Model Selection and checks
EA
12/08/2021
CurbEUenthusiasm_FINAL <- CurbEUenthusiasm_FINAL %>% mutate(lagged2y_MS_EU_diff = lagged2y_pct_support_cntr_cap_year - lagged2y_pct_EUsupport)
CurbEUenthusiasm_FINAL <- CurbEUenthusiasm_FINAL %>% mutate(lagged1y_MS_EU_diff = lagged_pct_support_cntr_cap_year - lagged_pct_EUsupport)Check variable correlation
Here, I subsample the vars which are used in the model ( see in the scripts by AK), generate corr matrix and plot it. This will allow avoiding the inclusion of highly correlated indicators, hence will also facilitate the process of identifying viable models
Seave out non correlated vars
The Correlation matrix above suggest a strong correlation between Public support lagged for 1 year and the Polarization measures (on both EU and the Country) lagged for one year. The inclusion of these variables together will substantially influence the quality of the results and inferences.
De Bryucker’s paper maintains that change in public opinion occurs prior to the process of politicization. This in turn points towards a lag structure for the variables capturing public support and vars measuring politicization ==> there should be a time-difference between them.
Controlling for EU and Country polarization: when both variables transformed using the same lag structure ( for instance 1Y lag), they are highly correlated at Rho= -0.69. Hence the inclusion of these indicators is only possible when and if their effect is hypothesized to appear at different points in time during the process of policy making
One line of reasoning: The Commission is a agent of the UNION, not specific countries. It is shown that the Commission intentionally designs/ includes questions into the EB surveys. Hence, it is possible that at the stage of preparing the proposal the overall patterns of polarization across the EU will influence the proposal content.
* However, as the stage of approval nears, the Commission has to account for the degree of polarization within Countries as it is likely to shape the chances of the ambitious legislation being approved/not approved. Hence, the Country Level polarization exerts its effects concurrently with the proposal being put forward.Second line of reasoning: The Commission is more likely to collect information about the country level polarization at the formulation stage to adjust the content and the authority expansion ambition to address the key preferences yet account for the cleavages between the states ==> hence the country polarization could exert influence at earlier stage;
- However, as the stage of putting the proposal forward nears, the polarization level across the Union would motivate the Commission to adjust the proposal more closely to the public preferences as the need for legitimizing the proposal and authority expansion increases. Also, aligning the proposed legislation and the Union’s public preferences grants the Commission’s proposal stronger justification in eyes of the CN and the EP. ( AGAIN: the COM is the agent of the UNION! at least technically=).
Checking the intuition from above for H1b
Dv: 2 Y lag;
IVS: lagged_pct_Salience_DK Lagged_no_trade_n_of_organiz_registered_per_year
Checking for: cntr_polarization, EUpolarizarion Lagged and not lagged
First argument
*The models below check the interaction with polarization. The output suggests that Lagged EU polarization slightly hinders he ambition of the Commission to expand the EU authority as the support for the EU action in the policy are grows. This effects shows by the decreasing slope of of the marginal effects at the highest level of polarization. This perhaps suggests that the Commission behaves more cautiously when the EU public is heavily divided, even if the EU engagement into the policy area is overall supported.
model1b_check <- glm(LR_Proposal_Probability ~ lagged2y_pct_support_cntr_cap_year +
lagged_pct_Salience_DK +
lagged2y_pct_support_cntr_cap_year*lagged_EUpolarizarion +
lagged_EUpolarizarion +
cntr_polarization+
Lagged_no_trade_n_of_organiz_registered_per_year +
Interinst_conflict +
nofterms_in_eurovoc +
competence_length+
factor(form) + factor(year) + factor(Country_num) ,
data = CurbEUenthusiasm_FINAL, family = quasibinomial('logit'))
#stargazer(model1b_check, type = "text")## Data were 'prettified'. Consider using `terms="lagged2y_pct_support_cntr_cap_year [all]"` to get smooth plots.
## Data were 'prettified'. Consider using `terms="lagged2y_pct_support_cntr_cap_year [all]"` to get smooth plots.
model1b2_check <- glm(LR_Proposal_Probability ~ lagged2y_pct_support_cntr_cap_year +
lagged_pct_Salience_DK +
lagged2y_pct_support_cntr_cap_year*cntr_polarization+
lagged_EUpolarizarion +
cntr_polarization+
Lagged_no_trade_n_of_organiz_registered_per_year +
Interinst_conflict +
nofterms_in_eurovoc +
competence_length+
factor(form) + factor(year) + factor(Country_num) ,
data = CurbEUenthusiasm_FINAL, family = quasibinomial('logit'))## Data were 'prettified'. Consider using `terms="lagged2y_pct_support_cntr_cap_year [all]"` to get smooth plots.
## Data were 'prettified'. Consider using `terms="lagged2y_pct_support_cntr_cap_year [all]"` to get smooth plots.
#### Model summaries with Not Lagged CNTR POlarization and Lagged 1Y EU Polarization - Underlying idea: The Commission at first checks the `moods’ across the Union at the prparation statge, and the Countries’ public support at the time of proposal.
stargazer(model1b_check, model1b2_check, type = "text",
omit=c("Country_num"), omit.labels = c("Country FE?"), title='Table 1: Testing for Polarization Lag and proxy', align = TRUE)##
## Table 1: Testing for Polarization Lag and proxy
## =====================================================================================
## Dependent variable:
## ----------------------------
## LR_Proposal_Probability
## (1) (2)
## -------------------------------------------------------------------------------------
## lagged2y_pct_support_cntr_cap_year 0.008*** -0.002
## (0.002) (0.002)
##
## lagged_pct_Salience_DK 0.004*** 0.004***
## (0.001) (0.001)
##
## lagged_EUpolarizarion 0.525** 0.263***
## (0.236) (0.058)
##
## cntr_polarization -0.186*** -0.949***
## (0.046) (0.229)
##
## Lagged_no_trade_n_of_organiz_registered_per_year -0.025* -0.026*
## (0.013) (0.013)
##
## Interinst_conflict -0.197*** -0.203***
## (0.020) (0.020)
##
## nofterms_in_eurovoc 0.042*** 0.042***
## (0.005) (0.005)
##
## competence_length -0.015*** -0.015***
## (0.001) (0.001)
##
## factor(form)Directive -0.450*** -0.442***
## (0.039) (0.039)
##
## factor(form)Regulation -0.294*** -0.286***
## (0.036) (0.036)
##
## factor(year)2011 0.036 0.031
## (0.032) (0.032)
##
## factor(year)2012 -0.217*** -0.229***
## (0.039) (0.039)
##
## factor(year)2013 -0.004 -0.049
## (0.040) (0.042)
##
## factor(year)2014 -0.719*** -0.749***
## (0.050) (0.051)
##
## factor(year)2015 0.122** 0.085
## (0.054) (0.055)
##
## factor(year)2016 -0.126*** -0.147***
## (0.039) (0.039)
##
## lagged2y_pct_support_cntr_cap_year:lagged_EUpolarizarion -0.005
## (0.003)
##
## lagged2y_pct_support_cntr_cap_year:cntr_polarization 0.011***
## (0.003)
##
## Constant 0.066 0.725***
## (0.187) (0.200)
##
## -------------------------------------------------------------------------------------
## Country FE? Yes Yes
## -------------------------------------------------------------------------------------
## Observations 12,195 12,195
## =====================================================================================
## Note: *p<0.1; **p<0.05; ***p<0.01Second Line of reasoning <- this one is more likely to be the case
The chunks below test if the EU polarization at T and lagged country polarization at t-1 affect the results. Chunk 8 test the effect of EU polarization in the probability of EU authority expansion in the proposal.
The results point towards the following: When the EU is divided at the time of proposal, the increasing support for the EU action will foster the Commissions ambitions and is likely to lead to a more authority expanding proposals. BUt if the Union’s society is homogeneous, the increase in public support has only marginal negative effects of the Commission’s ambition.
model1b_check2 <- glm(LR_Proposal_Probability ~ lagged2y_pct_support_cntr_cap_year +
lagged_pct_Salience_DK +
lagged2y_pct_support_cntr_cap_year*EUpolarizarion +
EUpolarizarion +
lagged_cntr_polarization+
Lagged_no_trade_n_of_organiz_registered_per_year +
Interinst_conflict +
nofterms_in_eurovoc +
competence_length+
factor(form) + factor(year) + factor(Country_num) ,
data = CurbEUenthusiasm_FINAL, family = quasibinomial('logit'))
#stargazer(model1b_check2, type = "text")#plot_model(model1b_check2, type = "pred")
plot_model(model1b_check2, type = 'int', mdrt.values = 'meansd') # This plots mean +/-1sd## Data were 'prettified'. Consider using `terms="lagged2y_pct_support_cntr_cap_year [all]"` to get smooth plots.
plot_model(model1b_check2, type = 'int')## Data were 'prettified'. Consider using `terms="lagged2y_pct_support_cntr_cap_year [all]"` to get smooth plots.
####Retesting with the interaction of the lagged polarization on the country level
- Results: increasing support with low polarization in the country increases the probability of EU authority expansion. Under High polarization on the country level, the public support has a more modest effect, BUT the COM tends to put forward more ambitious proposals in contrast to the environments of low polarization
model1b2_check2 <- glm(LR_Proposal_Probability ~ lagged2y_pct_support_cntr_cap_year +
lagged_pct_Salience_DK +
lagged2y_pct_support_cntr_cap_year*lagged_cntr_polarization +
EUpolarizarion +
lagged_cntr_polarization+
Lagged_no_trade_n_of_organiz_registered_per_year +
Interinst_conflict +
nofterms_in_eurovoc +
competence_length+
factor(form) + factor(year) + factor(Country_num) ,
data = CurbEUenthusiasm_FINAL, family = quasibinomial('logit'))
#stargazer(model1b_check2, type = "text")#plot_model(model1b_check2, type = "pred")
plot_model(model1b2_check2, type = 'int', mdrt.values = 'meansd') # This plots mean +/-1sd## Data were 'prettified'. Consider using `terms="lagged2y_pct_support_cntr_cap_year [all]"` to get smooth plots.
plot_model(model1b2_check2, type = 'int')## Data were 'prettified'. Consider using `terms="lagged2y_pct_support_cntr_cap_year [all]"` to get smooth plots.
stargazer(model1b_check2, model1b2_check2, type = "text",
omit=c("Country_num" ), omit.labels = c("Country FE?"),
title='Table 2: Testing for Polarization Lag and Proxy (CNTR Lag)', align = TRUE)##
## Table 2: Testing for Polarization Lag and Proxy (CNTR Lag)
## ========================================================================================
## Dependent variable:
## ----------------------------
## LR_Proposal_Probability
## (1) (2)
## ----------------------------------------------------------------------------------------
## lagged2y_pct_support_cntr_cap_year -0.006** 0.009***
## (0.002) (0.002)
##
## lagged_pct_Salience_DK 0.003*** 0.004***
## (0.001) (0.001)
##
## EUpolarizarion -1.427*** -0.251***
## (0.242) (0.048)
##
## lagged_cntr_polarization 0.194*** 0.584***
## (0.052) (0.203)
##
## Lagged_no_trade_n_of_organiz_registered_per_year -0.024* -0.022*
## (0.013) (0.013)
##
## Interinst_conflict -0.176*** -0.173***
## (0.020) (0.020)
##
## nofterms_in_eurovoc 0.044*** 0.044***
## (0.005) (0.005)
##
## competence_length -0.015*** -0.014***
## (0.001) (0.001)
##
## factor(form)Directive -0.445*** -0.451***
## (0.039) (0.039)
##
## factor(form)Regulation -0.283*** -0.292***
## (0.036) (0.036)
##
## factor(year)2011 0.015 0.019
## (0.032) (0.032)
##
## factor(year)2012 -0.267*** -0.248***
## (0.039) (0.039)
##
## factor(year)2013 -0.111*** -0.036
## (0.042) (0.040)
##
## factor(year)2014 -0.741*** -0.709***
## (0.050) (0.050)
##
## factor(year)2015 0.090* 0.135**
## (0.054) (0.054)
##
## factor(year)2016 -0.164*** -0.130***
## (0.039) (0.038)
##
## lagged2y_pct_support_cntr_cap_year:EUpolarizarion 0.017***
## (0.003)
##
## lagged2y_pct_support_cntr_cap_year:lagged_cntr_polarization -0.007**
## (0.003)
##
## Constant 1.161*** 0.077
## (0.217) (0.181)
##
## ----------------------------------------------------------------------------------------
## Country FE? Yes Yes
## ----------------------------------------------------------------------------------------
## Observations 12,251 12,251
## ========================================================================================
## Note: *p<0.1; **p<0.05; ***p<0.01EA: I believe that the specification allowing to control for the lagged CNTR polarization and the EU polarization at the time of the proposal given an opportunity to account for the proper Time-Lag structure, control for the competing pressure of polarization on the EU commission at different points in time ( and avoid multicollinearity).
