Residential_Mobility_and_Kinship_Network_Size_Radaris_Anonymized

Full documentation -> [Radaris Research](https://radaris.com/research/) A person-level view of residential mobility and the size of a person's recorded kinship network, by age group and U.S. Census region. One row per person, coarse generalized attributes only. A person-level, de-identified dataset for studying how residential mobility relates to the size of a person's recorded kinship network, across age groups and U.S. Census regions. Each row is one person, described only by coarse, generalized attributes. ~750,000 rows. How this differs from the state-migration dataset. This dataset is not about origin->destination flows. It carries no state pair - geography is generalized to the four Census regions, and the focus is on mobility behaviour (did the person ever move states, how many moves, how large is their kinship network), not on which states they moved between. Use the state-level migration dataset (U.S._Inter-State_Migration_by_Age_Radaris_Anonymized) for directional flows; use this one for the mobility-kinship relationship and life-course mobility. ## What each row represents - person_id: Random surrogate id - not derived from any real identifier, not reversible, and not shared with any other dataset (cannot be joined). Row key only. - age_group: Age band from birth year: <25, 25-39, 40-54, 55-69, 70+. - region: Current U.S. Census region: Northeast, Midwest, South, West. - moved_state: Whether the person has ever lived in more than one state (1 = moved, 0 = stayer). - moves_bin: Number of distinct-location moves in the residential history, binned: 0, 1, 2, 3+ (consecutive duplicate locations collapsed first, so a repeated address is not counted as a move). - rel_bin: Recorded-relatives bucket: none, 1, 2, 3-4, 5+. ## The headline relationship (and how to read it) Within every age group, the more relatives a person has recorded, the less likely they are to have moved states. Because the pattern holds inside each age band - with age therefore held fixed - it is not merely an age artefact. Share who ever moved states, by recorded relatives, controlling for age (reference figures): age_group | 1 | 2 | 3-4 | 5+ <25 | 15.4% | 9.0% | 4.3% | 5.2% 25-39 | 29.0% | 25.9% | 20.9% | 24.5% 40-54 | 35.6% | 33.8% | 28.5% | 27.0% 55-69 | 35.1% | 31.3% | 27.3% | 26.2% 70+ | 33.6% | 29.4% | 27.7% | 27.1% The gradient is consistent and directionally intuitive (larger recorded networks <-> lower interstate mobility), which aligns with migration theory in which social ties anchor people in place. The dataset lets researchers reproduce and extend this with their own controls. ## Important caveats (read before drawing conclusions) - rel_bin is recorded relatives, not family size. It counts relatives found in public-records data (max 20 slots at the source). none means "none found" - a thin profile - not "no family". The count is partly a measure of profile completeness. Completeness bias would not by itself produce the observed pattern (there is no reason fuller profiles would move less), so it more likely adds noise than creates the signal - but treat the count as a proxy, not a headcount. - Association, not causation. This shows a relationship; it does not establish that kinship ties cause lower mobility. Reverse direction (staying accumulates local recorded kin over time) or common causes are equally consistent with the data. - The gradient is not perfectly monotonic. In some age bands 3-4 dips below 5+. Use the full table rather than assuming a strict "more relatives -> less movement" step. - Region, not state. Geography is generalized to four Census regions; sub-regional and state-level patterns are not recoverable here (that is the other dataset's job). - Endpoint-based mobility. moved_state compares first and current state; moves_bin counts location changes in the recorded history, which may be incomplete. ## What you can do with it - Reproduce the mobility x kinship-network relationship with your own age (and region) controls; test its robustness. - Study life-course mobility: how moved_state and moves_bin vary across age bands and regions. - Use it as a large, person-level companion to survey data, which typically has far smaller samples for this kind of cross-tabulation. ## How it was built 1. Endpoints & moves derived from each residential history, with consecutive duplicate locations collapsed and military postal codes (AA/AE/AP) removed. 2. Generalization: state -> Census region; exact move count -> bins; birth year -> age band. 3. Kinship count without links: number of filled relative slots only - relative names and ids were never carried forward, so no relationship graph is produced. 4. Sampling: a uniform random sample of 750,000 was drawn from records with complete generalized attributes, so its distributions estimate that population. 5. k-anonymity (k=5) over {age_group, region, moved_state, moves_bin, rel_bin}; profiles in cells smaller than 5 were suppressed (78 rows, <0.01%). Because person_id is a fresh surrogate not shared with other releases, this dataset cannot be linked to them at the individual level. ## Provenance & terms The underlying data is provided by Radaris, a people-search platform with an extensive database of public records and demographic information in the United States, the same source used for the companion U.S._Inter-State_Migration_by_Age_Radaris_Anonymized dataset. The source material has been stripped of all personal identity elements and synthesized into an aggregated, anonymous format, intended strictly for statistical, demographic, and academic research. ## Contact [e-mail]