Skip to content
Contents
The Public Mind

    Type to search. Planned chapters match by title only.

    Chapter 4.4

    Perceptions of the economy

    Partisans are more likely to say the national economy got better when their own party holds the White House, by 34.0 percentage points on average in the American National Election Studies from 2012 to 2024, and the two parties swap places at every change of party.

    Affiliation
    NYU Stern
    Published
    DraftFigures and numbers may change as the analysis is rebuilt.what is still being checked

    Measured unemployment and inflation are not yet in this chapter, so the low-premium surveys are read through what the parties said and through presidential approval.

    In the American National Election Studies (ANES) of 2016, with a Democrat in the White House, 45.0%45.0% percent of partisans who said the economy got betterEstimandweighted share saying betterInterval95% CI 42.3% to 47.7%n1,934Weightsweight (VCF0009z)PopulationDemocrats including leanersPeriod2016SourceANES Cumulative Data File of Democrats and 11.2%11.2% percent of partisans who said the economy got betterEstimandweighted share saying betterInterval95% CI 9.6% to 13.1%n1,726Weightsweight (VCF0009z)PopulationRepublicans including leanersPeriod2016SourceANES Cumulative Data File of Republicans said the national economy had gotten better over the past year. In 2020, with a Republican in the White House, the shares were 5.2%5.2% percent of partisans who said the economy got betterEstimandweighted share saying betterInterval95% CI 4.3% to 6.3%n3,829Weightsweight (VCF0009z)PopulationDemocrats including leanersPeriod2020SourceANES Cumulative Data File and 36.2%36.2% percent of partisans who said the economy got betterEstimandweighted share saying betterInterval95% CI 34.1% to 38.4%n3,425Weightsweight (VCF0009z)PopulationRepublicans including leanersPeriod2020SourceANES Cumulative Data File. The president's party had the higher share in every ANES survey since 1980 and in every Cooperative Election Study (CES) survey since 2006. The difference between the parties, counted in pointsPercentage pointThe unit for the difference between two percentages. A change from one share to another is stated in points, not in percent, so that absolute and relative change are not confused., averaged 34.034.0 percentage pointsEstimandmean of the yearly premiums, each survey counted once; the interval covers sampling error within each survey onlyInterval95% CI 32.5 to 35.4n20,874Weightsweight (VCF0009z)PopulationANES respondents who identify with or lean toward a party and answered the itemPeriod2012 to 2024 (four surveys)SourceANES Cumulative Data File in the ANES from 2012 to 2024, and it was smaller than 1010 percentage pointsEstimandconstantWeightsnone (constant)Populationcutoff that separates the low-premium surveysPeriod1980 to 2024SourceANES Cumulative Data File points in 66 ANES survey yearsEstimandcountWeightsnone (count)PopulationANES surveys with a premium below the cutoffPeriod1980 to 2024SourceANES Cumulative Data File of the 1818 ANES survey yearsEstimandcountWeightsnone (count)PopulationANES surveys that ask the retrospective economy itemPeriod1980 to 2024SourceANES Cumulative Data File ANES surveys.

    The measure is one survey question. The ANES asks whether, over the past year, the nation's economy has gotten better, stayed about the same or gotten worse. The CES asks the same question with five answers, and this chapter counts somewhat better and much better as better. A respondent is a partisan if they name a party or say they lean toward one (leanersPartisan leanerA respondent who first says they are independent but, when asked, says they lean toward one party. Most figures group leaners with the party they lean toward, and each chapter says which grouping it uses. are included), and pure independents are left out. The in-party premium, called the premium below, is the share of the president's party who say better minus the share of the other party who say better, so it is signed toward the president's party. The gap is the Democratic share minus the Republican share, so it is signed by party. The analysis uses 42,86442,864 respondentsEstimandcount of respondents in the yearly premium cellsWeightsnone (count)PopulationANES respondents who identify with or lean toward a party and answered the itemPeriod1980 to 2024SourceANES Cumulative Data File ANES partisans, 568,229568,229 respondentsEstimandcount of respondents in the yearly premium cellsWeightsnone (count)PopulationCES respondents who identify with or lean toward a party and answered the itemPeriod2006 to 2024SourceCooperative Election Study CES partisans and, for the weekly figure, 324,560324,560 respondent-weeksEstimandcount of respondents in the weekly party cellsWeightsnone (count)PopulationNationscape respondents who identify as Democrats or Republicans and answered the itemPeriodJuly 2019 to January 2021SourceNationscape respondent-weeks in 8080 weekly wavesEstimandcountWeightsnone (count)PopulationNationscape waves with 500 or more Democrats and 500 or more RepublicansPeriodJuly 2019 to January 2021SourceNationscape waves of the Nationscape survey.Every estimate is weighted with the survey weightsSurvey weightsNumbers that scale each respondent so the sample matches the population on known traits such as age, race, education and region. A weighted average counts a respondent with a large weight for more than one with a small weight. of its survey. Each interval is a confidence intervalConfidence intervalA range around an estimate that would contain the true value in 95% of repeated samples if the sampling model held. The book reports 95% intervals unless it says otherwise. that treats the weighted sample as a simple random sample, using the Wilson formula [1] on the Kish effective sample sizeEffective sample sizeThe number of equally weighted respondents that would give the same precision as the weighted sample. It is smaller than the raw count when the weights vary a lot, and it sets how wide the confidence interval is. [2], and intervals that account for the sample design can be wider or narrower.

    Each party says the economy improved under its own president

    Figure 1 plots the two parties' answers in each ANES survey since 1980. The line for the party in the White House is above the other line in every survey, and the two lines cross at every change of party. The gap between Democrats and Republicans was +33.8 (95% CI 30.5 to 37.0)+33.8 percentage points, Democratic minus RepublicanEstimandDemocratic share saying better minus Republican share, weightedInterval95% CI 30.5 to 37.0n3,660Weightsweight (VCF0009z)PopulationANES respondents who identify with or lean toward a party and answered the itemPeriod2016SourceANES Cumulative Data File points in 2016, −31.0 (95% CI −33.3 to −28.7)−31.0 percentage points, Democratic minus RepublicanEstimandDemocratic share saying better minus Republican share, weightedInterval95% CI −33.3 to −28.7n7,254Weightsweight (VCF0009z)PopulationANES respondents who identify with or lean toward a party and answered the itemPeriod2020SourceANES Cumulative Data File in 2020 and +31.8 (95% CI 28.8 to 34.8)+31.8 percentage points, Democratic minus RepublicanEstimandDemocratic share saying better minus Republican share, weightedInterval95% CI 28.8 to 34.8n4,877Weightsweight (VCF0009z)PopulationANES respondents who identify with or lean toward a party and answered the itemPeriod2024SourceANES Cumulative Data File in 2024, so it reversed sign at each of the last two changes of party.The gap changes sign at all 66 changes of party in the White HouseEstimandcountWeightsnone (count)Populationchanges of the president's party between two surveysPeriod1980 to 2024SourceANES Cumulative Data File changes of party in the ANES and at all 33 changes of party in the White HouseEstimandcountWeightsnone (count)Populationchanges of the president's party between two surveysPeriod2006 to 2024SourceCooperative Election Study in the CES. At 44 changes of party in the White HouseEstimandcountWeightsnone (count)Populationtransfers at which the gap before and the gap after each have an interval on the side of zero that the president's party predictsPeriod1980 to 2024SourceANES Cumulative Data File of the 66 changes of party in the White HouseEstimandcountWeightsnone (count)Populationchanges of the president's party between two surveysPeriod1980 to 2024SourceANES Cumulative Data File ANES changes and 33 changes of party in the White HouseEstimandcountWeightsnone (count)Populationtransfers at which the gap before and the gap after each have an interval on the side of zero that the president's party predictsPeriod2006 to 2024SourceCooperative Election Study of the 33 changes of party in the White HouseEstimandcountWeightsnone (count)Populationchanges of the president's party between two surveysPeriod2006 to 2024SourceCooperative Election Study CES changes, the gap before and the gap after each have an interval on the side of zero that the president's party predicts. The two ANES exceptions are the change from 1980 to 1982, where the earlier gap was +1.4 (95% CI −0.6 to 3.4)+1.4 percentage points, Democratic minus RepublicanEstimandDemocratic share saying better minus Republican share, weightedInterval95% CI −0.6 to 3.4n1,341Weightsweight (VCF0009z)PopulationANES respondents who identify with or lean toward a party and answered the itemPeriod1980SourceANES Cumulative Data File points, and the change from 1992 to 1994, where the later gap was +4.0 (95% CI −1.2 to 9.3)+4.0 percentage points, Democratic minus RepublicanEstimandDemocratic share saying better minus Republican share, weightedInterval95% CI −1.2 to 9.3n1,560Weightsweight (VCF0009z)PopulationANES respondents who identify with or lean toward a party and answered the itemPeriod1994SourceANES Cumulative Data File. Both intervals include zero. The difference between the lines has a 95% intervalConfidence intervalA range around an estimate that would contain the true value in 95% of repeated samples if the sampling model held. The book reports 95% intervals unless it says otherwise. above zero in 1515 ANES survey yearsEstimandcountWeightsnone (count)PopulationANES surveys in which the 95% interval of the premium is above zeroPeriod1980 to 2024SourceANES Cumulative Data File of the 1818 ANES survey yearsEstimandcountWeightsnone (count)PopulationANES surveys that ask the retrospective economy itemPeriod1980 to 2024SourceANES Cumulative Data File surveys. The exceptions are 1980, 1990 and 1994, where the interval includes zero, and this chapter comes back to them below. The intervals for 2002 (3.0 (95% CI 0.2 to 5.9)3.0 percentage pointsEstimandin-party share saying better minus out-party share, weightedInterval95% CI 0.2 to 5.9n1,368Weightsweight (VCF0009z)PopulationANES respondents who identify with or lean toward a party and answered the itemPeriod2002SourceANES Cumulative Data File points) and 2008 (2.2 (95% CI 0.3 to 4.0)2.2 percentage pointsEstimandin-party share saying better minus out-party share, weightedInterval95% CI 0.3 to 4.0n2,013Weightsweight (VCF0009z)PopulationANES respondents who identify with or lean toward a party and answered the itemPeriod2008SourceANES Cumulative Data File points) start close to zero. An interval that accounted for the sample design could be wider, and those two surveys would then join the exceptions.