Retesting the hypotheses: Fracreg and main model specification
The models below re-test the main hypotheses using the FRACREG specification. THe outline of the variable combination is avaiable below and in the model summary tables.
DV: LR_Proposal_Probability
IV: +lagged2y_pct_support_cntr_cap_year
- lagged_pct_Salience_DK
- lagged_cntr_polarization
- Lagged_no_trade_n_of_organiz_registered_per_year
Result: Positive effect of public support for the EU action (consistent with most of the previous models).
model1_check <- glm(LR_Proposal_Probability ~ lagged2y_pct_support_cntr_cap_year +
lagged_pct_Salience_DK +
lagged_cntr_polarization+
Lagged_no_trade_n_of_organiz_registered_per_year +
EUpolarizarion +
Interinst_conflict +
nofterms_in_eurovoc +
competence_length+
factor(form) + factor(year) + factor(Country_num) ,
data = CurbEUenthusiasm_FINAL, family = quasibinomial('logit'))
plot_model(model1_check, type = "pred", terms='lagged2y_pct_support_cntr_cap_year')## Data were 'prettified'. Consider using `terms="lagged2y_pct_support_cntr_cap_year [all]"` to get smooth plots.
Interaction Hypotheses:
**H1a: Public Support*Salience**
At both high and low level of salience the increase in public support for policy are increases the probability of the EU authority expansion. However, the magnitude of the effect decreases when the salience is at the highest.
Possible explanation: Highly salient matters tend to generate more ambitious proposals, hence the degree to which an increase in support for the EU action strengthens the effect is limited
model1a_check <- glm(LR_Proposal_Probability ~ lagged2y_pct_support_cntr_cap_year +
lagged_pct_Salience_DK +
lagged_pct_Salience_DK*lagged2y_pct_support_cntr_cap_year+
lagged_cntr_polarization+
Lagged_no_trade_n_of_organiz_registered_per_year +
EUpolarizarion +
Interinst_conflict +
nofterms_in_eurovoc +
competence_length+
factor(form) + factor(year) + factor(Country_num) ,
data = CurbEUenthusiasm_FINAL, family = quasibinomial('logit'))
#stargazer(model1a_check, type = "text")## Data were 'prettified'. Consider using `terms="lagged2y_pct_support_cntr_cap_year [all]"` to get smooth plots.
## Data were 'prettified'. Consider using `terms="lagged2y_pct_support_cntr_cap_year [all]"` to get smooth plots.
H1b: Public Support X POlarization on the COuntry level lagged
- Results: At High polarization the increase in pubcli support has a positve but more moderate effect.
- When a country is not polarized ==> Increasing support for the EU actions substatially increases the probability of the EU authority expansion.
model1b_check <- glm(LR_Proposal_Probability ~ lagged2y_pct_support_cntr_cap_year +
lagged_pct_Salience_DK +
lagged2y_pct_support_cntr_cap_year*lagged_cntr_polarization +
lagged_cntr_polarization+
EUpolarizarion +
Lagged_no_trade_n_of_organiz_registered_per_year +
Interinst_conflict +
nofterms_in_eurovoc +
competence_length+
factor(form) + factor(year) + factor(Country_num) ,
data = CurbEUenthusiasm_FINAL, family = quasibinomial('logit'))
#stargazer(model1b_check, type = "text")## Data were 'prettified'. Consider using `terms="lagged2y_pct_support_cntr_cap_year [all]"` to get smooth plots.
## Data were 'prettified'. Consider using `terms="lagged2y_pct_support_cntr_cap_year [all]"` to get smooth plots.
Hypothesis H1c: Public Support and Actor Mobilization
Results: Increasing public support at no actor mobilization ==> increase in probability of authority expansion; High actor mobilization ==> decreasing probability of authority expansion even when public support is high.
Possible explanation: Competing demand + stoplight to the EU action (Stakeholders may not only communicate the wishes of the population to the EU, but also send info about the EU actions to the population)
model1c_check <- glm(LR_Proposal_Probability ~ lagged2y_pct_support_cntr_cap_year +
lagged_pct_Salience_DK +
lagged2y_pct_support_cntr_cap_year*Lagged_no_trade_n_of_organiz_registered_per_year +
lagged_cntr_polarization+
EUpolarizarion +
Lagged_no_trade_n_of_organiz_registered_per_year +
Interinst_conflict +
nofterms_in_eurovoc +
competence_length+
factor(form) + factor(year) + factor(Country_num) ,
data = CurbEUenthusiasm_FINAL, family = quasibinomial('logit'))
#stargazer(model1c_check, type = "text")## Data were 'prettified'. Consider using `terms="lagged2y_pct_support_cntr_cap_year [all]"` to get smooth plots.
## Data were 'prettified'. Consider using `terms="lagged2y_pct_support_cntr_cap_year [all]"` to get smooth plots.
###Include all models
stargazer(model1_check, model1a_check, model1b_check, model1c_check, type='text',
omit=c("Country_num"), omit.labels = c("Country FE?"), title="Table 3: Main results", align = TRUE)##
## Table 3: Main results
## ===========================================================================================================================
## Dependent variable:
## ---------------------------------------
## LR_Proposal_Probability
## (1) (2) (3) (4)
## ---------------------------------------------------------------------------------------------------------------------------
## lagged2y_pct_support_cntr_cap_year 0.005*** 0.010*** 0.009*** 0.006***
## (0.001) (0.003) (0.002) (0.001)
##
## lagged_pct_Salience_DK 0.004*** 0.008*** 0.004*** 0.004***
## (0.001) (0.002) (0.001) (0.001)
##
## lagged_cntr_polarization 0.138*** 0.127** 0.584*** 0.139***
## (0.051) (0.051) (0.203) (0.051)
##
## Lagged_no_trade_n_of_organiz_registered_per_year -0.023* -0.024* -0.022* 0.064
## (0.013) (0.013) (0.013) (0.045)
##
## EUpolarizarion -0.227*** -0.236*** -0.251*** -0.226***
## (0.047) (0.047) (0.048) (0.047)
##
## Interinst_conflict -0.175*** -0.176*** -0.173*** -0.174***
## (0.020) (0.020) (0.020) (0.020)
##
## nofterms_in_eurovoc 0.044*** 0.043*** 0.044*** 0.044***
## (0.005) (0.005) (0.005) (0.005)
##
## competence_length -0.014*** -0.015*** -0.014*** -0.014***
## (0.001) (0.001) (0.001) (0.001)
##
## factor(form)Directive -0.453*** -0.455*** -0.451*** -0.452***
## (0.039) (0.039) (0.039) (0.039)
##
## factor(form)Regulation -0.292*** -0.293*** -0.292*** -0.291***
## (0.036) (0.036) (0.036) (0.036)
##
## factor(year)2011 0.020 0.025 0.019 0.019
## (0.032) (0.032) (0.032) (0.032)
##
## factor(year)2012 -0.251*** -0.243*** -0.248*** -0.246***
## (0.039) (0.039) (0.039) (0.039)
##
## factor(year)2013 -0.048 -0.062 -0.036 -0.047
## (0.040) (0.041) (0.040) (0.040)
##
## factor(year)2014 -0.716*** -0.708*** -0.709*** -0.718***
## (0.050) (0.050) (0.050) (0.050)
##
## factor(year)2015 0.128** 0.140** 0.135** 0.129**
## (0.054) (0.054) (0.054) (0.054)
##
## factor(year)2016 -0.134*** -0.132*** -0.130*** -0.133***
## (0.038) (0.038) (0.038) (0.038)
##
## lagged2y_pct_support_cntr_cap_year:lagged_pct_Salience_DK -0.0001*
## (0.00003)
##
## lagged2y_pct_support_cntr_cap_year:lagged_cntr_polarization -0.007**
## (0.003)
##
## lagged2y_pct_support_cntr_cap_year:Lagged_no_trade_n_of_organiz_registered_per_year -0.001**
## (0.001)
##
## Constant 0.334** 0.063 0.077 0.316**
## (0.142) (0.201) (0.181) (0.142)
##
## ---------------------------------------------------------------------------------------------------------------------------
## Country FE? Yes Yes Yes Yes
## ---------------------------------------------------------------------------------------------------------------------------
## Observations 12,251 12,251 12,251 12,251
## ===========================================================================================================================
## Note: *p<0.1; **p<0.05; ***p<0.01ROBUSTENESS CHECKS
Following AK’s findings, the chunks below cross test the 3 way interactions, Betaregs; Models using Diff between CNTR public support and EU public support and Polarization Index
Beta-regs
# generate the variable => adjusted probability to run betareg
CurbEUenthusiasm_FINAL$adjusted_probability =CurbEUenthusiasm_FINAL$LR_Proposal_Probability-0.0001
###Models
### test hypothesis 1: public opinion in support of the EU authority in PA expansion results in more proposed authority expansion
#polarization: EU=no lag; CNTR= 1y lag
#actor expansion: MS, 1y lag
#salience: MS, 1y lag
#public opinion: %support, 2y lag
model1_robH1 <- betareg(adjusted_probability ~ lagged2y_pct_support_cntr_cap_year +
lagged_pct_Salience_DK +
lagged_cntr_polarization+
EUpolarizarion +
Lagged_no_trade_n_of_organiz_registered_per_year +
Interinst_conflict +
nofterms_in_eurovoc +
factor(form) + factor(year) + factor(Country_num),
data =CurbEUenthusiasm_FINAL)
#stargazer(model1, type = "text")
#plot_model(model1, type = "pred")
### test hypothesis 1a: public opinion*salience
model_robH1a <- betareg(adjusted_probability ~ lagged2y_pct_support_cntr_cap_year +
lagged2y_pct_support_cntr_cap_year* lagged_pct_Salience_DK+
lagged_pct_Salience_DK +
lagged_cntr_polarization+
EUpolarizarion +
Lagged_no_trade_n_of_organiz_registered_per_year +
Interinst_conflict +
nofterms_in_eurovoc +
factor(form) + factor(year) + factor(Country_num),
data = CurbEUenthusiasm_FINAL)
#stargazer(model2, type = "text")
#plot_model(model2, type = "pred")
#plot_model(model2, type = "int") #generates graph at salience = 0 and salience = 100, the values can be changed
### test hypothesis 1b: public opinion*polarization
model_robH1b <- betareg(adjusted_probability ~ lagged2y_pct_support_cntr_cap_year +
lagged2y_pct_support_cntr_cap_year*lagged_cntr_polarization+
lagged_pct_Salience_DK +
lagged_cntr_polarization+
EUpolarizarion +
Lagged_no_trade_n_of_organiz_registered_per_year +
Interinst_conflict +
nofterms_in_eurovoc +
factor(form) + factor(year) + factor(Country_num),
data = CurbEUenthusiasm_FINAL)
#stargazer(model3, type = "text")
#plot_model(model3, type = "pred")
#plot_model(model3, type = "int")
### test hypothesis 1c: public opinion*actor mobilization
model_robH1c=betareg(adjusted_probability ~ lagged2y_pct_support_cntr_cap_year + lagged2y_pct_support_cntr_cap_year*Lagged_no_trade_n_of_organiz_registered_per_year+
lagged_pct_Salience_DK +
lagged_cntr_polarization+
