Live tweeting the "Gridded Population Surveys" at #AAPOR
First is Dana Thomson (U of Twente) giving an introduction to gridded population sampling
Gridded population datasets publicly available
Not all are equal
One of the gridded population that should not be used for sampling, a couple should be used with cautious
Large increase used of gridded populations, especially in public opinion research
When NOT to use gridded population:
* Recent Census
* Population in small geography
* Data collection operational team doesn't have experience with basic mapping tools such as Google Maps
When to use gridded population sampling:
*Census grossly outdated or inaccurate
* Dangerous area
*Highly dynamic and complex
*Stratify by geo characteristics
Sample frame tools:
For Non-GIS users: GridSample, R (GridEZ)
GIS users: QGIS (PreEA)
State of the field:
* Tools for designing and implementing are still piecemeal, we need better tools
* Gridded population datasets, the most important component of this approach, are evolving and improving rapidly
* Area-microcensus designs are promising but need study
Manual for designing and implementing gridded population surveys: gridpopsurvey.com
Next is Yuliya Dudaronak and Stephanie Landas (ORB International) talking about Practical Implications of Using Grid Sample in Conflict and Desert Environments
Biggest issues in implementing gridded sampling compared to using Census Enumeration Areas: more difficult to identify grids (it's a alpha-numeric values, sometimes in the middle of nowhere)
Challenges in implementing gridded population sampling in some countries
* Varying levels of security
* Poor infrastructure
* Poor telephone coverage
* Socialcultural differences
Case study in Somalia GridSample vs. Census
Not a lot of differences with respect to demographics
Census v GridSample
Case by case decision
Next is Sarah Ford (US Department of State) presenting In Search of City Limits: Defining Urban Boundaries in Gridded Population Sampling
Defining urban boundaries with gridded population data is challenging
Not typically available in GIS format
Standardized degree of urbanization
Revised standards with 7 categories for urbanization
That's a huge uninhabited area!
Also a lot of that area is quite far from major roads
Next is Shannon Farley (Columbia) talking about Sampling Refugees in Uganda without a Census or Shapefile
Uganda hosts the 3rd largest population of refugees in the world, the 1st in Africa
Uganda Refugee Population-based HIV Impact Assessment (RUPHIA 2021)
Mylti-stage sample design
40 Enumeration area
Target pop: refugees 15 years and older
Assumed a HIV prevalence of 3% for sample design
Determined minimum number of HIV-positive respondents
MoE 1%
Gridded population sampling
Used grids (100m x 100m), combined or not, as PSUs
Missing about 7% of the population because of the sample frame of settlements zones available
Used WorldPop peanutButter (apps.worldpop.org/peanutButter) and the used gridzEZ in R to generate 2,636 gridEZ units of 100m x 100m grid cells
Issues in the field:
1 EA outside settlement boundaries
1 EA full of Uganda nationals
Next and final speaker is Giles Reid (Columbia) presenting Weighting Survey Data from a Gridded Population Survey: The Uganda Refugee Population-Based HIV Impact Assessment
This presentation discusses the weighting of the 2021 RUPHIA project from the previous presentation
Challenge 1: out-of-scope areas selected
Select a new EZ and adjust for the selection probability
Challenge 2: differences in estimated sizes and actual sizes
Challenge 3: Non-response adjustment using BART, because the sample size did not allow to use a CHAID, the standard method used by the team
Very high response rates!
@threadreaderapp compile unroll thread
Share this Scrolly Tale with your friends.
A Scrolly Tale is a new way to read Twitter threads with a more visually immersive experience.
Discover more beautiful Scrolly Tales like this.