Johnston] NAVAHO POPULATION 177 



Granting the severe limitations of these matching studies, it is 

 nevertheless evident that their resuhs lend further support to the con- 

 clusion that the reporting of vital events among the Navaho is 

 extremely unreliable, at least until the very recent past. 



MODEL AGE DISTRIBUTIONS 



The 10 model age distributions, shown in table 36, were developed 

 for purposes of comparison with the Navaho age distributions whose 

 summary characteristics are given in table 28. Each model is derived 

 from two basic parameters; an assumed life expectancy at birth (for 

 both sexes combined) and an assumed rate of natural increase. 



In selecting the values to be assigned to these parameters, an effort 

 was made to approximate conditions which might have obtained 

 among the Navaho population at different periods in the past. For 

 example. Model A, derived from an assumed life expectancy of 30 

 years and a rate of natural increase of 1.8 percent per year, was in- 

 tended to approximate conditions which might have existed during 

 the latter part of the 19th century. Similarly, Models B and C reflect 

 the age distribution of a population whose mortality level is still rel- 

 atively high, having a life expectancy of only 35 years. In Model B, 

 the accompanying growth rate is assumed to be 2.0 percent per year, 

 while in Model C, the assumed growth rate lies at 2.25 percent per 

 year. 



The remaining models are designed to approximate the effects of 

 progressive declines in the general level of mortality, in combination 

 Avith a constant or slightly declining level of fertility. In order to 

 facilitate comparisons between the model age distributions and those 

 of the Navaho population, the median age for each sex, the child- 

 woman ratio, and the implied crude birth rate derived from each 

 model are also given in table 36. 



Comparing the median ages of the 10 models with those given in 

 table 28, we find that most of the Navaho medians fall below the model 

 medians. This is especially striking in view of the fact that except 

 for Model </, the lowest model medians (obtained from Models A and 

 C) pui-port to reflect the combined effects of very high levels of fertil- 

 ity and mortality. The crude birth and death rates that are im- 

 plied in Model A are 52.2 and 34.2 per 1,000, respectively. The cor- 

 responding rates implied in Model C are 50.8 and 28.3 per 1,000, 

 respectively. If we assume that either of these models represents 

 plausible upper limits for the fertility and mortality levels experienced 



prolonged delay in registering a birth means that, in some instances, both the birth and 

 the deatli will go unrecorded simply because the infant died before his birth was recorded. 

 In areas experiencing substantial infant mortality, this kind of delayed registration could 

 result in a reduction of the death rate, as derived from reported deaths, by as much as 10 

 or 15 percent, while it would reduce the derived birth rate by no more than 3 or 4 percent. 

 See, in this connection, Winnie, 1959, ch. 13, p. 8. 



