Time to Discuss: Data to Crunch or Problems to Solve? A Rejoinder to Robert Thomson

Abstract

Testing the insights from qualitative case studies (Falkner et al. 2005) in quantitative models is a challenging task (see Thomson 2007). This article argues that in political science, our evaluation standards for quantitative analyses need to go far beyond the sophistication of calculations. They should include the quality of assumptions imputed, of data used, and of reasoning as to the political – as opposed to statistical – significance of regressions. When presenting statistical findings that counter solid qualitative work, scholars therefore are under an obligation also to present reasoned arguments for their case, taking into full account the qualitative findings on processes and mechanisms unveiled in these empirical studies.

Acknowledgements

This and my related work relies on intense cooperation with Oliver Treib, Miriam Hartlapp, and Simone Leiber who have not only co-authored Complying with Europe but have also published single-authored books in which the individual cases and qualitative findings of our project are discussed in even more depth for specific countries in each book (Hartlapp 2004; Leiber 2004; Treib 2004).

Notes

1. How much of ‘Social Europe’ actually trickles down into the member states? Are the intricate EU-level negotiations worthwhile at all? What are the problems during the implementation process? What can be done to improve compliance?

2. We certainly would have profited from even more ‘person power’ and time considering the workload we took upon ourselves.

3. Jointly with the first draft of another, co-authored paper by Robert Thomson (Thomson et al. 2005), this article was first presented at a workshop during the ECPR Joint Sessions in Granada.

4. We also observed how even under the condition of high misfit, non-compliance could be avoided in the ‘world of law observance’, as actors had learned from previous transposition processes (Falkner et al. 2005). An exclusively misfit-oriented perspective is not able to grasp this kind of adjustment of national actors to new circumstances (as, e.g., a new interpretation of implementation requirements by the European Commission or the ECJ). This is only possible if time sequences are taken into account.

5. We actually were extremely careful in drafting our final results: ‘In fact, the largest part of the observed transposition performance of member states may be explained by other factors. We can thus conclude that if there is any direct causal impact of the degree of misfit on member state compliance, the effect is undoubtedly much weaker than many scholars would expect, and the direction of this effect is sometimes even inverse to what has been suggested in the literature’ (Falkner et al. 2005: 291).

6. ‘Regional subsidiarity, by contrast, does not play any significant role in EU labour law since the competences are at the national (not regional) level in all member states. Only in a few member states are the regions at all involved in the transposition of EU labour law directives. Merely where there are public servants at the regional level, with specific regional labour law provisions, do the regions act as additional actors to the central governments, but only for their own public employees’ (Falkner et al. 2005: 9, n.5).

7. In the case of the working time directive, ‘some of the Austrian federal states had not yet issued legislation guaranteeing the 48-hour limit for their civil servants’ by the end of our study (Falkner et al. 2005: 111).

8. Indeed, the Complying with Europe team assumes that for fully correct transposition, where the last detail even on the lowest state level would have to be counted, decentralised law-making would account for more problems than in our study. Additionally, we found evidence that federalism matters for the stage of application and enforcement at the national level. However, Thomson consciously decided to calculate using our figures, and moreover focused on essential completeness as opposed to fully correct transposition.

9. Very hypothetically, Thomson is right, it could have been the case that ‘the form of involvement chosen depended partly on government policy-makers’ expectations regarding compliance problems' (p. 997). However, from our case studies we did not learn that governments chose to consult social partners or to give them co-decision rights because they wanted to achieve more or fewer months of delay in essentially correct transposition (this information forms Thomson's dependent variable).

10. Thomson et al. 2005. This text shares many problematic aspects with the text discussed here. At the same time, it is based on less sound data for measuring transposition timeliness than those used in the piece commented here, that is on CELEX data on the first notification for 24 directives, which is at best a measure for the delay in the domestic start of a transposition process, but certainly not for the timeliness of ‘transposition’ or of ‘compliance’. See Miriam Hartlapp's (2004) dissertation on why much official data, particularly Commission reports, are no valid measure for transposition and, even more so, for ‘compliance’ with EU directives. See also our forthcoming joint paper (Falkner and Hartlapp 2007).

11. He uses a ‘discretion ratio … multiplied by 100 to ease interpretation. This discretion ratio is the number of major provisions of a directive that grant member states discretionary powers divided by the total number of major provisions in the directive. A provision grants discretionary powers if it allows states to decide among two or more actions when transposing’ (Thomson 2007: 995).

12. Particularly in the field of EU compliance studies, the sheer correlations found by different authors are certainly not conclusive. To give just one example, some statistical studies ‘conclude that support for European integration is an important factor that facilitates compliance … , others do not … , while still others find a statistically significant negative correlation between these two variables …’ (Treib 2006).

13. Note that communicating the major insights from a problem-oriented qualitative perspective is an exercise similar to nailing a pudding to the wall so long as all attention qualitative studies receive from modellers is that the latter re-use the sheer figures without the relevant reasoning.