Constituency orientation in Irish politics: video statements of the candidates in the Irish general election 2016

ABSTRACT The constituency orientation of Irish politicians is a recurring topic in Irish political science. Its analysis has predominantly focused on TDs. This article uses a content analysis of candidate video statements in the 2016 general election in order to assess the strength of constituency orientation among Irish politicians. The results show a distinct geographical pattern, with a band of strong constituency orientation along the West coast extending into a corridor that reaches far into the East of the country. Differences between urban and rural areas and between centre and periphery shape this pattern, indicated by the weak constituency orientation in Dublin and Cork constituencies. Results also indicate differences between parties and some political statuses, while the gender of the candidates is of no relevance. Although the material does not permit a clear distinction between effects of political culture and short-term considerations, taken together the results indicate that localism in Irish politics matters, but in more complicated ways than usually depicted.


Introduction
The localism of Irish politics has been a recurring topic for academic observers and journalistic commentators for a long time. The strong constituency orientation of politicians is seen as one indication for this localism. Empirically, it has been measured either based on surveys of TDs or, more recently, the content analysis of parliamentary questions. This focus on TDs and their activities is based on an understanding of constituency orientation as a matter of role definition, even though the factor of political culture may influence this definition as a result of concerns about re-election. Although this certainly makes sense, it also implies a conceptual restriction. In elections, sitting TDs compete against candidates who are not members of the Dáil and who are not (yet) required to define the role of a TD for themselves. However, if the constituency orientation of Irish TDs is not so much a matter of individual choice but the result of a political culture that punishes a lack of constituency orientation, all electoral candidates can be expected to have an awareness of this culture and to take it into account in their campaigns.
A look at constituency orientation that analyses characteristics of candidates' electoral campaigns thus brings about a conceptual adjustment. It understands political culture as an ensemble of cognitive and evaluative models that result in generalized expectations (Cartocci, 2011), the awareness of which influences political behaviour also outside the Dáil, for example, in electoral campaigns. Moreover, the inclusion of general election candidates in an analysis of constituency orientation increases the potential number of cases and renders regional comparisons more feasible than surveys of TDs. This makes it possible to review the commonly held assumptions about the West of Ireland as the stronghold of localism and also facilitates distinguishing between parties with regard to the strength of constituency orientation.
The obvious question is whether there is empirical material available that renders such an analysis feasible. The run-up to the 2016 general election has provided such material in the form of a website, created by RTÉ News. It made available video statements of a large majority of the candidates in the election, which had been produced under relatively standardized conditions. The content analysis of these statements provides evidence with regard to the relative emphasis on national versus local issues in the candidates' self-presentations to potential voters.
The article will first give an overview of the research on the constituency orientation of TDs and its possible causes, both in general and with regard to the 2016 election. Then the material and the methods used to analyse it are presented in more detail. The presentation of the results and their discussion follow and are summed up in a short conclusion.

TDs and constituency orientation
The constituency focus of TDs is a well-known feature of the Irish political system. A number of studies have looked into the role orientations of TDs in order to gauge its strength. Wood and Young (1997) compared the constituency activity that junior MPs in Great Britain and Ireland had reported in a survey. They found that the Irish TDs gave higher numbers for various measures of constituency activity, such as hours per week devoted to the constituency, and that 40 per cent of the TDs would have preferred to do less constituency work and only 22.5 per cent to do more. Heitshusen et al. (2005) studied the constituency focus of MPs in five countries (UK, Canada, New Zealand, Australia, Ireland). Of the 41 Irish TDs that were interviewed for this study, 39 per cent ranked the constituency as their sole top focus and another 19 per cent named it their top focus along with another priority (Heitshusen et al., 2005: 39). In the PARTIREP survey, which asked about the role orientations of MPs in 15 European countries, 93.7 per cent of the interviewed Irish TDs indicated their regular engagement in constituency-oriented activities such as clinics, attendance at funerals or the publicizing of successes in attracting business and government funding to the constituency (André et al., 2014: 174). Martin (2011) used a content analysis of parliamentary questions in order to measure constituency orientation. For the period 1997-2002, he found the expected focus on the constituency to some degree but also noted that more than half of all written parliamentary questions concerned a non-local issue. The results lead him to suspect that TDs may slightly overstate their constituency focus in interviews.
Election studies have shown that Irish voters tend to expect a strong orientation towards the constituency. For example, in 2007before the financial and economic crisis temporarily drew more attention to issues at the national level  -40 per cent of survey respondents indicated that 'choosing a candidate to look after the needs of the constituency' had been their most important consideration when they decided on their vote in the general election (Marsh, 2008: 113-114).