    Figure 1

    Each party is more likely to say the economy improved when its own party holds the White House

    Share of partisans (leaners included) saying the national economy got better over the past year, percent, by party; ANES, 1980 to 2024

    Said the economy got better (% of partisans)

    Show data
    Data for Each party is more likely to say the economy improved when its own party holds the White House
    Survey yearGroupingGroupPresidentShare saying better (%)Interval lowInterval highn
    1980partyDemocratsCarter4.33.15.9818
    1980partyRepublicansCarter2.91.74.7523
    1980rolePresident's partyCarter4.33.15.9818
    1980roleOther partyCarter2.91.74.7523
    1982partyDemocratsReagan6.65.08.6758
    1982partyRepublicansReagan22.819.126.9435
    1982rolePresident's partyReagan22.819.126.9435
    1982roleOther partyReagan6.65.08.6758
    1984partyDemocratsReagan25.923.328.61,013
    1984partyRepublicansReagan63.860.667.0868
    1984rolePresident's partyReagan63.860.667.0868
    1984roleOther partyReagan25.923.328.61,013
    1986partyDemocratsReagan17.715.520.21,043
    1986partyRepublicansReagan32.329.135.7755
    1986rolePresident's partyReagan32.329.135.7755
    1986roleOther partyReagan17.715.520.21,043
    1988partyDemocratsReagan10.88.912.9920
    1988partyRepublicansReagan29.126.032.3801
    1988rolePresident's partyReagan29.126.032.3801
    1988roleOther partyReagan10.88.912.9920
    1990partyDemocratsBush3.52.54.81,000
    1990partyRepublicansBush3.72.55.3710
    1990rolePresident's partyBush3.72.55.3710
    1990roleOther partyBush3.52.54.81,000
    1992partyDemocratsBush2.41.73.51,217
    1992partyRepublicansBush8.16.510.0918
    1992rolePresident's partyBush8.16.510.0918
    1992roleOther partyBush2.41.73.51,217
    1994partyDemocratsClinton37.233.641.0819
    1994partyRepublicansClinton33.229.637.0741
    1994rolePresident's partyClinton37.233.641.0819
    1994roleOther partyClinton33.229.637.0741
    1996partyDemocratsClinton48.644.952.3887
    1996partyRepublicansClinton26.222.630.2650
    1996rolePresident's partyClinton48.644.952.3887
    1996roleOther partyClinton26.222.630.2650
    1998partyDemocratsClinton53.749.557.8647
    1998partyRepublicansClinton38.433.843.3467
    1998rolePresident's partyClinton53.749.557.8647
    1998roleOther partyClinton38.433.843.3467
    2000partyDemocratsClinton49.645.953.3879
    2000partyRepublicansClinton26.322.830.1676
    2000rolePresident's partyClinton49.645.953.3879
    2000roleOther partyClinton26.322.830.1676
    2002partyDemocratsBush3.72.35.8703
    2002partyRepublicansBush6.74.89.4665
    2002rolePresident's partyBush6.74.89.4665
    2002roleOther partyBush3.72.35.8703
    2004partyDemocratsBush10.17.713.1585
    2004partyRepublicansBush42.637.947.4479
    2004rolePresident's partyBush42.637.947.4479
    2004roleOther partyBush10.17.713.1585
    2008partyDemocratsBush1.40.82.41,362
    2008partyRepublicansBush3.52.25.6651
    2008rolePresident's partyBush3.52.25.6651
    2008roleOther partyBush1.40.82.41,362
    2012partyDemocratsObama48.846.451.33,091
    2012partyRepublicansObama9.47.911.21,992
    2012rolePresident's partyObama48.846.451.33,091
    2012roleOther partyObama9.47.911.21,992
    2016partyDemocratsObama45.042.347.71,934
    2016partyRepublicansObama11.29.613.11,726
    2016rolePresident's partyObama45.042.347.71,934
    2016roleOther partyObama11.29.613.11,726
    2020partyDemocratsTrump5.24.36.33,829
    2020partyRepublicansTrump36.234.138.43,425
    2020rolePresident's partyTrump36.234.138.43,425
    2020roleOther partyTrump5.24.36.33,829
    2024partyDemocratsBiden36.533.939.32,546
    2024partyRepublicansBiden4.73.66.22,331
    2024rolePresident's partyBiden36.533.939.32,546
    2024roleOther partyBiden4.73.66.22,331
    The line for the president's party is the higher one in every survey year, and the two lines cross at every change of party in the White House. The shaded background marks the president's party. Bands are 95% intervals. The second view groups respondents by whether they share the president's party.Note: The ANES ran no survey in 2006, 2010, 2014, 2018 or 2022, so each line joins survey years.Note: Interviews before 2012 were mainly face-to-face; web samples begin in 2012 (see the mode comparison later in the chapter).Source: American National Election Studies Cumulative Data File, 1980-2024; weighted (VCF0009z)

    The CES asks the item every year, so it shows how fast the reversal happens (Figure 2). Between 2016 and 2017 the gap moved from +43.6+43.6 percentage points, Democratic minus RepublicanEstimandDemocratic share saying better minus Republican share, weightedInterval95% CI 42.7 to 44.6n50,685Weightsweight (year-specific)PopulationCES respondents who identify with or lean toward a party and answered the itemPeriod2016SourceCooperative Election Study to −48.3−48.3 percentage points, Democratic minus RepublicanEstimandDemocratic share saying better minus Republican share, weightedInterval95% CI −50.4 to −46.2n13,851Weightsweight (year-specific)PopulationCES respondents who identify with or lean toward a party and answered the itemPeriod2017SourceCooperative Election Study points. Between 2020 and 2021 it moved from −38.7−38.7 percentage points, Democratic minus RepublicanEstimandDemocratic share saying better minus Republican share, weightedInterval95% CI −39.7 to −37.7n48,334Weightsweight (year-specific)PopulationCES respondents who identify with or lean toward a party and answered the itemPeriod2020SourceCooperative Election Study to +37.9+37.9 percentage points, Democratic minus RepublicanEstimandDemocratic share saying better minus Republican share, weightedInterval95% CI 36.4 to 39.3n19,809Weightsweight (year-specific)PopulationCES respondents who identify with or lean toward a party and answered the itemPeriod2021SourceCooperative Election Study, and between 2008 and 2009 from −1.3−1.3 percentage points, Democratic minus RepublicanEstimandDemocratic share saying better minus Republican share, weightedInterval95% CI −1.6 to −0.9n28,482Weightsweight (year-specific)PopulationCES respondents who identify with or lean toward a party and answered the itemPeriod2008SourceCooperative Election Study to +19.2+19.2 percentage points, Democratic minus RepublicanEstimandDemocratic share saying better minus Republican share, weightedInterval95% CI 17.7 to 20.7n11,426Weightsweight (year-specific)PopulationCES respondents who identify with or lean toward a party and answered the itemPeriod2009SourceCooperative Election Study. The 95% interval of the premium is above zero in 1919 CES survey yearsEstimandcountWeightsnone (count)PopulationCES surveys in which the 95% interval of the premium is above zeroPeriod2006 to 2024SourceCooperative Election Study of the 1919 CES survey yearsEstimandcountWeightsnone (count)PopulationCES surveys that ask the retrospective economy itemPeriod2006 to 2024SourceCooperative Election Study CES surveys. The 2008 gap is small because no more than 2.1%2.1% percent of partisans who said the economy got betterEstimandlarger of the two party sharesWeightsweight (year-specific)Populationpartisans of either partyPeriod2008SourceCooperative Election Study of either party said the economy got better in the autumn of 2008, a survey taken up again below.

