Online Appendix
CODING SCHEME OF THE TERRISS_DATASET (Adapted Version)
- Why this Coding Scheme?
This coding scheme has been developed to analyse the political discourse of parties on the territorial dimension. The coded documents are party electoral manifestos. When not available, leaflets and other programmatic documents are analysed instead. The principle underpinning the coding scheme is that the analysis of party manifestos can be useful in deciphering the attitudesadopted by political parties in electoral competition. In other words, the focus is on the rhetoric of a party and on the image they convey to voters on a specific issue. This might be different from the actual position of the party, as well as that of individual party members. Furthermore, the attitude adopted in a manifesto is often the result of bargaining and compromise among divergent currents of opinion within a party (and especially between parties in the case of coalition documents).
The codebook outlined here is an adapted version of the original coding scheme (Basile, 2014).. It therefore does not include some elements that are not relevant to the analysis undertaken in this article. The complete dataset and codebook is available upon request from the author.
- How are Manifestos Coded?
According to Krippendorf (2004, 18), content analysis is a “research technique for making replicable and valid inferences from texts (...) to the context of their use” (emphasis added).Moreover, content analysis is often described as an unobtrusive and non reactive method (Ibid), to the extent that the systematic and neutral procedures used in the coding process imply a narrow scope of interpretation on the investigator's part. However, a degree of subjective evaluation from the coder is difficult to exclude completely. The coderindividually reads the text and decides to apply codes, although according to strict and clearly defined rules. For this reasons, content analysis can be understood as a way of quantifying information that is intrinsically qualitative.
This latter point leads to a further aspect, namely the choice of a hand-coded procedure rather than computer-assisted content analysis. Computerised techniques have the merit of reducing the probability of codererrors, and permittingtheanalysis of a large amount of text with vast cost savings. Nevertheless, manual coding is preferred to automated coding for theoretical and methodological reasons. Firstly, the study of a single issue presupposes a deep knowledge of it. Accordingly, only a coder who has been trained to a certain level of familiarity with the issue can observe some of the more “subtle” references to it. Secondly, the use of software is based on the input of a set of keywords that can then be searched for in a document. The problem here is that words like “region”, “governance” “subsidiary principle” and “autonomy” can be used in different contexts from that of territorial politics. For instance, the notion of region may identify the “European” or the “Mediterranean region”, as an expression of a supranational entity; similarly, the subsidiary principle can be understood as a sharing of competences with non-state authorities other than the sub-central governments. Even the term decentralisation can referto other processes that do not necessarily involve the redistribution of resources to regional and local levels. For instance, in the 1994 manifesto of the Partito dei Democratici di Sinistra 1994, in the passage: “(...) che prevedano un decentramento della proprietà e dei poteri di decisione”[1], the term decentralisation is used with reference to market and economic policies. Hence, as in a study focusing on one single political dimension it is important to reduce to the greatest possible extent any misinterpretation of the meanings of words, the preference has been to ground the coding procedure in the investigator’s informed evaluation. The price to pay, of course, is reduced reliability and more labour intensive work; this has required narrowing the analysis to one single case (Italy). As always, it is a matter of choices and trade-offs, as reflected in the necessity to balance research goals with practical constraints.
The coding procedure is based on a two-stage analysis. Firstly, a spreadsheet is developed for each manifesto, where sentences are split and codes are applied according to each category. Then, the data is aggregated according to the rules that are specified in the following sections and inputted into a second spreadsheet, which constitutes the final TERRID_DATASET, available in Excel®, STATA® and SPSS® format.
- Spreadsheet n.1: Applying codes to the document
In the first step, each manifesto is read and divided into quasi-sentences(hereinafter: QSs; see below for the rules for the division into quasi-sentences). Then, a code is applied to each relevant quasi-sentence, according to four main categories, namely: Territorial Issues, Policy Proposal, Territorial level and Frame. For the purposes of the present article, this appendix will focus on two of these domains, namely Territorial Issues and Frames.In defining categories and codes, two criteria are followed: firstly, the categories must be exhaustive, so that each coded unit should have an appropriate category in which it can be included; secondly, the categories and levels should be mutually exclusive. In other words, there must only be one code applicable to each coded unit (Neuendorf 2002).