EUpolarizarion +
Lagged_no_trade_n_of_organiz_registered_per_year +
Interinst_conflict +
nofterms_in_eurovoc +
factor(form) + factor(year) + factor(Country_num),
data = CurbEUenthusiasm_FINAL)
stargazer(model1_robH1, model_robH1a, model_robH1b, model_robH1c, type = "text" ,
omit=c("Country_num"), omit.labels = c("Country FE?"), title='Table 4: Robstness checks: BetaRegs', align = TRUE)##
## Table 4: Robstness checks: BetaRegs
## ===========================================================================================================================
## Dependent variable:
## ---------------------------------------
## adjusted_probability
## (1) (2) (3) (4)
## ---------------------------------------------------------------------------------------------------------------------------
## lagged2y_pct_support_cntr_cap_year 0.007*** 0.003 0.010*** 0.007***
## (0.001) (0.002) (0.002) (0.001)
##
## lagged_pct_Salience_DK 0.007*** 0.004** 0.007*** 0.007***
## (0.001) (0.002) (0.001) (0.001)
##
## lagged_cntr_polarization 0.228*** 0.235*** 0.658*** 0.229***
## (0.049) (0.049) (0.196) (0.049)
##
## EUpolarizarion -0.298*** -0.291*** -0.322*** -0.297***
## (0.046) (0.046) (0.047) (0.046)
##
## Lagged_no_trade_n_of_organiz_registered_per_year -0.033** -0.032** -0.032** 0.068
## (0.013) (0.013) (0.013) (0.043)
##
## Interinst_conflict -0.068*** -0.069*** -0.066*** -0.068***
## (0.018) (0.018) (0.018) (0.018)
##
## nofterms_in_eurovoc 0.041*** 0.041*** 0.041*** 0.041***
## (0.005) (0.005) (0.005) (0.005)
##
## factor(form)Directive -0.323*** -0.323*** -0.322*** -0.322***
## (0.038) (0.038) (0.038) (0.038)
##
## factor(form)Regulation -0.240*** -0.239*** -0.240*** -0.239***
## (0.035) (0.035) (0.035) (0.035)
##
## factor(year)2011 0.043 0.040 0.043 0.043
## (0.031) (0.031) (0.031) (0.031)
##
## factor(year)2012 -0.200*** -0.205*** -0.197*** -0.195***
## (0.038) (0.038) (0.038) (0.038)
##
## factor(year)2013 -0.059 -0.049 -0.047 -0.058
## (0.039) (0.039) (0.039) (0.039)
##
## factor(year)2014 -0.733*** -0.737*** -0.727*** -0.735***
## (0.047) (0.047) (0.047) (0.047)
##
## factor(year)2015 -0.028 -0.035 -0.021 -0.027
## (0.052) (0.052) (0.052) (0.052)
##
## factor(year)2016 -0.214*** -0.214*** -0.210*** -0.212***
## (0.037) (0.037) (0.037) (0.037)
##
## lagged2y_pct_support_cntr_cap_year:lagged_pct_Salience_DK 0.00004
## (0.00003)
##
## lagged2y_pct_support_cntr_cap_year:lagged_cntr_polarization -0.006**
## (0.003)
##
## lagged2y_pct_support_cntr_cap_year:Lagged_no_trade_n_of_organiz_registered_per_year -0.002**
## (0.001)
##
## Constant -0.672*** -0.468** -0.920*** -0.691***
## (0.126) (0.186) (0.168) (0.126)
##
## ---------------------------------------------------------------------------------------------------------------------------
## Country FE? Yes Yes Yes Yes
## ---------------------------------------------------------------------------------------------------------------------------
## Observations 12,251 12,251 12,251 12,251
## R2 0.055 0.055 0.056 0.056
## Log Likelihood 673.834 674.850 676.335 676.933
## ===========================================================================================================================
## Note: *p<0.1; **p<0.05; ***p<0.01Margins’ plots for the Beta regressions
Model 1: Without interaction
Model 2: Public Support X Salience
plot_model(model_robH1a, type='pred', terms='lagged2y_pct_support_cntr_cap_year')
plot_model(model_robH1a, type='int', mdrt.values = 'meansd') # plotting mean +/- 1sd
plot_model(model_robH1a, type='int')
Model 3: Public Support X POlarization
Model 4: Public Support x Actor Moblization
plot_model(model_robH1c, type='pred', terms='lagged2y_pct_support_cntr_cap_year')
plot_model(model_robH1c, type='int', mdrt.values = 'meansd') # plotting mean +/- 1sd
plot_model(model_robH1c, type='int')
## Polarization Index: Robustness checks Following De Bryucker, the polarization index is calculated as following: PI=Salience*( Actor Mobilization+Polarization)
## Generate cntr _polarization in % first to at leasthave some scale comparable
CurbEUenthusiasm_FINAL=CurbEUenthusiasm_FINAL %>% mutate(lagged_pct_cntr_polariz=100*lagged_cntr_polarization,
Polit_Index=lagged_pct_Salience_DK*(Lagged_no_trade_n_of_organiz_registered_per_year+
lagged_pct_cntr_polariz))
var_label(CurbEUenthusiasm_FINAL ) <- list( Polit_Index="Lagged_PoliticizationIndex")
##Beta reg with PI
beta_PI=betareg(adjusted_probability ~ lagged2y_pct_support_cntr_cap_year + Polit_Index+ EUpolarizarion + Interinst_conflict +
nofterms_in_eurovoc + competence_length+
factor(form) + factor(year) + factor(Country_num),
data = CurbEUenthusiasm_FINAL)
beta_PI_inter=betareg(adjusted_probability ~ lagged2y_pct_support_cntr_cap_year + Polit_Index+
lagged2y_pct_support_cntr_cap_year*Polit_Index+
EUpolarizarion +
Interinst_conflict +
nofterms_in_eurovoc + competence_length+
factor(form) + factor(year) + factor(Country_num),
data = CurbEUenthusiasm_FINAL)
###Fracreg
frac_PI= glm(LR_Proposal_Probability ~ lagged2y_pct_support_cntr_cap_year +
Polit_Index+
EUpolarizarion +
Interinst_conflict +
nofterms_in_eurovoc +
competence_length+
factor(form) + factor(year) + factor(Country_num) ,
data = CurbEUenthusiasm_FINAL, family = quasibinomial('logit'))
frac_PI_inter= glm(LR_Proposal_Probability ~ lagged2y_pct_support_cntr_cap_year +
Polit_Index+
lagged2y_pct_support_cntr_cap_year*Polit_Index+
EUpolarizarion +
Interinst_conflict +
nofterms_in_eurovoc +
competence_length+
factor(form) + factor(year) + factor(Country_num) ,
data = CurbEUenthusiasm_FINAL, family = quasibinomial('logit'))
stargazer(beta_PI, beta_PI_inter, frac_PI, frac_PI_inter, type='text',
omit=c("Country_num"), omit.labels = c("Country FE?"), title='Table 5: Effects of Politicization Index', align = TRUE)##
## Table 5: Effects of Politicization Index
## ==============================================================================================
## Dependent variable:
## -----------------------------------------------
## adjusted_probability LR_Proposal_Probability
## beta glm: quasibinomial
## link = logit
## (1) (2) (3) (4)
## ----------------------------------------------------------------------------------------------
## lagged2y_pct_support_cntr_cap_year 0.006*** 0.010*** 0.005*** 0.010***
## (0.001) (0.002) (0.001) (0.002)
##
## Polit_Index 0.00003*** 0.0001*** 0.00002*** 0.0001***
## (0.00001) (0.00002) (0.00001) (0.00002)
##
## EUpolarizarion -0.262*** -0.297*** -0.270*** -0.314***
## (0.045) (0.046) (0.047) (0.048)
##
## Interinst_conflict -0.183*** -0.181*** -0.169*** -0.167***
## (0.019) (0.019) (0.020) (0.020)
##
## nofterms_in_eurovoc 0.035*** 0.035*** 0.042*** 0.042***
## (0.005) (0.005) (0.005) (0.005)
##
## competence_length -0.015*** -0.015*** -0.015*** -0.016***
## (0.001) (0.001) (0.001) (0.001)
##
## factor(form)Directive -0.384*** -0.384*** -0.466*** -0.465***
## (0.037) (0.037) (0.039) (0.039)
##
## factor(form)Regulation -0.250*** -0.251*** -0.302*** -0.304***
## (0.035) (0.035) (0.036) (0.036)
##
## factor(year)2011 0.058* 0.065** 0.016 0.025
## (0.030) (0.030) (0.031) (0.032)
##
## factor(year)2012 -0.210*** -0.195*** -0.261*** -0.242***
## (0.037) (0.038) (0.039) (0.039)
##
## factor(year)2013 -0.057 -0.055 -0.088** -0.085**
## (0.036) (0.036) (0.038) (0.038)
##
## factor(year)2014 -0.600*** -0.578*** -0.713*** -0.686***
## (0.047) (0.048) (0.050) (0.050)
##
## factor(year)2015 0.117** 0.138*** 0.138*** 0.164***
## (0.052) (0.052) (0.054) (0.054)
##
## factor(year)2016 -0.145*** -0.136*** -0.147*** -0.135***
## (0.037) (0.037) (0.038) (0.038)
##
## lagged2y_pct_support_cntr_cap_year:Polit_Index -0.00000*** -0.00000***
## (0.00000) (0.00000)
##
## Constant 0.659*** 0.417*** 0.728*** 0.422***
## (0.102) (0.129) (0.106) (0.135)
##
## ----------------------------------------------------------------------------------------------
## Country FE? Yes Yes Yes Yes
## ----------------------------------------------------------------------------------------------
## Observations 12,251 12,251 12,251 12,251
## R2 0.080 0.081
## Log Likelihood 835.941 840.394
## ==============================================================================================
## Note: *p<0.1; **p<0.05; ***p<0.01The Plots for PI Models
- Margin plots are extremely messy! Best to be avoided!
Plot Betareg Polarization INdex and Public Support
plot_model(beta_PI, type='pred', terms=c ('lagged2y_pct_support_cntr_cap_year'))
plot_model(beta_PI, type='pred', terms=c ('Polit_Index'))
Plot Betareg Polarization INdex as interaction with PublicSupport
plot_model(beta_PI_inter, type='pred', terms=c ('lagged2y_pct_support_cntr_cap_year'))
plot_model(beta_PI_inter, type='pred', terms=c ('Polit_Index'))
plot_model(beta_PI_inter, type='int', mdrt.values = 'meansd') # plotting mean +/- 1sd
plot_model(beta_PI_inter, type='int')
Plot Fracreg Polarization Index and PublicSupport
plot_model(frac_PI, type='pred', terms=c ('lagged2y_pct_support_cntr_cap_year'))## Data were 'prettified'. Consider using `terms="lagged2y_pct_support_cntr_cap_year [all]"` to get smooth plots.
plot_model(frac_PI, type='pred', terms=c ( 'Polit_Index'))## Data were 'prettified'. Consider using `terms="Polit_Index [all]"` to get smooth plots.
Plot Fracreg Polarization Index as interaction with Public Support
plot_model(frac_PI_inter, type='pred', terms=c ('lagged2y_pct_support_cntr_cap_year'))## Data were 'prettified'. Consider using `terms="lagged2y_pct_support_cntr_cap_year [all]"` to get smooth plots.
plot_model(frac_PI_inter, type='pred', terms=c ( 'Polit_Index'))## Data were 'prettified'. Consider using `terms="Polit_Index [all]"` to get smooth plots.
plot_model(frac_PI_inter, type='int', mdrt.values = 'meansd') # plotting mean +/- 1sd## Data were 'prettified'. Consider using `terms="lagged2y_pct_support_cntr_cap_year [all]"` to get smooth plots.
plot_model(frac_PI_inter, type='int')## Data were 'prettified'. Consider using `terms="lagged2y_pct_support_cntr_cap_year [all]"` to get smooth plots.