    Figure 2

    In the CES the president's party has the higher share in every year, and the lines cross at each change of party in the White House

    Share of partisans (leaners included) saying the national economy got better over the past year, percent, by party; CES, 2006 to 2024

    Said the economy got better (% of partisans)

    Show data
    Data for In the CES the president's party has the higher share in every year, and the lines cross at each change of party in the White House
    Survey yearPartyPresidentShare saying better (%)Interval lowInterval highn
    2006DemocratsBush7.26.87.716,316
    2006RepublicansBush70.369.471.114,998
    2007DemocratsBush2.11.62.84,240
    2007RepublicansBush25.623.727.63,671
    2008DemocratsBush0.90.71.114,626
    2008RepublicansBush2.11.92.413,856
    2009DemocratsObama24.423.125.75,973
    2009RepublicansObama5.24.65.95,453
    2010DemocratsObama36.035.137.023,997
    2010RepublicansObama5.95.46.423,350
    2011DemocratsObama24.323.025.69,165
    2011RepublicansObama4.94.25.67,285
    2012DemocratsObama55.955.056.924,943
    2012RepublicansObama5.34.95.820,488
    2013DemocratsObama46.144.547.87,387
    2013RepublicansObama10.59.411.85,253
    2014DemocratsObama52.851.953.825,131
    2014RepublicansObama11.911.312.618,807
    2015DemocratsObama56.354.657.96,480
    2015RepublicansObama12.911.714.24,792
    2016DemocratsObama52.251.453.030,201
    2016RepublicansObama8.68.19.120,484
    2017DemocratsTrump26.625.328.18,200
    2017RepublicansTrump75.073.376.55,651
    2018DemocratsTrump23.322.624.126,244
    2018RepublicansTrump83.482.784.121,346
    2019DemocratsTrump26.325.127.78,081
    2019RepublicansTrump49.347.750.86,363
    2020DemocratsTrump5.75.36.128,686
    2020RepublicansTrump44.443.545.319,648
    2021DemocratsBiden45.544.346.711,808
    2021RepublicansBiden7.66.98.48,001
    2022DemocratsBiden25.024.325.828,158
    2022RepublicansBiden4.23.84.620,127
    2023DemocratsBiden41.940.643.210,420
    2023RepublicansBiden8.57.89.38,738
    2024DemocratsBiden52.551.553.528,671
    2024RepublicansBiden5.04.55.521,191
    The lines cross between 2008 and 2009, between 2016 and 2017 and between 2020 and 2021, so the reversal happens within one year of each change of party. The shaded background marks the president's party. Bands are 95% intervals and are narrow because the samples are large.Source: Cooperative Election Study cumulative common content, 2006-2024; weighted (year-specific weight)

    The ANES premium is higher after 2012, on thin evidence

    Figure 3 plots the in-party premium in both surveys. In the six ANES surveys from 1980 to 1990 the premium averaged 14.814.8 percentage pointsEstimandmean of the yearly premiums, each survey counted once; the interval covers sampling error within each survey onlyInterval95% CI 13.3 to 16.2n9,644Weightsweight (VCF0009z)PopulationANES respondents who identify with or lean toward a party and answered the itemPeriod1980 to 1990 (six surveys)SourceANES Cumulative Data File points, and in the four surveys from 2012 to 2024 it averaged 34.034.0 percentage pointsEstimandmean of the yearly premiums, each survey counted once; the interval covers sampling error within each survey onlyInterval95% CI 32.5 to 35.4n20,874Weightsweight (VCF0009z)PopulationANES respondents who identify with or lean toward a party and answered the itemPeriod2012 to 2024 (four surveys)SourceANES Cumulative Data File, a difference of 19.2 (95% CI 4.8 to 33.7)19.2 percentage pointsEstimanddifference between the two era means with each survey as one observation; Welch t interval on the yearly premiums (6 and 4 surveys)Interval95% CI 4.8 to 33.7Weightsweight (VCF0009z)PopulationANES respondents who identify with or lean toward a party and answered the itemPeriod2012 to 2024 minus 1980 to 1990SourceANES Cumulative Data File points.Each average counts every survey once, so a survey with more respondents does not weigh more. The six earlier surveys are 1980, 1982, 1984, 1986, 1988 and 1990, and the four later ones are 2012, 2016, 2020 and 2024. The interval on the difference treats each survey as one observation and lets the two periods differ in spread (a Welch t interval). Sampling error alone gives 19.2 (95% CI 17.2 to 21.3)19.2 percentage pointsEstimanddifference between the two era means, standard errors combined in quadrature; the interval covers sampling error within each survey onlyInterval95% CI 17.2 to 21.3Weightsweight (VCF0009z)PopulationANES respondents who identify with or lean toward a party and answered the itemPeriod2012 to 2024 minus 1980 to 1990SourceANES Cumulative Data File for the same difference, an interval that describes these ten surveys and not a typical survey of each period.

    The earlier average includes 1980 and 1990, when no more than 4.3%4.3% percent of partisans who said the economy got betterEstimandlargest of the four shares (both parties, both surveys)Weightsweight (VCF0009z)Populationpartisans of either partyPeriod1980 and 1990SourceANES Cumulative Data File of either party said the economy got better. Leaving those two surveys out, the average for 1982 to 1988 is 21.821.8 percentage pointsEstimandmean of the yearly premiums, each survey counted once; the interval covers sampling error within each survey onlyInterval95% CI 19.7 to 23.8n6,593Weightsweight (VCF0009z)PopulationANES respondents who identify with or lean toward a party and answered the itemPeriod1982 to 1988 (four surveys)SourceANES Cumulative Data File points, and the later average is 12.2 (95% CI −4.3 to 28.8)12.2 percentage pointsEstimanddifference between the two era means with each survey as one observation; Welch t interval on the yearly premiums (4 and 4 surveys)Interval95% CI −4.3 to 28.8Weightsweight (VCF0009z)PopulationANES respondents who identify with or lean toward a party and answered the itemPeriod2012 to 2024 minus 1982 to 1988SourceANES Cumulative Data File points higher, an interval that includes zero when each survey is one observation.

    Most of the difference between the averages comes from the president's party. Its share saying better averaged 26.0%26.0% percent of partisans who said the economy got betterEstimandmean of the yearly shares, each survey counted onceInterval95% CI 24.8% to 27.2%n9,644Weightsweight (VCF0009z)Populationpartisans of the president's partyPeriod1980 to 1990 (six surveys)SourceANES Cumulative Data File in the earlier surveys and 41.6%41.6% percent of partisans who said the economy got betterEstimandmean of the yearly shares, each survey counted onceInterval95% CI 40.4% to 42.9%n20,874Weightsweight (VCF0009z)Populationpartisans of the president's partyPeriod2012 to 2024 (four surveys)SourceANES Cumulative Data File in the later ones, while the other party's share went from 11.2%11.2% percent of partisans who said the economy got betterEstimandmean of the yearly shares, each survey counted onceInterval95% CI 10.4% to 12.0%n9,644Weightsweight (VCF0009z)Populationpartisans of the other partyPeriod1980 to 1990 (six surveys)SourceANES Cumulative Data File to 7.7%7.7% percent of partisans who said the economy got betterEstimandmean of the yearly shares, each survey counted onceInterval95% CI 6.9% to 8.4%n20,874Weightsweight (VCF0009z)Populationpartisans of the other partyPeriod2012 to 2024 (four surveys)SourceANES Cumulative Data File.

    Figure 3

    The ANES premium averages higher from 2012 on, while the CES premium ranges from 1.3 to 63.1 points

    In-party premium, percentage points; partisans including leaners; ANES 1980 to 2024 and CES 2006 to 2024

    Show data
    Data for The ANES premium averages higher from 2012 on, while the CES premium ranges from 1.3 to 63.1 points
    Survey yearSurveyPresidentPremium (points)Interval lowInterval highn
    1980ANESCarter1.4−0.63.41,341
    1982ANESReagan16.211.820.51,193
    1984ANESReagan38.033.842.11,881
    1986ANESReagan14.610.518.61,798
    1988ANESReagan18.314.622.11,721
    1990ANESBush0.2−1.62.01,710
    1992ANESBush5.73.77.62,135
    1994ANESClinton4.0−1.29.31,560
    1996ANESClinton22.317.127.61,537
    1998ANESClinton15.38.921.61,114
    2000ANESClinton23.318.128.51,555
    2002ANESBush3.00.25.91,368
    2004ANESBush32.527.038.01,064
    2008ANESBush2.20.34.02,013
    2012ANESObama39.436.442.45,083
    2016ANESObama33.830.537.03,660
    2020ANESTrump31.028.733.37,254
    2024ANESBiden31.828.834.84,877
    2006CESBush63.162.164.031,314
    2007CESBush23.521.525.67,911
    2008CESBush1.30.91.628,482
    2009CESObama19.217.720.711,426
    2010CESObama30.129.131.247,347
    2011CESObama19.418.020.916,450
    2012CESObama50.649.551.745,431
    2013CESObama35.633.637.612,640
    2014CESObama40.939.842.143,938
    2015CESObama43.441.345.411,272
    2016CESObama43.642.744.650,685
    2017CESTrump48.346.250.413,851
    2018CESTrump60.159.061.147,590
    2019CESTrump22.920.924.914,444
    2020CESTrump38.737.739.748,334
    2021CESBiden37.936.439.319,809
    2022CESBiden20.820.021.748,285
    2023CESBiden33.431.934.919,158
    2024CESBiden47.546.448.749,862
    1980ANES averagen/a14.8n/an/a9,644
    1990ANES averagen/a14.8n/an/a9,644
    2001ANES averagen/an/an/an/an/a
    2012ANES averagen/a34.0n/an/a20,874
    2024ANES averagen/a34.0n/an/a20,874
    The ANES premium is higher from 2012 on than from 1980 to 1990, and the CES premium is higher than the ANES premium in each year from 2012 on in which both ask the item. Dashed segments mark the ANES average of the yearly premiums: 14.8 points for 1980 to 1990 and 34.0 for 2012 to 2024. Bands are 95% intervals.Note: The ANES adds web interviews in 2012; the CES is an online panel throughout. Mode comparisons follow in the chapter.Note: The averages count each survey once and carry no band here: the interval of an average covers sampling error within each survey only, and the yearly premiums vary by more than that.Source: American National Election Studies Cumulative Data File, 1980-2024, and Cooperative Election Study cumulative common content, 2006-2024; weighted (VCF0009z for the ANES, year-specific weight for the CES)