All the sentences that are related to territorial issues are coded under the Territorial Issue category. In particular, the coder, having identified the relevance of the statement for the topic of the research, has to choose among five codes. 100 is applied when the quasi-sentence generically refers to territorial issues, be they related to cultural or institutional aspects[2], but it does not express any supportive or negative stance on it. For instance: "Le Regioni vanno sempre più assumendo un ruolo quasi statuale, svolgendo competenze di grande valenza politica trasferite dallo Stato, e un tempo di esclusiva competenza legislativa nazionale”[3] (Electoral manifesto of Rifondazione Comunista, 2001). This sentence describes the current territorial system and the distribution of powers within the state, but it does not provide any clear indication of potential party support for a process of transferring powers downwards, just as it does not provide information about the views of a party about the cultural distinctiveness of sub-central authorities. Different codes are applied when the QS refers to what has been defined as the Institutional Dimension. The latter includes all the statements dealing with the territorial redistribution of competencies and resources in several policy fields, constitutional reform, centre-periphery relations and so on. AQS will be coded 101 if it expresses preference for a delegation of competencies downwards; otherwise, as 103 if a QS indicates support for strengthening the central level of government. In contrast, the Cultural Dimension identifies those parts of the document that are devoted to the question of promoting and protecting local and regional identity, culture and symbols (code 102) or national ones (code 104).
Table i provides a summary of these codes.
Table i. Territorial Issues
Code / Title / Description1 / Decentralist Issue (DECIS1)
100 / Neutral / QSs containing statements concerning the territorial organisation of the state, the distribution of power and resources such as: state-regions relationship, the coordination and control of the state over the sub-national entities, state financial aids to regions and local authorities. Use this code when the document refers in a general or ambiguous way to these issues and it does not express any specific supportive or negative stance on it. Example (RC 2001: ‘Le Regioni vanno sempre più assumendo un ruolo quasi statuale, svolgendo competenze di grande valenza politica trasferite dallo Stato, e un tempo di esclusiva competenza legislativa nazionale’[4])
101 / Institutional pro-decentralist argument / Statements expressing support for the redistribution of powers to sub-national levels; statements describing the current centralist asset as negative.
102 / Cultural Regionalism / QSs related to linguistic and cultural issues, namely: support for preservation of local and regional symbols and customs; protection of minorities; reference to regional or local identities.
103 / Institutional centralist argument / Similar to 101 but negative: support for further centralisation of the state structure or for the current centralist asset; preservation of the unity of the state. Claims stating that powers should be kept at central level in order to protect the major ‘national interest’.
104 / Cultural Nationalism / Similar to 102 but negative; reference to national identity and to the national unity as a value to be protected.
Sometimes, the quasi-sentences coded as Territorial Issue might also contain a reference to a specific topic associated with it, to a particular cause or pressure that is deemed to trigger the process of devolution, as well as to a particular actor involved in the process of devolution. In this case, the codes of the category Frame will be applied. The same QS might thus be subjected to multiple coding, as each sentence related to the territorial issue can also contain a reference to a specific frame.For each frame, the coder identifies whether it expresses a neutral stance or, on the contrary, a pro/contra attitude towards sub-central authorities. Accordingly, codes might be neutral, pro or contra the sub-national levels. In this case, each frame has an initial two/three-digit code, ranging from 30 to 103; then, a second digit should be applied (0 if neutral, 1 if pro-sub-national level, 2 if otherwise).
Table ii. Frames (Code from 30 to 103 + 0, 1 or 2, if neutral, decentralist or centralist)[5]
30-10 / Frames/arguments (FRAME)30 -Subsidiary principle
300 / Neutral / This frame should be used only when the subsidiary principle is explicitly mentioned, otherwise, another code, such as 900-902 should be applied.
