Methods in Epidemiology: Population Size Estimation (Advances in Experimental Medicine and Biology, 1333)

Preface
Contents
1 Review of Size Estimation Methods
	1.1 The Importance of Population Size Estimation for Key Groups in the Context of HIV
	1.2 Direct Population Size Estimation Methods
		1.2.1 Population-Based Surveys
		1.2.2 Census/Enumeration Method
		1.2.3 Nomination Method
		1.2.4 Capture-Recapture Method
		1.2.5 Multiplier Method
	1.3 Indirect Population Size Estimation Methods
		1.3.1 Cross-Wise Method
		1.3.2 Network Scale-up Method
		1.3.3 Proxy Respondent Method
	1.4 Advantages of Network Scale-up Over Other Size Estimation Methods
	1.5 Requirements of Network Scale-up
		1.5.1 Network Size
		1.5.2 Preliminary Size Estimation
		1.5.3 Correction Factors
	References
2 Methods to Estimate the Average Social Network Size
	2.1 Introduction
	2.2 Direct Estimation of C
		2.2.1 Global Method
		2.2.2 Summation Method
	2.3 Indirect Estimation of C: Reference Groups Approach
		2.3.1 Incorporating Multiple Reference Groups
			2.3.1.2 Means of Sums Estimator
	2.4 Considerations in Selecting Appropriate Reference Groups
		2.4.1 Visibility
		2.4.2 Barrier Effect
		2.4.3 Cognitive Error
		2.4.4 Attractiveness Bias
		2.4.5 Racal Bias
		2.4.6 Conclusion
	2.5 Calibration of Calculated Average Social Network Size
	2.6 Which Factors Affects Average Network Size?
		2.6.1 Summation versus Traditional Reference Group Method
		2.6.2 Missing Data
		2.6.3 Joint Effect of Estimator, Definition of Ratio, and Its Tolerable Range
		2.6.4 Joint Effect of Estimator, Missing Data, and Calibration
		2.6.5 Digit Preference
		2.6.6 Effect of Type of Network Size and Geographical Zone
		2.6.7 Conclusion
	2.7 Network Size Variation at the Subnational Level
	2.8 Aggregation of Different Estimates
	2.9 Effect of Demographic Characteristics on Network Size
		2.9.1 Conclusion
	2.10 Age-Sex Distribution of the Network Size
	2.11 Confidence Interval Calculation for Average Network Size
	References
3 Estimating the Size of Hidden Groups
	3.1 Crude Estimations of the Size of Hidden Groups
	3.2 Correction Factors
	3.3 Visibility Factor
		3.3.1 Methods to Estimate the Visibility Factor
		3.3.2 Popularity Ratio
	3.4 Influence of Missing Data on Size Estimations of Hidden Groups
		3.4.1 Types of Missing Data
		3.4.2 Ad Hoc Approaches to Adjust for Missing Data
		3.4.3 Likelihood-Based Approaches to Impute Missing Data
	3.5 Which Factors Influence Replies to Sensitive Questions
		3.5.1 Background Characteristics
		3.5.2 Definition, Frequency of Acts, and Question Order Effect
		3.5.3 Data Collection Methods
		3.5.4 Matching Between Interviewers and Interviewees
			3.5.4.1 Conclusions
	3.6 Methodological Considerations in the Size Estimation of Gender and Age-Specific Stigmatized Behaviors: Abortion
		3.6.1 Conclusions
	References
4 Data Smoothing, Extrapolation, and Triangulation
	4.1 Smoothing of District Level Estimates
		4.1.1 Spatial Smoothing by Weighted Head-Banging Algorithm
		4.1.2 Spatial Smoothing Based on Kernel Functions
		4.1.3 Empirical Bayesian Smoothing of Prevalence by Beta-Binomial Model
		4.1.4 Empirical Bayes Smoothing of Standard Morbidity Ratio (SMR) by Poisson-Gamma Model
		4.1.5 Practical Challenges in Bayesian Smoothing of NSU Studies
	4.2 National Estimates by Extrapolation from Subnational-Level Estimates
		4.2.1 How to Select Provinces?
		4.2.2 Proportion-Based Extrapolation
		4.2.3 Extrapolation by Linear Regression
			4.2.3.1 Pitfalls in Linear Regression Extrapolation
			4.2.3.2 Extrapolation by Spatial Regression
	4.3 Triangulating Results
	4.4 Conclusion
	References