TOBITS
library(AER)
#H1
tobit1=AER:: tobit(LR_Proposal_Probability~ lagged2y_pct_support_cntr_cap_year +
lagged_pct_Salience_DK +
lagged_pct_cntr_polariz+
EUpolarizarion +
Lagged_no_trade_n_of_organiz_registered_per_year +
Interinst_conflict +
nofterms_in_eurovoc +
factor(form) + factor(year) + factor(Competence) + factor(Country_num), right=1,
data = CurbEUenthusiasm_FINAL)
summary(tobit1)##
## Call:
## AER::tobit(formula = LR_Proposal_Probability ~ lagged2y_pct_support_cntr_cap_year +
## lagged_pct_Salience_DK + lagged_pct_cntr_polariz + EUpolarizarion +
## Lagged_no_trade_n_of_organiz_registered_per_year + Interinst_conflict +
## nofterms_in_eurovoc + factor(form) + factor(year) + factor(Competence) +
## factor(Country_num), right = 1, data = CurbEUenthusiasm_FINAL)
##
## Observations: (4913 observations deleted due to missingness)
## Total Left-censored Uncensored Right-censored
## 12250 0 12223 27
##
## Coefficients:
## Estimate Std. Error z value
## (Intercept) 0.2761730 0.0325291 8.490
## lagged2y_pct_support_cntr_cap_year 0.0014649 0.0001994 7.348
## lagged_pct_Salience_DK 0.0017003 0.0001786 9.521
## lagged_pct_cntr_polariz 0.0003566 0.0001239 2.879
## EUpolarizarion -0.0754438 0.0116608 -6.470
## Lagged_no_trade_n_of_organiz_registered_per_year -0.0078371 0.0032755 -2.393
## Interinst_conflict -0.0153648 0.0047287 -3.249
## nofterms_in_eurovoc 0.0123029 0.0012528 9.820
## factor(form)Directive -0.0992705 0.0094621 -10.491
## factor(form)Regulation -0.0652281 0.0087721 -7.436
## factor(year)2011 -0.0106656 0.0078032 -1.367
## factor(year)2012 -0.0721163 0.0095586 -7.545
## factor(year)2013 -0.0184218 0.0097873 -1.882
## factor(year)2014 -0.2111952 0.0119968 -17.604
## factor(year)2015 -0.0044157 0.0130709 -0.338
## factor(year)2016 -0.0560596 0.0093596 -5.990
## factor(Competence)1 0.0653744 0.0093192 7.015
## factor(Competence)2 0.1770356 0.0147094 12.036
## factor(Country_num)2 0.0284922 0.0232208 1.227
## factor(Country_num)3 -0.0033991 0.0173600 -0.196
## factor(Country_num)4 -0.0024573 0.0173507 -0.142
## factor(Country_num)5 0.0005698 0.0173455 0.033
## factor(Country_num)6 -0.0061248 0.0174057 -0.352
## factor(Country_num)7 -0.0006777 0.0174240 -0.039
## factor(Country_num)8 0.0028188 0.0174105 0.162
## factor(Country_num)9 0.0090630 0.0175492 0.516
## factor(Country_num)11 -0.0107843 0.0174094 -0.619
## factor(Country_num)12 0.0039743 0.0173511 0.229
## factor(Country_num)13 -0.0039522 0.0174003 -0.227
## factor(Country_num)16 0.0055504 0.0175260 0.317
## factor(Country_num)17 0.0003460 0.0174453 0.020
## factor(Country_num)18 0.0012910 0.0174875 0.074
## factor(Country_num)19 -0.0114095 0.0174498 -0.654
## factor(Country_num)20 -0.0000215 0.0174272 -0.001
## factor(Country_num)21 0.0047937 0.0174241 0.275
## factor(Country_num)22 -0.0044344 0.0173861 -0.255
## factor(Country_num)23 0.0010962 0.0174026 0.063
## factor(Country_num)24 -0.0001651 0.0173968 -0.009
## factor(Country_num)25 0.0089688 0.0174160 0.515
## factor(Country_num)26 0.0018694 0.0174027 0.107
## factor(Country_num)27 -0.0084440 0.0173970 -0.485
## factor(Country_num)28 -0.0060128 0.0174067 -0.345
## factor(Country_num)29 0.0016164 0.0174846 0.092
## factor(Country_num)30 0.0059362 0.0175017 0.339
## factor(Country_num)32 0.0649274 0.0245679 2.643
## Log(scale) -1.3489563 0.0063998 -210.783
## Pr(>|z|)
## (Intercept) < 2e-16 ***
## lagged2y_pct_support_cntr_cap_year 2.02e-13 ***
## lagged_pct_Salience_DK < 2e-16 ***
## lagged_pct_cntr_polariz 0.00399 **
## EUpolarizarion 9.81e-11 ***
## Lagged_no_trade_n_of_organiz_registered_per_year 0.01673 *
## Interinst_conflict 0.00116 **
## nofterms_in_eurovoc < 2e-16 ***
## factor(form)Directive < 2e-16 ***
## factor(form)Regulation 1.04e-13 ***
## factor(year)2011 0.17168
## factor(year)2012 4.53e-14 ***
## factor(year)2013 0.05981 .
## factor(year)2014 < 2e-16 ***
## factor(year)2015 0.73549
## factor(year)2016 2.10e-09 ***
## factor(Competence)1 2.30e-12 ***
## factor(Competence)2 < 2e-16 ***
## factor(Country_num)2 0.21982
## factor(Country_num)3 0.84477
## factor(Country_num)4 0.88737
## factor(Country_num)5 0.97380
## factor(Country_num)6 0.72492
## factor(Country_num)7 0.96898
## factor(Country_num)8 0.87138
## factor(Country_num)9 0.60555
## factor(Country_num)11 0.53562
## factor(Country_num)12 0.81883
## factor(Country_num)13 0.82032
## factor(Country_num)16 0.75148
## factor(Country_num)17 0.98417
## factor(Country_num)18 0.94115
## factor(Country_num)19 0.51321
## factor(Country_num)20 0.99902
## factor(Country_num)21 0.78323
## factor(Country_num)22 0.79868
## factor(Country_num)23 0.94977
## factor(Country_num)24 0.99243
## factor(Country_num)25 0.60657
## factor(Country_num)26 0.91445
## factor(Country_num)27 0.62741
## factor(Country_num)28 0.72977
## factor(Country_num)29 0.92634
## factor(Country_num)30 0.73448
## factor(Country_num)32 0.00822 **
## Log(scale) < 2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Scale: 0.2595
##
## Gaussian distribution
## Number of Newton-Raphson Iterations: 3
## Log-likelihood: -891.4 on 46 Df
## Wald-statistic: 959 on 44 Df, p-value: < 2.22e-16### test hypothesis 1a: public opinion*salience : Also not much difference in the effects
tobit2=AER:: tobit(LR_Proposal_Probability~ lagged2y_pct_support_cntr_cap_year +
lagged_pct_Salience_DK +
lagged2y_pct_support_cntr_cap_year* lagged_pct_Salience_DK+
lagged_pct_cntr_polariz+
EUpolarizarion +
Lagged_no_trade_n_of_organiz_registered_per_year +
Interinst_conflict +
nofterms_in_eurovoc +
competence_length +
factor(form) + factor(year) + factor(Country_num), right=1, data = CurbEUenthusiasm_FINAL)
