![Sample our Politics & International Relations journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/5947/TOC-Subject-Sample-2021-Politics-International-Relations.jpg"})
ABSTRACT
Drawing on an original survey of voters and parties, this article examines the policy space in Irish politics in the context of the 2016 general election. Exploratory and confirmatory factor analyses show that four broad ideological dimensions structure voters’ policy views across a range of salient issues. These are an economic dimension, a cultural dimension, a religious dimension and an austerity dimension. Comparing the location of voters and parties on these dimensions, gaps in the policy space are identified where voters are not represented by any party. Most noticeably, a significant segment of the electorate is found to have left-wing views on economic issues but conservative/authoritarian views on the cultural dimension, and this combination is currently not offered by any of the existing political parties. The article also highlights areas where political parties are out of step with the views of their own voters.
Acknowledgements
The author would like to thank Achillefs Papageorgiou, Maura Adshead, John Costello, David Farrell, Eoin O’Malley and Gary Murphy for their contribution to the data collection and survey design for this study.
Disclosure statement
No potential conflict of interest was reported by the author.
Notes
* The data used in this article can be accessed at the following links: https://dataverse.harvard.edu/dataset.xhtml?persistentId=doi:10.7910/DVN/FYRHS9 and https://dataverse.harvard.edu/dataset.xhtml?persistentId=doi:10.7910/DVN/CB2PZA
1. In exit polls at the 2007, 2011 and 2016 elections, the percentage of respondents who said that policy was the most important factor influencing their decision was, respectively: 25 per cent, 41 per cent and 33 per cent.
2. This question format departs from the more usual Likert format, where respondents are asked to indicate the extent to which they agree or disagree with a statement. The Likert format was used in trials of the Which Candidate survey, but it proved to be unsuitable when it came to eliciting public responses from political parties and candidates. There was a strong tendency for candidates to agree with statements that might sound appealing to certain sections of the electorate, even when this implied contradicting a previous answer.
3. There are two exceptions to this. An item on abortion had four answer categories, but for the purpose of the analysis in this paper these categories are collapsed into three. In addition, an item on budget priorities had three unordered categories; for the purpose of this paper this is recoded into a dichotomous measure (see Appendix).
4. As discussed below, in some instances (20 out of a total of 136 cases) parties did not provide an official response to a particular question, and the party position is instead coded based on the written explanation provided by the party. These coded positions do not always match the position of the median candidate from the party (there are five cases where a difference is found). However, the written explanation of the party position is a more reliable indicator of the party position than the median candidate position due to the low number of candidate responses on some of these questions.
5. Voters were free to go back and change their answers to the policy questions; however, only their initial answers are recorded in the data set examined here.
6. There were 22 items in total in the survey. One item, on the referendum on the 8th Amendment to the Constitution, is excluded because it is logically related to another item on abortion (if someone favours liberalising abortion, this implies they also favour repealing the 8th Amendment). Some other items are also closely related to one another, but are logically and conceptually distinct. There is a danger when using items in factor analysis that are very highly correlated with one another that a factor will emerge simply because of the similarity of these items. For this reason, the exploratory factor analysis was repeated while excluding such items, but this did not affect the results.
7. Not reported; available from the author on request.
8. According to Brown (Citation2006), a model should have a CFI value over .90 (acceptable fit) or .95 (good fit); RMSEA below .08 (acceptable fit) or .06 (good fit); and SRMR below .08.
9. Given how the items are worded, the orientation of these dimensions are: right–left (economic dimension); authoritarian–libertarian (cultural dimension); secular–religious (religious dimension) and pro-/anti-austerity (austerity dimension).
10. While the party positions were not included in the factor analysis, the positions of the parties on these four dimensions is calculated in the same way as the positions of the voters – that is, taking the weighted average of the party's positions across the various items that make up the dimension in question.
11. Cronbach's alpha measures reliability or internal consistency of a composite scale, and is a function of the number of items included in the scale and the correlations between them. The alpha scores for the four dimensions identified here range from 0.70 to 0.79. Given the relatively small number of items used for each dimension, these alpha scores do not imply that the scales should be rejected on reliability grounds.
12. The analysis below was replicated with missing values imputed by multiple imputation; this did not change the substantive findings.
13. The corresponding figures for the other parties are: AAA/PBP 40 per cent; Fianna Fáil 36 per cent; Fine Gael 32 per cent; Renua 25 per cent; Social Democrats 25 per cent; Labour 17 per cent and Greens 10 per cent.
14. As with the data on party positions, missing data were coded based on the written comments provided by the candidates. However, unlike the data on party positions, this was not possible to do in all cases.
15. To test this, respondents who selected each of the answer categories on these questions were compared in terms of their left–right self-placement. For each item, the average left–right self-placement of respondents who selected the middle option was between the average left–right self-placements of respondents who selected the two extreme options, as we should expect.
16. Many of the policy questions address similar issues but the answer categories were ordered in opposing directions, so respondents should not be expected to answer the same way on all items.
17. The 8th Amendment introduced a constitutional ban on abortion by acknowledging the right to life of the unborn (with due regard to the equal right to life of the mother).
18. As the Which Candidate survey was collected before the election and was designed to provide information to voters, many respondents were undecided (8,950 respondents from the matched sample gave a preference, while 11,234 did not).
19. A party identification variable was included in the matching procedure, but the vast majority of respondents (69 per cent in INES and 77 per cent in Which Candidate) did not consider themselves to be close to any political party.