The strong constituency orientation in national-level politics has been attributed to a number of factors. Among them is the small size of society (Gallagher & Komito, 2010: 247-248) as well as the centralized and inaccessible state apparatus, which renders the use of TDs as brokers necessary or at least attractive for some citizens (Komito, 1984;O'Leary, 2011). The electoral system is seen as encouraging a focus on constituency work in order to cultivate a personal vote, especially due to its potential for intra-party competition (Carty, 1981;Martin, 2010). Finally, it is suggested that a political culture of dependency, which was the result of a peasant society under a foreign rule (Higgins, 1982), still influences common notions about the TD's role as a broker for individual constituents and as a promoter for the constituency vis-à-vis the state (Gallagher & Komito, 2010: 246-247).
At the same time, many studies have acknowledged differences within Ireland regarding the strength of constituency focus. The distinction between urban centre and rural periphery areas has been used in order to make sense of the localism in national politics. This perspective is related to the literature on patron-client ties and clientelism, which had been used especially in the 1970s to make sense of the brokerage activities of TDs and put them in a comparative theoretical framework stressing the incomplete transition from an agrarian to an industrialized society (Gibbon & Higgins, 1974). Garvin (1974) regarded the localism as an element of a political style that originally predominated in the rural West of Ireland but could spread to the whole country as a result of the overwhelming success of Fianna Fáil in the 1930s.T he allusions to modernization theory present in these arguments were already criticized in the 1980s, when Komito (1984) pointed out that the problem-solving potential of brokerage may well extend to modern society. The suspected cleavage between Dublin and peripheral regions is nowadays mostly conceived as a matter of geographical distance between the capital and the constituency. In their study on junior legislators, Wood and Young (1997) found a positive correlation between this distance and the days per week spent in the constituency. They also suggest that 'the local cultures in the most distant areas […] require legislative representatives to be more visibly "here, among us" than is true in the more accessible parts of the country' (Wood & Young, 1997: 226). Martin (2011) measured a possible effect of a centre-periphery cleavage both with a dummy variable for Dublin constituencies and with a variable that used the contiguous distance of constituencies from Dublin. His results show that distance from Dublin has a positive effect on the local focus of parliamentary questions. However, they do not support the assumption that the constituencies that are most peripheral to Dublin have the most localism. It is actually not clear whether the presumed centre-periphery cleavage should be theorized as a matter of distance from Dublin as the centre of political decision-making or as a matter of distance from Dublin as an urban area. In the latter case, one would expect that, for example, some constituencies in Cork might display lower levels of constituency orientation, similar to Dublin constituencies. If we consider the local orientation of TDs partly or even predominantly a result of voters' expectations (Wood & Young, 1997), there are good reasons to suspect that voters in urban areas, which typically display higher social mobility, may both feel less attached to their constituency and be less familiar with local TDs and the brokerage services they are prepared to provide.
Pointing to political culture in order to explain the local orientation of TDs is one way of stating the relevance of such expectations. The reference to culture in this regard expresses the relative stability and ubiquity of the latter (Prager, 1986). In other words, the prevalence of certain expectations can be expected and behaviour is adjusted accordingly. If political culture works in this way, its effects will not be limited to TDs but potentially concern all candidates for the Dáil, whether eventually successful or not.
To the extent to which candidates behave as vote-seekers or office-seekers, their election campaigns are likely to take into account expectations about their potential voters' expectations. This is especially plausible since campaign statements do not actually commit the speaker to anything. Of course, not all candidates who present themselves in a general election do so in order to win a seat in the Dáil. Especially among the group of Independents, there are also single-issue candidates who do not hope to actually become a TD but want to raise awareness of a specific topic and 'reclusive mavericks' (Weeks, 2011: 21). Yet, in any constituency the number of candidates who enter the race as voteseekers or office-seekers will be higher than the number of seats.
A possible localist political culture should consequently be discernible not only in the statements and behaviour of TDs, but also in election campaigns of other candidates. This broadens the number of relevant cases when looking for geographical patterns. The look at all election candidates instead of TDs has the advantage of permitting a separation between the support for a particular party, which may or may not happen to favour a strong constituency orientation, and general expectations within a constituency regarding the local orientation of national-level politicians. If all the candidates in certain constituencies on average display a stronger constituency orientation than in others, this would indicate differences with regard to political culture. Although such differences are unlikely to be completely unrelated to the role orientations of past and incumbent TDs in this constituency, causeeffect relationships are impossible to make out in this regard. Instead of information on the role orientation of TDs, differences indicate the presence of generalized expectations in the sense of a political culture, the homogeneity of which is an open question.