summary(tobit2)##
## Call:
## AER::tobit(formula = LR_Proposal_Probability ~ lagged2y_pct_support_cntr_cap_year +
## lagged_pct_Salience_DK + lagged2y_pct_support_cntr_cap_year *
## lagged_pct_Salience_DK + lagged_pct_cntr_polariz + EUpolarizarion +
## Lagged_no_trade_n_of_organiz_registered_per_year + Interinst_conflict +
## nofterms_in_eurovoc + competence_length + factor(form) +
## factor(year) + factor(Country_num), right = 1, data = CurbEUenthusiasm_FINAL)
##
## Observations: (4912 observations deleted due to missingness)
## Total Left-censored Uncensored Right-censored
## 12251 0 12224 27
##
## Coefficients:
## Estimate Std. Error
## (Intercept) 5.229e-01 4.718e-02
## lagged2y_pct_support_cntr_cap_year 2.310e-03 5.902e-04
## lagged_pct_Salience_DK 1.757e-03 4.853e-04
## lagged_pct_cntr_polariz 3.128e-04 1.236e-04
## EUpolarizarion -5.629e-02 1.155e-02
## Lagged_no_trade_n_of_organiz_registered_per_year -5.746e-03 3.257e-03
## Interinst_conflict -4.334e-02 4.861e-03
## nofterms_in_eurovoc 1.045e-02 1.245e-03
## competence_length -3.604e-03 2.008e-04
## factor(form)Directive -1.089e-01 9.403e-03
## factor(form)Regulation -7.044e-02 8.698e-03
## factor(year)2011 6.741e-03 7.738e-03
## factor(year)2012 -5.977e-02 9.486e-03
## factor(year)2013 -1.500e-02 9.874e-03
## factor(year)2014 -1.708e-01 1.198e-02
## factor(year)2015 3.391e-02 1.316e-02
## factor(year)2016 -3.245e-02 9.339e-03
## factor(Country_num)2 2.248e-02 2.308e-02
## factor(Country_num)3 -2.047e-03 1.724e-02
## factor(Country_num)4 -9.557e-04 1.723e-02
## factor(Country_num)5 -1.794e-04 1.723e-02
## factor(Country_num)6 -4.015e-03 1.729e-02
## factor(Country_num)7 -1.973e-04 1.730e-02
## factor(Country_num)8 9.236e-04 1.729e-02
## factor(Country_num)9 6.162e-03 1.743e-02
## factor(Country_num)11 -6.816e-03 1.729e-02
## factor(Country_num)12 1.937e-03 1.724e-02
## factor(Country_num)13 -3.361e-03 1.728e-02
## factor(Country_num)16 4.681e-03 1.742e-02
## factor(Country_num)17 7.644e-04 1.732e-02
## factor(Country_num)18 2.247e-03 1.737e-02
## factor(Country_num)19 -8.157e-03 1.733e-02
## factor(Country_num)20 2.237e-04 1.731e-02
## factor(Country_num)21 2.674e-03 1.731e-02
## factor(Country_num)22 -2.326e-03 1.727e-02
## factor(Country_num)23 6.263e-04 1.728e-02
## factor(Country_num)24 -1.106e-03 1.728e-02
## factor(Country_num)25 4.020e-03 1.731e-02
## factor(Country_num)26 -4.897e-04 1.729e-02
## factor(Country_num)27 -5.433e-03 1.728e-02
## factor(Country_num)28 -3.603e-03 1.729e-02
## factor(Country_num)29 -3.931e-04 1.737e-02
## factor(Country_num)30 3.101e-03 1.739e-02
## factor(Country_num)32 5.409e-02 2.466e-02
## lagged2y_pct_support_cntr_cap_year:lagged_pct_Salience_DK -1.275e-05 7.062e-06
## Log(scale) -1.356e+00 6.399e-03
## z value Pr(>|z|)
## (Intercept) 11.082 < 2e-16 ***
## lagged2y_pct_support_cntr_cap_year 3.915 9.05e-05 ***
## lagged_pct_Salience_DK 3.620 0.000295 ***
## lagged_pct_cntr_polariz 2.531 0.011374 *
## EUpolarizarion -4.873 1.10e-06 ***
## Lagged_no_trade_n_of_organiz_registered_per_year -1.764 0.077653 .
## Interinst_conflict -8.915 < 2e-16 ***
## nofterms_in_eurovoc 8.389 < 2e-16 ***
## competence_length -17.947 < 2e-16 ***
## factor(form)Directive -11.576 < 2e-16 ***
## factor(form)Regulation -8.099 5.54e-16 ***
## factor(year)2011 0.871 0.383622
## factor(year)2012 -6.301 2.97e-10 ***
## factor(year)2013 -1.519 0.128830
## factor(year)2014 -14.254 < 2e-16 ***
## factor(year)2015 2.576 0.009995 **
## factor(year)2016 -3.474 0.000512 ***
## factor(Country_num)2 0.974 0.330034
## factor(Country_num)3 -0.119 0.905488
## factor(Country_num)4 -0.055 0.955771
## factor(Country_num)5 -0.010 0.991693
## factor(Country_num)6 -0.232 0.816318
## factor(Country_num)7 -0.011 0.990905
## factor(Country_num)8 0.053 0.957390
## factor(Country_num)9 0.353 0.723720
## factor(Country_num)11 -0.394 0.693443
## factor(Country_num)12 0.112 0.910511
## factor(Country_num)13 -0.195 0.845778
## factor(Country_num)16 0.269 0.788205
## factor(Country_num)17 0.044 0.964805
## factor(Country_num)18 0.129 0.897068
## factor(Country_num)19 -0.471 0.637919
## factor(Country_num)20 0.013 0.989689
## factor(Country_num)21 0.154 0.877230
## factor(Country_num)22 -0.135 0.892829
## factor(Country_num)23 0.036 0.971093
## factor(Country_num)24 -0.064 0.948974
## factor(Country_num)25 0.232 0.816328
## factor(Country_num)26 -0.028 0.977401
## factor(Country_num)27 -0.314 0.753171
## factor(Country_num)28 -0.208 0.834919
## factor(Country_num)29 -0.023 0.981949
## factor(Country_num)30 0.178 0.858437
## factor(Country_num)32 2.194 0.028251 *
## lagged2y_pct_support_cntr_cap_year:lagged_pct_Salience_DK -1.805 0.071065 .
## Log(scale) -211.891 < 2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Scale: 0.2577
##
## Gaussian distribution
## Number of Newton-Raphson Iterations: 3
## Log-likelihood: -803.8 on 46 Df
## Wald-statistic: 1148 on 44 Df, p-value: < 2.22e-16tobit3=AER:: tobit (LR_Proposal_Probability~ lagged2y_pct_support_cntr_cap_year +
lagged_pct_Salience_DK +
lagged2y_pct_support_cntr_cap_year* lagged_pct_cntr_polariz+
lagged_pct_cntr_polariz +
EUpolarizarion +
Lagged_no_trade_n_of_organiz_registered_per_year +
Interinst_conflict +
nofterms_in_eurovoc +
competence_length+
factor(form) + factor(year) + factor(Country_num) ,
right=1, left=0,
data = CurbEUenthusiasm_FINAL)
summary(tobit3)##
## Call:
## AER::tobit(formula = LR_Proposal_Probability ~ lagged2y_pct_support_cntr_cap_year +
## lagged_pct_Salience_DK + lagged2y_pct_support_cntr_cap_year *
## lagged_pct_cntr_polariz + lagged_pct_cntr_polariz + EUpolarizarion +
## Lagged_no_trade_n_of_organiz_registered_per_year + Interinst_conflict +
## nofterms_in_eurovoc + competence_length + factor(form) +
## factor(year) + factor(Country_num), left = 0, right = 1,
## data = CurbEUenthusiasm_FINAL)
##
## Observations: (4912 observations deleted due to missingness)
## Total Left-censored Uncensored Right-censored
## 12251 0 12224 27
##
## Coefficients:
## Estimate
## (Intercept) 5.149e-01
## lagged2y_pct_support_cntr_cap_year 2.295e-03
## lagged_pct_Salience_DK 9.441e-04
## lagged_pct_cntr_polariz 1.479e-03
## EUpolarizarion -6.029e-02
## Lagged_no_trade_n_of_organiz_registered_per_year -5.486e-03
## Interinst_conflict -4.251e-02
## nofterms_in_eurovoc 1.055e-02
## competence_length -3.553e-03
## factor(form)Directive -1.081e-01
## factor(form)Regulation -7.014e-02
## factor(year)2011 5.600e-03
## factor(year)2012 -6.056e-02
## factor(year)2013 -8.681e-03
## factor(year)2014 -1.713e-01
## factor(year)2015 3.315e-02
## factor(year)2016 -3.179e-02
## factor(Country_num)2 2.104e-02
## factor(Country_num)3 -1.658e-03
## factor(Country_num)4 -1.916e-03
## factor(Country_num)5 1.436e-03
## factor(Country_num)6 -3.507e-03
## factor(Country_num)7 7.369e-04
## factor(Country_num)8 1.684e-03
## factor(Country_num)9 6.594e-03
## factor(Country_num)11 -5.591e-03
## factor(Country_num)12 2.932e-03
## factor(Country_num)13 -2.593e-03
## factor(Country_num)16 7.002e-03
## factor(Country_num)17 5.181e-04
## factor(Country_num)18 3.184e-03
## factor(Country_num)19 -9.616e-03
## factor(Country_num)20 1.859e-03
## factor(Country_num)21 3.933e-03
## factor(Country_num)22 -1.574e-03
## factor(Country_num)23 1.116e-03
## factor(Country_num)24 -2.789e-04
## factor(Country_num)25 4.636e-03
## factor(Country_num)26 1.335e-03
## factor(Country_num)27 -5.216e-03
## factor(Country_num)28 -2.880e-03
## factor(Country_num)29 7.697e-04
## factor(Country_num)30 5.210e-03
## factor(Country_num)32 4.494e-02
## lagged2y_pct_support_cntr_cap_year:lagged_pct_cntr_polariz -1.711e-05
## Log(scale) -1.356e+00
## Std. Error z value
## (Intercept) 4.405e-02 11.690
## lagged2y_pct_support_cntr_cap_year 4.557e-04 5.035
## lagged_pct_Salience_DK 1.822e-04 5.183
## lagged_pct_cntr_polariz 4.912e-04 3.012
## EUpolarizarion 1.177e-02 -5.123
## Lagged_no_trade_n_of_organiz_registered_per_year 3.257e-03 -1.684
## Interinst_conflict 4.862e-03 -8.742
## nofterms_in_eurovoc 1.245e-03 8.475
## competence_length 1.985e-04 -17.902
## factor(form)Directive 9.401e-03 -11.493
## factor(form)Regulation 8.695e-03 -8.066
## factor(year)2011 7.712e-03 0.726
## factor(year)2012 9.449e-03 -6.410
## factor(year)2013 9.815e-03 -0.884
## factor(year)2014 1.194e-02 -14.344
## factor(year)2015 1.311e-02 2.529
## factor(year)2016 9.345e-03 -3.402
## factor(Country_num)2 2.307e-02 0.912
## factor(Country_num)3 1.724e-02 -0.096
## factor(Country_num)4 1.723e-02 -0.111
## factor(Country_num)5 1.723e-02 0.083
## factor(Country_num)6 1.729e-02 -0.203
## factor(Country_num)7 1.730e-02 0.043
## factor(Country_num)8 1.728e-02 0.097
## factor(Country_num)9 1.742e-02 0.378
## factor(Country_num)11 1.730e-02 -0.323
## factor(Country_num)12 1.723e-02 0.170
## factor(Country_num)13 1.728e-02 -0.150
## factor(Country_num)16 1.740e-02 0.402
## factor(Country_num)17 1.732e-02 0.030
## factor(Country_num)18 1.736e-02 0.183
## factor(Country_num)19 1.733e-02 -0.555
## factor(Country_num)20 1.731e-02 0.107
## factor(Country_num)21 1.730e-02 0.227
## factor(Country_num)22 1.727e-02 -0.091
## factor(Country_num)23 1.728e-02 0.065
## factor(Country_num)24 1.728e-02 -0.016
## factor(Country_num)25 1.730e-02 0.268
## factor(Country_num)26 1.729e-02 0.077
## factor(Country_num)27 1.728e-02 -0.302
## factor(Country_num)28 1.729e-02 -0.167
## factor(Country_num)29 1.736e-02 0.044
## factor(Country_num)30 1.739e-02 0.300
## factor(Country_num)32 2.442e-02 1.840
## lagged2y_pct_support_cntr_cap_year:lagged_pct_cntr_polariz 7.115e-06 -2.404
## Log(scale) 6.399e-03 -211.906
## Pr(>|z|)
## (Intercept) < 2e-16 ***
## lagged2y_pct_support_cntr_cap_year 4.77e-07 ***
## lagged_pct_Salience_DK 2.19e-07 ***
## lagged_pct_cntr_polariz 0.00260 **
## EUpolarizarion 3.00e-07 ***
## Lagged_no_trade_n_of_organiz_registered_per_year 0.09211 .
## Interinst_conflict < 2e-16 ***
## nofterms_in_eurovoc < 2e-16 ***
## competence_length < 2e-16 ***
## factor(form)Directive < 2e-16 ***
## factor(form)Regulation 7.27e-16 ***
## factor(year)2011 0.46781
## factor(year)2012 1.46e-10 ***
## factor(year)2013 0.37646
## factor(year)2014 < 2e-16 ***
## factor(year)2015 0.01144 *
## factor(year)2016 0.00067 ***
## factor(Country_num)2 0.36164
## factor(Country_num)3 0.92337
## factor(Country_num)4 0.91145
## factor(Country_num)5 0.93361
## factor(Country_num)6 0.83925
## factor(Country_num)7 0.96603
## factor(Country_num)8 0.92239
## factor(Country_num)9 0.70512
## factor(Country_num)11 0.74652
## factor(Country_num)12 0.86491
## factor(Country_num)13 0.88074
## factor(Country_num)16 0.68742
## factor(Country_num)17 0.97614
## factor(Country_num)18 0.85450
## factor(Country_num)19 0.57901
## factor(Country_num)20 0.91446
## factor(Country_num)21 0.82023
## factor(Country_num)22 0.92738
## factor(Country_num)23 0.94851
## factor(Country_num)24 0.98712
## factor(Country_num)25 0.78868
## factor(Country_num)26 0.93844
## factor(Country_num)27 0.76272
## factor(Country_num)28 0.86773
## factor(Country_num)29 0.96464
## factor(Country_num)30 0.76447
## factor(Country_num)32 0.06575 .
## lagged2y_pct_support_cntr_cap_year:lagged_pct_cntr_polariz 0.01621 *
## Log(scale) < 2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Scale: 0.2577
##
## Gaussian distribution
## Number of Newton-Raphson Iterations: 3
## Log-likelihood: -802.5 on 46 Df
## Wald-statistic: 1151 on 44 Df, p-value: < 2.22e-16### test hypothesis 1c: public opinion*actor mobilization
tobit4= AER::tobit (LR_Proposal_Probability~ lagged2y_pct_support_cntr_cap_year +
lagged_pct_Salience_DK +
lagged2y_pct_support_cntr_cap_year*Lagged_no_trade_n_of_organiz_registered_per_year+
lagged_pct_cntr_polariz+
EUpolarizarion +
Lagged_no_trade_n_of_organiz_registered_per_year +
Interinst_conflict +
nofterms_in_eurovoc +
factor(form) + factor(year) + competence_length + factor(Country_num),
right=1, left=0,
data = CurbEUenthusiasm_FINAL)
summary(tobit4)##
