Web appendix (Weathering the Crisis: Evidence of Diffuse Support for the EU from a Six-Wave Dutch Panel) This is the accepted version of the appendix to the following article article. Please, cite the article as Ringlerova Zuzana. Weathering the crisis: Evidence of diffuse support for the EU from a six-wave Dutch panel. European Union Politics. 2015;16(4):558-576. doi: 10.1177/1465116515588964 For direct quotes, please, use the version published by the publisher. Discussion of the Dutch and the EU-wide trend in support for the European Parliament There may be a concern, however, that the Dutch case significantly differs from the EU as a whole and that conclusions drawn from this analysis have only a limited application to the entire European Union. To address this concern, I use data from the Eurobarometer surveys to compare support for the European Parliament in the Netherlands and support for the European Parliament in the EU as a whole (see Figure A in the appendix). There is somewhat higher support for the European Parliament in the Netherlands than in the EU as a whole. The over-time trend in support for the European Parliament in the Netherlands, however, is parallel to the trend in the European Union as a whole. This comparison therefore suggests that the over-time dynamics in support for the European Parliament in the Netherlands are not radically different from dynamics of support in the European Union as a whole. Q_ c cc CD Q. O LU O g c CD o CD Q_ 100 ■ 90 ■ 80 ■ 70 ■ 60 ■ Date of the Eurobarometer survey The Netherlands -----The EU Figure A. Percentage of respondents who tend to trust the European Parliament. (Source: All available Eurobarometer surveys between October 2007 and October 2012.: Available at http://ec.europa.eu/public_opinion/cf/stepl.cfm. Accessed 7/7 2013.1 use all available Eurobarometer surveys between October 2007 and October 2012.1 used the following Eurobarometer survey question: And, for each of them, please tell me if you tend to trust it or tend not to trust it? The European Parliament'. I compared percentages of respondents who indicated that they tend to trust the European Parliament.) Measurement of support for the Dutch government and support for democracy Support for the Dutch government and support for democracy are measured within the same battery of questions as support for the European Parliament. Respondents are asked the following question: 'Can you indicate, on a scale from 0 to 10, how much confidence you personally have in each of the following institutions? Dutch government/Democracy'. The variable is measured on a scale from 0 to 10. For the purposes of this analysis, the variable was rescaled to a scale from 0 to 100 (with low values indicating low confidence and high values indicating high confidence). 100- 90- 80- 1 (mean) 70-60- tical suppor 50-40- Polil 30-20-10-o- n-1-1-1-1-1- 2007 2008 2009 2010 2011 2012 Year - Confidence: Eur. Parliament (respondents participating in all six waves) — — — — ■ Confidence: Dutch government (respondents participating in all six waves) -----Confidence: Democracy (respondents participating in all six waves) Figure B. Mean values of political support over time. iooH 90- 80 -\ 70- 60- 50- 40- 30- 20- 10- o- -I-1- 2 3 Years since the first wave Confidence: Eur. Parliament (respondents participating in all six waves) -----Confidence: Dutch government (respondents participating in all six waves) Confidence: Democracy (respondents participating in all six waves) Figure C. Percentage of respondents who give the same response as in the first wave of the panel. 100- -i-1-1 I r 1 2 3 4 5 Years since the first wave - Confidence: Eur. Parliament (respondents participating in all six waves) -----Confidence: Dutch government (respondents participating in all six waves) Confidence: Democracy (respondents participating in all six waves) Figure D. Percentage changing by 20 points or less compared to the first wave of the Discussion of the assumptions of the dynamic panel data model The Arellano-Bond estimator assumes that there is no serial correlation in errors. A violation of this assumption would be indicated by a significant test statistic in the second-order autocorrelation test (this test statistic is reported at the bottom of Table 2). In the model of support for the European Parliament as well as in the model of support for democracy, the test statistic is sufficiently small (and therefore not statistically significant). The models of confidence in the European Parliament therefore satisfy this assumption. The test statistic for the model of support for the Dutch government is statistically significant, indicating violation of this assumption. However, the test statistic depends on sample size and it easily becomes statistically significant when the sample size is large (as in the present case). I drew a 50% random sample from the data and replicated the dynamic panel data model for support for the Dutch government. See Table A for the results. The substantive results did not change and the model satisfied the assumption of no autocorrelation in errors. Table A. Dynamic panel data model. National Government Political support (lagged) 0.159 (0.056)** 2009 -3.299 (0.928)** 2010 0.720 (1.145) 2011 -4.303 (1.382)** 2012 -8.335 (1.538)** Autocorrelation test: lst-Order -9.95** Autocorrelation test: 2nd-Order 1.19 N 1,846 Note: Dependent variable: Political support. Data: 50% random draw from the full LISS dataset. Coefficients from a one-step Arrellano-Bond (1991) model (estimated using Stata 12 xtabond command). Standard errors in parentheses. Statistical significance levels: * p<0.05; ** p<0.01. Values for autocorrelation test represent the autocorrelation test statistic.