At the same time, political culture is not a deterministic concept. On the one hand, one might expect that serious election candidates will respect the exigencies of political culture and give their campaign a strongly local touch in a localist political culture. On the other hand, the localism of Irish politics has been a perennial subject of criticism (Gallagher & Komito, 2010: 242). This type of criticism has become more common than ever as a reaction to the financial and economic crisis (Murphy, 2015). The behaviour of candidates in the 2016 election is therefore especially interesting. It indicates how years of economic crisis and austerity may have affected candidates' expectations about what kind of appeal works best. Ex ante, an explicitly declared break with the presumed localist outlook of TDs could have seemed appealing at least to some candidates.
The choice between a national and a local focus of campaign messages is likely to differ according to the party to which a candidate belongs. Once again, different lines of reasoning could claim ex ante plausibility for the election 2016. If sitting TDs and candidates for the parties in government, Fine Gael and Labour, feared the voters' retribution for years of harsh austerity, they may have been tempted to stress their commitment to local issues. However, if candidates from these parties believed that voters perceived and appreciated the return to growth in the Irish economy (Lynch, 2016) and could be convinced to reward the governing parties for it, a focus on this and other national issues would have seemed sensible. Candidates from Fianna Fáil faced the conundrum that a strong appeal to the level of the constituency could be regarded as new evidence of a clientelistic stance, which critical commentators had identified as one cause of the crisis (Carey, 2010;O'Toole, 2011) and with which Fianna Fáil has traditionally been associated more than other parties (Garvin, 1974;Prager, 1986). On the other hand, to the extent to which this past indicated an actual strength of the party it seemed by no means an obvious choice to turn away from it, especially after one legislative period in opposition. In contrast, candidates from small parties like the Greens or the newly established Social Democrats could be expected to focus on national-level policy messages.
Apart from geographical and party differences, an exploration of the focus of appeal that incumbent TDs and candidates choose in the run-up to a general election permits looking for further relevant differences. The political status of the candidate at the time of the campaign could be one of them. Local councillors running for a Dáil seat may want to build on their reputation gained for dealing with local issues and could thus be expected to show a stronger constituency orientation than incumbent TDs. Within the group of incumbents, those with a ministerial office may on the one hand want to highlight national issues. On the other hand, a localist political culture may incline them to stress constituency matters even more in order to address concerns for which they no longer sufficiently care.
Finally, potential differences with regard to gender deserve attention. The introduction of gender quotas for the 2016 election incentivized parties to select more female candidates. Not only did this render the issue of women's under-representation in politics more salient. It also raises the question how female candidates perceive voters' expectations in terms of constituency orientation and whether there are differences to male candidates. Studies on possible influences of gender on the behaviour of parliamentarians are sparse and focus on legislative debates, priorities and effectiveness (Bird, 2005;Paxton et al., 2007: 272-274;Bicquelet et al., 2012). Catalano (2009: 53) mentions research on gendered behaviour in constituency work as a desideratum. With regard to the constituency orientation of female compared to male electoral candidates, two opposing arguments are potentially relevant. On the one hand, women still tend to bear a much larger share of caretaking work at home, which makes a career as public representative especially challenging (Catalano, 2009: 49). As constituency-related work is known to be extremely time-consuming, female candidates might prefer to focus on broader, national issues. On the other hand, studies on other countries have found that women often enter politics as a result of previous involvement in voluntary work (Paxton et al., 2007: 268). The grassroots nature of such experience could foster the constituency orientation of female electoral candidates.
Due to the initial plausibility of different arguments regarding the direction of influences of the various factors discussed above, the hypotheses for the following analysis are non-directional, except for those regarding the role of the candidates' location on the constituency map. Here I expect to find that generalized expectations regarding the emphasis on constituency matters influence the content of individual statements. In other words, I expect candidates from constituencies in which the overall level of constituency orientation is higher to display a stronger constituency focus in their statements, compared to candidates from constituencies with a lower average level of constituency orientation (H1a). I expect a relationship between this aspect of political culture and geographical patterns, in such a way that candidates in constituencies in the vicinity of Dublin focus less on the constituency level in their campaign messages than candidates in constituencies with a peripheral location in relation to Dublin (H1b) and that candidates from urban constituencies focus less on the constituency level, compared to candidates from rural areas (H1c).
I further expect to find differences in the strength of constituency orientation between candidates from different parties (H2). In regard to the influence of political status, I expect differences in constituency orientation between local councillors and sitting TDs as well as between sitting TDs and ministers (H3). Finally, I also expect to find differences between male and female candidates in terms of the strength of constituency orientation (H4).

