Democratic
(Dis)Connections

Lab directed by Sébastien Parker

Research

Research

01  ·  Over time

Cohorts and social change

Over time, which divides narrow across cohorts and which widen?

More about this work: Cohorts and social change

An animation compares two groups of Americans born from 1930 to 1990, one General Social Survey question at a time, for eight of the more than 100 questions in our study. Each curve traces social change, or change across successive cohorts, in the predicted share of a group giving the answer shown, estimated separately for each group, and the distance between the two curves is the gap between the groups. Bands are 95 percent confidence intervals, and gaps are in percentage points. Democrats and Republicans include those who lean toward a party. First come four questions on which the gap between the groups widened from those born in 1930 to those born in 1990. By party, the gap on allowing legal abortion if a woman wants it for any reason goes from Republicans 4 points higher to Democrats 51 points higher. By religious attendance, the gap on saying same-sex relations are not wrong at all goes from 9 points to 41, higher among those attending less than weekly. By party, the gap on attributing Black people’s worse jobs, income, and housing mainly to discrimination goes from 6 points to 59, higher among Democrats. By party, the gap on supporting preference for Black people in hiring and promotion goes from 5 points to 48, higher among Democrats. Then come four questions on which the gap narrowed. By race, the gap on supporting preference for Black people in hiring and promotion goes from 40 points to 5, higher among Black Americans. By race, the gap on attributing Black people’s worse jobs, income, and housing mainly to discrimination goes from 41 points to 16, higher among Black Americans. By education, the gap on allowing a gay man to make a speech in the community goes from 24 points to 4, higher among those with a bachelor’s degree or more. By education, the gap on allowing a speech by someone who believes Black people are genetically inferior goes from those with a bachelor’s degree or more 23 points higher to those with less 5 points higher.

02  ·  Among people

Social ties and belief systems

Among people, which connections bridge divides and which deepen them?

More about this work: Social ties and belief systems

An animation in three parts. First, networks of political positions are compared across New Brunswick, Nova Scotia, and Newfoundland and Labrador. Each line is a partial correlation between two positions, holding the others constant. A solid line means that people higher on one position tend to be higher on the other, and a dashed line that they tend to be lower on the other. The lines link a person’s answers, not people. Thicker lines are stronger links. Each province shows its stronger half of links (35 of 71, 43 of 86, and 31 of 63), and the left–right position is ringed. A line of text names the positions most connected to the others, which stay among the strongest across resamples of respondents. In New Brunswick these are left–right self-placement, carbon tax, environmental regulation, and openness to immigration; in Nova Scotia, carbon tax and environmental regulation; and in Newfoundland and Labrador, carbon tax, environmental regulation, and environmental spending. Left–right self-placement is less connected in Nova Scotia and in Newfoundland and Labrador than in New Brunswick. Second comes an illustration of a simulated discussion network after Dokshin and Parker (2026), with 80 agents in two parties, blue and orange. Blue agents use strategic framing. When many ties cross party lines, their framing persuades orange agents too, and opinion moves toward consensus. With partisan homophily and backlash, blue agents persuade one another and orange agents push back. A line of text gives the share of cross-partisan ties, 48 percent, then 11. Third, a network links confidence in four French institutions (the government, political parties, the justice system, and the army) to approval of three forms of government (a strong leader, experts, and army rule), in 2013, 2024, and 2025. A solid line means that more confidence goes with more approval, and a dashed line with less, holding the other five answers constant. Thicker lines are stronger links. Confidence in the justice system goes with less approval of a strong leader and of army rule, and confidence in the army with more. These links point the same way in all twelve surveys from 2013 to 2025, and in most of them they are clearly different from zero. The French estimates are drafts.

03  ·  Across places

Places and public voice

Across places, who engages in democratic processes, and whose voices are amplified or muted?

More about this work: Places and public voice

A sequence of maps. France’s départements are ranked on the class composition index from our study of the Grand Débat, from less advantaged in blue to more advantaged in amber. The map then zooms into Paris and its three neighbouring départements. Communes and Paris arrondissements are shaded by median income adjusted for household size in 2018, against the median for metropolitan France of €21,730, from €10,900 or less in blue to €43,500 or more in amber. Each département’s label gives its Grand Débat contributors per 10,000 residents, from 72.6 in Paris and 58.8 in Hauts-de-Seine to 34.4 in Val-de-Marne and 15.7 in Seine-Saint-Denis. Neighbourhoods across Atlantic Canada, then in Halifax and on the island of Montréal, are shaded by median household income after taxes in 2020, from $36,500 or less in blue to $146,000 or more in amber, against Canada’s median of $73,000. Grey marks areas where Statistics Canada withholds the figure, and sparsely settled areas are drawn fainter.

Three charts from our study of France’s Grand Débat. First, coverage. About a third of the answers (35 percent) lie in groups of similar answers that the official summary’s categories do not cover. For comparison, we use summaries in the same format, with as many categories, written from samples of 300 answers to each question. They leave out 28 percent, and the official summary leaves out more in 65 of the 75 questions. That gap depends on how the categories are counted. Counted subcategory by subcategory, the official summary leaves out less. Second, terms. In 69 of the 75 questions, the official summary is closer than the average answer to a vocabulary of rules, implementation, and the allocation of resources. Third, direction. Where the official summary differs from what people wrote, it leans toward the answers from more advantaged départements, and not only because they sent more answers. Its lean is 0.19 on a scale from −1 to 1, averaged over the 73 questions with enough answers, above the range we get by shuffling class composition across départements. The written summaries lean the same way and about as much (0.18), and headings taken from the answers a little less (0.15), so the lean is not particular to the official summary.

People

Sébastien Parker

Sébastien Parker

Director

Assistant Professor of Sociology and Political Science, University of Prince Edward Island

Student collaborators

  • 01 · Over time

    Tom Chen

    PhD student in Sociology, University of Minnesota

    Tom studies inequality and political divides, with a focus on how poverty and health are organized across cities. His work examines how the educational divide in political attitudes widens or narrows with age and across birth cohorts, in a collaborative project with the Social Change Lab at the University of Toronto.

    Website

  • 01 · Over time

    Andraya Gallant

    MA student in Sociology, Queen’s University

    Andraya studies trust in public institutions, with a focus on policing and policy. With the lab, which she joined as an undergraduate at UPEI, she examines how values and confidence in institutions change across Canadian cohorts.

  • 02 · Among people

    Ana Brinkerhoff

    Master of Public Health student, École des hautes études en santé publique (EHESP), Rennes

    Ana studies public institutions and the people they serve, with a focus on health policy for aging populations. With the lab, she examines how confidence in representative institutions and the justice system relates to approval of democratic and alternative forms of government in France.

  • 03 · Across places

    Lydia Doyle

    Undergraduate student, University of Prince Edward Island

  • 03 · Across places

    Aidan Murphy

    Undergraduate student, University of Prince Edward Island

  • 03 · Across places

    Kelly Tuttle

    Undergraduate student, University of Prince Edward Island

  • 03 · Across places

    Maria MacKinnon

    Undergraduate student, University of Prince Edward Island

Write to us