260 likes | 421 Views
Modelling propensity to move house after job change using event history analysis and GIS. Marie-Hélène Vandersmissen (CRAD, Laval University), Anne-Marie Séguin (INRS-UCS), Marius Thériault (CRAD, Laval University) and Christophe Claramunt (Naval Academy Research Institute, France).
E N D
Modelling propensity to move house after job change using event history analysis and GIS Marie-Hélène Vandersmissen (CRAD, Laval University), Anne-Marie Séguin (INRS-UCS), Marius Thériault (CRAD, Laval University) and Christophe Claramunt (Naval Academy Research Institute, France) 2nd MCRI/GEOIDE PROCESSUS Colloquium on the Foundations of Integrated Land-Use and Transportation Models Toronto, June 12-15 2005
Introduction • Transportation land-use modelling must consider decision-making behaviour of urban actors using disaggregate data in order to relate • Activity location, home choice, commuting and travel decision • Household, individual and professional profiles of persons • Probabilistic discrete choice theory is becoming the central issue of urban and transport modelling research • Implemented using logistic and Cox regression techniques • Aimed at modelling individual’s and household’s behaviour • Need for spatio-temporal GIS for analysing urban and transport systems where • Uncertainties exist in the system (aggregation is not straightforward) • Emergent behaviour occurs • Decision rules for individuals and households are intricate • System processes are time-path and location dependent • Future system state depends partly on past and current states 2nd MCRI/PROCESSUS Colloquium, Toronto, June 12-15, 2005
Our project in the MCRI programme 2nd MCRI/PROCESSUS Colloquium, Toronto, June 12-15, 2005
Purpose • Emergent residential behaviours of individual actors in context of profound social changes in the work sphere • Long term-view in the analysis of the relationship between social changes in the work sphere and these behaviours Social changes Long-term dynamics of residential location behaviour Travel behaviour 2nd MCRI/PROCESSUS Colloquium, Toronto, June 12-15, 2005
Objective and Research Issues • Estimate the propensity for professional workers to move house after a change of workplace • How many will move house during the following job episode? • For how long will they delay that decision? • What are the factors significantly influencing that move house decision? 2nd MCRI/PROCESSUS Colloquium, Toronto, June 12-15, 2005
Data: The 1996 Retrospective Survey for Quebec City • Survey collecting, in one interview, information about all changes occurred over a long period of time, since their departure of the respondent’s parental home • Spatially stratified sample of two cohorts of professional workers • 418 respondents living in Quebec CMA in 1995 • Two cohorts (mid-thirty and mid-forty) • 224 women; 194 men • 112 women and 100 men in their mid-thirty • 112 women and 94 men in their mid-forty • Reporting on significant events occurred during their life time describing • Residential trajectory (every home occupied with their location) • Household trajectory (each change in the household’s composition) • Professional trajectory (each change in employer, each workplace) • Collecting dates of every change 2nd MCRI/PROCESSUS Colloquium, Toronto, June 12-15, 2005
Complex mix of real world phenomena Personal Biography Complex Evolution Processes Leading to at least one episode Change in status Combining facts describing a specific aspect of life Set of related lifelines using application-specific semantic relationships 2nd MCRI/PROCESSUS Colloquium, Toronto, June 12-15, 2005
Management of Evolution in Trajectories • We developped a generic spatio-temporal datamodel to handle historical orderings and querying patterns of facts in order to produce flat files needed for event-history analysis Application semantics Historical ordering of facts Facts : events and episodes Location of facts 2nd MCRI/PROCESSUS Colloquium, Toronto, June 12-15, 2005
Specifying spatial distance condition Specifying duration condition Specifying spatial location condition Specifying patterns of facts Specifying temporal conditions Specifying time ordering Specifying target fact Specifying other status condition Spatio-temporal Query of Patterns of Facts within Trajectories • We developped a query interface combining georelational GIS capabilities and temporal historical ordering of facts using ODBC links 2nd MCRI/PROCESSUS Colloquium, Toronto, June 12-15, 2005
