On Jan 29, Germany's CDU/CSU broke a longstanding taboo #Brandmauer by accepting votes from the far-right AfD. How did citizens respond? A preliminary analysis of an ongoing survey suggests that the conservatives might have actually benefited from their move! 1/5

The change in voting propensities stands in contrast to heavy scrutiny of the CDU/CSU and widespread protests following the vote. However, the result holds across various specifications and for similar outcomes, and appears to go hand-in-hand with losses for the AfD. 2/5

@anselmhager To be clear the dots just represent binned averages (~3h), we have over 1,000 respondents within 24h of the vote. This is the estimate of the effect. As mentioned, this holds for related outcomes, suggesting a consistent pattern.

@beckerbastian @anselmhager Thanks for sharing! Could you explain a little more about the data and the graph. How many respondents per point estimate? What does vote propensity mean on a scale of 1-10?

@beckerbastian @anselmhager Interessant! Gibt es hier einen klar definierten Treatment-Zeitpunkt oder eher einen Zeitraum (Ankündigung des Vorgehens bis Abstimmung im BT)? Mich wundert der negative Trend kurz vor dem hier gewählten Threshold. Wie stark treibt der die Ergebnisse?

@beckerbastian @anselmhager At what level of stat. significance?

@beckerbastian @anselmhager Those lines look overfitted though? The jump essentially relies a couple of low points right before the cutoff? I would suggest fitting less flexible curves.

@beckerbastian @richten_nach @anselmhager gibts da genauere zahlen ob das real ist? ist ja offenbar weniger als die amplitude 1./3./4. februar, also rein optisch sieht das nach 0 effekt aus