301 / Subsidiary positive
302 / Subsidiary negative
40 - European Union and Regionalism
400 / Neutral / The EU can be mentioned as a positive framework in which decentralisation should be achieved; similarly, ideas such as the ‘Europe of People’ and ‘Europe of the Regions’ and positive references to multilevel governance require the application of code 401.
Code 402 applies when there are negative mentions of the role of EU in triggering a process of disaggregation of the territorial unity and/or the perspective of a Europe of the People rather than ‘of States’ is negatively perceived.
401 / EU positive
402 / EU negative
50- National Unity
500 / Neutral / The value of national unity (against the enhancement of local powers) can be perceived as negative or positive.
501 / National unity negative
502 / National unity positive
60- Efficiency and responsibility of public administration
600 / Neutral / Arguments concerning the efficiency of the public administration and its modernisation. Decentralisation can be seen as a valuable instrument to achieve greater state efficiency (601).
On the contrary, in order to address the problem of state inefficiency, other statements claim that it is necessary to further strengthen the central, state structure, rather than decentralise (602).
601 / State efficiency pro-decentralisation
602 / State efficiency anti-decentralisation
70-Centre-periphery cleavage/territorial solidarity/cohesion
700 / Neutral / Generic reference to the socio-economic gap between north and south (richer and poorer regions)
701 / Territorial solidarity and national cohesion negative / Strong federalist argument: resources should be kept where they are produced (i.e. the richer regions).
702 / Territorial solidarity and national cohesion positive / Although decentralist processes seem to be inevitable, they should be pursued within an overall framework of territorial solidarity; in particular it should be maintained the ‘centralist’ principle of redistribution of resources from the richer to the poorer regions.
80- Autonomist parties
800 / Neutral / Generic reference to the emergence of autonomist parties in the political system
801 / Autonomist positive / Autonomist parties are seen as a positive force in triggering processes of decentralisation.
802 / Autonomist negative / Autonomist parties are mentioned as disaggregating forces, a threat for the national unity.
90- Democracy
900 / Neutral / Generic reference to the democratic advantages of a better territorial organisation.
901 / Democracy positive / The centralised state does not allow democratic control of the citizens over the resources allocated. Hence, more autonomy is an advantage for the democratic control of the citizens over the decision process.
902 / Democracy negative / The purported democratic virtues of decentralisation are denied.
100- Liberalist argument
1000 / Neutral / State competencies are described as they are: e.g. many powers for the central state.
1001 / Liberalism / A ‘lighter’ state is pursued. In order to achieve this goal, a redistribution of competencies downward is invoked.
1002 / Interventionism / Interventionist state: this implies that resources and powers should be kept as much as possible at central level.
101- Identity argument/territorial attachment
1010 / Neutral / Generic reference to identities.
1011 / Local Identity / Decentralisation is invoked with reference to the need for protection of local identities.
1012 / National identity / Reference to national identity and values.
102- Cooperative federalism
1020 / Neutral / Generic reference to the kind of cooperation among territorial levels.
1021 / Clear separation of powers / Clear separation of powers across levels is envisaged.
1022 / Cooperative federalism / Principle of cooperative federalism. A clear separation of competencies should not be pursued.
103- Political foes
1030 / Neutral / Generic reference to the interaction with other political actors on the territorial issues.
1031 / Foes as obstacle to decentralisation / Political adversaries have been an obstacle to the full implementation of decentralist reforms.
1032 / Foes as supporters of decentralisation / Decentralisation is negatively perceived, while political enemies are deemed to be the responsible of a process of territorial disaggregation.
All these coded documents are stored as Excel® files and they also contain a basic range of identifying variables, which can be useful in case of merging of all the documents:
ID: Each document preliminarily identified by a unique number starting with the electoral year and a progressive numeration: Yyyy + party code (e.g. 199407 is the ID for Forza Italia in 1994 and 200621 for Ulivo in 2006). See Table iii, below, for the codes of the parties.