## Call:
## AER::tobit(formula = LR_Proposal_Probability ~ lagged2y_pct_support_cntr_cap_year +
## lagged_pct_Salience_DK + lagged2y_pct_support_cntr_cap_year *
## Lagged_no_trade_n_of_organiz_registered_per_year + lagged_pct_cntr_polariz +
## EUpolarizarion + Lagged_no_trade_n_of_organiz_registered_per_year +
## Interinst_conflict + nofterms_in_eurovoc + factor(form) +
## factor(year) + competence_length + factor(Country_num), left = 0,
## right = 1, data = CurbEUenthusiasm_FINAL)
##
## Observations: (4912 observations deleted due to missingness)
## Total Left-censored Uncensored Right-censored
## 12251 0 12224 27
##
## Coefficients:
## Estimate
## (Intercept) 5.768e-01
## lagged2y_pct_support_cntr_cap_year 1.367e-03
## lagged_pct_Salience_DK 9.453e-04
## Lagged_no_trade_n_of_organiz_registered_per_year 1.544e-02
## lagged_pct_cntr_polariz 3.383e-04
## EUpolarizarion -5.412e-02
## Interinst_conflict -4.293e-02
## nofterms_in_eurovoc 1.051e-02
## factor(form)Directive -1.082e-01
## factor(form)Regulation -6.993e-02
## factor(year)2011 5.517e-03
## factor(year)2012 -6.030e-02
## factor(year)2013 -1.162e-02
## factor(year)2014 -1.732e-01
## factor(year)2015 3.153e-02
## factor(year)2016 -3.254e-02
## competence_length -3.540e-03
## factor(Country_num)2 2.765e-02
## factor(Country_num)3 -3.078e-03
## factor(Country_num)4 -1.368e-03
## factor(Country_num)5 -8.597e-05
## factor(Country_num)6 -5.121e-03
## factor(Country_num)7 -5.578e-04
## factor(Country_num)8 7.859e-04
## factor(Country_num)9 6.585e-03
## factor(Country_num)11 -7.678e-03
## factor(Country_num)12 3.545e-03
## factor(Country_num)13 -4.542e-03
## factor(Country_num)16 5.967e-03
## factor(Country_num)17 4.804e-04
## factor(Country_num)18 2.263e-03
## factor(Country_num)19 -1.018e-02
## factor(Country_num)20 9.004e-05
## factor(Country_num)21 2.590e-03
## factor(Country_num)22 -3.325e-03
## factor(Country_num)23 -2.009e-04
## factor(Country_num)24 -1.817e-03
## factor(Country_num)25 4.733e-03
## factor(Country_num)26 -8.050e-04
## factor(Country_num)27 -6.625e-03
## factor(Country_num)28 -4.805e-03
## factor(Country_num)29 -3.986e-04
## factor(Country_num)30 3.195e-03
## factor(Country_num)32 4.834e-02
## lagged2y_pct_support_cntr_cap_year:Lagged_no_trade_n_of_organiz_registered_per_year -3.165e-04
## Log(scale) -1.356e+00
## Std. Error
## (Intercept) 3.440e-02
## lagged2y_pct_support_cntr_cap_year 1.982e-04
## lagged_pct_Salience_DK 1.822e-04
## Lagged_no_trade_n_of_organiz_registered_per_year 1.073e-02
## lagged_pct_cntr_polariz 1.229e-04
## EUpolarizarion 1.149e-02
## Interinst_conflict 4.858e-03
## nofterms_in_eurovoc 1.245e-03
## factor(form)Directive 9.402e-03
## factor(form)Regulation 8.696e-03
## factor(year)2011 7.713e-03
## factor(year)2012 9.458e-03
## factor(year)2013 9.725e-03
## factor(year)2014 1.193e-02
## factor(year)2015 1.309e-02
## factor(year)2016 9.338e-03
## competence_length 1.985e-04
## factor(Country_num)2 2.326e-02
## factor(Country_num)3 1.725e-02
## factor(Country_num)4 1.723e-02
## factor(Country_num)5 1.723e-02
## factor(Country_num)6 1.729e-02
## factor(Country_num)7 1.731e-02
## factor(Country_num)8 1.729e-02
## factor(Country_num)9 1.743e-02
## factor(Country_num)11 1.730e-02
## factor(Country_num)12 1.724e-02
## factor(Country_num)13 1.729e-02
## factor(Country_num)16 1.740e-02
## factor(Country_num)17 1.733e-02
## factor(Country_num)18 1.737e-02
## factor(Country_num)19 1.734e-02
## factor(Country_num)20 1.731e-02
## factor(Country_num)21 1.731e-02
## factor(Country_num)22 1.727e-02
## factor(Country_num)23 1.729e-02
## factor(Country_num)24 1.728e-02
## factor(Country_num)25 1.730e-02
## factor(Country_num)26 1.729e-02
## factor(Country_num)27 1.728e-02
## factor(Country_num)28 1.729e-02
## factor(Country_num)29 1.737e-02
## factor(Country_num)30 1.739e-02
## factor(Country_num)32 2.440e-02
## lagged2y_pct_support_cntr_cap_year:Lagged_no_trade_n_of_organiz_registered_per_year 1.534e-04
## Log(scale) 6.399e-03
## z value
## (Intercept) 16.767
## lagged2y_pct_support_cntr_cap_year 6.897
## lagged_pct_Salience_DK 5.189
## Lagged_no_trade_n_of_organiz_registered_per_year 1.438
## lagged_pct_cntr_polariz 2.752
## EUpolarizarion -4.709
## Interinst_conflict -8.837
## nofterms_in_eurovoc 8.440
## factor(form)Directive -11.504
## factor(form)Regulation -8.042
## factor(year)2011 0.715
## factor(year)2012 -6.376
## factor(year)2013 -1.194
## factor(year)2014 -14.515
## factor(year)2015 2.409
## factor(year)2016 -3.485
## competence_length -17.831
## factor(Country_num)2 1.189
## factor(Country_num)3 -0.178
## factor(Country_num)4 -0.079
## factor(Country_num)5 -0.005
## factor(Country_num)6 -0.296
## factor(Country_num)7 -0.032
## factor(Country_num)8 0.045
## factor(Country_num)9 0.378
## factor(Country_num)11 -0.444
## factor(Country_num)12 0.206
## factor(Country_num)13 -0.263
## factor(Country_num)16 0.343
## factor(Country_num)17 0.028
## factor(Country_num)18 0.130
## factor(Country_num)19 -0.587
## factor(Country_num)20 0.005
## factor(Country_num)21 0.150
## factor(Country_num)22 -0.192
## factor(Country_num)23 -0.012
## factor(Country_num)24 -0.105
## factor(Country_num)25 0.274
## factor(Country_num)26 -0.047
## factor(Country_num)27 -0.383
## factor(Country_num)28 -0.278
## factor(Country_num)29 -0.023
## factor(Country_num)30 0.184
## factor(Country_num)32 1.981
## lagged2y_pct_support_cntr_cap_year:Lagged_no_trade_n_of_organiz_registered_per_year -2.062
## Log(scale) -211.897
## Pr(>|z|)
## (Intercept) < 2e-16
## lagged2y_pct_support_cntr_cap_year 5.31e-12
## lagged_pct_Salience_DK 2.11e-07
## Lagged_no_trade_n_of_organiz_registered_per_year 0.150349
## lagged_pct_cntr_polariz 0.005927
## EUpolarizarion 2.49e-06
## Interinst_conflict < 2e-16
## nofterms_in_eurovoc < 2e-16
## factor(form)Directive < 2e-16
## factor(form)Regulation 8.84e-16
## factor(year)2011 0.474462
## factor(year)2012 1.82e-10
## factor(year)2013 0.232310
## factor(year)2014 < 2e-16
## factor(year)2015 0.016002
## factor(year)2016 0.000492
## competence_length < 2e-16
## factor(Country_num)2 0.234626
## factor(Country_num)3 0.858349
## factor(Country_num)4 0.936704
## factor(Country_num)5 0.996018
## factor(Country_num)6 0.767143
## factor(Country_num)7 0.974288
## factor(Country_num)8 0.963738
## factor(Country_num)9 0.705510
## factor(Country_num)11 0.657115
## factor(Country_num)12 0.837034
## factor(Country_num)13 0.792779
## factor(Country_num)16 0.731667
## factor(Country_num)17 0.977877
## factor(Country_num)18 0.896310
## factor(Country_num)19 0.557313
## factor(Country_num)20 0.995850
## factor(Country_num)21 0.881026
## factor(Country_num)22 0.847364
## factor(Country_num)23 0.990727
## factor(Country_num)24 0.916253
## factor(Country_num)25 0.784398
## factor(Country_num)26 0.962861
## factor(Country_num)27 0.701499
## factor(Country_num)28 0.781129
## factor(Country_num)29 0.981695
## factor(Country_num)30 0.854179
## factor(Country_num)32 0.047550
## lagged2y_pct_support_cntr_cap_year:Lagged_no_trade_n_of_organiz_registered_per_year 0.039161
## Log(scale) < 2e-16
##
## (Intercept) ***
## lagged2y_pct_support_cntr_cap_year ***
## lagged_pct_Salience_DK ***
## Lagged_no_trade_n_of_organiz_registered_per_year
## lagged_pct_cntr_polariz **
## EUpolarizarion ***
## Interinst_conflict ***
## nofterms_in_eurovoc ***
## factor(form)Directive ***
## factor(form)Regulation ***
## factor(year)2011
## factor(year)2012 ***
## factor(year)2013
## factor(year)2014 ***
## factor(year)2015 *
## factor(year)2016 ***
## competence_length ***
## factor(Country_num)2
## factor(Country_num)3
## factor(Country_num)4
## factor(Country_num)5
## factor(Country_num)6
## factor(Country_num)7
## factor(Country_num)8
## factor(Country_num)9
## factor(Country_num)11
## factor(Country_num)12
## factor(Country_num)13
## factor(Country_num)16
## factor(Country_num)17
## factor(Country_num)18
## factor(Country_num)19
## factor(Country_num)20
## factor(Country_num)21
## factor(Country_num)22
## factor(Country_num)23
## factor(Country_num)24
## factor(Country_num)25
## factor(Country_num)26
## factor(Country_num)27
## factor(Country_num)28
## factor(Country_num)29
## factor(Country_num)30
## factor(Country_num)32 *
## lagged2y_pct_support_cntr_cap_year:Lagged_no_trade_n_of_organiz_registered_per_year *
## Log(scale) ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Scale: 0.2577
##
## Gaussian distribution
## Number of Newton-Raphson Iterations: 3
## Log-likelihood: -803.3 on 46 Df
## Wald-statistic: 1149 on 44 Df, p-value: < 2.22e-16Three Way Interactions
- In this section we explore the effects of three way interactions between Public Support, Salience and Country POlarization
#Control for EU polarization
model_3way1= glm(LR_Proposal_Probability ~ lagged2y_pct_support_cntr_cap_year +
lagged_pct_Salience_DK +
lagged_pct_support_cntr_cap_year*lagged_pct_Salience_DK*lagged_cntr_polarization +
lagged_cntr_polarization +
EUpolarizarion +
Lagged_no_trade_n_of_organiz_registered_per_year +
Interinst_conflict +
nofterms_in_eurovoc +
factor(form) + factor(year) + factor(Country_num) +
competence_length, data = CurbEUenthusiasm_FINAL, family = quasibinomial('logit'))
plot_model(model_3way1, type='int', mdrt.values = 'meansd') # plotting mean +/- 1sd## Data were 'prettified'. Consider using `terms="lagged_pct_Salience_DK [all]"` to get smooth plots.## Data were 'prettified'. Consider using `terms="lagged_pct_support_cntr_cap_year [all]"` to get smooth plots.## Data were 'prettified'. Consider using `terms="lagged_pct_Salience_DK [all]"` to get smooth plots.
## Data were 'prettified'. Consider using `terms="lagged_pct_Salience_DK [all]"` to get smooth plots.## [[1]]
##
## [[2]]
##
## [[3]]
##
## [[4]]
plot_model(model_3way1, type='int')## Data were 'prettified'. Consider using `terms="lagged_pct_Salience_DK [all]"` to get smooth plots.## Data were 'prettified'. Consider using `terms="lagged_pct_support_cntr_cap_year [all]"` to get smooth plots.## Data were 'prettified'. Consider using `terms="lagged_pct_Salience_DK [all]"` to get smooth plots.
## Data were 'prettified'. Consider using `terms="lagged_pct_Salience_DK [all]"` to get smooth plots.## [[1]]
##
## [[2]]
##
## [[3]]
##
## [[4]]
Here we drop the indicator for the EU polarization
#Drop the EU polarization
model_3way2= glm(LR_Proposal_Probability ~ lagged2y_pct_support_cntr_cap_year +
lagged_pct_Salience_DK +
lagged_pct_support_cntr_cap_year*lagged_pct_Salience_DK*lagged_cntr_polarization +
lagged_cntr_polarization +
Lagged_no_trade_n_of_organiz_registered_per_year +
Interinst_conflict +
nofterms_in_eurovoc +
factor(form) + factor(year) + factor(Country_num) +
competence_length, data = CurbEUenthusiasm_FINAL, family = quasibinomial('logit'))
plot_model(model_3way2, type='int', mdrt.values = 'meansd') # plotting mean +/- 1sd## Data were 'prettified'. Consider using `terms="lagged_pct_Salience_DK [all]"` to get smooth plots.## Data were 'prettified'. Consider using `terms="lagged_pct_support_cntr_cap_year [all]"` to get smooth plots.## Data were 'prettified'. Consider using `terms="lagged_pct_Salience_DK [all]"` to get smooth plots.
## Data were 'prettified'. Consider using `terms="lagged_pct_Salience_DK [all]"` to get smooth plots.## [[1]]
##
## [[2]]
##
## [[3]]
##
## [[4]]
plot_model(model_3way2, type='int')## Data were 'prettified'. Consider using `terms="lagged_pct_Salience_DK [all]"` to get smooth plots.## Data were 'prettified'. Consider using `terms="lagged_pct_support_cntr_cap_year [all]"` to get smooth plots.## Data were 'prettified'. Consider using `terms="lagged_pct_Salience_DK [all]"` to get smooth plots.
## Data were 'prettified'. Consider using `terms="lagged_pct_Salience_DK [all]"` to get smooth plots.## [[1]]
##
## [[2]]
##
## [[3]]
##
## [[4]]
stargazer(model_3way1, model_3way2, type='text', omit = ('Country_num'), omit.labels = c("Country FE?"), title='Table 7: Testing 3- way interations', align = TRUE)##
## Table 7: Testing 3- way interations
## =============================================================================================================
## Dependent variable:
## ----------------------------
## LR_Proposal_Probability
## (1) (2)
## -------------------------------------------------------------------------------------------------------------
## lagged2y_pct_support_cntr_cap_year 0.004*** 0.005***
## (0.001) (0.001)
##
## lagged_pct_Salience_DK -0.049*** -0.047***
## (0.010) (0.009)
##
## lagged_pct_support_cntr_cap_year -0.047*** -0.038***
## (0.010) (0.009)
##
## lagged_cntr_polarization -4.201*** -3.667***
## (0.895) (0.821)
##
## EUpolarizarion -0.280***
## (0.054)
##
## Lagged_no_trade_n_of_organiz_registered_per_year -0.022* -0.018
## (0.013) (0.013)
##
## Interinst_conflict -0.180*** -0.150***
## (0.020) (0.019)
##
## nofterms_in_eurovoc 0.044*** 0.049***
## (0.005) (0.005)
##
## factor(form)Directive -0.448*** -0.400***
## (0.039) (0.038)
##
## factor(form)Regulation -0.284*** -0.291***
## (0.036) (0.036)
##
## factor(year)2011 0.017 0.005
## (0.032) (0.031)
##
## factor(year)2012 -0.248*** -0.345***
## (0.039) (0.037)
##
## factor(year)2013 -0.072* -0.049
## (0.042) (0.039)
##
## factor(year)2014 -0.712*** -0.704***
## (0.051) (0.046)
##
## factor(year)2015 0.143*** 0.070
## (0.054) (0.054)
##
## factor(year)2016 -0.103** -0.144***
## (0.041) (0.040)
##
## competence_length -0.015*** -0.013***
## (0.001) (0.001)
##
## lagged_pct_Salience_DK:lagged_pct_support_cntr_cap_year 0.001*** 0.001***
## (0.0001) (0.0001)
##
## lagged_pct_support_cntr_cap_year:lagged_cntr_polarization 0.036*** 0.031***
## (0.008) (0.008)
##
## lagged_pct_Salience_DK:lagged_cntr_polarization 0.048*** 0.044***
## (0.010) (0.010)
##
## lagged_pct_Salience_DK:lagged_pct_support_cntr_cap_year:lagged_cntr_polarization -0.0004*** -0.0003***
## (0.0001) (0.0001)
##
## Constant 4.800*** 3.776***
## (0.917) (0.825)
##
## -------------------------------------------------------------------------------------------------------------
## Country FE? Yes Yes
## -------------------------------------------------------------------------------------------------------------
## Observations 12,251 12,953
## =============================================================================================================
## Note: *p<0.1; **p<0.05; ***p<0.01Alternative operationalizations to Public Opinion ==> Results hold!
It is possible that our results are affected by the way we calculate public support for the EU policy action. To ensure the robustness of the main results, we turn to a different operationalization. In the models that follow, the level of public support for EU action in a policy area is measured 1) as a difference between the Country level of support and the EU average level of support. This approach permits us to examine how the deviations form the mean level of support within a country may shape the probability of the authority expansion the Commission envisions.
