Skip to content
Back to search results

Historical Social Mobility Tables

Introduced by
Robert Erikson and John H. Goldthorpe; log-linear methods from Leo Goodman
Year
1992
Type
descriptive-tabular
Family
social-mobility-measurement
Font size

What does it do?

Historical social mobility tables measure how much a person's social position depended on the position of their parents in past societies.

The core device is the mobility table: a cross-tabulation of origin class (typically the father's) against destination class (the child's), built from linked parent-child pairs drawn from marriage registers, censuses, or genealogies. Following the framework Erikson and Goldthorpe codified for modern sociology and that historians adapted using HISCLASS, the table is analysed not by raw movement, which is dominated by changes in the class structure itself, but by odds ratios and log-linear models that isolate relative mobility, the strength of association between origins and destinations net of structural change. This distinction between absolute and relative mobility lets historians ask whether genuine fluidity, equality of opportunity, rose or fell across industrialization, migration, and demographic transition, independent of how the shape of the class structure shifted.

The intuition behind the method

Suppose far more people end up in the middle class than their fathers occupied. Is that because society became more open, or simply because industrialization created many middle-class jobs that someone had to fill? Raw mobility cannot tell these apart, because it mixes two things: the changing shape of the class structure (structural mobility) and the underlying tightness of the link between where you start and where you end (relative or exchange mobility). The mobility table separates them. By looking at odds ratios, comparing, say, the relative chances of a professional's versus a labourer's son of reaching the professional class, you measure the strength of inheritance independent of how many such positions exist. A society can show huge absolute movement while relative chances stay rigidly unequal. Log-linear models formalise this, fitting the association pattern in the table so historians can compare fluidity across cohorts, regions, or centuries on a common, structure-free footing.

History

Mobility tables matured in twentieth-century sociology, culminating in Erikson and Goldthorpe's The Constant Flux (1992), which used log-linear models on cross-national data to argue that relative mobility, social fluidity, was broadly similar across industrial societies. Leo Goodman's development of log-linear methods supplied the statistical machinery. Historians adopted the framework as digitised marriage registers, censuses, and genealogies made it possible to build origin-destination tables for the past, with HISCLASS providing comparable class codes. Projects on the Netherlands, Britain, Sweden, and elsewhere have applied these tools to test whether industrialization increased fluidity, often finding more persistence and complexity than modernization theory predicted, and pushing mobility measurement deep into the pre-industrial era.

When to use it

Use mobility tables when you have linked origin-destination class pairs and want to measure social fluidity in a historical population, especially to ask whether opportunity rose or fell across industrialization, migration, or reform. The log-linear apparatus is essential whenever you need to separate genuine changes in openness from changes in the class structure itself, and whenever comparing mobility across societies or periods with different class distributions. It requires reasonably large samples for stable odds-ratio and log-linear estimation. Prefer surname methods when individual linkage is impossible but group representation is observable, and prefer raw rates only for description, not for claims about underlying equality of opportunity.

Limitations and considerations

  • Requires linked parent-child pairs, which are hard to construct historically
  • Marriage-register samples over-represent grooms and under-represent women
  • Small samples yield sparse tables and unstable log-linear estimates
  • Captures only a single origin-destination snapshot, not lifetime trajectories

Source conditions do not replace independent scientific review or validation by an analysis engine.

Strengths

  • Cleanly separates structural from relative (exchange) mobility
  • Odds ratios are insensitive to changing class sizes, enabling fair comparison
  • Log-linear models support formal tests of changing fluidity over time
  • Builds directly on standardized HISCLASS origin and destination codes

Applications

  • Testing whether industrialization raised social fluidity in past societies
  • Comparing intergenerational mobility across regions and centuries
  • Measuring class persistence using marriage-register father-son pairs
  • Studying the mobility effects of migration and urbanization historically
  • Assessing how demographic and educational change reshaped opportunity

How it works

How it works

1

The analysis begins with a sample of origin-destination pairs, most often a father's class linked to his child's class at a comparable life stage. Historically these come from marriage registers that record both groom's and father's occupations, or from record linkage joining children to parents across censuses. Each occupation is coded to a class scheme, usually HISCLASS, so origins and destinations are measured on the same ladder, the prerequisite for a meaningful table.

