Sympathy thermometers
Posted 3/6/2008 14:13 (#37)
Subject: Sympathy thermometers

New user

Posts: 1

Turnout will be a major subject in the 2009 EP election, because it will differ greatly between countries (a high of 80% plus and a low of less than 40% is likely) and because within the great majority of countries there are usually substantial differences, with the EP turnout being as low as half that of the most recent general election and EP election. The analysis of turnout will not only be of concern to political scientists (Franklin, 2004) but also be of major public policy interest to candidates for the European Parliament, EU policymakers, national and EU parties and the media. Given the results of recent referendums on EU constitutional issues, interpretations of turnout in 2009 will be of particular political relevance to the debate on popular (dis)satisfaction with the EU.

The existing literature on why people do not vote, which the Voter Survey should monitor as in earlier elections, needs to be complemented by the analysis of what difference, if any, it makes to representation in the European Parliament at the party and the national level. This requires the provision of appropriate data and models of analysis.

The existing Voter Survey has a number of relevant questions for analyzing non-voters. It could be further strengthened by adding ‘sympathy’ or ‘feeling thermometers’, which ask for all major parties on the ballot (party leaders could also be considered). The thermometer items make it possible to determine both the strength of sympathy or antipathy toward a party but also the extent to which individuals hold an exclusive sympathy or have similar attachments to more than one party. The value of the scale for analyzing both non-voters and voters is demonstrated in Pettersen and Rose (2007).

A particular advantage of the question is that it is asked independently of questions about vote choice or turnout. This serves to minimise bias when the item is used in estimations of counterfactual voter behaviour. To determine partisan differences between voters and non-voters requires imputing party preferences to non-voters. Bernhagen and Marsh (2007) have demonstrated how this can be done with data of the Voter Survey type by applying the method of multiple imputation of missing data developed by Gary King et al. (2001). We propose to undertake this type of analysis again once the 2009 EES survey data is available.

To determine the impact of non-voting on representation in the European Parliament requires converting preferences of the electorate (that is, voters + non-voters) into seats. This is a complex process varying with the electoral system of each country (more or less proportional), the number of seats it has in the EP (cf. district magnitude), the number of parties contesting an election and the ‘leverage’ of non-voters (that is, what % of ballots they would add to the total count if they were to vote). A model that takes all of these considerations into account has been developed by Kohler and Rose (2008) and applied to national elections across Europe.

The additional electoral system data is available in EJPR Yearbooks and, for 10 new EU member states in Rose and Munro (2009). A memorandum on this dimension of contextual data has been submitted to the Context strand of the study. The same data base can be used to address the scale of national over and under-representation, which is especially large among small countries, e.g. one MEP per 81,000 persons in Malta and Luxembourg; 140,000 persons in Cyprus; 205,000 in Estonia; 280,000 in Slovenia and about 370,000 in Latvia and Lithuania. It also is substantial between medium size EU members, for example, 474,000 persons per MEP in Belgium and 660,000 in the Netherlands. (For comparative data based on American states, see Ladewig and Jasinski, 2008).

We believe that the above recommendations meet the EES criteria of integrating survey with contextual data, taking advantage of the extraordinary character of the 2009 EP election and offering a ‘big bang for the buck’ by requiring only fine tuning of the Voter Survey and efficient assemblage of the Contextual Data base.

Patrick Bernhagen (Department of Politics & International Relations, University of Aberdeen)
Richard Rose (Centre for the Study of Public Policy, University of Aberdeen)

APPENDIX: Example 'feeling thermometer' question from the 2002 Irish national Election Study:

I’d now like to ask you how you feel about some Irish politicians, using what we call the “feeling thermometer”. The feeling thermometer works like this: [Int. Show Card C1]

If you have a favourable feeling (a warm feeling) towards a POLITICIAN you should place him/her somewhere between 50 and 100 degrees;

If you have an unfavourable feeling (a cold feeling) towards a POLITICIAN, you should place him/her somewhere between 0 and 50 degrees; and

If you don’t feel particularly warm or cold (have no feeling towards the politician at all) then you should place him/her at 50 degrees

Where would you place these Irish politicians?