Data and method
The empirical analysis is based on short video statements of 488 candidates in the run-up to the general election on 26 February 2016, that is, 88.7 per cent of all candidates. RTÉ News had initiated the so-called Project 500, in which they had asked candidates from all the constituencies to provide a statement with a maximum duration of one minute. As soon as the elections had been called, the videos were published online and have remained accessible ever since (http://www.rte.ie/news/election-2016/candidates). In addition, the website also presented short profiles of the 62 candidates who did not give a video statement. Those without a statement are partly high-profile party politicians or well-known independents, but mainly unknown candidates running as independents. In the case of a few of the unknown candidates it is possible that they declared their intention to run so late that RTÉ was not able to produce a video, although recordings were still added between the first presentation of the website and the days immediately before the election. At any rate, 62.3 per cent of the cases without a video are candidates without a prior formal political role at the national or local level; among the cases with a video this share is only 39.1 per cent (Appendix 1).
Controlling for prior political experience, there are few significant differences between candidates with a video and candidates without one (Appendix 1). Among those without a prior political role, men are significantly more frequent in the group without a video than in the group who provided videos. In terms of party, the only worrying difference concerns the independent candidates with a prior formal political role. Their share among the 488 candidates with videos is significantly lower than among the 62 for whom no video is available. It is not clear whether RTÉ invested less effort in approaching independent candidates or whether they simply chose more often not to be included. Nevertheless, there are videos for 39 independent candidates with a prior formal political role, which is a sufficiently broad basis to analyse this group.
The video statements obviously constitute only a small part of the campaign efforts of the candidates. Depending on their political status, candidates are likely to have assessed the electoral value of such a statement very differently. For relatively unknown candidates, it may have appeared as a unique opportunity to reach a considerable number of potential voters. Wellknown candidates on the other hand may have reasoned that it would do no harm to add a video to the other campaign efforts and could help to reach younger people who often use new media. In any case, there is no reason to suspect that candidates chose to record a video but did not think carefully about the message they wanted it to convey. Consequently, the choice to talk about national issues, matters pertaining to the constituency or both is an indication of both their own priorities and their expectations about what kind of message might appeal to the voters. The parameters of the recording with regard to length, camera work and background design as well as the method of dissemination were uniform for all candidates. This makes the statements an excellent material for comparisons between (groups of) candidates.
The content analysis is based on the written transcripts of the complete statements, except for passages in Irish. No candidate provided an Irish-only statement and most of those who used Irish did so only briefly. There is no indication that English and Irish passages might have contradicted each other, which is why it seems acceptable to focus only on the former.
The aim of the analysis was to distinguish between statements with an exclusive focus on matters of national politics, statements addressing issues and concerns at the constituency level only, and statements combining a local and a national outlook. Considering the incentives provided by the electoral system, combined references to constituency matters and national politics do not come as a surprise. In contrast, statements with either an exclusive constituency focus or an exclusive national focus are remarkable, because they deviate from this pattern. They either disregard the national dimension of the general election or the systemic incentives to cultivate a personal vote (Carty, 1981;Martin, 2010).
In a first step, I coded sentences that clearly address issues at the local level, the national level or both accordingly. Only phrases that explicitly refer to problems, pledges or achievements at the respective level were coded in this way. Many video statements contain content that does not meet these criteria. They give, for example, information on the personal and/or professional background of the candidate or express commitments to certain values, like honesty or hard work. Although a reference to personal roots in the constituency is a common feature of such passages, this was not in itself considered as an indication of a constituency focus. After all, the local pedigree of candidates is a ubiquitous feature of the Irish political system.
Based on the coding with regard to the local and/or national purview of sentences within the statements, I classified every statement as either exclusively addressing the level of the constituency, the level of national politics or both. Only statements that did not contain any references to national issues were included in the first category and only statements that did not refer to matters on the level of the constituency at all were included in the second.
Although the coding rules are straightforward, the coding process has an interpretative dimension that cannot be completely eliminated. In order to deal with it, I coded all statements twice, with an interval of several months between the two coding rounds. When there was a discrepancy between the two coding rounds, I carefully considered which code seemed most appropriate. Since the data comprise 88.7 per cent of all election candidates, significance levels given in the data analysis do not have a probabilistic meaning in relation to the population of cases. Instead they can be interpreted as a general measure of robustness of the results, also against interpretative ambiguities in coding.