    The higher level from 2012 does not come from the web samples that the ANES added that year. The premium was already 32.5 (95% CI 27.0 to 38.0)32.5 percentage pointsEstimandin-party share saying better minus out-party share, weightedInterval95% CI 27.0 to 38.0n1,064Weightsweight (VCF0009z)PopulationANES respondents who identify with or lean toward a party and answered the itemPeriod2004SourceANES Cumulative Data File points in 2004, when interviews were mainly face-to-face, and Figure 4 shows that face-to-face interviews alone give 35.1 (95% CI 29.9 to 40.4)35.1 percentage pointsEstimandin-party share saying better minus out-party share, weighted with the mode-specific weightInterval95% CI 29.9 to 40.4n1,805Weightsweight_ftfPopulationANES face-to-face respondents who identify with or lean toward a partyPeriod2012SourceANES Cumulative Data File points in 2012, 38.4 (95% CI 32.3 to 44.5)38.4 percentage pointsEstimandin-party share saying better minus out-party share, weighted with the mode-specific weightInterval95% CI 32.3 to 44.5n1,042Weightsweight_ftfPopulationANES face-to-face respondents who identify with or lean toward a partyPeriod2016SourceANES Cumulative Data File in 2016 and 33.3 (95% CI 26.9 to 39.7)33.3 percentage pointsEstimandin-party share saying better minus out-party share, weighted with the mode-specific weightInterval95% CI 26.9 to 39.7n966Weightsweight_ftfPopulationANES face-to-face respondents who identify with or lean toward a partyPeriod2024SourceANES Cumulative Data File in 2024. The smallest of the three is 33.333.3 percentage pointsEstimandsmallest of the three face-to-face premiumsWeightsweight_ftfPopulationANES face-to-face respondents who identify with or lean toward a partyPeriod2012, 2016 and 2024SourceANES Cumulative Data File points, against an average of 14.814.8 percentage pointsEstimandmean of the yearly premiums, each survey counted once; the interval covers sampling error within each survey onlyInterval95% CI 13.3 to 16.2n9,644Weightsweight (VCF0009z)PopulationANES respondents who identify with or lean toward a party and answered the itemPeriod1980 to 1990 (six surveys)SourceANES Cumulative Data File from 1980 to 1990. The two interview modesSurvey modeHow respondents were interviewed: in person, by phone, on paper or online. Mode can change answers, so a change in mode can look like a change in opinion. do not give the same premium. Face-to-face minus web is −6.7 (95% CI −13.0 to −0.3)−6.7 percentage points, face-to-face minus webEstimandface-to-face premium minus web premium, standard errors combined in quadratureInterval95% CI −13.0 to −0.3Weightsweight_ftf and weight_webPopulationANES respondents who identify with or lean toward a partyPeriod2012SourceANES Cumulative Data File points in 2012, +6.3 (95% CI −0.9 to 13.5)+6.3 percentage points, face-to-face minus webEstimandface-to-face premium minus web premium, standard errors combined in quadratureInterval95% CI −0.9 to 13.5Weightsweight_ftf and weight_webPopulationANES respondents who identify with or lean toward a partyPeriod2016SourceANES Cumulative Data File in 2016 and +1.6 (95% CI −6.2 to 9.5)+1.6 percentage points, face-to-face minus webEstimandface-to-face premium minus web premium, standard errors combined in quadratureInterval95% CI −6.2 to 9.5Weightsweight_ftf and weight_webPopulationANES respondents who identify with or lean toward a partyPeriod2024SourceANES Cumulative Data File in 2024, so the difference reaches 6.76.7 percentage pointsEstimandlargest absolute difference between the face-to-face and web premiumsWeightsweight_ftf and weight_webPopulationANES respondents who identify with or lean toward a partyPeriod2012, 2016 and 2024SourceANES Cumulative Data File points and changes sign between 2012 and 2016. Only the 2012 interval excludes zero.

    The two surveys differ in level. The CES premium is higher than the ANES premium in every year in which both ask the item after 2008, by 11.2 (95% CI 8.1 to 14.4)11.2 percentage points, CES minus ANESEstimandCES premium minus ANES premium, standard errors combined in quadratureInterval95% CI 8.1 to 14.4Weightsweight (CES year-specific) and weight (ANES VCF0009z)Populationpartisans with an answer in both surveysPeriod2012SourceCooperative Election Study points in 2012, 9.9 (95% CI 6.5 to 13.2)9.9 percentage points, CES minus ANESEstimandCES premium minus ANES premium, standard errors combined in quadratureInterval95% CI 6.5 to 13.2Weightsweight (CES year-specific) and weight (ANES VCF0009z)Populationpartisans with an answer in both surveysPeriod2016SourceCooperative Election Study in 2016, 7.7 (95% CI 5.2 to 10.2)7.7 percentage points, CES minus ANESEstimandCES premium minus ANES premium, standard errors combined in quadratureInterval95% CI 5.2 to 10.2Weightsweight (CES year-specific) and weight (ANES VCF0009z)Populationpartisans with an answer in both surveysPeriod2020SourceCooperative Election Study in 2020 and 15.7 (95% CI 12.5 to 19.0)15.7 percentage points, CES minus ANESEstimandCES premium minus ANES premium, standard errors combined in quadratureInterval95% CI 12.5 to 19.0Weightsweight (CES year-specific) and weight (ANES VCF0009z)Populationpartisans with an answer in both surveysPeriod2024SourceCooperative Election Study in 2024. The questions differ in the number of answers, and the surveys differ in mode, sample and weighting. This chapter compares the shape of each series and does not treat the levels as interchangeable.

    Figure 4

    Face-to-face interviews alone give an ANES premium of at least 33.3 points in 2012, 2016 and 2024, against an average of 14.8 points from 1980 to 1990

    In-party premium, percentage points, by survey and interview mode; partisans including leaners; 2012, 2016, 2020 and 2024

    • ANES
    • CES
    Show data
    Data for Face-to-face interviews alone give an ANES premium of at least 33.3 points in 2012, 2016 and 2024, against an average of 14.8 points from 1980 to 1990
    YearSurvey and modePremium (points)Interval lowInterval highn
    2012ANES all modes39.436.442.45,083
    2012ANES face-to-face35.129.940.41,805
    2012ANES web41.838.245.43,278
    2012CES online panel50.649.551.745,431
    2016ANES all modes33.830.537.03,660
    2016ANES face-to-face38.432.344.51,042
    2016ANES web32.128.335.92,618
    2016CES online panel43.642.744.650,685
    2020ANES all modes31.028.733.37,254
    2020CES online panel38.737.739.748,334
    2024ANES all modes31.828.834.84,877
    2024ANES face-to-face33.326.939.7966
    2024ANES web31.727.236.21,855
    2024CES online panel47.546.448.749,862
    In each of the four years the CES premium lies to the right of every ANES premium. The face-to-face and web premiums differ in both directions across the three years in which the ANES has mode weights. Lines are 95% intervals.Note: The ANES has no mode weights in 2020, so that year shows the all-mode estimate only.Note: The ANES all-mode estimate for 2024 also includes panel and paper respondents, who have no mode weight.Source: American National Election Studies Cumulative Data File, 1980-2024, and Cooperative Election Study cumulative common content, 2006-2024; weighted (VCF0009z for the ANES, year-specific weight for the CES); weight_ftf and weight_web for the face-to-face and web rows

    The CES cannot say whether the premium rose, because its series starts in 2006. Its premium runs from 1.3 (95% CI 0.9 to 1.6)1.3 percentage pointsEstimandin-party share saying better minus out-party share, weightedInterval95% CI 0.9 to 1.6n28,482Weightsweight (year-specific)PopulationCES respondents who identify with or lean toward a party and answered the itemPeriod2008SourceCooperative Election Study points in 2008 to 63.1 (95% CI 62.1 to 64.0)63.1 percentage pointsEstimandin-party share saying better minus out-party share, weightedInterval95% CI 62.1 to 64.0n31,314Weightsweight (year-specific)PopulationCES respondents who identify with or lean toward a party and answered the itemPeriod2006SourceCooperative Election Study in 2006. By presidency the average premium runs between 29.329.3 percentage pointsEstimandsmallest of the four presidency means of the yearly premiums (Bush 2006 to 2008, Obama 2009 to 2016, Trump 2017 to 2020, Biden 2021 to 2024), each survey counted onceWeightsweight (year-specific)PopulationCES respondents who identify with or lean toward a party and answered the itemPeriod2006 to 2024, by presidencySourceCooperative Election StudyNoteNo interval: the yearly premiums within a presidency differ by more than sampling error, and a presidency has only 3 to 8 surveys. and 42.542.5 percentage pointsEstimandlargest of the four presidency means of the yearly premiums, each survey counted onceWeightsweight (year-specific)PopulationCES respondents who identify with or lean toward a party and answered the itemPeriod2006 to 2024, by presidencySourceCooperative Election StudyNoteNo interval: the yearly premiums within a presidency differ by more than sampling error, and a presidency has only 3 to 8 surveys. points (Figure 5), and within each presidency the highest and lowest yearly premiums differ by at least 26.726.7 percentage pointsEstimandsmallest, across the four presidencies, of the largest minus the smallest yearly premium within the presidencyWeightsweight (year-specific)PopulationCES respondents who identify with or lean toward a party and answered the itemPeriod2006 to 2024, by presidencySourceCooperative Election Study points.The CES fields a large survey in each even year, when there is a national election, and a smaller one in each odd year. Its premium averages 39.739.7 percentage pointsEstimandmean of the yearly premiums in the even-year surveys, each survey counted oncen441,268Weightsweight (year-specific)PopulationCES respondents who identify with or lean toward a party and answered the itemPeriod2006, 2008 and every second year to 2024SourceCooperative Election StudyNoteNo interval: it is the unweighted mean of 10 yearly premiums. points in the even-year surveys and 31.531.5 percentage pointsEstimandmean of the yearly premiums in the odd-year surveys, each survey counted oncen126,961Weightsweight (year-specific)PopulationCES respondents who identify with or lean toward a party and answered the itemPeriod2007, 2009 and every second year to 2023SourceCooperative Election StudyNoteNo interval: it is the unweighted mean of 9 yearly premiums. in the odd-year surveys, and the cumulative file gives no reason for the difference. A straight line fitted to the yearly premiums has a slope of 0.53 (95% CI −0.85 to 1.91)0.53 percentage points per yearEstimandleast-squares slope of the yearly premium on the survey year, one point per survey, t intervalInterval95% CI −0.85 to 1.91n568,229Weightsweight (year-specific)PopulationCES respondents who identify with or lean toward a party and answered the itemPeriod2006 to 2024SourceCooperative Election Study points per year over 2006 to 2024 in the CES, an interval that includes zero and the ANES slope of 0.49 (95% CI 0.01 to 0.97)0.49 percentage points per yearEstimandleast-squares slope of the yearly premium on the survey year, one point per survey, t intervalInterval95% CI 0.01 to 0.97n42,864Weightsweight (VCF0009z)PopulationANES respondents who identify with or lean toward a party and answered the itemPeriod1980 to 2024SourceANES Cumulative Data File points per year over 1980 to 2024. The ANES interval clears zero by a small margin, and without 1980 and 1990 the ANES slope is 0.37 (95% CI −0.16 to 0.89)0.37 percentage points per yearEstimandleast-squares slope of the yearly premium on the survey year, one point per survey, t intervalInterval95% CI −0.16 to 0.89n39,813Weightsweight (VCF0009z)PopulationANES respondents who identify with or lean toward a party and answered the itemPeriod1982 to 2024, without 1980 and 1990SourceANES Cumulative Data File points per year.

