American Jewish Population Project - Pennsylvania Report
NOTES
American Jewish Population Project
1 Methodology: Individual-level data from all surveys were combined using Bayesian multilevel modeling with poststratification.
Poststratification included geographic distributions of respondents by ZIP Codes within congressional districts, and demographic
characteristics of age, educational attainment, race/ethnicity, population density, as well as interactions of age by educational attainment,
population density by age, and population density by educational attainment. Modeling is based to Jewish adults who self-identify as
Jewish when asked about their religion. Estimates of "Total Jewish Adults" are obtained by adding to the model-based estimate,
independent estimates of the percentage of Jewish adults who do not identify religiously as Jewish. This percentage can range from a low
of 10% to a high of 30% depending on the region.
2 State-level Jewish adult totals are estimated from AJPP 2020 models and adjustments for Jewish adults who do not identify religiously
as Jewish.
3 Partisan lean of Independents was estimated using a design-based pooled analysis method in which each survey's original survey
weights were adjusted for survey specific designs and sample sizes. This method is not as sensitive to estimation of rare populations as
the Bayesian methods used for the main Jewish population estimates but provides an initial ballpark estimate of the groups of interest.
Follow-up studies will compare these estimates to those derived from more fully developed Bayesian model-based estimates.
4 Congressional district competitive scores from Cook Political Report House Race Ratings (Nov 2, 2020) and FiveThirtyEight's Partisan
Lean (Oct 19, 2020); data accessed January 2021.
Brandeis
STEINHARDT SOCIAL
RESEARCH INSTITUTE
6View entire presentation