I defined a variable for the individual constituency orientation of a candidate (ICO), which can take three values: 1 for candidates with a statement focusing on the national level only, 2 for candidates addressing both levels and 3 for candidates exclusively talking about matters pertaining to the constituency. The interpretation of the ICO as metric permits the calculation of the average constituency orientation (ACO) in a given constituency as the average of the ICOs of all candidates in that constituency. The ACO is taken as indicating an important aspect of the political culture in the constituency, namely a more or less pronounced expectation of localism. A grouping of the constituencies in terms of their ACOs permits to draw a nuanced picture of the prevalence of generalized localist expectations in Ireland. The results are reported in Section 4.
The interpretation of the ICO as a metric variable also lends itself to analyses using multiple regression. As the characteristics of the variable violate certain statistical assumptions of ordinary least square (OLS) regression analysis, such a course is not without problems. As a robustness test, I therefore calculated three binary logistic regression models, merging the values ICO = 2 and ICO = 3 into one category in order to create a binary variable for localism. The results are almost identical in terms of which coefficients are significant and the direction of significant effects. To facilitate the analysis, in Section 5 I interpret the results of the OLS regression, point out deviations from the results of the binary logistic regression and present the results of the latter in Appendix 2.
To measure the impact of geography on constituency orientation, I employ two additional measures in the regression analysis. One is the variable Periphery, which was proposed by Martin (2011). It measures the contiguous distance of a constituency from Dublin. Contiguity to Dublin means that the variable takes the value 1. Constituencies that are contiguous to these constituencies score 2, and so on, the most peripheral constituency scoring 6.
The second variable for geography is Rurality. It takes the value 1 for all constituencies that are contiguous to predominantly urban constituencies, namely to the Dublin constituencies, Limerick City and Cork South-Central. Constituencies that encompass both a city and more rural parts also receive the value 1 (Waterford, Galway West). For predominantly rural constituencies with only smaller towns, the variable scores 2.

Results for the ACO in the constituencies
The ACO varies considerably between the Irish constituencies. It is lowest (1.2) in Dublin Bay South and Dublin North-West and highest (2.3) in Laois and Roscommon-Galway. The lowest value of the ACO outside the Dublin region is 1.4 for Cork South-Central and Cork East. The highest value of the ACO in a Dublin constituency is 1.7 for Dublin Bay North.
Although more than 90 per cent of all electoral candidates provided a video statement, there are a number of constituencies where such statements were available for less than 75 per cent of all candidates, for example, Kerry, Donegal or Mayo. Against this background and also in order to facilitate an interpretation in geographical terms, it is helpful to form broader categories that can be interpreted in meaningful ways. An easy way to do this is to divide the constituencies in three groups, each of which cover an equal range between the lowest and the highest ACO. An ACO from 1.2 to 1.5 is thus interpreted as a weak constituency orientation, an ACO from 1.6 to 1.9 as a medium constituency orientation and an ACO from 2.0 to 2.3 as a strong constituency orientation.
Applying these three categories to a constituency map of Ireland renders interesting results (Figure 1). It shows a band of constituencies with strong constituency orientation along the Western seaboard as well as a corridor that shares this feature and stretches from the West coast right through the middle of the country far into the East. This pattern contrasts with the Dublin region, where most constituencies display a weak constituency orientation, and several Cork constituencies, where constituency orientation is also weak. This indicates that less emphasis on matters of the constituency is an effect of urban centres and not Dublin in particular. Furthermore, Limerick City and Galway East display only a medium constituency orientation, which is lower than in all the constituencies surrounding them. This also suggests that the distinction between urban and rural areas is relevant when it comes to constituency orientation (although Galway City is actually part of the Galway West constituency). Since Waterford and smaller towns do not display a similar effect, the urban factor seems to be relevant only for cities above a certain size.
There is no obvious way in which the geographical pattern is linked to the strength or weakness of specific parties. Constituencies with the same ACO display widely diverging election results. Among the group with weak constituency orientation are two Cork constituencies where Fianna Fáil was considerably stronger than Fine Gael as well as a third Cork constituency and the Dublin constituencies, wherewith the exception of Dublin Fingal -Fine Gael received more first preference votes than Fianna Fáil. Among the constituencies with strong constituency orientation, there are six in which Fianna Fáil got more first preference votes than Fine Gael, five in which the picture is reversed and one in which both parties received approximately the same percentage of first preference votes. The look at the electoral support for the smaller parties or independent candidates does not reveal a geographical pattern that fits the variation in constituency orientation either.
On the other hand, parties seem to be sensitive to some degree to differences regarding localist expectations. The parties' ACO values can be grouped according to the three constituency groups (Table 1). Results are not reported, if there were fewer than five candidates for a party in one of the constituency groups. Since the independent candidates are a category, but not a group with some inner coherence or even campaign strategy, the differences among them between the three constituency groups can be expected to simply mirror the distinction between high, medium and low levels of constituency orientation. In contrast, the parties display distinct patterns.
