold? 56% MMP more proportional than FPTP? 54%
What don t we know? who formed the last govt Labour 83% Green 4% Progressive 39% United Future 7% Labour/Progressive 38% Labour/Progressive/UF 9%
NZ: What do we know? Ideology Can place both National and Labour on a left-right axis: 79% Correctly? 8% (64% of overall sample) Can place Nat, Lab, and UF: 64% Correctly? 95% (60% of overall sample) Extracted from New Zealand Election Survey 005
Method Generate each ignorance variable, adding one point for each incorrect answer Aggregate variables by: Sum Sum of z-scores Principal component analysis
Ignorance measures 500 000 500 000 500 0 Ideology MMP Quiz Last Govt MP 0 3 4 5 6
Explaining ignorance highlights Interest in politics, seeking out news reduce ignorance; newspaper and internet most effective Ignorance decreasing in age University education and farming each reduce ignorance by about half s.d. Gender, ethnicity, income matter Not placing self on right-left index, rightwing ideology both increase ignorance.
Table : Correlates of ignorance Lack of interest in politics 0.7 (=very interested; 4= not at all) [5.0]** Seek out news about politics? -0. (0=no, =yes) [.56]* Pay no attention to political news 0.5 (0=no, =yes) [4.4]** Have internet -0. (0=no, =yes) [3.3]** Follow political news in newspaper -0. (0=no, =yes) [.65] Follow political news in newspaper -0.06 (=once per week; 7=every day) [3.43]** Member community service group -0.79 (0=no, =yes) [3.7]** Age -0.03 (years) [5.69]** gender of respondent -0.8 (0=female, =male) [7.9]** Some tertiary education -0.33 (0=no, =yes) [3.8]** University educated -0.387 (0=no, =yes) [8.49]** Technical occupation -0.94 (0=no, =yes, manual labour omitted) [4.44]** Clerical occupation -0.38 (0=no, =yes, manual labour omitted) [4.83]** Service occupation -0.8 (0=no, =yes, manual labour omitted) [3.43]** Farming occupation -0.4 (0=no, =yes, manual labour omitted) [4.56]** Parents expressed political preference -0.47 (0=no, =yes) [4.65]** European ethnicity -0.05 (0=no, =yes) [.49]* Maori ethnicity 0.406 (0=no, =yes) [9.84]** Household income $58,900-$87,599-0.05 (0=no, =yes) [.38]* Household income $87,600-$9,999-0.79 (0=no, =yes) [3.40]** Household income >$0,000-0.79 (0=no, =yes) [4.9]** partner retired -0.48 (0=no, =yes) [.94]** Opinion of previous govt performance -0.055 ( = v. good; 4=v. bad; don t know dropped) [.38]* Left-wing ideology -0.88 (0=no, =yes; omitted category don t know ) [7.85]** Centrist ideology -0.478 (0=no, =yes; omitted category don t know ) [.86]** Right-wing ideology -0.509 (0=no, =yes; omitted category don t know ) [9.90]** Dependent variable: Principal component of Ignorance Measures. 845 obs, adj R-sq. 0.47
Table 3: Ignorance and policy preference Policy variable Protect environment even if lowers income (=yes, 7=no) GM foods relatively safe (=strongly agree; 5=strongly disagree) Death penalty reinstatement (=strongly agree; 5 = strongly disagree) Tax and redistribution (=more redist; 7=less) Environmental spending (=much more; 5=much less) Health spending (=much more; 5=much less) Defence spending (=much more; 5=much less) Assist low income families (=much more; 5=much less) Reduce taxes in general (=strongly support; 5=strongly oppose) Work for unemployment benefits (=strongly agree; 5=strongly disagree) Welfare makes people lazy and dependent (=strongly agree; 5=strongly disagree) Big business has too much power (=strongly agree; 5=strongly disagree) Trade unions have too much power (=strongly agree; 5=strongly disagree) Policy Mean, (Std Dev) 3.40 (.46).93 (.09).70 (.40) 4.4 (.60).5 (0.77).86 (0.73).73 (0.97).48 (0.99).3 (0.97).8 (.0).4 (.7).5 (.03) 3.3 (0.97) Predicted Ignorance (t-stat) -.959 (.5)*.75 (.0) 0.354 (0.8) 0.70 (0.46) -.98 (.67) -0.307 (0.45).73 (.38) -0.58 (0.9) 0.07 (0.08).506 (.55).543 (.4) -0.987 (.03) 0.6 (0.66) Residual Ignorance (t-stat) 0.04 (0.37) 0.077 (.56)* -0.59 (7.59)** -0.3 (.70)** -0.044 (.5)* -0.097 (5.33)** -0.07 (.08) -0.085 (3.77)** -0.086 (3.63)** -0.09 (4.35)** -0.3 (4.48)** -0.0 (4.09)** -0.033 (.34) Left Ideology (t-stat) -.67 (3.46)** 0.69 (.55) 0.46 (0.95) -0.457 (0.86) -0.65 (.58)** -0.44 (0.60) 0.835 (.57)* -0.67 (0.90) 0.487 (.55) 0.9 (.68).0 (.66)** -0.694 (.07)* 0.547 (.64) Right Ideology (t-stat) 0.46 (.63) -0.3 (3.)** -0.73 (3.3)** 0.737 (7.43)** 0.063 (.35) 0.07 (.66) -0.76 (.97)** 0.97 (3.63)** -0.367 (6.44)** -0.73 (.84)** -0.6 (3.74)** 0.336 (5.6)** -0.8 (3.7)** Adjusted R 0. 0. 0.0 0.9 0. 0.09 0. 0.8 0.9 0.5 0.7 0. 0.