Bernhagen, P. and Marsh, M., 2007. ‘The Partisan Effects of Low Turnout’, Electoral Studies, 26, 548-560.

Franklin, M., 2004. Voter Turnout and the Dynamics of Electoral Competition in Established Democracies since 1945. New York: Cambridge UP.

King, G., et al., 2001. ‘Analyzing Incomplete Political Science Data’, American Political Science Review, 95, 49-69.

Kohler, Uli, and Rose, R., 2008. ‘Under What Conditions Can Non-Voters Affect an Election Outcome?’ (under review).

Ladewig, J. W. and Jasinski, M.P., 2008. ‘On the Causes and Consequences of and Remedies for Interstate Malapportionment of the U.S. House of Representatives’, Perspectives on Politics, 6, 89-107.

Pettersen, P.A. and Rose, L.E., 2007. ‘The Dog That Didn't Bark’, Electoral Studies, 26, 574-588.

Rose, R. and Munro, N., 2009. Elections and Parties in New European Democracies. Colchester: ECPR Press, 2nd edition of a book initially published by CQ Press, Washington DC, 2003.
Top of the page Bottom of the page
Posted 15/6/2008 21:31 (#75 - in reply to #37)
Subject: RE: Sympathy thermometers


Posts: 11

Because they require a question for each party, a battery of this kind is very expensive for the EES. We already have a question for each party regarding propensity to vote for it, position in left-right terms and position on European unification. We might well want to add another battery on position in GAL-TAN terms. The question of whether a feeling thermomiter adds significantly to the vote propensity variables in terms of the objectives mentioned in this post needs to be carefully considered. Arguably, knowing how likely it is that each respondent would 'ever' vote for each party already provides a better means of estimating the likely vote of non-voters, but the matter should certainly be discussed.
Top of the page Bottom of the page
Posted 16/7/2008 16:26 (#95 - in reply to #37)
Subject: RE: Sympathy thermometers

Posts: 26
Location: University of Oxford, UK
Dear Dr Bernhagen and Professor Rose,

Thank you for submitting this proposal to the Open Forum. The PIREDEU Steering Committee met at the end of June to evaluate each of the proposals. We assessed them on the basis of whether they met the following criteria:

* An explicit argument about why the proposed question/coding category merited inclusion in one or more of the PIREDEU data components.

* An explicit argument about the conceptual and theoretical foundations of the question/coding category.

* An explicit case for how the question/coding category facilitates integration and linking of several data components. The PIREDEU Steering Committee preferred proposals that allowed for conceptual integration across the five data components (i.e. voter survey, candidate survey, media study, manifestos and contextual data).

* An explicit consideration of how the proposed question/coding category linked with questions/coding categories in past data collection efforts.

The PIREDEU Steering Committee favoured proposals that ensured over time and across instrument comparability. Moreover, given that the voter and candidate surveys can only contain a limited number of question items, priority was given to proposals with succinct question formats.

On this basis we ranked each question in the proposal as follows:

(1) The proposed item will be included in data collection instrument
(2) High priority proposal that will be included if space and time constraints permit
(3) Proposal can only be included if additional funding is secured
(4) Proposed item is not a priority

The item(s) from your proposal received the following ranking(s):

Instrument: Voter Survey

I'd now like to ask you how you feel about some [COUNTRY] politicians, using what we call the “feeling thermometer”. Where would you place these [COUNTRY] politicians?
Ranking: 3

Rationale: The steering committee decided to add a series of candidate/leadership evaluation questions, but these will not necessarily include feeling themometers, as these are very lengthy. Moreover, the Propensity to Vote (PTV) questions which will be included will capture respondents’ feelings toward support for the party.

Thank you again for your participation in this process. We hope that you will continue to use the Open Forum to comment on the questionnaires/codebook that will be posted online on the Forum in the autumn.

Yours sincerely,

Mark Franklin
Chair of the PIREDEU Steering Committee

Sara Hobolt
Deputy Chair of the PIREDEU Steering Committee
Top of the page Bottom of the page