Formula
\{(O_i, D_i)\}_{i=1}^{N}\ \text{origin-destination class pairs}
2

The pairs are cross-tabulated into a square mobility table whose rows index origin classes and columns destination classes, each cell counting people of a given origin who reached a given destination. The diagonal records immobility, staying in the origin class; off-diagonal cells record upward or downward movement. The table's marginals encode the origin and destination distributions, that is, the changing shape of the class structure between generations.

Formula
T = [n_{jk}],\ n_{jk} = \#\{i: O_i=j, D_i=k\}
3

From the table the analyst computes absolute mobility rates. Outflow percentages, row-based, show where children of a given origin ended up; inflow percentages, column-based, show the origins of those reaching a given destination. Total mobility is the share off the diagonal. These descriptive rates are intuitive but conflate structural and relative mobility, so they describe the lived experience of movement without yet isolating the openness of the society.

Formula
\text{outflow}_{jk} = n_{jk}/n_{j\cdot},\quad \text{inflow}_{jk} = n_{jk}/n_{\cdot k}
4

To capture relative mobility the analyst turns to odds ratios, which compare the relative chances of reaching one destination versus another for people from two different origins. Odds ratios are margin-insensitive: they are unaffected by changes in the size of classes, so they measure the pure association between origins and destinations. A table of odds ratios summarises the structure of inequality of opportunity independent of how many positions each class contains.

Formula
\theta_{jk} = \frac{n_{jj} n_{kk}}{n_{jk} n_{kj}}
5

Finally log-linear models formalise the analysis by modelling the logged cell counts as functions of origin and destination main effects plus an origin-by-destination association term. Specifying and constraining that association, through models of constant fluidity, uniform difference, or topological patterns, lets the analyst test whether relative mobility is stable or changing across cohorts, regions, or periods, and to quantify how the strength of social inheritance evolved net of structural shifts.

Formula
\log m_{jk} = \mu + \lambda_j^O + \lambda_k^D + \lambda_{jk}^{OD}

Common pitfalls

  • Interpreting high absolute mobility as openness without removing structural effects
  • Fitting log-linear models to sparse tables with many near-empty cells
  • Ignoring linkage bias that systematically misses the most mobile individuals
  • Comparing tables coded with different class schemes as if commensurable

Frequently asked questions

What is the difference between absolute and relative mobility?

Absolute mobility is the raw share of people who end up in a different class from their origin, heavily driven by changes in the class structure, the creation or destruction of positions. Relative mobility measures the strength of association between origins and destinations net of those structural changes, captured by odds ratios. A society can show high absolute mobility while relative chances remain rigidly unequal, so the two must be distinguished.

Why use log-linear models instead of simple percentages?

Percentages, outflow and inflow rates, conflate the effect of changing class sizes with the underlying openness of society. Log-linear models decompose the table into origin and destination margins plus an association term, isolating relative mobility and allowing formal tests of whether fluidity changed across cohorts or differed between societies. They make comparisons fair when the compared populations have different class distributions.

How do historians get origin-destination pairs?

The classic source is the marriage register, which records both the groom's occupation and his father's, yielding an origin-destination pair at marriage. Alternatively, record linkage joins children to parents across successive censuses or in genealogies. Each method has biases, marriage registers favour grooms and miss women, linkage misses the most mobile, so the construction of pairs is itself a careful, error-prone stage of the analysis.

Related methods

Sources

  1. Erikson, R., & Goldthorpe, J. H. (1992). The Constant Flux: A Study of Class Mobility in Industrial Societies. Clarendon Press.

    ISBN: 9780198273837
    Copied.
  2. van Leeuwen, M. H. D., & Maas, I. (2011). HISCLASS: A Historical International Social Class Scheme. Leuven University Press.

    ISBN: 9789058678577
    Copied.