## H1: public opinion in support of the EU authority in PA expansion results in more proposed authority expansion
model1_diff <- glm(LR_Proposal_Probability ~ lagged2y_MS_EU_diff +
lagged_pct_Salience_DK +
lagged_pct_cntr_polariz+
Lagged_no_trade_n_of_organiz_registered_per_year +
EUpolarizarion +
Interinst_conflict +
nofterms_in_eurovoc +
factor(form) +
factor(year) + factor(Country_num) +competence_length,
data = CurbEUenthusiasm_FINAL, family = quasibinomial('logit'))
##H1a: Support * Salience
model1a_diff <- glm(LR_Proposal_Probability ~ lagged2y_MS_EU_diff +
lagged_pct_Salience_DK +
lagged2y_MS_EU_diff*lagged_pct_Salience_DK +
lagged_pct_cntr_polariz+
Lagged_no_trade_n_of_organiz_registered_per_year +
EUpolarizarion +
Interinst_conflict +
nofterms_in_eurovoc +
factor(form) +
factor(year) + factor(Country_num) +competence_length,
data = CurbEUenthusiasm_FINAL, family = quasibinomial('logit'))
plot_model(model1a_diff, type='int', mdrt.values = 'meansd') # plotting mean +/- 1sd## Data were 'prettified'. Consider using `terms="lagged2y_MS_EU_diff [all]"` to get smooth plots.
plot_model(model1a_diff, type='int')## Data were 'prettified'. Consider using `terms="lagged2y_MS_EU_diff [all]"` to get smooth plots.
##H1b: Support* Polarization
model1b_diff <- glm(LR_Proposal_Probability ~ lagged2y_MS_EU_diff +
lagged_pct_Salience_DK +
lagged2y_MS_EU_diff*lagged_pct_cntr_polariz +
lagged_pct_cntr_polariz+
Lagged_no_trade_n_of_organiz_registered_per_year +
EUpolarizarion +
Interinst_conflict +
nofterms_in_eurovoc +
factor(form) +
factor(year) + factor(Country_num) +competence_length,
data = CurbEUenthusiasm_FINAL, family = quasibinomial('logit'))
plot_model(model1b_diff, type='int', mdrt.values = 'meansd') # plotting mean +/- 1sd## Data were 'prettified'. Consider using `terms="lagged2y_MS_EU_diff [all]"` to get smooth plots.
plot_model(model1b_diff, type='int')## Data were 'prettified'. Consider using `terms="lagged2y_MS_EU_diff [all]"` to get smooth plots.
##H1c: Support*Actor mobilization
model1c_diff <- glm(LR_Proposal_Probability ~ lagged2y_MS_EU_diff +
lagged_pct_Salience_DK +
lagged2y_MS_EU_diff* Lagged_no_trade_n_of_organiz_registered_per_year +
lagged_pct_cntr_polariz+
Lagged_no_trade_n_of_organiz_registered_per_year +
EUpolarizarion +
Interinst_conflict +
nofterms_in_eurovoc +
factor(form) +
factor(year) + factor(Country_num) +competence_length,
data = CurbEUenthusiasm_FINAL, family = quasibinomial('logit'))
plot_model(model1c_diff, type='int', mdrt.values = 'meansd') # plotting mean +/- 1sd## Data were 'prettified'. Consider using `terms="lagged2y_MS_EU_diff [all]"` to get smooth plots.
plot_model(model1c_diff, type='int')## Data were 'prettified'. Consider using `terms="lagged2y_MS_EU_diff [all]"` to get smooth plots.
stargazer(model1_diff, model1a_diff, model1b_diff, model1c_diff, type='text',
omit=c("Country_num"), omit.labels = c("Country FE?"), title="Table 8: Retesting the modes using a different DV operationalization", align = TRUE)##
## Table 8: Retesting the modes using a different DV operationalization
## ============================================================================================================
## Dependent variable:
## ---------------------------------------
## LR_Proposal_Probability
## (1) (2) (3) (4)
## ------------------------------------------------------------------------------------------------------------
## lagged2y_MS_EU_diff 0.006*** 0.011*** 0.003 0.006***
## (0.001) (0.003) (0.003) (0.001)
##
## lagged_pct_Salience_DK 0.003*** 0.003*** 0.003*** 0.003***
## (0.001) (0.001) (0.001) (0.001)
##
## lagged_pct_cntr_polariz 0.001* 0.001* 0.001* 0.001*
## (0.0005) (0.0005) (0.0005) (0.0005)
##
## Lagged_no_trade_n_of_organiz_registered_per_year -0.024* -0.024* -0.024* -0.027
## (0.013) (0.013) (0.013) (0.017)
##
## EUpolarizarion -0.278*** -0.286*** -0.282*** -0.279***
## (0.048) (0.048) (0.048) (0.048)
##
## Interinst_conflict -0.188*** -0.188*** -0.186*** -0.188***
## (0.020) (0.020) (0.020) (0.020)
##
## nofterms_in_eurovoc 0.043*** 0.043*** 0.043*** 0.043***
## (0.005) (0.005) (0.005) (0.005)
##
## factor(form)Directive -0.444*** -0.447*** -0.446*** -0.445***
## (0.039) (0.039) (0.039) (0.039)
##
## factor(form)Regulation -0.293*** -0.295*** -0.295*** -0.294***
## (0.036) (0.036) (0.036) (0.036)
##
## factor(year)2011 0.078*** 0.077** 0.076** 0.079***
## (0.030) (0.030) (0.030) (0.030)
##
## factor(year)2012 -0.188*** -0.187*** -0.191*** -0.188***
## (0.038) (0.038) (0.038) (0.038)
##
## factor(year)2013 0.023 0.017 0.019 0.023
## (0.039) (0.039) (0.039) (0.039)
##
## factor(year)2014 -0.693*** -0.689*** -0.694*** -0.692***
## (0.050) (0.050) (0.050) (0.050)
##
## factor(year)2015 0.202*** 0.199*** 0.197*** 0.202***
## (0.053) (0.053) (0.054) (0.053)
##
## factor(year)2016 -0.098*** -0.103*** -0.100*** -0.098***
## (0.038) (0.038) (0.038) (0.038)
##
## competence_length -0.015*** -0.015*** -0.015*** -0.015***
## (0.001) (0.001) (0.001) (0.001)
##
## lagged2y_MS_EU_diff:lagged_pct_Salience_DK -0.0001*
## (0.00004)
##
## lagged2y_MS_EU_diff:lagged_pct_cntr_polariz 0.00005
## (0.00004)
##
## lagged2y_MS_EU_diff:Lagged_no_trade_n_of_organiz_registered_per_year 0.0004
## (0.002)
##
## Constant 0.804*** 0.844*** 0.828*** 0.806***
## (0.122) (0.124) (0.124) (0.123)
##
## ------------------------------------------------------------------------------------------------------------
## Country FE? Yes Yes Yes Yes
## ------------------------------------------------------------------------------------------------------------
## Observations 12,251 12,251 12,251 12,251
## ============================================================================================================
## Note: *p<0.1; **p<0.05; ***p<0.01Models with 1 Y lag for Public support and T0 for the rest of politicization (Dropped EU polarization).
In this section we unpack to what extent the lag structure may influence our results. Unlike previous studies ( see De Bryucker), we do not focus on the approved legislation. Instead we examine the proposals and their ambitiousness at the earlier stage of the policy making process. Hence, one could argue that it takes less time for the public preference to be transmitted to the EU agenda-setter, and for politicization to enhance/suppress the motivations of the COmmission to align its legislative activities with the will of the public.
#H1
model_1yLag= glm(LR_Proposal_Probability ~ lagged_pct_support_cntr_cap_year +
lagged_pct_Salience_DK +
EUpolarizarion +
lagged_cntr_polarization +
Lagged_no_trade_n_of_organiz_registered_per_year +
Interinst_conflict +
nofterms_in_eurovoc +
factor(form) + factor(year) + factor(Country_num) +
competence_length, data = CurbEUenthusiasm_FINAL, family = quasibinomial('logit'))
#H1a: 1y Lag Support * 1y Lag Salience
model_1yLag_a= glm(LR_Proposal_Probability ~ lagged_pct_support_cntr_cap_year +
lagged_pct_Salience_DK +
lagged_pct_support_cntr_cap_year *lagged_pct_Salience_DK +
EUpolarizarion +
lagged_cntr_polarization +
Lagged_no_trade_n_of_organiz_registered_per_year +
Interinst_conflict +
nofterms_in_eurovoc +
factor(form) + factor(year) + factor(Country_num) +
competence_length, data = CurbEUenthusiasm_FINAL, family = quasibinomial('logit'))
#H1b: 1y Lag Support* 1 Lag Polariz
model_1yLag_b= glm(LR_Proposal_Probability ~ lagged_pct_support_cntr_cap_year +
lagged_pct_Salience_DK +
lagged_pct_support_cntr_cap_year *lagged_cntr_polarization +
EUpolarizarion +
lagged_cntr_polarization +
Lagged_no_trade_n_of_organiz_registered_per_year +
Interinst_conflict +
nofterms_in_eurovoc +
factor(form) + factor(year) + factor(Country_num) +
competence_length, data = CurbEUenthusiasm_FINAL, family = quasibinomial('logit'))
#H1c 1 Y Lag Support * 1 Y Lag Actors
model_1yLag_c= glm(LR_Proposal_Probability ~ lagged_pct_support_cntr_cap_year +
lagged_pct_Salience_DK +
lagged_pct_support_cntr_cap_year * Lagged_no_trade_n_of_organiz_registered_per_year +
EUpolarizarion +
lagged_cntr_polarization +
Lagged_no_trade_n_of_organiz_registered_per_year +
Interinst_conflict +
nofterms_in_eurovoc +
factor(form) + factor(year) + factor(Country_num) +
competence_length, data = CurbEUenthusiasm_FINAL, family = quasibinomial('logit'))
stargazer( model_1yLag, model_1yLag_a, model_1yLag_b, model_1yLag_c, type='text',
omit=c("Country_num"), omit.labels = c("Country FE?"), title="Table 9: Retesting with 1 year Lag for Dv and for main IVs.", align = TRUE)##
## Table 9: Retesting with 1 year Lag for Dv and for main IVs.
## =========================================================================================================================
## Dependent variable:
## ---------------------------------------
## LR_Proposal_Probability
## (1) (2) (3) (4)
## -------------------------------------------------------------------------------------------------------------------------
## lagged_pct_support_cntr_cap_year 0.001 -0.007*** 0.003 0.001
## (0.002) (0.002) (0.002) (0.002)
##
## lagged_pct_Salience_DK 0.002*** -0.006*** 0.002*** 0.002***
## (0.001) (0.002) (0.001) (0.001)
##
## EUpolarizarion -0.265*** -0.225*** -0.261*** -0.264***
## (0.047) (0.048) (0.047) (0.047)
##
## lagged_cntr_polarization -0.014 0.105 0.193 -0.014
## (0.099) (0.102) (0.227) (0.099)
##
## Lagged_no_trade_n_of_organiz_registered_per_year -0.020 -0.019 -0.020 -0.029
## (0.013) (0.013) (0.013) (0.059)
##
## Interinst_conflict -0.164*** -0.161*** -0.165*** -0.164***
## (0.019) (0.019) (0.019) (0.019)
##
## nofterms_in_eurovoc 0.052*** 0.052*** 0.052*** 0.052***
## (0.005) (0.005) (0.005) (0.005)
##
## factor(form)Directive -0.513*** -0.510*** -0.513*** -0.513***
## (0.038) (0.038) (0.038) (0.038)
##
## factor(form)Regulation -0.395*** -0.387*** -0.397*** -0.395***
## (0.035) (0.035) (0.035) (0.035)
##
## factor(year)2011 0.001 0.007 0.002 0.0005
## (0.029) (0.029) (0.029) (0.029)
##
## factor(year)2012 -0.298*** -0.292*** -0.298*** -0.298***
## (0.037) (0.037) (0.037) (0.037)
##
## factor(year)2013 -0.083** -0.092** -0.080** -0.083**
## (0.038) (0.038) (0.038) (0.038)
##
## factor(year)2014 -0.633*** -0.654*** -0.639*** -0.633***
## (0.047) (0.048) (0.048) (0.047)
##
## factor(year)2015 -0.127*** -0.120*** -0.125*** -0.127***
## (0.044) (0.044) (0.044) (0.044)
##
## factor(year)2016 -0.215*** -0.202*** -0.219*** -0.215***
## (0.036) (0.036) (0.036) (0.036)
##
## competence_length -0.014*** -0.014*** -0.014*** -0.014***
## (0.001) (0.001) (0.001) (0.001)
##
## lagged_pct_support_cntr_cap_year:lagged_pct_Salience_DK 0.0001***
## (0.00003)
##
## lagged_pct_support_cntr_cap_year:lagged_cntr_polarization -0.003
## (0.003)
##
## lagged_pct_support_cntr_cap_year:Lagged_no_trade_n_of_organiz_registered_per_year 0.0001
## (0.001)
##
## Constant 0.900*** 1.264*** 0.755*** 0.900***
## (0.183) (0.198) (0.232) (0.184)
##
## -------------------------------------------------------------------------------------------------------------------------
## Country FE? Yes Yes Yes Yes
## -------------------------------------------------------------------------------------------------------------------------
## Observations 13,104 13,104 13,104 13,104
## =========================================================================================================================
## Note: *p<0.1; **p<0.05; ***p<0.011 year Lagged, No Interaction Model
plot_model(model_1yLag, type='pred', terms= 'lagged_pct_support_cntr_cap_year') # plotting mean +/- 1sd## Data were 'prettified'. Consider using `terms="lagged_pct_support_cntr_cap_year [all]"` to get smooth plots.