    Figure 5

    The CES premium averages between 29.3 and 42.5 points by presidency, and the yearly premiums within a presidency differ by at least 26.7 points

    Average, lowest and highest yearly in-party premium in each presidency, percentage points; partisans including leaners; CES, 2006 to 2024

    PresidencyYearsAverage premium (points)Lowest yearly premiumHighest yearly premium
    Bush2006 to 200829.31.363.1
    Obama2009 to 201635.419.250.6
    Trump2017 to 202042.522.960.1
    Biden2021 to 202434.920.847.5
    Show data
    Data for The CES premium averages between 29.3 and 42.5 points by presidency, and the yearly premiums within a presidency differ by at least 26.7 points
    highestlowestmeannpresidentsurveysyears
    63.051.2729.2967,707Bush32006 to 2008
    50.6119.1835.36239,189Obama82009 to 2016
    60.0722.9242.50124,219Trump42017 to 2020
    47.5320.8434.91137,114Biden42021 to 2024
    Each row is one presidency. The average counts each yearly survey once, and the lowest and highest values are the smallest and largest yearly premiums in that presidency. The averages carry no interval because the yearly premiums within a presidency differ by more than sampling error.Source: Cooperative Election Study cumulative common content, 2006-2024; weighted (year-specific weight)

    The premium shrinks when the president's party says worse

    In 66 ANES survey yearsEstimandcountWeightsnone (count)PopulationANES surveys with a premium below the cutoffPeriod1980 to 2024SourceANES Cumulative Data File of the 1818 ANES survey yearsEstimandcountWeightsnone (count)PopulationANES surveys that ask the retrospective economy itemPeriod1980 to 2024SourceANES Cumulative Data File ANES surveys (1980, 1990, 1992, 1994, 2002 and 2008) the premium is below 1010 percentage pointsEstimandconstantWeightsnone (constant)Populationcutoff that separates the low-premium surveysPeriod1980 to 2024SourceANES Cumulative Data File points, and it is at most 5.75.7 percentage pointsEstimandlargest premium among the surveys below the cutoffWeightsweight (VCF0009z)PopulationANES respondents who identify with or lean toward a party and answered the itemPeriodthe low-premium surveysSourceANES Cumulative Data File. In the other 1212 ANES survey yearsEstimandcountWeightsnone (count)PopulationANES surveys with a premium at or above the cutoffPeriod1980 to 2024SourceANES Cumulative Data File surveys it is 14.614.6 percentage pointsEstimandsmallest premium among the surveys at or above the cutoffWeightsweight (VCF0009z)PopulationANES respondents who identify with or lean toward a party and answered the itemPeriodthe other surveysSourceANES Cumulative Data File or more. In every low survey except 1994, no more than 8.1%8.1% percent of partisans who said the economy got betterEstimandlargest in-party share among those surveysWeightsweight (VCF0009z)Populationpartisans of the president's partyPeriodthe five low-premium surveys other than 1994SourceANES Cumulative Data File of the president's party said the economy got better, against at least 22.8%22.8% percent of partisans who said the economy got betterEstimandsmallest in-party share among those surveysWeightsweight (VCF0009z)Populationpartisans of the president's partyPeriodthe surveys with a premium at or above the cutoffSourceANES Cumulative Data File in every one of the other surveys (Figure 6). The premium cannot exceed the share of the president's own party who say better, so a survey in which even the president's partisans mostly say worse leaves little room for it.

    Figure 6

    The premium is below 10 points in 6 ANES surveys, and in all but 1994 no more than 8.1% of the president's party said the economy got better

    In-party premium against the share of the president's party saying the economy got better, one point per ANES survey, 1980 to 2024; percent and percentage points

    • Premium below 10 points
    • Premium of 10 points or more
    Show data
    Data for The premium is below 10 points in 6 ANES surveys, and in all but 1994 no more than 8.1% of the president's party said the economy got better
    Survey yearPresident's party saying better (%)Premium (points)Interval lowInterval highn
    19804.31.4−0.63.41,341
    198222.816.211.820.51,193
    198463.838.033.842.11,881
    198632.314.610.518.61,798
    198829.118.314.622.11,721
    19903.70.2−1.62.01,710
    19928.15.73.77.62,135
    199437.24.0−1.29.31,560
    199648.622.317.127.61,537
    199853.715.38.921.61,114
    200049.623.318.128.51,555
    20026.73.00.25.91,368
    200442.632.527.038.01,064
    20083.52.20.34.02,013
    201248.839.436.442.45,083
    201645.033.830.537.03,660
    202036.231.028.733.37,254
    202436.531.828.834.84,877
    Surveys at the left are those in which the president's own party was least likely to say the economy got better, and the premium cannot exceed that share, which the dashed y = x line marks. The horizontal rule separates the 6 surveys with a premium below 10 points, which are labelled by year. 1994 lies apart from the other low surveys.Note: The chart shows each survey's estimate without its interval; the data table gives the interval of each premium.Source: American National Election Studies Cumulative Data File, 1980-2024; weighted (VCF0009z)

    In 1990, 72.3%72.3% percent of partisans who said the economy got worseEstimandweighted share saying worseInterval95% CI 68.8% to 75.4%n710Weightsweight (VCF0009z)Populationpartisans of the president's partyPeriod1990SourceANES Cumulative Data File of the president's party and 75.8%75.8% percent of partisans who said the economy got worseEstimandweighted share saying worseInterval95% CI 73.0% to 78.4%n1,000Weightsweight (VCF0009z)Populationpartisans of the other partyPeriod1990SourceANES Cumulative Data File of the other party said the economy got worse, and in 2008 the shares were 87.6%87.6% percent of partisans who said the economy got worseEstimandweighted share saying worseInterval95% CI 84.3% to 90.2%n651Weightsweight (VCF0009z)Populationpartisans of the president's partyPeriod2008SourceANES Cumulative Data File and 93.9%93.9% percent of partisans who said the economy got worseEstimandweighted share saying worseInterval95% CI 92.0% to 95.3%n1,362Weightsweight (VCF0009z)Populationpartisans of the other partyPeriod2008SourceANES Cumulative Data File. The parties still differed about the president in these years. The approval premium, the share of the president's party approving of the president's job minus the share of the other party, was at least 34.934.9 percentage pointsEstimandsmallest yearly premium in approval of the president's jobWeightsweight (VCF0009z)PopulationANES partisans who gave an approve or disapprove answerPeriod1980 to 2024SourceANES Cumulative Data File points in every ANES survey, including 1990 and 2008.

    The 1994 survey does not fit that account. 37.2%37.2% percent of partisans who said the economy got betterEstimandweighted share saying betterInterval95% CI 33.6% to 41.0%n819Weightsweight (VCF0009z)Populationpartisans of the president's partyPeriod1994SourceANES Cumulative Data File of the president's party and 33.2%33.2% percent of partisans who said the economy got betterEstimandweighted share saying betterInterval95% CI 29.6% to 37.0%n741Weightsweight (VCF0009z)Populationpartisans of the other partyPeriod1994SourceANES Cumulative Data File of the other party said the economy got better, and the premium of 4.0 (95% CI −1.2 to 9.3)4.0 percentage pointsEstimandin-party share saying better minus out-party share, weightedInterval95% CI −1.2 to 9.3n1,560Weightsweight (VCF0009z)PopulationANES respondents who identify with or lean toward a party and answered the itemPeriod1994SourceANES Cumulative Data File points has an interval that includes zero. Approval of Clinton divided the parties as usual, by 55.0 (95% CI 50.3 to 59.7)55.0 percentage pointsEstimandin-party share approving of the president's job minus out-party share, weightedInterval95% CI 50.3 to 59.7n1,509Weightsweight (VCF0009z)PopulationANES partisans who gave an approve or disapprove answerPeriod1994SourceANES Cumulative Data File points. The ANES interviewed its 1994 sample after the November election, but it did the same in 1982, 1986, 1990 and 1998, so timing does not separate 1994 from those surveys. These data cannot say whether the state of the economy or a feature of the sample made the premium small.In only 11 ANES survey yearsEstimandcountWeightsnone (count)PopulationANES surveys in which the share of the other party saying better exceeds the 1994 sharePeriod1980 to 2024SourceANES Cumulative Data File other ANES survey, 1998, was the other party's share saying better higher in the sample than in 1994, at 38.4% (95% CI 33.8% to 43.3%)38.4% percent of partisans who said the economy got betterEstimandweighted share saying betterInterval95% CI 33.8% to 43.3%n467Weightsweight (VCF0009z)Populationpartisans of the other partyPeriod1998SourceANES Cumulative Data File, and that interval overlaps the 1994 interval.

