##--##--##--##--##--##--##
## Appendix: Table A5
##--##--##--##--##--##--##

# Status (excluded, not excluded)

nv_status = lm(success ~  EPR_STATUS_EXCL + EPR_GROUPSIZE + POP_LOG_LAG_EXT + GDPPC_LOG_LAG_EXT + PASTNV + PASTV + VDEM_POLYARCHY_LAG + VDEM_PHYSINT_LAG + EPR_TEK_EGIP + EPR_DOWNGRADED5 + HORIZ_INEQ + NVYEARS + I(NVYEARS^2) + I(NVYEARS^3) + VYEARS + I(VYEARS^2) + I(VYEARS^3) , data= EBCR_EPR_NAVCO2[EBCR_EPR_NAVCO2$INIT_NV_ONSET==1,])  
v_status = lm(success ~  EPR_STATUS_EXCL + EPR_GROUPSIZE + POP_LOG_LAG_EXT + GDPPC_LOG_LAG_EXT + PASTNV + PASTV + VDEM_POLYARCHY_LAG + VDEM_PHYSINT_LAG + EPR_TEK_EGIP + EPR_DOWNGRADED5 + HORIZ_INEQ + NVYEARS + I(NVYEARS^2) + I(NVYEARS^3) + VYEARS + I(VYEARS^2) + I(VYEARS^3) , data= EBCR_EPR_NAVCO2[EBCR_EPR_NAVCO2$INIT_V_ONSET==1,])  

preds_nv = ggpredict(nv_status, terms = c("EPR_STATUS_EXCL"))
preds_nv$tactic = "Non-violent"
preds_v = ggpredict(v_status, terms = c("EPR_STATUS_EXCL"))
preds_v$tactic = "Violent"

preds_status = rbind(preds_nv, preds_v)
preds_status$tactic <- factor(preds_status$tactic, levels = c("Violent", "Non-violent"))
preds_status$Status = "Excluded"
preds_status$Status[preds$x==0] = "Not excluded"
preds_status$type = "status"

## Size (above and below the mean of the distribution)

EBCR_EPR_NAVCO2$small_size = NA
EBCR_EPR_NAVCO2$small_size[EBCR_EPR_NAVCO2$EPR_GROUPSIZE < mean(EBCR_EPR_NAVCO2$EPR_GROUPSIZE, na.rm=T)] = 1
EBCR_EPR_NAVCO2$small_size[EBCR_EPR_NAVCO2$EPR_GROUPSIZE >= mean(EBCR_EPR_NAVCO2$EPR_GROUPSIZE, na.rm=T)] = 0

nv_size = lm(success ~  small_size  + POP_LOG_LAG_EXT + GDPPC_LOG_LAG_EXT + PASTNV + PASTV + VDEM_POLYARCHY_LAG + VDEM_PHYSINT_LAG + EPR_TEK_EGIP + EPR_DOWNGRADED5 + HORIZ_INEQ + NVYEARS + I(NVYEARS^2) + I(NVYEARS^3) + VYEARS + I(VYEARS^2) + I(VYEARS^3) , data= EBCR_EPR_NAVCO2[EBCR_EPR_NAVCO2$INIT_NV_ONSET==1,])  
v_size = lm(success ~  small_size  + POP_LOG_LAG_EXT + GDPPC_LOG_LAG_EXT + PASTNV + PASTV + VDEM_POLYARCHY_LAG + VDEM_PHYSINT_LAG + EPR_TEK_EGIP + EPR_DOWNGRADED5 + HORIZ_INEQ + NVYEARS + I(NVYEARS^2) + I(NVYEARS^3) + VYEARS + I(VYEARS^2) + I(VYEARS^3) , data= EBCR_EPR_NAVCO2[EBCR_EPR_NAVCO2$INIT_V_ONSET==1,])  

preds_nv = ggpredict(nv_size, terms = c("small_size"))
preds_nv$tactic = "Non-violent"
preds_v = ggpredict(v_size, terms = c("small_size"))
preds_v$tactic = "Violent"

preds_size = rbind(preds_nv, preds_v)
preds_size$tactic <- factor(preds_size$tactic, levels = c("Violent", "Non-violent"))
preds_size$Status = "Group size < mean"
preds_size$Status[preds_size$x==0] = "Group size >= mean"
preds_size$type = "size"

preds_all = as.data.frame(rbind(preds_status, preds_size))
preds_all = preds_all[,c("predicted", "std.error", "conf.low", "conf.high",
                         "tactic", "Status")]
table_a5 = xtable(preds_all, digits=2)
print(table_a5, file="Tables/table_A5.tex")