New Population Estimates for Kibera at 200,000 to 250,000

Study by Stephano Morras PhD
Title Mapping the unmapped
Department of Sociology and Social Research
Universita Degli Studi di Milano - Bicocca, Milano. Italy.
Download the study here

A new study conducted by a group of Italian Specialists led by Stephano Morras estimates the population of Kibera to be between 200,000 to 250,000. They do this by first calculating the estimates for Kianda Village which they find to be 15,219. They then extrapolate the results from Kianda’s population to arrive at the following conclusion:

Final note: Looking upon the data reported above and considering the type, dimension and distribution of the buildings observed in Kianda is typically the same in the whole of Kibera, it is possible to make a guess about the numerical dimension of its population. Considering that the area of Kibera is set between 2.3 to 2.5 square km, the total population living in the slum can be most likely estimated between 200,000 to 250,000.

I have re-looked at the abstract of this study and I am not convinced (I have not seen the full text though). I think that its problem lies with the methodology (It points to ethnographic methods), data analysis and ultimately the inferences. Goldthorpe J. H's book (2006, 2nd ed) entitled on sociology, numbers and the integration of research and theory speaks much about social research and its inferential problems owing to its qualitative nature and use of inferential logic. These researchers will have considerable difficulties to defend their results. The inference is certainly problematic. The researchers note about the informality of Kibera and the fact that its estimates are not official. The Kenya bureau of statistics in their Kenya Integrated Household Budget Surveys (KIHBS) and countless other studies show that the average household number in Kenya is 6 which is placed at 3.5 in Kianda by the Morras study. I am tempted to conclude that this study's inferences are flawed.

My opinion is that this study might suffer from; one a problem of data integrity and secondly non transparent inferential rules and hence not well defensible and third a non probabilistic sample (a probabilistic sample should be open, transparent and well codified). The abstract makes an indication of the use of Mills' classical methods of logic or Bullien's Algebraic inferential method or alternatively the use of truth tables to arrive at the inferences. The weaknesses of these methods is that they cannot deal with the issues of endogeneity, have no allowable degree of error, their inferences, their explanatory and predictive power are not robust to minor adjustments in variables.

The above problems are always experienced by case studies, ethnographic studies, qualitative or grounded studies. This study is an example of a "Grounded study." The study grounds itself in Kianda and using for example Mills' logic infers that Kibera's reality is grounded in the reality of Kianda. Ethnographic studies can only used to make inferences about the subject of study or its specific mechanisms (within effects) but have considerable difficulties to make general inferences. The problem of endogeneity would generate a pseudo solution or misleading solutions and conclusions. Cases where ethnographic methods would have fitted well would be in rare and transient phenomenons such as the post electoral violence, the financial bubbles, revolutions etc. By explaining the characteristics of the phenomenon, ethnographic and case studies help people to develop a better intuition about a particular phenomenon.

Other issues to look at in this research include the sampling issues (probabilistic or judgmental) , response rates, bias in selection and response and non response bias, interaction effects, the testable hypotheses and the scope of the study. The great problem from the inference above is what has been referred by Goldthorpe above as a homogeneity assumption i.e. grounding the inferences from the findings of the case study, which is a big problem.

I would be interested if Stephano would reply to my criticism. Maybe they will help me to undo the "population myth" belief that I have lived with for a long time. Secondly I agree with him on the fact that a million people is far too much but disagree on the 250,000 people.