The ACO for Fianna Fáil is high in constituencies that have on average a strong or a medium constituency orientation. Only in constituencies with weak constituency orientation do the Fianna Fáil candidates express a markedly lower concern for constituency matters in their video statements. The same pattern applies to Fine Gael although the ACO of its candidates is already a little lower in constituencies with a medium constituency orientation. Since the group of constituencies with weak constituency orientation is predominantly made up of urban areas, the strategy of the two largest parties appears to follow an urban/rural distinction. Green Party candidates display an ACO well below 2 in all the constituency groups. Nevertheless, on a lower level than Fianna Fáil and Fine Gael they seem to follow this urban/rural distinction as well.
The Labour party candidates show a graded emphasis on constituency matters; in constituencies in which this emphasis is on average strong their constituency orientation is even a little more pronounced than that of Fianna Fáil  and Fine Gael candidates. Sinn Féin candidates display a pronounced difference regarding the stressing of constituency matters between constituencies with a strong constituency orientation and the rest of the country. In the former, their ACO is only a little less pronounced than that of Fianna Fáil or Fine Gael candidates. In all other constituencies it is, however, very low.

Multivariate results
In order to test the hypotheses presented in Section 2, I calculated three OLS regression models ( Table 2). The three models differ in the way in which effects of the constituency on the ICO are operationalized. Model 1 uses the ACO to characterize the constituency, model 2 employs the Periphery variable and model 3 the Rurality variable. Party effects are calculated using Fine Gael as the reference category, and effects of political status are calculated in relation to the status of outgoing TD. The effects are overall small, but occur in a data set comprising almost 90 per cent of the relevant population. Significance levels are therefore interpreted as indicating robust results.
All three variables used to gauge the effect of the constituency as the political environment in which a candidate is running display robust results. They indicate a positive effect of both peripheral location in relation to Dublin and rurality. Effects of the Periphery variable are, however, not significant for all levels of periphery in the logistic regression model (Appendix 2). Moreover, the effect does not increase in size with increasing distance from Dublin. The Average Constituency Orientation variable displays a relatively strong effect, compared to the other variables in the respective model. Its direction is positive. All else holding equal, an increase of the ACO by one unit would raise the ICO of a candidate by 0.9 units. However, it is important to keep in mind that some correlation is to be expected in any case, as the ICO of a candidate enters the calculation of the ACO in her constituency.
All three models consistently show that candidates from Sinn Féin and the Green Party as well as independent candidates have a lower constituency orientation than Fine Gael candidates, holding all else equal. This effect is even more pronounced for candidates from the Anti-Austerity Alliance/ People Before Profit as well as for the category of other, small parties. The results indicate that Labour candidates also have a lower constituency orientation than Fine Gael candidates. However, this effect is smaller than in the cases of the other parties and it does not reach significance level in the logistic regression (Appendix 2). There is no robust difference between Fianna Fáil and Fine Gael or between members of the Independent Alliance and Fine Gael regarding the strength of constituency orientation.
Local councillors have a stronger constituency orientation compared to outgoing TDs. The results also show a similar effect for senators. However, in the logistic regression this effect reaches significance in only one of the three models. Results for the other political status categories lack robustness and signs differ between the OLS regression and the logistic regression models.
The coefficients for the variable Female are very small, have varying signs and do not reach conventional significance levels in all calculated models.

Discussion
The findings confirm and expand the results of other studies regarding the geographical pattern of constituency orientation in Ireland. Both the mapping of average constituency orientations in the constituencies and the results of the multivariate analysis indicate that there are differences with regard to the average level of constituency orientation in the constituencies and that these have an impact on the ICO of candidates (hypothesis H1a). The results show an urban-rural divide in terms of the perceived necessity and inclination to bring local issues into a general election campaign (hypothesis H1c). There is a clear difference between most Dublin and Cork constituencies on the one hand and the rest of the country on the other, with candidates from the former displaying low levels of localism in their statements, compared to the latter. But not only the two largest cities stand out. The medium constituency orientation in Limerick City amidst other constituencies with strong constituency orientation also fits the pattern. The medium constituency orientation in Galway East could be due to an urban influence as well, although Galway City is part of the Galway West constituency, where a possible urban influence seems to be cancelled out by the large rural hinterland.
Not only the Rurality variable but also the Periphery variable is linked to the strength of candidates' constituency orientation, which partly supports hypothesis H1b. The pattern is, however, more complicated than a simple difference between Western periphery and Eastern centre. Apart from in the West of Ireland there is a corridor of strong constituency orientation, which reaches far into the East of the country. If an emphasis on constituency matters is an indication of a perception of remoteness to national-level concerns, it does not seem to be directly related to distance from Dublin. Since the empirical material used does not provide hints as to the stability of the discovered pattern, different interpretations seem initially plausible.
