That video does a very good job of explaining this. As child mortality decreases, family size goes down. I'm sure the data is manipulated to suit someone's purposes, but all seems reasonable to me...
http://www.ted.com/talks/lang/en/hans_rosling_shows_the_best...
http://www.ted.com/talks/lang/en/hans_rosling_reveals_new_in...
The point is, when you have incredibly poor country, excessive influence of religion, poor health and a high child mortality rate, you have more children, both planned and unplanned.
Religion has very little to do with the number of babies per woman. All the religions in the world are fully [able] to maintain their values and adapt to this new world. </quote> http://www.ted.com/talks/lang/en/hans_rosling_at_state.html
For example, here is a graph I created of infant mortality vs children per woman: http://www.bit.ly/QTPNBk (starts at 1900, press play to see changes to current)
Note that each axis has a small 'various sources' button under it that will cite specific sources.
You may also enjoy Hans Rosling's TED Talk on global population growth, as he gets into the details pretty thoroughly. https://www.ted.com/talks/hans_rosling_on_global_population_...
Here's an "abstract" of the article, which hints at my synopsis: http://www.newyorker.com/archive/2005/10/24/051024fa_fact_sp...