CALIFORNIA STATE UNIVERSITY LONG BEACH
PPA 696--RESEARCH METHODS:
BINGHAM & FELBINGER CH. 5
  1. BIBLIOGRAPHIC INFORMATION
    1. Author: F. Skidmore
    1. Title: Overview of the Seattle-Denver Income Maintenance Experiment Final Report
    1. Source: U.S. Dept. of Health and Human Services, May 1983

      2.  
  1. SUMMARY OF THE RESEARCH
    1. PROBLEM STATEMENT:
What would be the effect of a negative income tax on the labor force participation of lower income family members?
    1. BACKGROUND:
Could cash welfare programs be replaced by a general income-transfer program?
    1. HYPOTHESIS:
A cash transfer program will decrease the labor force participation of low-income persons
    1. MEASUREMENT OF VARIABLES
      1. Dependent variable: hours worked per year
      1. Independent variable(s): cash transfer program; counseling and training allowance program.
      1. Control variable(s): marital status
    1. RESEARCH DESIGN:
An experimental design with multiple types of treatments and multiple levels of each type, to test the effects of higher or lower levels of payment; called a factorial design. In all there were 48 possible combinations of treatments for the 3-year plan and 48 for the 5-year plan. This is the design for each level of payment; there were 12 different possible levels of payment.
Group First Obs. Payments Counseling Training Allowance Final 

Obs.
G-1 O1 X X X On
G-2 O1 X X On
G-3 O1 X On
G-4 O1 X X On
G-5 O1 X On
Control O1 On
 
    1. SAMPLING:
5,000 low-income families randomly assigned to either 3-year or 5-year treatments; to levels of payment; and to counseling only or counseling with training allowances.
    1. INSTRUMENTATION:
Data collected from self-report as well as from public records
    1. DATA COLLECTION/ETHICS:
Data collected three times per year and for one additional year after leaving the program; families were originally told the program would last 20 years but it was terminated after 5 years.
    1. DATA ANALYSIS:
Various techniques were used (not reported in the article)
    1. CONCLUSIONS:
1) Cash transfers reduce the number of hours worked for husbands, wives, and female heads of households

2) Counseling/training also reduced the number of hours worked

3) The program increased the divorce rate among all ethnic groups

  1. CRITIQUE
    1. Possible Threats to Internal Validity
      1. History:
effects are controlled by control group
      1. Maturation:
effects are controlled by control group
      1. Testing:
participants may have altered their behavior due to testing
      1. Instrumentation:
Experimental groups learned more quickly how to correctly complete the required reports than the control groups, so hours of work had to be used instead of gross earnings to control for errors in reporting between the two types of groups.
      1. Regression Artifact:
Families were equally needy
      1. Selection bias:
Families tended to be larger than in the general population; no disabled family heads; earnings between $9,000 and $11,000 per year; participation in counseling and training was voluntary, meaning there may have been some self-selection bias.
      1. Experimental Mortality:
20% of husbands, 15% of wives, and 15% of single heads of household dropped out of the study in the first 30 months; dropouts may have had different hours of work; researchers attempted to estimate differences for those who dropped out.
      1. Design contamination:
All participants were still eligible for other public assistance programs (AFDC, Food Stamps, unemployment, etc) which may have influenced the behavior of the participants.
    1. Possible Threats to External Validity
      1. unique program features:
Not all possible combinations of income levels, counseling, and training were tested, so results apply only to those on which data exists.
      1. experimental arrangements:
Conducted in natural settings; counseling and training were carried out at Community Colleges
      1. other threats:
Program had a limited time span, which would not be true of a national entitlement program.