    Weekly answers moved before the 2021 inauguration

    The weekly Nationscape survey shows how quickly answers can change (Figure 7). In late February 2020, 74.9% (95% CI 71.7% to 77.8%)74.9% percent of partisans who said the economy got betterEstimandweighted share saying better in one weekly waveInterval95% CI 71.7% to 77.8%n1,967Weightsweight (Nationscape survey weight)PopulationRepublicans among Nationscape respondentsPeriod2020-02-17 to 2020-02-23SourceNationscape of Republican identifiers said the economy got better, and by mid-April the share was 22.0% (95% CI 19.2% to 25.1%)22.0% percent of partisans who said the economy got betterEstimandweighted share saying better in one weekly waveInterval95% CI 19.2% to 25.1%n1,654Weightsweight (Nationscape survey weight)PopulationRepublicans among Nationscape respondentsPeriod2020-04-13 to 2020-04-19SourceNationscape. The Democratic share went from 20.0%20.0% percent of partisans who said the economy got betterEstimandweighted share saying better in one weekly waveInterval95% CI 17.4% to 22.7%n2,366Weightsweight (Nationscape survey weight)PopulationDemocrats among Nationscape respondentsPeriod2020-02-17 to 2020-02-23SourceNationscape to 12.1%12.1% percent of partisans who said the economy got betterEstimandweighted share saying better in one weekly waveInterval95% CI 10.0% to 14.5%n1,857Weightsweight (Nationscape survey weight)PopulationDemocrats among Nationscape respondentsPeriod2020-04-13 to 2020-04-19SourceNationscape over the same weeks. A Republican was president, so the premium is the Republican share minus the Democratic share, and it went from 54.9 (95% CI 50.9 to 58.9)54.9 percentage pointsEstimandRepublican share saying better minus Democratic share in one weekly wave (a Republican was president), standard errors combined in quadratureInterval95% CI 50.9 to 58.9Weightsweight (Nationscape survey weight)PopulationNationscape respondents who identify as Democrats or Republicans and answered the itemPeriod2020-02-17 to 2020-02-23SourceNationscape points to 10.0 (95% CI 6.2 to 13.7)10.0 percentage pointsEstimandRepublican share saying better minus Democratic share in one weekly wave (a Republican was president), standard errors combined in quadratureInterval95% CI 6.2 to 13.7Weightsweight (Nationscape survey weight)PopulationNationscape respondents who identify as Democrats or Republicans and answered the itemPeriod2020-04-13 to 2020-04-19SourceNationscape, close to the 1010 percentage pointsEstimandconstantWeightsnone (constant)Populationcutoff that separates the low-premium surveysPeriod1980 to 2024SourceANES Cumulative Data File-point cutoff that marks the low-premium surveys above. The pandemic was declared in March, and the survey cannot separate a change in conditions from a change in what partisans say.

    The levels differ from the other two surveys. In 2020 the share of Republicans saying better was 36.2%36.2% percent of partisans who said the economy got betterEstimandweighted share saying betterInterval95% CI 34.1% to 38.4%n3,425Weightsweight (VCF0009z)PopulationRepublicans including leanersPeriod2020SourceANES Cumulative Data File in the ANES and 44.4%44.4% percent of partisans who said the economy got betterEstimandweighted share saying betterInterval95% CI 43.5% to 45.3%n19,648Weightsweight (year-specific)PopulationRepublicans including leanersPeriod2020SourceCooperative Election Study in the CES, against 28.2%28.2% percent of partisans who said the economy got betterEstimandweighted share saying better in one weekly waveInterval95% CI 25.2% to 31.4%n2,383Weightsweight (Nationscape survey weight)PopulationRepublicans among Nationscape respondentsPeriod2020-10-26 to 2020-11-01SourceNationscape in Nationscape in the last full week before the election, and the Democratic shares were 5.2%5.2% percent of partisans who said the economy got betterEstimandweighted share saying betterInterval95% CI 4.3% to 6.3%n3,829Weightsweight (VCF0009z)PopulationDemocrats including leanersPeriod2020SourceANES Cumulative Data File, 5.7%5.7% percent of partisans who said the economy got betterEstimandweighted share saying betterInterval95% CI 5.3% to 6.1%n28,686Weightsweight (year-specific)PopulationDemocrats including leanersPeriod2020SourceCooperative Election Study and 10.3%10.3% percent of partisans who said the economy got betterEstimandweighted share saying better in one weekly waveInterval95% CI 8.5% to 12.5%n2,424Weightsweight (Nationscape survey weight)PopulationDemocrats among Nationscape respondentsPeriod2020-10-26 to 2020-11-01SourceNationscape. The surveys differ in mode, sample and timing, and Nationscape leaves out leaners, so this figure is read for its shape.

    Figure 7

    The Republican share saying better fell from 74.9% to 22.0% within two months in 2020, and the parties' lines met before the inauguration

    Share of Democratic and Republican identifiers saying the national economy got better over the past year, percent, by interview week; Nationscape, July 2019 to January 2021

    Said the economy got better (% of identifiers)

