Twenty years later the 1922 - 1922 dip due to war can still be seen clearly. The scars of a people are written in their demography. There is now a second dip, again in the 30 year old group.
The same features can be seen in the male & female separated distributions. On average the male - female birthing patterns are normal at around 50 %each. No major problems are indicated
The dip which was present in the 1971 30 year old group is still very marked, now in the 45 year old group. This second dip of course corresponds with the second world war. Few males were left in the village to help generate children.
In this histogram I have combine the aggregated male & female histograms for each census into one graph. This is to get a sense of the changing population growth. The series are not offset so a direct comparison of the numbers in each age set for each census year is possible.
This age histogram separates males and female. The population is growing at around 2.8 - 3.2% in accordance with Stirling's figures. From about 1925 onwards the population growth rate starts to increase. The biggest increases occur after 1950 slowing down again as we come into the mid 80's. This growth spurt corresponds with a dramatic drop in infant mortality probably bought along by the introduction of antibiotics and better health care. (See p 8.).
I calculated some statistics based on the database data and came up with the following:
1950 Infant Mortality rates approx. 3.0% (Stirling, 1965)
1950 Child Woman Ratio =.30331754 compared with Britain (1985) =.2591
1950 Crude Birth Rate ~ 39 (3.9%) Compared with England + Wales (1974) = 13.1 (1.3%) and Bangladesh (1974), 48.3 (4.83%)
1950 General Fertility Rate = 71.16 compared with USA (1985) = 66.1% & Bangladesh (1974) = 246.7
These are 1950 male and females age histograms separated. I decided to use another graphical representation to bring out the features and there are some interesting results. This graph represent the three census years for 1950; 1971 & 1986. The populations have been staggered in the graph so as to reveal continuity across age ranges i.e the 1950 population is seen here occupying their real position in the 1986 census. The population dips due to the first & second world wars can be seen as can the 'normal death' slope off in the 60 + age group.There is also a drop in the male and female bands of the over 30's in 1986 possibly indicating permanent migration from the village. Other points of interest are marked on the graphs.
Figs. 18 & 19
These are essentially the same graphs as 16 & 17 but presented this way are a little clearer for direct comparisons. They raise some interesting questions.
These two graphs use a log scale to allow comparisons between migrant and village populations, number of household heads and the mean size of families. The important feature here is that the average house hold size remains fairly constant at around 6 members. There is a drop in this figure in the migrant data at first but every indication is that the preferred household size will rise again to around 6. Notice the steep migrant outflow between 1950 and 1971. Thereafter it slowed down. This can be partially accounted for due to a clamp down on Germany work permits. In the early 70's the waiting list was around 2 million. Notice also that the village population although growing has not grown much between 1971 and 1986. This indicates that most new population members are leaving the village.
This is an important graph as it shows the change in the average land area available to each person. It is a meaningless figure as far as the real villages are concerned but as an aggregated variable it has a useful analytic purpose As I will argue later this index is fundermental to Stirling's migration model and is related to box number 17 on the 'wiring diagram'.
This graph shows the approximate distribution in 1950 of cultivatable land across households. It is derived from the database and differs from what one might expect given Stirling's 1950 classification of house holds. (cf. p15)
This spreadsheet gives the primary index data for evaluating an eventual simulation.
The main tables are the 'Lucy villstats' table which was corrected from the 1991 'Labour Migration in Turkey' table after more analysis of the database, and the 1994 'Choosing' Spouses' table which shows changes in male & female migration experiences.
Taken with the demographic data above these three sources will provide the control tables against which Stirling's migration model will be evaluated.
Formalising the model
Having examined the 'wiring diagram' and having found suitable indexes from the census data, the next stage in testing a Stirling model is to choose one and formalise it.
At its simplest Stirling's model of migration only takes into account boxes 10, 16, 17, 12 & 18. of the 'wiring diagram'.
The Next Stage
For the purposes of this course grain study we can aggregate the conceptual boxes even more. The yellow circle around the Pendular and Permanent migrant objects shows that I want to bring these two together. Similarly I will disregard the 'national' effects boxes represented here by the purple line, and concern myself only with the 'health services' box. This leaves the following:
The Basic Model
The next stage is to represent these conceptual entities with formal variables aggregated at appropriate levels and write pseudo code to program a simple deterministic cyclical simulation.
The Theory Behind The Model
The number of persons per household seems to remain quite constant (Empirical data) so as the population increases due to decreased infant mortality rates for example, so do the no of households and the average land per household drops. Therefore the average land available per person drops and at a certain point they or the household has to leave. These leads to increased migration for obvious 'ecological' reasons. (See Causal chain 1) "choice and change" (1974). Stirling had predicted this in 1963 (Stirling, 1963).
This will be the first course grain approach to deconstructing the rest of his
models. If I can find consistency between the simple simulation and Stirling's census
data it will provide a powerful argument that there is something substantial behind