male, near the age of 30, most likely from the traditional or central region, and having
been recovered during a summer month.
Table 3. Descriptive Statistics for Independent Variables
Ind. Variable (N = 752)
Mean/Percent
Std. Dev.
Min.
Max
FY
2003.33
2.982
1990
2007
TOPD
0.414
0.493
0
1
Summer
0.772
0.419
0
1
Female
0.233
0.423
0
1
Age
30.59
10.46
3
66
North
0.147
0.354
0
1
Central
0.291
0.455
0
1
Traditional
0.199
0.399
0
1
South-Southeast
0.252
0.434
0
1
Non-Mexican
0.110
0.313
0
1
Multinomial Logistic Regression and Results
This section introduces and discusses the inferential analysis used in the study.
Although ideal, estimating the probability of a UBC death occurring would be nearly
impossible given the unavailability of data regarding UBCs who successfully cross as
well as the relatively limited data available on UBC bodies that are actually recovered.
Although I am not able to estimate an overall probability of a UBC death occurring, I am
able to estimate the probability of a UBC having died a specific way given that she or he
died and was recovered in a southern Arizona. Thus, the focus of this study is to assign
recovered UBC bodies the probability of a confirmed cause of death given a set of
demographic characteristics as well as where, when and by whom the body was
recovered.
As previously mentioned, series of log likelihood ratio tests were utilized to arrive
at the final model. Each theoretically relevant independent variable was incrementally
added to the model and was tested for overall improvement of the model. All
independent variables used in the analysis improved overall fit of the model, thus
justifying their inclusion. Also, a series of interaction effects that were related to
biological sex and region of origin were tested, however, none yielded statistically
significant results. This was most likely related to the low frequencies of interaction
effect variables.
Table 4 describes the multinomial logistic regression results for the analysis
sample. Coefficients for three causes of death were estimated: exposure, motor vehicle
11
Convention