    Show data
    Data for The Republican share saying better fell from 74.9% to 22.0% within two months in 2020, and the parties' lines met before the inauguration
    Week startsWeek endsPartyShare saying better (%)Interval lowInterval highn
    2019-07-182019-07-21Democrats16.313.519.71,741
    2019-07-182019-07-21Republicans70.266.074.11,445
    2019-07-222019-07-28Democrats14.812.317.72,252
    2019-07-222019-07-28Republicans66.462.370.21,750
    2019-07-292019-08-04Democrats14.912.517.72,346
    2019-07-292019-08-04Republicans66.162.269.71,814
    2019-08-052019-08-11Democrats14.311.817.21,959
    2019-08-052019-08-11Republicans64.060.167.71,654
    2019-08-122019-08-18Democrats12.39.915.11,970
    2019-08-122019-08-18Republicans64.660.468.61,565
    2019-08-192019-08-25Democrats12.510.414.92,707
    2019-08-192019-08-25Republicans66.963.570.22,152
    2019-08-262019-09-01Democrats11.08.913.61,993
    2019-08-262019-09-01Republicans64.460.468.21,571
    2019-09-022019-09-08Democrats11.49.314.02,119
    2019-09-022019-09-08Republicans62.258.465.91,685
    2019-09-092019-09-15Democrats11.59.214.32,002
    2019-09-092019-09-15Republicans64.760.668.61,632
    2019-09-162019-09-22Democrats12.810.415.72,171
    2019-09-162019-09-22Republicans66.061.969.91,709
    2019-09-232019-09-29Democrats11.49.214.22,044
    2019-09-232019-09-29Republicans64.560.568.41,741
    2019-09-302019-10-06Democrats12.810.915.13,266
    2019-09-302019-10-06Republicans62.759.366.02,579
    2019-10-072019-10-13Democrats13.611.016.81,732
    2019-10-072019-10-13Republicans62.157.666.31,377
    2019-10-142019-10-20Democrats15.712.619.31,397
    2019-10-142019-10-20Republicans68.463.972.71,251
    2019-10-212019-10-27Democrats13.110.815.92,337
    2019-10-212019-10-27Republicans66.162.269.81,911
    2019-10-282019-11-03Democrats16.413.819.42,427
    2019-10-282019-11-03Republicans68.464.771.92,030
    2019-11-042019-11-10Democrats14.712.217.61,936
    2019-11-042019-11-10Republicans69.766.173.11,769
    2019-11-112019-11-17Democrats12.910.915.32,565
    2019-11-112019-11-17Republicans68.164.871.22,181
    2019-11-182019-11-24Democrats17.113.721.31,156
    2019-11-182019-11-24Republicans69.865.274.11,158
    2019-11-252019-12-01Democrats14.712.417.32,604
    2019-11-252019-12-01Republicans69.866.572.92,303
    2019-12-022019-12-08Democrats13.611.016.81,566
    2019-12-022019-12-08Republicans68.964.872.71,490
    2019-12-092019-12-15Democrats14.011.916.42,706
    2019-12-092019-12-15Republicans69.766.472.82,316
    2019-12-162019-12-22Democrats18.916.421.72,289
    2019-12-162019-12-22Republicans70.066.773.21,996
    2019-12-232019-12-29Democrats22.419.625.51,913
    2019-12-232019-12-29Republicans73.469.876.61,526
    2019-12-302020-01-05Democrats17.815.720.03,131
    2019-12-302020-01-05Republicans73.170.575.52,804
    2020-01-062020-01-12Democrats19.316.322.71,572
    2020-01-062020-01-12Republicans73.769.977.21,401
    2020-01-132020-01-19Democrats21.718.924.72,195
    2020-01-132020-01-19Republicans70.166.673.41,845
    2020-01-202020-01-26Democrats15.913.618.52,370
    2020-01-202020-01-26Republicans70.867.573.81,972
    2020-01-272020-02-02Democrats17.814.921.01,714
    2020-01-272020-02-02Republicans73.169.276.61,417
    2020-02-032020-02-09Democrats15.312.818.12,143
    2020-02-032020-02-09Republicans73.269.676.41,897
    2020-02-102020-02-16Democrats20.717.923.91,935
    2020-02-102020-02-16Republicans73.269.776.51,823
    2020-02-172020-02-23Democrats20.017.422.72,366
    2020-02-172020-02-23Republicans74.971.777.81,967
    2020-02-242020-03-01Democrats19.016.322.11,778
    2020-02-242020-03-01Republicans68.965.372.31,682
    2020-03-022020-03-08Democrats16.414.019.22,264
    2020-03-022020-03-08Republicans66.062.269.51,767
    2020-03-092020-03-15Democrats14.011.716.62,054
    2020-03-092020-03-15Republicans58.955.262.51,803
    2020-03-162020-03-22Democrats8.97.211.02,313
    2020-03-162020-03-22Republicans44.540.948.11,837
    2020-03-232020-03-29Democrats10.38.512.62,097
    2020-03-232020-03-29Republicans37.033.840.41,918
    2020-03-302020-04-05Democrats9.98.511.53,657
    2020-03-302020-04-05Republicans31.128.733.63,108
    2020-04-062020-04-12Democrats7.26.18.54,597
    2020-04-062020-04-12Republicans28.126.030.23,908
    2020-04-132020-04-19Democrats12.110.014.51,857
    2020-04-132020-04-19Republicans22.019.225.11,654
    2020-04-202020-04-26Democrats7.45.89.41,846
    2020-04-202020-04-26Republicans21.518.624.61,566
    2020-04-272020-05-03Democrats6.95.39.01,784
    2020-04-272020-05-03Republicans24.221.327.41,684
    2020-05-042020-05-10Democrats6.65.28.32,556
    2020-05-042020-05-10Republicans21.418.924.22,138
    2020-05-112020-05-17Democrats10.48.412.91,735
    2020-05-112020-05-17Republicans18.115.421.11,583
    2020-05-182020-05-24Democrats9.58.011.32,806
    2020-05-182020-05-24Republicans23.320.925.82,503
    2020-05-252020-05-31Democrats7.66.09.71,799
    2020-05-252020-05-31Republicans21.118.424.11,587
    2020-06-012020-06-07Democrats6.34.98.12,035
    2020-06-012020-06-07Republicans20.117.523.01,757
    2020-06-082020-06-14Democrats6.44.98.12,197
    2020-06-082020-06-14Republicans22.419.825.22,000
    2020-06-152020-06-21Democrats5.94.67.62,214
    2020-06-152020-06-21Republicans18.716.321.41,898
    2020-06-222020-06-28Democrats6.55.18.32,337
    2020-06-222020-06-28Republicans17.215.019.81,991
    2020-06-292020-07-05Democrats5.64.27.41,961
    2020-06-292020-07-05Republicans21.218.524.21,785
    2020-07-062020-07-12Democrats4.83.46.71,666
    2020-07-062020-07-12Republicans20.717.624.21,458
    2020-07-132020-07-19Democrats9.07.710.54,172
    2020-07-132020-07-19Republicans22.420.324.73,373
    2020-07-202020-07-26Democrats6.04.87.33,562
    2020-07-202020-07-26Republicans21.019.023.23,379
    2020-07-272020-08-02Democrats7.35.99.02,640
    2020-07-272020-08-02Republicans14.612.616.82,423
    2020-08-032020-08-09Democrats5.33.97.21,739
    2020-08-032020-08-09Republicans19.817.022.91,504
    2020-08-102020-08-16Democrats6.65.08.61,849
    2020-08-102020-08-16Republicans22.119.325.31,659
    2020-08-172020-08-23Democrats5.64.37.42,320
    2020-08-172020-08-23Republicans21.318.824.12,237
    2020-08-242020-08-30Democrats6.95.29.11,865
    2020-08-242020-08-30Republicans26.623.130.41,506
    2020-08-312020-09-06Democrats7.65.89.81,875
    2020-08-312020-09-06Republicans23.320.426.51,898
    2020-09-072020-09-13Democrats8.36.510.51,966
    2020-09-072020-09-13Republicans24.321.227.71,669
    2020-09-142020-09-20Democrats9.77.911.92,430
    2020-09-142020-09-20Republicans28.425.431.62,069
    2020-09-212020-09-27Democrats8.56.511.11,809
    2020-09-212020-09-27Republicans24.020.727.61,629
    2020-09-282020-10-04Democrats8.36.310.91,665
    2020-09-282020-10-04Republicans27.123.730.91,673
    2020-10-052020-10-11Democrats6.44.98.42,205
    2020-10-052020-10-11Republicans20.417.623.52,058
    2020-10-122020-10-18Democrats9.07.211.32,127
    2020-10-122020-10-18Republicans26.923.730.41,871
    2020-10-192020-10-25Democrats9.97.812.61,897
    2020-10-192020-10-25Republicans28.024.431.91,669
    2020-10-262020-11-01Democrats10.38.512.52,424
    2020-10-262020-11-01Republicans28.225.231.42,383
    2020-11-022020-11-08Democrats9.98.012.22,050
    2020-11-022020-11-08Republicans33.630.237.11,808
    2020-11-092020-11-15Democrats14.111.916.62,165
    2020-11-092020-11-15Republicans30.827.534.21,790
    2020-11-162020-11-22Democrats14.311.817.11,931
    2020-11-162020-11-22Republicans25.622.229.31,496
    2020-11-232020-11-29Democrats11.910.014.12,520
    2020-11-232020-11-29Republicans27.324.430.52,079
    2020-11-302020-12-06Democrats15.313.217.82,639
    2020-11-302020-12-06Republicans21.318.724.32,209
    2020-12-072020-12-13Democrats18.115.221.51,630
    2020-12-072020-12-13Republicans24.120.228.31,131
    2020-12-142020-12-20Democrats13.611.615.82,713
    2020-12-142020-12-20Republicans21.418.824.22,125
    2020-12-212020-12-27Democrats14.212.116.52,387
    2020-12-212020-12-27Republicans21.018.423.81,975
    2020-12-282020-12-30Democrats10.27.913.11,334
    2020-12-282020-12-30Republicans18.615.222.6924
    2021-01-122021-01-16Democrats18.215.221.71,647
    2021-01-122021-01-16Republicans18.615.222.51,222
    2021-01-212021-01-24Democrats18.315.721.21,788
    2021-01-212021-01-24Republicans13.811.316.71,404
    2021-01-252021-01-31Democrats23.620.327.21,358
    2021-01-252021-01-31Republicans13.310.616.51,155
    The Republican line falls from 74.9% in late February 2020 to 22.0% in mid-April, reaches a local high of 33.6% in the election week and falls from 28.2% in the last full week before the election to 18.6% in mid-January, when it meets the Democratic line. The shaded background marks the president's party. Bands are 95% intervals.Note: Party is the file's three-category party identification, so leaners are not counted with a party. Nationscape recruits online respondents who chose to take part, and the intervals treat it as a probability sample.Note: The levels differ from the other two surveys: in 2020 the Republican share saying better is 28.2% in the last full week before the election, against 36.2% in the ANES and 44.4% in the CES, so this figure is read for its shape.Note: The Republican share in the election week (33.6%) is above the share in the week before (28.2%), and its 95% interval overlaps the intervals of the weeks on either side.Note: No wave was fielded between the end of December 2020 and the middle of January 2021, and the line is not drawn across that gap.Source: Democracy Fund + UCLA Nationscape, July 2019 to January 2021; weighted (survey weight)

    The change of president shows up before the inauguration. From the last full week before the election to mid-January the Democratic share rose from 10.3%10.3% percent of partisans who said the economy got betterEstimandweighted share saying better in one weekly waveInterval95% CI 8.5% to 12.5%n2,424Weightsweight (Nationscape survey weight)PopulationDemocrats among Nationscape respondentsPeriod2020-10-26 to 2020-11-01SourceNationscape to 18.2%18.2% percent of partisans who said the economy got betterEstimandweighted share saying better in one weekly waveInterval95% CI 15.2% to 21.7%n1,647Weightsweight (Nationscape survey weight)PopulationDemocrats among Nationscape respondentsPeriod2021-01-12 to 2021-01-16SourceNationscape and the Republican share fell from 28.2%28.2% percent of partisans who said the economy got betterEstimandweighted share saying better in one weekly waveInterval95% CI 25.2% to 31.4%n2,383Weightsweight (Nationscape survey weight)PopulationRepublicans among Nationscape respondentsPeriod2020-10-26 to 2020-11-01SourceNationscape to 18.6%18.6% percent of partisans who said the economy got betterEstimandweighted share saying better in one weekly waveInterval95% CI 15.2% to 22.5%n1,222Weightsweight (Nationscape survey weight)PopulationRepublicans among Nationscape respondentsPeriod2021-01-12 to 2021-01-16SourceNationscape, so the lines met before the new president took office. The Republican share in the election week itself, 33.6%33.6% percent of partisans who said the economy got betterEstimandweighted share saying better in one weekly waveInterval95% CI 30.2% to 37.1%n1,808Weightsweight (Nationscape survey weight)PopulationRepublicans among Nationscape respondentsPeriod2020-11-02 to 2020-11-08SourceNationscape, is above the share in the week before, but the intervals overlap. By the last week of January the shares were 23.6%23.6% percent of partisans who said the economy got betterEstimandweighted share saying better in one weekly waveInterval95% CI 20.3% to 27.2%n1,358Weightsweight (Nationscape survey weight)PopulationDemocrats among Nationscape respondentsPeriod2021-01-25 to 2021-01-31SourceNationscape and 13.3%13.3% percent of partisans who said the economy got betterEstimandweighted share saying better in one weekly waveInterval95% CI 10.6% to 16.5%n1,155Weightsweight (Nationscape survey weight)PopulationRepublicans among Nationscape respondentsPeriod2021-01-25 to 2021-01-31SourceNationscape. Several things changed between November and January, including the election result, and the survey cannot say which of them moved answers.