1 year Lagged models with interaction of Public Suport X Salience
plot_model(model_1yLag_a, type='int', mdrt.values = 'meansd') # plotting mean +/- 1sd## Data were 'prettified'. Consider using `terms="lagged_pct_support_cntr_cap_year [all]"` to get smooth plots.
plot_model(model_1yLag_a, type='int')## Data were 'prettified'. Consider using `terms="lagged_pct_support_cntr_cap_year [all]"` to get smooth plots.
1 year Lagged Models wih interaction of Public Support X Polarization (H1b)
plot_model(model_1yLag_b, type='int', mdrt.values = 'meansd') # plotting mean +/- 1sd## Data were 'prettified'. Consider using `terms="lagged_pct_support_cntr_cap_year [all]"` to get smooth plots.
plot_model(model_1yLag_b, type='int')## Data were 'prettified'. Consider using `terms="lagged_pct_support_cntr_cap_year [all]"` to get smooth plots.
1 year Lagged Models wih interaction of Public Support X Actor Mobilization (H1b)
plot_model(model_1yLag_c, type='int', mdrt.values = 'meansd') # plotting mean +/- 1sd## Data were 'prettified'. Consider using `terms="lagged_pct_support_cntr_cap_year [all]"` to get smooth plots.
plot_model(model_1yLag_c, type='int')## Data were 'prettified'. Consider using `terms="lagged_pct_support_cntr_cap_year [all]"` to get smooth plots.
Controllling for Package deals ==> Main results do not change!
To rule out that our results are driven by the specific of the legislative act, we introduce additional variable into the model which allows us to control for the package deal. The new indicator is a dummy variable which takes a value of 1 if the legislative act is a pack of a package legislation. The introduction of this variable doe snot affect our key results. ( here model set up is the same as in the main section)
model1_diff <- glm(LR_Proposal_Probability ~ lagged2y_pct_support_cntr_cap_year +
lagged_pct_Salience_DK +
lagged_pct_cntr_polariz+
Lagged_no_trade_n_of_organiz_registered_per_year +
EUpolarizarion +
Interinst_conflict +
nofterms_in_eurovoc +
factor(form) + package_deals+
factor(year) + factor(Country_num) +competence_length,
data = CurbEUenthusiasm_FINAL, family = quasibinomial('logit'))
##H1a: Support * Salience
model1a_diff <- glm(LR_Proposal_Probability ~ lagged2y_pct_support_cntr_cap_year +
lagged_pct_Salience_DK +
lagged2y_pct_support_cntr_cap_year*lagged_pct_Salience_DK +
lagged_pct_cntr_polariz+
Lagged_no_trade_n_of_organiz_registered_per_year +
EUpolarizarion +
Interinst_conflict +
nofterms_in_eurovoc +
factor(form) +package_deals+
factor(year) + factor(Country_num) +competence_length,
data = CurbEUenthusiasm_FINAL, family = quasibinomial('logit'))
plot_model(model1a_diff, type='int', mdrt.values = 'meansd') # plotting mean +/- 1sd## Data were 'prettified'. Consider using `terms="lagged2y_pct_support_cntr_cap_year [all]"` to get smooth plots.
plot_model(model1a_diff, type='int')## Data were 'prettified'. Consider using `terms="lagged2y_pct_support_cntr_cap_year [all]"` to get smooth plots.
##H1b: Support* Polarization
model1b_diff <- glm(LR_Proposal_Probability ~ lagged2y_pct_support_cntr_cap_year +
lagged_pct_Salience_DK +
lagged2y_pct_support_cntr_cap_year*lagged_pct_cntr_polariz +
lagged_pct_cntr_polariz+
Lagged_no_trade_n_of_organiz_registered_per_year +
EUpolarizarion +
Interinst_conflict +
nofterms_in_eurovoc +
factor(form) +package_deals+
factor(year) + factor(Country_num) +competence_length,
data = CurbEUenthusiasm_FINAL, family = quasibinomial('logit'))
plot_model(model1b_diff, type='int', mdrt.values = 'meansd') # plotting mean +/- 1sd## Data were 'prettified'. Consider using `terms="lagged2y_pct_support_cntr_cap_year [all]"` to get smooth plots.
plot_model(model1b_diff, type='int')## Data were 'prettified'. Consider using `terms="lagged2y_pct_support_cntr_cap_year [all]"` to get smooth plots.
##H1c: Support*Actor mobilization
model1c_diff <- glm(LR_Proposal_Probability ~ lagged2y_pct_support_cntr_cap_year +
lagged_pct_Salience_DK +
lagged2y_pct_support_cntr_cap_year* Lagged_no_trade_n_of_organiz_registered_per_year + lagged_pct_cntr_polariz+
Lagged_no_trade_n_of_organiz_registered_per_year +
EUpolarizarion +
Interinst_conflict +
nofterms_in_eurovoc +
factor(form) +package_deals+
factor(year) + factor(Country_num) +competence_length,
data = CurbEUenthusiasm_FINAL, family = quasibinomial('logit'))
plot_model(model1c_diff, type='int', mdrt.values = 'meansd') # plotting mean +/- 1sd## Data were 'prettified'. Consider using `terms="lagged2y_pct_support_cntr_cap_year [all]"` to get smooth plots.
plot_model(model1c_diff, type='int')## Data were 'prettified'. Consider using `terms="lagged2y_pct_support_cntr_cap_year [all]"` to get smooth plots.
stargazer(model1_diff, model1a_diff, model1b_diff, model1c_diff, type='text',
omit=c("Country_num"), omit.labels = c("Country FE?"), title="Table 10: Controlling for `Package Deals`", align = TRUE)##
## Table 10: Controlling for `Package Deals`
## ===========================================================================================================================
## Dependent variable:
## ---------------------------------------
## LR_Proposal_Probability
## (1) (2) (3) (4)
## ---------------------------------------------------------------------------------------------------------------------------
## lagged2y_pct_support_cntr_cap_year 0.006*** 0.010*** 0.010*** 0.006***
## (0.001) (0.003) (0.002) (0.001)
##
## lagged_pct_Salience_DK 0.004*** 0.008*** 0.004*** 0.004***
## (0.001) (0.002) (0.001) (0.001)
##
## lagged_pct_cntr_polariz 0.001*** 0.001*** 0.006*** 0.001***
## (0.001) (0.001) (0.002) (0.001)
##
## Lagged_no_trade_n_of_organiz_registered_per_year -0.024* -0.024* -0.023* 0.063
## (0.013) (0.013) (0.013) (0.045)
##
## EUpolarizarion -0.204*** -0.213*** -0.229*** -0.203***
## (0.047) (0.048) (0.049) (0.047)
##
## Interinst_conflict -0.186*** -0.187*** -0.184*** -0.185***
## (0.020) (0.020) (0.020) (0.020)
##
## nofterms_in_eurovoc 0.045*** 0.045*** 0.045*** 0.045***
## (0.005) (0.005) (0.005) (0.005)
##
## factor(form)Directive -0.468*** -0.470*** -0.466*** -0.467***
## (0.039) (0.039) (0.039) (0.039)
##
## factor(form)Regulation -0.295*** -0.297*** -0.295*** -0.294***
## (0.036) (0.036) (0.036) (0.036)
##
## package_deals 0.129*** 0.130*** 0.130*** 0.129***
## (0.025) (0.025) (0.025) (0.025)
##
## factor(year)2011 -0.015 -0.010 -0.015 -0.015
## (0.032) (0.033) (0.032) (0.032)
##
## factor(year)2012 -0.256*** -0.248*** -0.253*** -0.251***
## (0.039) (0.039) (0.039) (0.039)
##
## factor(year)2013 -0.055 -0.070* -0.043 -0.054
## (0.040) (0.041) (0.041) (0.040)
##
## factor(year)2014 -0.702*** -0.694*** -0.696*** -0.704***
## (0.050) (0.050) (0.050) (0.050)
##
## factor(year)2015 0.092* 0.104* 0.099* 0.092*
## (0.054) (0.055) (0.054) (0.054)
##
## factor(year)2016 -0.149*** -0.147*** -0.145*** -0.148***
## (0.038) (0.039) (0.039) (0.038)
##
## competence_length -0.015*** -0.015*** -0.015*** -0.015***
## (0.001) (0.001) (0.001) (0.001)
##
## lagged2y_pct_support_cntr_cap_year:lagged_pct_Salience_DK -0.0001**
## (0.00003)
##
## lagged2y_pct_support_cntr_cap_year:lagged_pct_cntr_polariz -0.0001**
## (0.00003)
##
## lagged2y_pct_support_cntr_cap_year:Lagged_no_trade_n_of_organiz_registered_per_year -0.001**
## (0.001)
##
## Constant 0.320** 0.028 0.051 0.302**
## (0.142) (0.201) (0.182) (0.142)
##
## ---------------------------------------------------------------------------------------------------------------------------
## Country FE? Yes Yes Yes Yes
## ---------------------------------------------------------------------------------------------------------------------------
## Observations 12,251 12,251 12,251 12,251
## ===========================================================================================================================
## Note: *p<0.1; **p<0.05; ***p<0.01Conclusion
The level of public support for EU action consistency increases the probability of EU authority expansion across models; ==> H1 is supported
Salience: the average effect appears to reduce the effect of public support (see model 2 in Table 3.). HOwever, Marginal effects suggest that higher levels of salience coupled with the increasing public support do increase the probability of authority expansion. However, the magnitude of the effect s more moderate than for the policies of low salience but highly supported by the population. Here one should be cautious concluding the interaction effects is definitively constraining on the prospect of authority expansion, as highly salient domain appear to have higher chance of having the EU involved in contrast to non-salient issues even at the low levels of support. Hence, a less steep slope of the effect is hardly surprising as the extent to which a policy can broaden the EU powers and successfully pass through the policy making process is constrained. ==> Hence some support for the H1a.
Country Polarization: Whilst polarization of the Countries’ societies appear to have a positive effect on the probability of EU authority expansion, the interaction effect requires more attention and more careful interpretation. The Regression analysis suggests a negative association between polarization on the country level and the level of public support for the EU action in a policy area. However, upon a more careful examination, it appears that at higher level of polarization, the probability of EU authority expansion increase as the support within a policy area grows, yet at a more modest rate in contrast to the environment characterized by low polarization . In other words, the effect of polarization suggests a positive trend yer the magnitude to which polarization fosters the effect of the increasing public support differs by the degree of divisions the society faces. ==> H1b some support for positive relationship which gets weaker
Actor Mobilization: The results suggest that the public support tends to increase the chance of more authority expanding proposal being put forward if there is no or very low actor mobilization takes place. Under condition of low actor mobilization, the Commission, can adjust the preferences to the demands of the population without being torn between competing demands of the various interests groups. However, it is worth noting that the the probability of authority expansion seems to be high when actors are highly mobilized and the public support for the policy action is low. These contrasting effects of the public opinion and actors mobilization cast a shadow of doubt onto the perspective that the interests groups could be seen as mediators of the public preferences to the EU institutions. In fact it appears when interest groups and organizations mobilize, and the public supports more EU action within a policy area, the competing pressures on the EU policy makers rather curb the policy making enthusiasm of the EU COmmission. ==>> H1c==> supported