An impression of being neglected and therefore forced to speak up for the constituency may be linked to aspects of political representation. In Ireland, having a minister among the constituencies' TDs is often seen as an indication of influence on national politicswhich may then also benefit the constituency (Suiter & O'Malley, 2014). However, the inclusion of a variable distinguishing between constituencies with and without senior or junior ministers did not improve the calculated models and the respective coefficients did not reach significance. The assumption that candidates in constituencies without a minister believed that voters especially worried about their constituency being left behind and therefore stressed constituency matters in the statements is thus not supported by the data.
It is possible that effects of long-term political culture are overlaid with shorter-term reactions to the financial crisis. Against the background of the existing literature, the strong constituency orientation in constituencies of the Midlands (Laois, Offaly) and the Mid-East (Kildare South) is particularly surprising. Looking at the changes in disposable income per person from 2005 to 2010, we find, however, that the Midlands and the Mid-East were those regions that faced the biggest decreases of −7.3 and −7.7 per cent, respectively (Central Statistics Office, 2013: 53). In such an environment, candidates may have experimented with stressing the concern for their own constituency even without a strongly entrenched localist culture. Unfortunately, the available data do not allow distinguishing between long-term cultural expectations and short-term reactions to the crisis with any certainty.
In line with hypothesis H2, the multivariate analysis has found differences in the strength of constituency orientation between candidates from different parties, although not for the two biggest parties Fine Gael and Fianna Fáil. Except for the Independent Alliance, all other parties and groups show less inclination to focus on the constituency level than Fine Gael. The least established parties and groups, that is the Anti-Austerity Alliance/People Before Profit and a number of small parties, grouped together for the analysis, run candidates with especially low constituency orientation. These results are in line with McGraw's (2016) observation that minor parties in Ireland generally stress cohesive policy views much more than the major parties, which tend to opt for catchall appeals.
The comparison of parties between the constituencies with strong, medium and weak constituency orientation suggests that Irish parties differ in their efforts to adapt to a putative localist culture. Both Fianna Fáil and Fine Gael display less constituency concern in the two urban centres Dublin and Cork and more in the rest of the country. Candidates of the Green Party follow this pattern as well, although they generally address constituency matters less than Fianna Fáil and Fine Gael. This is in line with the traditionally greater importance of a general ideological framework within the left-wing parties. At the same time Green candidates seem somewhat prepared to adapt to the exigencies of a localistic political culture. The Labour Party candidates go much further in this adaptation. In constituencies where constituency orientation is generally strong, they display on average an even slightly stronger constituency orientation than Fianna Fáil and Fine Gael candidates. The pattern for Sinn Féin indicates an adaptation to localistic expectations in those constituencies with a strong constituency orientation and an emphasis on national-level issues in the rest of the country. In sum, all parties appear to take the possibility of localistic expectations into consideration.
The political status of candidates had partly the effects expected in hypothesis H3. Candidates who were local councillors at the time of the election tended to focus more on constituency level issues than outgoing TDs. The localistic orientation of many TDs has often been attributed to the fact that most of them served at the local level before running for the national parliament (Gallagher & Komito, 2010: 248). The results indicate that this localistic outlook is somewhat moderated once TDs have spent time in nationallevel politics. Contrary to expectations, the multivariate analysis does not show relevant differences between sitting TDs and ministers running in the election. This result does not necessarily contradict findings by Suiter and O'Malley (2014) and Martin (2016), who have shown the ability of ministers to target their constituencies with special resources and the electoral benefits they draw from their office. Considering the general importance ascribed to having a minister in the constituency, such candidates may well have come to the conclusion that it is unnecessary to stress their local credentials in a statement to constituents. On the other hand, the results suggest that senators running for the Dáil tend to feel such a necessity, which means that their constituency orientation in the statements is stronger than that of outgoing TDs.
At first sight, it may be surprising that candidates without any previous political office at the local or national level do not differ from TDs when it comes to the strength of constituency orientation. In fact, a bivariate analysis of political status and ICO shows such an effect. In the multivariate analysis it is superseded by the influence of the party variables, since many candidates running as Independents are at the same time candidates without a prior formal role.
Hypothesis H4 is not confirmed. There is no relevant difference between male and female candidates. This indicates that gender-related experiences in other realms of life are not translated into different expectations about what voters prefer to hear in terms of a local or a national message in a general campaign statement. This finding suggests the actual relevance of a general political culture of constituency orientation.
