Applying Thematic Analysis to Analyse Press Coverage in Cross-Country Comparative Research: A Qualitative Study Protocol

Abstract

The aim of this exploratory study is to investigate similarities and differences in how the press report on school education and policy in England and Germany. For the purpose of analysing relevant regional and national press coverage, we will create four data sets. Inspired by Bourdieu’s theories of the state and work in relation to news media, and applying thematic analysis, we will investigate which specific contents are reported and how issues are covered. To provide contextual information we will also survey who is given voice in the data and which article types are prevalent. Time frame and sample size for this study are determined by balancing the partly conflicting goals of (a) paying due attention to the exploratory ethos of the investigation, (b) analysing a relatively large number of articles, and (c) working with a manageable size of the data set. On the basis of the empirical findings, we will draw conclusions about the relationship between educational policy and the press, adopting a cross-country comparative perspective.

Background

In this exploratory study, we are interested in the relationship between educational policy and the press in two liberal capitalist Western democracies: England and Germany. Adopting a comparative case study research design, this study will investigate similarities and differences in how the press report on school education and policy in the two countries under study. As there are striking differences in the domestic organization of the press, which partly reflects the centralized English and federal German character of state formation (Humphreys, 1996; Potschka, 2012), drawing on Bourdieu’s theories of the state (Lingard & Rawolle, 2004) and his work in relation to news media (Benson, 1999), we will distinguish between regional- and national press coverage within each country. Thereby, we will create four data sets ¹ and focus on three comparative layers: first, English regional press coverage versus German regional press coverage, second, English national press coverage versus German national press coverage, and, third, the complete English press coverage under study versus the complete German press coverage under study (see Figure 1). The third comparative layer also involves the comparison between regional and national press coverage within each country. The study thus sets out to generate in-depth understandings of individual national contexts as well as structured analyses and comparisons of the two countries.

Figure 1.

Data sets and comparative layers.

Research Questions

This study seeks to answer the following main research question:

• How does the regional and national press in England and Germany cover issues of school education?

The main research question involves three sub-questions:

• Which specific contents are reported and how are issues covered?

• Who is given voice in the press coverage?

• Which article types are prevalent?

On the basis of the empirical findings, we will draw conclusions about the relationship between educational policy and the press, adopting a cross-country comparative perspective (Livingstone, 2003).

Literature Search and Originality of the Study

There is a large body of scholarship investigating either the coverage of one particular newspaper or magazine (e.g. Cohen, 2010) or how the press reports on one specific educational issue (e.g. Bierbaum, 2021). Thereby, press coverage of the Programme for International Student Assessment (PISA), mostly in single countries (e.g. Baroutsis & Lingard, 2017; Hu, 2022), but sometimes also in comparative research (e.g. Hopfenbeck & Görgen, 2017; Waldow et al., 2014), is one focal point of interest.

However, studies which analyse press coverage about school education are scarce. With regards to the two countries under study, notable exemptions include Warmington and Murphy’s (2007) study of the British media debate about A-level results, Allgaier’s (2010) investigation of British press coverage about a controversy within the science education curriculum, Kelly et al.’s (2021) investigation of COVID-19-related media coverage in the UK, Herzog et al.’s (2022a) study of how different actor groups position themselves in the German press and Blömeke’s (2005) and Köller et al.’s (2019) investigations of how teachers are portrayed in German weekly news magazines. Still, to our knowledge, no study has followed an outright exploratory approach to investigate press coverage about school education and – what is more – there is a dearth of cross-country comparative research.

Comparing Countries and Regions

In this study, we aim to contrast specific instances of a given phenomenon in order to grasp the idiosyncrasies of each case (Yin, 2018). In Hallin and Mancini’s (2017) seminal typology for the comparison of media systems and follow-up work, England and Germany are treated as prime examples for the liberal- and democratic corporatist model. ² Investigating cases from the liberal model – England interestingly also shares some characteristics with the democratic corporatist model – is particularly valuable because there is a general trend that most countries media systems, including the German one, develop in the direction of this model (which is characterized by limited state intervention, a strong role of the market and strong journalistic professionalization).

