I have data with death rates for each year 1921 to 2014; and for each age 0 to 110 as below:
Age X1921 X1922 X1923 ...
0 0,059987 0,045943 0,055646
1 0,012064 0,008851 0,009971
2 0,005779 0,003822 0,004315
3 0,002889 0,002971 0,002576
4 0,003254 0,002268 0,002199
...
I want to model the distribution of death rates for a specific age over the years (which is time series analysis). By looking the mortality curves for ages, it can be seen that around the age 15-18, there is a breaking point (or jump) and this completely differentiates the distribution of death rate into two different data structure (because mortality increases by getting older).
However, although this is seen by visually; I want to conduct a statistical test which can give me the exact age for the jump. Actually I performed Chow test, but since I want to find cut off value for age, I have taken independent variable as the age (mean death rate for total years ~ age) not the years which means this can not be suitable for Chow test (because I think it is not time series somehow).
I would be very greateful if you can give me an advise what analyses I should try.
Thank you very much
