University of Delaware
|Topic:||Stability and Equilibrium in Structural Equation Models|
In most social research and much medical/biological work, controlled experiments are not feasible. This constraint has stimulated development of analytic methods for teasing out "cause-and-effect" relationships in the absence of experiments.
Most of the work in structural modeling does not incorporate explicitly time into the models, and neither do many discussions of causation. In fact, most empirical work using the structural-equation framework is done with cross-sectional data. One way to view a system of cross-sectional structural equations is through the lens of a dynamic system in equilibrium. From the vantagepoint of a set of linear difference or differential equations, cross-sectional estimation of structural equations generates inconsistent parameter estimates. The bias is most troubling when the system is not in equilibrium at the time the date were collected.
This presentation reports the results of simulations designed to assess the impacts on parameter estimates of using cross-sectional data obtained when the dynamic system was still in flux. We find that the cross-sectional estimates depend greatly on the point in the time path when the system is measured and whether the time path to equilibrium is monotonic or oscillatory.
|Thursday, April 13, 2000 |
|Time:||Social hour -- 6:00 pm|
Dinner -- 6:30 pm
Speaker -- 7:15 pm
|Place:||Newark Holiday Inn|
|Menu:||Appetizer: V-8 juice |
Salad: Tossed salad
Entree: Stuffed shrimp over rice pilaf, Vegetables du jour
Dessert: Pear belle Helene
|Cost:||$20 for members, $5 for students|
Reservations to Jane Buck by noon, Monday, April 10, 2000.
Please note: Be sure to RSVP by this date, to assure a meal is reserved for you.
Return to Delaware Chapter ASA home page.