Methodology: Event History Analysis • Ordinary multiple regression is ill-suited to the analysis of biographies • Censoring: refers to the fact that the value of a variable may be unknown at the time of survey • Considering time varying explanatory factors • Need to consider time-varying information to study the effect of job change on house moving • Event history analysis can handle such a problem (survival tables and logistic regression) • The query interface enhance data restructuring needed for this kind of statistical analysis 2nd MCRI/PROCESSUS Colloquium, Toronto, June 12-15, 2005
Event History Analysis (Cox Regression) • Survival tables are using conditional probabilities to estimate the mean proportion of people experiencing some change in their life after a significant event occurs, computing the time delay after a specified enabling event • Specific conditions may influence propensity to change • Requires a combination of survival tables and logistic regression to estimate the marginal effect of other personal attributes on the probability that an event occurs • Event History Analysis to model specific variations of the probability of state transition through time for individuals considering independent variables describing their personal situation on other lifelines 2nd MCRI/PROCESSUS Colloquium, Toronto, June 12-15, 2005
Probability to move home after a job change: 2nd MCRI/PROCESSUS Colloquium, Toronto, June 12-15, 2005
Number of pair of events (change of job-workplace versus moving house or not) Basic statistics • 380 respondents (on 418) had a change of job or workplace at least once during during their career • 411 respondents moved their home at least once after departure from parent’s home • 1056 changes of job or workplace within or towards the Quebec CMA (321 persons) • 458 of those changes of workplace were followed by at least one move house during the subsequent employment episode-stability of job and workplace • 598 of those changes of workplace were not followed by any move house during the subsequent employment episode (231 persons) 2nd MCRI/PROCESSUS Colloquium, Toronto, June 12-15, 2005
Basic variables for the Event History analysis 2nd MCRI/PROCESSUS Colloquium, Toronto, June 12-15, 2005
Descriptive Statistics 2nd MCRI/PROCESSUS Colloquium, Toronto, June 12-15, 2005
Empirical Results: 1. Cross-tables X2: 1,281 ddl:1 P: 0,258 X2: 0,495 ddl:1 P: 0,482 2nd MCRI/PROCESSUS Colloquium, Toronto, June 12-15, 2005
X2: 89,601 ddl:4 P< 0,000 C= 0,280 X2: 19,192 ddl:2 P< 0,000 C= 0,134 2nd MCRI/PROCESSUS Colloquium, Toronto, June 12-15, 2005
X2: 131,327 ddl: 4 P< 0,000 C= 0,333 X2=152,63 ddl: 2 P< 0,000 C= 0,355 2nd MCRI/PROCESSUS Colloquium, Toronto, June 12-15, 2005
Empirical Results: 2. Event History Analysis Tests of Model coeff. X2: 845,29 Df: 22 Sig.: 0,000
For how long will they delay that decision? 2nd MCRI/PROCESSUS Colloquium, Toronto, June 12-15, 2005
Discussion and Conclusion • Results given by Event History Analysis: • How many will move house during the following job episode? • On 418 respondents, 271 moved home after a job change (64,8%) • For how long will they delay that decision? • Probability of changing home after a job change =0,2 after ~2 years • What are the factors significantly influencing that move house decision? • Tenure • Co-tenant Tenant + • Owner Tenant + • Tenant Owner + • Tenant Tenant + • Gender (man) + • Increased Distance home job + • Number of Children - • Previous home duration - • Change of Home Neighbourood • New Suburbs + Fringe - • Old Suburbs - • Core - 2nd MCRI/PROCESSUS Colloquium, Toronto, June 12-15, 2005
Retrospective Survey • Inaccuracy of responses (limitations of human memory with elapsed time) • Memory distorsions (individual’s account of the event) • But people tend to remember major events (year of residential move, job change) • Results reflect situation in 80’s and 90’s • To the best of our knowledge, this type of application is original (residential move after a job change • Positive contribution to transportation land-use modelling (Quebec) • The query interface could be also used to analyse patterns of activity/travel decision coming from our panel surveys (Quebec & Toronto) and OD surveys • Next stage: Elaborate separate models for owners and tenants 2nd MCRI/PROCESSUS Colloquium, Toronto, June 12-15, 2005