Apart from the national level, we will focus on two emblematic regions: South-West England and Schleswig-Holstein. Both regions are suitable for comparison as they feature a sufficient degree of similarities and differences. Both are rural coastal regions, remote from the capital with resemblant number of inhabitants. Taking into account the average household income, both regions are poor. They exhibit ‘the worst educational outcomes for disadvantaged young people in the [UK]’ (Sim & Major, 2022, p. 3) and, in the case of Schleswig-Holstein, an exceptionally high proportion of young adults, aged 30–35 years, who lack occupational qualifications (Hollstein et al., 2021, p. 65). This makes them prime cases for investigating issues of educational inequality and the public good (Rawolle & Lingard, 2022, p. 164).

Method

It is to be expected that a crucial difference between regional and national press coverage relating to school education within each country and across countries, lies in the contents and issues covered. To go beyond a mere quantitative mapping of content keywords, we will subject the articles from the four data sets to a thematic analysis using Braun and Clarke’s (2022; 2006) ‘organic’ approach. Having become familiar with the data, open coding will be performed in an inductive manner across the four data sets and initial codes will be generated. Individual sentences and paragraphs will be designated as data items for coding (Braun & Clarke, 2022). In the following steps the codes will be reworked and refined, assessed in relation to the whole data set, and sorted into themes by collating all data relevant to each of them. Themes will be systematically reviewed, and given a name and definition. A theme, the core concept in thematic analysis, refers to ‘an abstract entity that brings meaning and identity to a recurrent experience and its variant manifestation. [… It] unifies the nature or basis of the experience into a meaningful whole’ (DeSantis & Ugarriza, 2000, p. 362). Themes thus go beyond mere repetitions of keywords from the data. They are actively crafted and interpretative efforts are invested in their development (Herzog et al., 2019). Whole press articles will serve as research units for the thematic analysis, which will be conducted from a constructionist perspective, aiming to contextualize the societal discourses and sociocultural conditions that affect the meanings within the examined written accounts (Braun & Clarke, 2006, p. 85). Furthermore, the analysis, which will be conducted at the semantic level, will be informed by the theoretical underpinning of the study.

To answer sub-question two (Who is given voice?) we will investigate the prevalence of actors in all press articles under study across the four data sets (Beckers & Van Aelst, 2019). First, we will list all actors who are either cited with direct speech or given voice in that their views and statements are indirectly cited. In the next step, we will create categories – whenever applicable, also sub-categories – for the various types of actors (see Herzog & Scerbinina, 2021). To address sub-question 3 (article types) we will distinguish between various types of positioned articles and more objective information-centered articles (Franklin, 2008). The former take a particular stance in an area of contention and debate whereas the latter report factual information in an apparently neutral, objective and balanced manner.

Sampling

To obtain an adequate representation of the regional press landscape in Schleswig-Holstein we will gather coverage from the two daily newspapers with the highest local circulation Lübecker Nachrichten and Kieler Nachrichten, including their local newspaper editions Holsteiner Zeitung, Ostholsteiner Zeitung and Stockelsdorfer Nachrichten/Bad Schwartauer Nachrichten. Additionally, we will collect national press coverage from Frankfurter Allgemeine Zeitung and Frankfurter Allgemeine Sonntagszeitung. In England we will gather regional press coverage in South-West England from the print editions of two daily newspapers Western Morning News and Evening Standard and two weekly newspapers Exmouth Journal and Express and Echo. We will collect national press coverage from The Guardian and The Observer.

In alignment with the exploratory ethos of the study – and in contrast to virtually all studies mentioned in the literature review above –, we will choose an inductive proceeding for the selection of press articles and their inclusion in four data sets. All articles which cover issues of school education in the specified time frame (see below) will be included in the sample. Coverage of kindergarten, nursery school, university, and vocational training issues do not fall within the scope of this study and will be excluded. Historical articles (which, for example, exclusively deal with school reforms in the 20th century and do not make links to current developments) will be excluded. Articles in the regional press which cover relevant issues of school education but focus exclusively on other countries or regions without making links to the two regions under study will be excluded. After conducting a trial, it is to be expected that following these criteria may result in disproportionally more regional German than English press articles. To mitigate this imbalance only German press articles with 300 or more words will be included in the sample.

With regards to Germany, articles will be sampled and accessed via the fee-based Genios online archive, which lists all articles per newspaper edition. ³ When clicking on the title of an article a short snippet, which mentions contents and the number of words, appears (Herzog et al., 2022b; see Figure 2). From reading the snippet it has to be clear that the article is relevant in terms of the sample criteria. The article in Figure 2 refers to Ukrainian refugees in Schleswig-Holstein but there is no mention of school-related matters. As a consequence, it will not be included in data set 3.