The overall constituency orientation of the recorded campaign statements does not seem excessive though. Even the highest average constituency orientations did not diverge far from the middle ground of balancing local and national appeal. Considering the incentives that the Irish electoral system offers the strength of national-level appeals could even seem surprising. It may partly be due to the nature of the analysed material. Both the fact that the video statements were initiated and produced by RTÉ and their release on a website are cues that suggest a perspective that reaches beyond the constituency. Therefore, the results should not be taken as an objective measure of the strength of localism in Irish politics. This qualification does, however, not diminish the value of comparing cases within this material and the interpretations based on these comparisons.

Conclusion
The article has made use of a new source in order to reassess some common assumptions about localism and constituency orientation in Irish politics. The analysis of a specific type of electoral campaign message does not provide a comprehensive picture regarding the emphasis that candidates in the Irish general election 2016 gave national and local issues, respectively. However, the analysed video statements have the advantage of being easily comparable due to the similar circumstances in which they were produced. Moreover, a large majority of all the candidates who ran in the election provided a video statement. These two points render them a highly valuable empirical material, notwithstanding the fact that candidates probably assessed their relative importance compared to other elements of their respective campaigns very differently.
Although this kind of material is only available for the most recent general election, its analysis permits some tentative conclusions with regard to Irish political culture and its supposed localist character. First, there is a clear difference between urban centres above a certain sizeespecially Dublin and Cork, but to some degree also Galway and Limerickand the rest of the country in the way candidates present themselves with regard to the relevance of national versus local issues. The picture is complicated by further differences outside the urban centres, which do not fit a simple distinction between centre and periphery. The corridor of strong constituency orientation that stretches from the Western coast through the middle of the country into the East indicates a more nuanced pattern, to which effects of long-term political culture and more immediate reactions to especially severe consequences of the economic crisis may contribute. Overall the results indicate that constituency orientation should be understood not just as a matter of role definition for TDs, but as a reaction to generalized expectations that influence the style of political campaigns of electoral candidates in general.
Second, parties make a difference for the candidates' emphasis on local or national issues. To some extent, this is no surprise since the official party strategies for the 2016 election differed and candidates were certainly influenced by them. However, apart from explicit strategic choices regarding an unequivocal national-level message versus a campaign style that stresses local canvassing party ideology seems to play a role. It appears to counteract assumptions about voters' localist expectations to some degree. The effect is most pronounced in the case of small and newcomer parties, but also visible for the Green Party, Sinn Féin and to some degree Labour. At the same time, in constituencies with a strong ACO even Labour and Green candidates chose a markedly stronger constituency focus than their party colleagues elsewhere. This indicates that awareness of a political culture favouring localism is a relevant factor for electoral candidates, even when their party political background seems to favour a national outlook.
Third, the results confirm that a previous role as local councillor strengthens the constituency orientation of candidates. However, the difference between such candidates and outgoing TDs in terms of constituency orientation also suggests that a localistic outlook is moderated once politicians are actually involved in national politics in the Dáil. On the other hand, this does not seem to be an effect of a political role at the national level as such, which is indicated by the stronger tendency towards constituency orientation that senators display, compared to TDs. Future, more comprehensive comparisons between members of both Houses of the Oireachtas may therefore further our understanding of how politicians perceive and react to localist expectations.
Fourth, the analysis has paid explicit attention to the possibility of differences in terms of gender. It thus takes a first step to advance the literature on influences of gender on political behaviour in the neglected area of constituency work. The fact that the analysis did not find such differences in campaign-related statements is an important result. It still raises the question to what extent the actual practice of constituency work and other local work of political candidates may be gendered nevertheless, which must be left to future research.
Based on the use of a new type of data, the results of the study provide a fresh perspective on the well-known topic of Irish political culture. They suggest that localism indeed matters, but in more nuanced and complicated ways than usually depicted. The nature of the analysed data prevents any conclusion as to the temporal stability of the observed patterns. This limits their generalizability and indicates the need for studies that take into account different points in time. Obviously, it is not easy to address this need, considering the unique character of the data material used here. In any case, the article shows that the focus on electoral candidates and their constituency campaigns is a way to advance the research on the constituency orientation and localism in Irish politics and beyond.

Appendices
Appendix 1. Significant differences between candidates with and without video For the binary logistic regression, two values of the individual constituency orientation (ICO) (ICO = 2 and ICO = 3) were merged in order to create a binary variable for localism yes/no as the dependent variable.   Note: Coefficients from binary logistic regression models with standard errors in parentheses. **p < .05; *p < .1.