Different identification codes have been assigned to each party and/or coalition. For example, the Ulivo has a different code to the DS alone. It can therefore happen that, for the same year, both the coalition document and the single party manifesto have been coded. In this way it is possible to compare the differences in political rhetoric between coalition and party documents.
Docu_ID: Party name and year (e.g. AN_1994).
Type: Type of document (e.g. Electoral Manifesto; other programmatic document).
Type_ID: 10 = Electoral Manifesto
20 = other programmatic document.This is used when the document is something other than an official party manifesto, such as congress final reports and leaflets. An example is the ‘Manifesto of Fiuggi’ by AN in 1994. It is a programmatic document delivered on the occasion of the founding congress of the party rather than an electoral manifesto. Such documents have been analysed when an electoral manifesto was not available for that specific election year.
3.1.Rules For The Division In Quasi-Sentences
The present coding scheme relies on the same general rules that have been adopted by the Comparative Manifesto Project (CMP) group and by the Comparative Agendas Project (CAP). These rules are as follows:
-Each QS should contain an argument, i.e. a subject, verb and object;
- When a sentence is long and contains either one subject and many verbs or many subjects and one verb, it should be split;
- When different arguments share the subject, they should be split (‘Vogliamo un governo che risponda alle sfide dell’immigrazione/combatta la criminalità’ …)[6];
- When different arguments share the verb, they should be split (‘Vogliamo una politica estera forte/ un forte esercito/innovazione tecnologica’)[7].
- Do not divide the sentence if it contains a rhetorical series of names, adjectives and verbs, just to further emphasise the concept;
- In cases of list of arguments introduced by the ‘:’, divide the sentences;
- The goal of a party should be divided from the QS referring to the means proposed to address it (‘Possiamo ridurre la disoccupazione/perseguendo attivamente dei piani occupazionali’)[8].
As a general rule, the QS should be a unit of significance by itself. If it is split, it should be understandable and express a single concept.When the coder has to choose whether to divide a sentence or not, the basic distinction between position arguments and frames is kept in mind. The same sentence might contain a declaration of stance plus a justification for it; in such a case, the sentence should be split.
1
- Spreadsheet n. 2: Aggregate data
4.1.Elaboration of the aggregate data
The raw data, obtained through the coding of each manifesto, are then elaborated into a second spreadsheet. This contains variables measuring salience, position and frames. The unit of analysis (cases) of this comprehensive dataset are all the parties analysed over the time span chosen, in order to allow both interparty and time-series comparisons.
4.1.1.Salience
Followingan extensive scholarly literature which hasanalysed manifestos, salience is calculated as the percentage of quasi-sentences related to the territorial issue for each manifesto. It is calculated by summing up all the quasi-sentences dealing with territorial issues in each political document; this number is then divided by the total number of quasi-sentences in which the manifesto can be split and multiplied by 100, to obtain a percentage.
The formula to calculate the overall salience of the statements related to territorial issues is:
Salience =
where “Territorial” indicates all the statements dealing with the territorial issue. In order to calculate the salience of pro-periphery statements, only the quasi-sentences coded as 101 and 102 will be taken into account. On the other hand, by counting the frequency of codes 103 and 104, the salience of pro-centre sentences can be calculated.
4.1.2.Direction of attitudes (or Position)
Datasets based on the counting of quasi-sentences assume that, following salience theory, the greater the relevance of the theme, the more the attitude towards it will be favourable. Nevertheless, a detailed description of the preferences of parties on a single dimension of the political space necessarily requires a step further in the analysis. The relevance of a theme such as the territorial issue in electoral debate may compel a party to adopt a stance on it, despite its reluctance or lack of real interest. Accordingly, beyond the emphasis attributed to the question, it is necessaryalso to estimate the overall direction of party attitudes on an issue.