Kristallnacht PAP 4 update 4, 7/21/2023: Revise reference to OLS to clarify that I estimate the multivariate models using a generalized least squares regression model with a spatial lag, using the *lagsarlm* package in R. Kristallnacht PAP 4 update 3, 7/19/2023: Revise non-jewish_pop_share visibility models to include a jewish_pop_share variable for all models. Kristallnacht PAP 4 update 2, 7/7/2023: Remove reference to Liang-Zeger standard errors on line 675. Key updates for Kristallnacht PAP 4: - Updated project title - Revised spatial join procedures with data from Voigtlander and Voth (2012), Adena et al. (2015), and---a new addition---Knowles et al. (2014) - Added model variables on (1) 1932 - 3 NSDAP membership (Adena et al. 2015, originally from Falter and Hanisch); (2) distance to nearest SS concentration camp (Knowles et al. 2014); SPD, KPD, and NSDAP + DNVP performance in elections in 1928, 1930, and 1933 - Streamlined multivariate spatial-lag models using a for-loop format, as well as the procedure for storing coefficient estimates, standard errors, p-values, and confidence intervals from each model Kristallnacht PAP 3 update 3: Adding Models 32 and 33, which estimate the bivariate relationship between (1) support for antisemitic parties and (2) the presence of a kosher butcher, respectively, and synagogue attacks. Kristallnacht PAP 2 update 3: Adding the radio access variable (ss_n373) to Model 17, from which I accidentally omitted it. See lines 1811 - 1812. Kristallnacht PAP 3 update 2: Adding Model 31, the bivariate relationship between Jewish population share and synagogue attacks. Kristallnacht PAP 3 update 1: Removing "religious_center_1933" as a control variable in Model 30, which centers on the effect of kosher butchers on synagogue attacks during Kristallnacht. Kristallnacht PAP 2 update 2: Separating Model 21 into Model 21a, which tests the interaction between an active religious center and the presence of a kosher butcher, and Model 21b, which tests the interaction between an active religious center and support for antisemitic parties. Kristallnacht PAP 2 update 1: Correcting R code lines 988 - 91, to correct select coding errors in the original diverse_tactics variable. The additional code to correct this error is as follows: \### correct NAs for syns_attacked_in_town_dummy == 1, change to 0s kristallnacht <- kristallnacht %>% mutate(diverse_tactics = ifelse(syns_attacked_in_town_dummy == 1 & is.na(diverse_tactics), 0, diverse_tactics)) -- Kristallnacht PAP update 5: Correcting R code lines 482 - 491, which scale-transformed variables in the negative binomial model that should not have been scale-transformed, per the text of the PAP. Kristallnacht PAP update 4: Correcting a computational error associated with standardized regression coefficients, resulting in a switch from the lm.beta R package to the manual use of the scale command for all OLS variables. Kristallnacht PAP update 3: Correcting an incorrect reference to Alfred Hugenberg's name on p. 20. Kristallnacht PAP update 2: Correcting incorrect formatting on the right-hand margins.