    Belief and expression give two readings of the gap

    The premium shows that partisans answer differently. It does not show why. One reading is that partisans perceive different economies. Bartels found that Democrats and Republicans differ in their perceptions of events that have an objective answer [3]. A panel around the 2006 midterm election found that partisans' economic assessments moved apart after control of Congress changed, and the authors attribute the movement to different views of the parties' economic competence [4].

    A second reading is that partisans give congenial answers when asked, even when they know or could infer less flattering facts. In two surveys with experiments, monetary payments and appeals for accuracy reduced the partisan gap in reports of economic conditions [5]. Small payments for correct answers also narrowed partisan gaps on other factual questions [6]. Behavior gives mixed evidence. One study found that consumption changed after a presidential election in line with a county's partisan makeup [7], and a later study of administrative spending data found that partisan shifts in expectations after the 2008 and 2016 elections had no effect on household spending [8].

    Nothing in the ANES, CES or Nationscape items can separate these readings. Chapter 2.1, Feelings toward the other party covers how partisans feel about each other. Chapter 6.2, Trust in government (planned) asks whether trust in the federal government also follows the party in the White House. Views about economic policy, such as trade and taxes, are a different subject from perceptions of how the economy performed, and Chapter 4.6, Trade and economic populism (planned) takes them up. Chapter 1.3, Sort dates by attitude dates when abortion, religion and racial resentment became partisan divides and finds no sort date for economic perceptions, whose gap follows the party of the president.

    What these data cannot show

    The surveys record what partisans say in one interview, so they cannot separate different information from a wish to praise one's own side. A partisan who answers "worse" may have read different news, may live in a region or work in an industry where the year went differently, may hold a different standard for what counts as better, or may be signaling loyalty. The question asks about the nation over one year, and the data hold no measure of what actually happened to unemployment, prices or income. The chapter reads the low-premium surveys through the share of each party saying worse and through presidential approval, and it cannot say how much of any survey's premium reflects conditions.

    The comparison of 1980 to 1990 with 2012 to 2024 also mixes party attachment with conditions. The earlier surveys include 1980 and 1990, when no more than 4.3%4.3% percent of partisans who said the economy got betterEstimandlargest of the four shares (both parties, both surveys)Weightsweight (VCF0009z)Populationpartisans of either partyPeriod1980 and 1990SourceANES Cumulative Data File of either party said the economy got better, which lowers the earlier average, and the analysis cannot say what the premium would have been in those years with better conditions. The comparison rests on six and four surveys (four and four without 1980 and 1990), and the premium differs across surveys by more than sampling error, so the comparison of 1982 to 1988 with 2012 to 2024 cannot separate a higher premium from year-to-year variation. The CES starts in 2006, so it cannot check whether the premium was lower before then, and its own slope is too imprecise to confirm or rule out a rise since. Party identification is measured in the same interview as the economic judgment. A respondent whose view of the economy shapes their party identification would appear as a partisan with a matching view, and repeated cross-sections cannot separate that from partisans changing their view.

    The intervals for shares, premiums and gaps treat each weighted sample as a simple random sample. Real surveys select respondents in clusters, such as households in the same area, and in strata, and that can make the true intervals wider or narrower. The CES and Nationscape recruit online respondents who chose to join a panel, and their weights adjust for known traits only. The ANES has no weights for interview mode in 2020, so the mode comparison covers 2012, 2016 and 2024. The Nationscape party measure leaves out leaners, and its weekly estimates overlap the pandemic, so they cannot isolate the effect of any single event. The CES premium differs between even-year and odd-year surveys for reasons the file does not give.

    Data sources

    • ANES Cumulative Data FileCumulative Data File 1948-2024, surveys from 1980 to 2024: retrospective economy (VCF0870), approval of the president's job (VCF0450), party identification (VCF0301), president in officeYears: 1948-2024 (33 study years; even years only after 2004; no 2006, 2010 or 2014 study) · Weights: weight (VCF0009z); weight_ftf and weight_web for the mode comparison · License: Free download after accepting the ANES terms of use; cite the release.
    • Cooperative Election StudyCumulative common content 2006-2024: retrospective economy (economy_retro), 7-point party identificationYears: 2006-2024 (every year; self-reported turnout is available in the 10 even years) · Weights: weight (year-specific) · License: See the provider's terms of use.
    • NationscapeWeekly waves July 2019 to January 2021: retrospective economy (economy_better_scale), 3-category party identificationYears: 2019-07-18 to 2021-04-15 (85 weekly waves) · Weights: weight (survey weight) · License: Free after registration and acceptance of the Voter Study Group terms; also on Harvard Dataverse.

    All data sources

    Methods

    Sample: ANES Cumulative Data File respondents in the 18 surveys from 1980 to 2024 that ask whether the national economy got better, stayed about the same or got worse over the past year (VCF0870), who identify with or lean toward a party (7-point party identification, 1 to 3 Democratic and 5 to 7 Republican; pure independents are left out) and carry a positive weight: 42,864 respondents. CES cumulative common content respondents in the 19 surveys from 2006 to 2024 (item economy_retro, recoded so that somewhat better and much better count as better and 'not sure' is missing; party by the 7-point scale with leaners): 568,229 respondents. Nationscape weekly waves from July 2019 to January 2021 (economy_better_scale, 3 = better), Democratic and Republican identifiers only because the file's 3-category party identification has no leaners: 324,560 respondent-weeks in 80 waves. Estimand: for each survey year, the weighted share of partisans who say the economy got better, by party and by relation to the president; the in-party premium is the share among partisans of the president's party minus the share among partisans of the other party, in percentage points. The president's party is the party of the president in office at the time of interview (the ANES cumulative file's president variable; for the CES, which is fielded each autumn, the president in office that autumn: Bush 2006 to 2008, Obama 2009 to 2016, Trump 2017 to 2020, Biden 2021 to 2024). Weights: ANES weight (VCF0009z); weight_ftf and weight_web for the interview-mode comparison in 2012, 2016 and 2024 (2008 was face-to-face only and 2020 has no mode weights); CES year-specific weight; Nationscape survey weight. Intervals: 95% intervals; a share uses the Wilson interval on the Kish effective sample size; a premium or gap combines the two normal standard errors in quadrature; an average of yearly estimates counts each survey once and combines the yearly standard errors in quadrature, so its interval covers sampling error within each survey only and not the spread between surveys (the CES presidency averages carry no interval and are shown with their lowest and highest yearly values); the difference between two periods is also given with a Welch t interval on the yearly premiums, which treats each survey as one observation and lets the two periods differ in spread; the difference between the face-to-face and web premiums combines the two standard errors in quadrature; a slope is the least-squares slope of the yearly premium on the survey year with a t interval on the number of surveys minus two. The intervals for shares, premiums, gaps, era averages and mode differences are model based (a weighted simple random sample) and ignore clustering and stratification; the Welch and slope intervals take the yearly premiums as the observations. Suppression: a party-year cell needs at least 100 respondents and an effective sample of at least 100, a survey needs at least 150 respondents on the item, and a Nationscape party-week cell needs at least 500 respondents; no ANES or CES cell fell below a floor (the smallest cell holds 435 respondents), and 80 weekly Nationscape waves meet the weekly floor while the thin waves after January 2021 do not. Item differences: the ANES item has three answers and the CES item five, so levels differ across the two surveys. Timing: the ANES surveys of 1982, 1986, 1990, 1994 and 1998 have post-election interviews only (no respondent has a pre-election interview in the cumulative file).

    References

    1. Wilson, Edwin B. (1927). Probable Inference, the Law of Succession, and Statistical Inference. Journal of the American Statistical Association 22(158): 209-212. doi:10.1080/01621459.1927.10502953
    2. Kish, Leslie (1965). Survey Sampling. John Wiley and Sons.
    3. Bartels, Larry M. (2002). Beyond the Running Tally: Partisan Bias in Political Perceptions. Political Behavior 24(2): 117-150. doi:10.1023/A:1021226224601
    4. Gerber, Alan S.; Huber, Gregory A. (2010). Partisanship, Political Control, and Economic Assessments. American Journal of Political Science 54(1): 153-173. doi:10.1111/j.1540-5907.2009.00424.x
    5. Prior, Markus; Sood, Gaurav; Khanna, Kabir (2015). You Cannot Be Serious: The Impact of Accuracy Incentives on Partisan Bias in Reports of Economic Perceptions. Quarterly Journal of Political Science 10(4): 489-518. doi:10.1561/100.00014127
    6. Bullock, John G.; Gerber, Alan S.; Hill, Seth J.; Huber, Gregory A. (2015). Partisan Bias in Factual Beliefs about Politics. Quarterly Journal of Political Science 10(4): 519-578. nowpublishers.com/article/Details/QJPS-14074
    7. Gerber, Alan S.; Huber, Gregory A. (2009). Partisanship and Economic Behavior: Do Partisan Differences in Economic Forecasts Predict Real Economic Behavior?. American Political Science Review 103(3): 407-426.
    8. Mian, Atif R.; Sufi, Amir; Khoshkhou, Nasim (2023). Partisan Bias, Economic Expectations, and Household Spending. Review of Economics and Statistics 105(3): 493-510. direct.mit.edu/rest/article/105/3/493/100979/Partisan-Bias-Economic-Expectations-and-Household

    How to cite

    @misc{singh2026perceptionsoftheeconomy,
  author = {Singh, Vishal},
  title = {Perceptions of the economy},
  year = {2026},
  url = {https://vishalsingh.org/politics/perceptions-of-the-economy},
  note = {In The Public Mind. Published 30 September 2026.}
}