Economic thinking There should be a law to further reduce pay differences between women and men High income tax makes people less willing to work hard The government should control wages / prices by law; introduce import controls Immigration is good for the NZ economy Minimum wages reduce the creation of new jobs The government should provide a job for everyone who wants one.
Table 4: Understanding economics. Political ignorance measure (Mean zero, sd ; increasing measure) Residual ignorance measure Predicted ignorance measure -0. [8.77]** -0. [8.74]** Have internet (0=no, =yes) 0.67 [3.47]** 0.93 [4.0]** gender of respondent 0.69 0.39 (0=female, =male) [6.50]** [7.76]** Some tertiary education 0.09 0.39 (0=no, =yes; less than secondary omitted) [.4]* [3.08]** University educated 0.36 0.447 (0=no, =yes; less than secondary omitted) [6.4]** [7.70]** Home owner 0.6 0.3 (0=no, =yes) [.30]* [.56]* European ethnicity 0.35 0.6 (0=no, =yes) [.43]* [.88]** Maori ethnicity -0.335-0.4 (0=no, =yes) [6.0]** [7.8]** Household income $87,600-$9,999 0.5 0.66 (0=no, =yes) [.84] [.44]* High household income 0.4 0.483 (0=no; = income > $0,000) [5.8]** [6.69]** Opinion of previous govt performance -0.073-0.06 (=very good; 4 = very bad; don t know dropped) [.30]* [.89] Left-wing ideology -0.55-0.079 (0=no, =yes) [3.0]** [.65] Right-wing ideology 0.7 0.3 (0=no, =yes) [.]* [.3]* State of the economy -0.004-0.004 (=very good; 5 = very bad) [0.5] [0.5] Current household financial situation vs last year -0.09-0.09 (= a lot better ; 5 = a lot worse) [.5] [.5] Expectation for national economy next year 0.085 0.085 (= a lot better ; 5 = a lot worse) [3.55]** [3.55]** Absolute value of t-statistics in brackets; * sig at 5%; ** sig at %; N=33 and adjusted R-squared 0.3 in all specifications -0. [8.76]** -0.88 [0.97] 0.089 [0.75] 0. [0.58] 0.0 [0.6] 0.06 [0.30] 0.076 [.0] 0.06 [0.5] -0.075 [0.] 0.004 [0.0] 0.38 [0.9] -0.3 [.80] -0.384 [.] 0.098 [.68] -0.004 [0.5] -0.08 [.50] 0.086 [3.57]**
Correlates of economic thinking Ignorance Male University Maori ethnicity European ethnicity High income Left wing Right wing N=33; R =0.3-0. (8.77)** 0.7 (6.50)** 0.36 (6.4)** -0.34 (6.0)** 0.4 (.43)* 0.4 (5.8)** -0.6 (3.0)** 0. (.)**
Ignorance and Voting Behavior Dependent Variable Ignorance coefficient (z-statistic) Not voting.4% (0=voted in 005; = didn t) (5.0)** Voted Labour 3% (.0)* Voted National 0.9% (0.80) Econ Index coefficient (z-stat) 0.3% (.40) -.5% (.08)* 4.5% (4.63)** Left wing ideology (z-stat) -0.7% (.33) 0.6% (7.4)** -.% (0.75)** Rightwing (z-stat) -.5% (.36)* -.6% (6.3)** 9.7% (7.4)** Pseudo R 0.4 0.5 0.44 Voted Green -.% (3.4)** -0.% (0.44) 6% (6.44)** -.6% (.6)** 0.3 Voted ACT 0% (0.66) -0.% (0.07) (dropped) 0.3% (0.89) 0.4 Voted NZ First -0.% (0.4) -% (.4)* -3.% (3.73)** -.9% (.4)* 0. Voted United Future -0.3% (.90)** -0.% (0.98) -0.4% (.40)* -0.4% (.45)* 0.3 Voted Maori -0.3% (.50) -0.3% (.04)* 0.4% (0.9) -0.4% (.05) 0.40
Cueing Results 9 specifications cases where cueing group membership augments ignorance significant 7 cases where cueing group membership attenuates ignorance significant
Discussion Political ignorance matters Cueing helps little Ignorance itself as a demographic characteristic with particular interests?