After the nuclear accident at Three Mile Island in Spring 1979, mothers in the area were surveyed four times to determine their levels of stress. 115 mothers living within 5 miles of the plant were assigned to group LT5, another 152 mothers living between 5 and 10 miles away were assigned to group GT5. Data are from Fienberg SE, Bromet EJ, Follmann D, Lambert D, May SM 1985. Longitudinal analysis of categorical epidemiological data. Environmental Health Perspectives, 63, 241-248.

 

For this analysis, each mother at each time point is regarded as a random example of the fixed stress effect of one of the two groups. The analysis estimates the mean stress level of each group of mothers at each time point by averaging their stress levels as recorded in their ratings.

A better analytical technique is in the Arithmetic Test example.

 

Facets specifications and data (in file 3mile.txt):

 

Title = Three Mile Island - Mother's Stress (Fienberg et al.)

Facets = 1  ; only facet is group at time point

Noncenter=1  ; ratings on elements define their own origin

Convergence = .001,.0001 ; very precise convergence criteria

Vertical = 1A,1N,1* ; display facet 1 by name, number and distribution

Yardstick = 0,10 ; vertical columns as needed, lines 10 per logit.

Model = ?,Stress ; scale of stress for mothers

Rating (or partial credit) scale = Stress,R3 ; Stress has three levels

1=Low

2=Medium

3=High

*

Labels =

1=Group/time points

1=LT5 Winter 1979 ;Find out the effect for each group/time point

2=LT5 Spring 1980

3=LT5 Fall 1981

4=LT5 Fall 1982

5=GT5 Winter 1979

6=GT5 Spring 1980

7=GT5 Fall 1981

8=GT5 Fall 1982

*

data =

R2,1-4,1,1,1,1 ; 2 LT5 Mothers always low stress: R2 indicates datum replicated twice

|

1-4,2,1,1,2 ; 1 LT5 mother

R1,1-4,2,1,2,1 ; R1 is not needed here - shown for illustration

|

R7,5-8,3,3,3,3 ; R7   7 GT5 Mothers always high stress