Figure 2.

News snippets in the Genios online archive.

Time Frame and Sample Size

We set out to conduct an explorative study using thematic analysis. It is not the goal of thematic analysis scholarship, rooted in an interpretivist paradigm, to make generalizations. Instead, the aim is to provide a rich account of the data and, including contextual information, to embed the data analysis in a coherent narrative. In this exploratory study, we conceive the concept of saturation – which may involve stopping data collection when no new codes or themes emerge (Braun & Clarke, 2019) – in a way as unsuitable because the issues covered in the press change over time. What is more, Braun and Clarke’s (2016, p. 739) thematic analysis approach is at odds with quantitative logic, which for instance underlies statistic modelling of sample size (Fugard & Potts, 2015; Rahdes et al., 2023 in press). To balance the partly conflicting goals of (a) paying due attention to the exploratory ethos of the investigation, (b) analysing a relatively large number of articles, and (c) working with a manageable size of the data set, we looked out for guidance, and, for this reason, searched the Scopus database for articles published in communication journals which used the keyword ‘thematic analysis’ in the abstract and analysed press coverage. We found 37 articles which analysed news articles, news stories, editorials and opinion articles published in print and online newspaper editions. From these 37 articles 21 articles analysed only press coverage and 16 articles combined the analysis of press coverage with other primary or secondary data. For the academic articles that analysed only press coverage, the respective sample size ranged between 32 and 286 press articles (median: 150 articles). The remaining academic articles, whose data sets included both press articles and other data, analysed between 46 and 510 press articles (median: 190 articles). ⁴

Balancing the above-mentioned goals, we decided that, in total, our four data sets should ideally not exceed 400 press articles. From a sampling trial, in England we expect to find about 20 articles per month in the regional press coverage and 40 articles per month in the national press coverage (investigating the newspapers specified above). In Germany, we expect to find 50 articles per month in the regional press and 15 articles in the national press. We thus determined that the sampling period should amount to 3 months and not coincide with school holidays in any of the countries. Eventually, the period 1 April 2022 to 30 June 2022 was chosen. As the a priori determination of sample size in exploratory research is problematic (Sim et al., 2018), we will retain some flexibility to adjust the time frame, if analytically required.

Ethics

In this study we are exclusively dealing with secondary data and, apart from the researchers, there are no participants involved. The study does not include any procedure which involves danger, harm, distress or discomfort to research participants.

Rigour

Throughout the research process we will adhere to the established quality and trustworthiness criteria for (reflexive) thematic analyses. In particular this will involve clearly explicating the analytic positions and assumptions, giving each data item thorough attention in the coding process, consistently coding in a thorough, inclusive and comprehensive manner, ‘crafting’ internally coherent and distinctive themes (which do not overlap), actually analysing (interpreting) the data, ensuring that the analytic claims are founded in the data and positioning the researchers as active in the research process (Braun & Clarke, 2022, p. 269; Nowell et al., 2017). Throughout the various stages of this study, the researchers will regularly discuss progress and problems in the analytic process and treat the data from each country consistently, paying due consideration to the challenges of cross-national comparative research. These challenges include not taking phenomena in one’s own national context as given or self-evident, consistently using a common theoretical framework (including unitary interpretations and applications of concepts), and, in terms of methodology and research design, finding the right balance between homogenizing and particularizing data collection and analysis in each country.

To counteract these risks throughout the research process, following Wendt (2020), two analytical strategies will be pursued: collective project design and alternating between analytical closeness and distance. The former involves the two researchers sharing a common frame of references and conjointly developing a common research design which facilitates cross-national collaboration. The latter will be achieved, first, by systematically relating and comparing idiosyncrasies of the researcher’s own national context (analytical closeness) to the other (more distanced) national context, second, by conjointly engaging into data analysis, and, third, by working directly with empirical material from the other country (Wendt, 2020, p. 248). Various authors have stressed that in-depth understandings of political and cultural contexts are preconditions for successful qualitative cross-national comparative research (e.g. Gómez & Kuronen, 2011; Chapple & Ziebland, 2018). Both researchers involved in this study are well trained in qualitative methodologies. Each researcher is based in and obtains ‘deep’ inside knowledge of one of the two countries under study while both previously carried out comparative research which involved the other country respectively.