Wright and Reeskens suggested the nationalistic perspective influences individuals’ attitudes toward welfare (2013, 1443). Furthermore, Johnston, Banting, Kymlicka and Soroka (2010) insisted that social cohesion is strengthened by not only nationalism, but also the social welfare system (p. 350). However, contemporary, social inequality has been gradually increasing. The redistribution of wealth has been decreasing continuously among developed countries (p. 349). Due to the urgency of this social issue, this paper seeks to examine how the level of nationalism correlates with the public opinion of welfare in the UK. In accordance with Wright and Reeskens (2013), people from different socio-cultural contexts have different attitudes toward the welfare state (p. 1443). Up to this date, only a few studies on this topic conducted in the UK. Therefore, this geographical area is selected as a unit of analysis.
Defining the Concept
The study aims to learn how the level of nationalism correlates with attitudes toward welfare. The independent variable is the level of nationalism. The dependent variable refers to attitudes toward welfare because this is dependent on the degree of nationalism the respondent feels. In addition, gender is used as a control variable which will be explained in detail in the finding section.
Hypotheses
H0: The Level of Nationalism does not correlate with Attitudes toward Welfare.
H1: The Level of Nationalism correlates with Attitudes toward Welfare.
Weighting dataset
Before explaining this part, I would like to clarify that all variables will be weighted, then it will be recoded. These are to prepare the variables before starting the analysis.
This report uses the data from the 2018 British Social Attitudes (BSA) Survey. Weighting the dataset is required because weighted variables better represent the population (UK Data Service, 2014, 3). For the derived variables: bestnatu2 and Rsex , their weighting variable is WtFactor . For the self-completion questionnaire variables, the indicators that will be used to compute WelfarismIndex : UnempJob , SocHelp , DoleFidl , WelfFeet and WelfHelp will be weighted with WtFactorSC .
Variables (all are recoded)
Independent Variable: bestnatu2 What nationality best describes a respondent?
Even if there is Northern Irish in the answer choices of Q1021: What nationality best describes a respondent? (bestnatu2), this variable cannot be found in the data file of BSA 18. Thus, the similar variable bestnatu2 is used; however, there is a limitation in its answer choices. There is no Northern Irish in the choice even if the research question focuses on the UK as the level of analysis, not the Great Britain. After weighting, the variable bestnatu2 is recoded (further details in Appendix 3) because too many answer choices make it difficult to present the analysis.
Dependent Variable: WelfarismIndex Attitudes toward welfare
WelfarismIndex is recoded, then computed using five indicators (further details in Appendix 1 and 2). Buckley and King-Hele (2015, 33) suggested variables should be recoded when “the sample is spread too thinly over many cells”. Thus, before creating WelfarismIndex , variables are recoded. After that, they are checked for internal consistency among each other, which their Cronbach’s alpha is 0.79. This means they are reliable for computing WelfarismIndex .
Control Variable: Rsex What is the biological sex of a respondent?
“What is your sex-assigned-at-birth?” is used to ask which biological sex the respondent is. This is to avoid the complexity of socially constructed types of gender: LGBTQIA+. For the limitation of the variable, as explained by Blackless, Charuvastra, Derryck, Fausto-Sterling, Lauzanne and Lee (2000, 159), approximately 1.8% of the global population is intersex; however, there is no such choice in the answer sheet. For why Rsex needs to be recoded, as explained by Buckley and King-Hele (2015, 33), when “there are [too] many response categories”, the answer choices should be categorized to simplify the analysis. Therefore, these response options are recoded due to their too-much-detailed nature. (For any “recode” responses, please see Appendix 4).
Before starting the statistical analysis, it is important for a researcher to know which research paradigm he/she decides to follow. Therefore, I will explain my philosophical stance before proceeding with the analysis.
This report uses the STATA program to analyze data from the BSA 2018 survey. Therefore, it uses a quantitative data analysis method. According to Moroi (2021, 131), there are four main components of philosophical assumptions in quantitative research: ontology, epistemology, methodology, and axiology. In accordance with Moon and Blackman (2014), there are two main branches of philosophy in social science research: Ontology and epistemology, which both affect the philosophical perspectives of a researcher (p. 1168-1170). Jackson also pointed out the philosophical perspective influences how researchers choose their research methodologies (2013). Hence the relationship between the ontological, epistemological, and methodological perspectives of this report will be shown in Diagram 1.
Ontology simply means what a researcher aims to study (The University of Warwick, 2017, para. 1), and “epistemology [means] how do we create knowledge” (Moon and Blackman, 2014, 1169). Because Moon and Blackman also claimed ontology and epistemology are interlinked with one another and should not be explained separately (p. 1170). Therefore, I will explain both terminologies in an interlinked manner. The ontological perspective that will be adopted in this paper is objectivism. As explained by Given (2008, para. 2), “objectivist ontology [emphasizes that] to objectively know the world, there must be a real objective, definite world”. This could be interpreted as there is only one truth to discover, which links with the perspective of positivist epistemology. Positivists believe there is no need to interpret the underlying meaning of the data from respondents as they are fact generators.
Hence, the datafile of BSA 18 will be treated as an absolute fact. The data will be used without seeking to understand the underlying meanings of the data. All in all, the positivist perspective will be epistemologically adopted in this paper. Due to the positivist nature of this report, I will axiologically adopt an objective view throughout the analysis since, as mentioned by Moroi (2021), in positivism, objectivity is preferable to subjectivity (p. 130). Therefore, as an analyst of this paper, I will view data objectively and think of respondents as fact generators.
Diagram 1: How the research paradigm influences my choice of quantitative essay
To better understand the data, I will begin this section by exploring each variable through the univariate analysis.
Levels of Nationalism
Importantly, all variables are weighted with the appropriate weighting factors. Then, all weighted variables are recoded. After these preparations of data, the univariate analysis begins.
Table 1: Levels of Nationalism
Source: BSA Dataset 2018
Table 1 shows the distribution of participants’ levels of nationalism. 46.4% of participants feel more identified with collective identity (Britishness) than national identity; 40.7% of participants feel more identified with national identity (e.g. Scottish, Welsh, English) than collective identity; 11.1% of participants feel more identified with other identities.
Table 2: The measures of central tendency
N = 3,879 Source: BSA Dataset 2018
According to Manikandan (2011), when there is nominal data, it is best to use mode to measure its central tendency. Hence, I will use mode as the measure of central tendency because the Level of Nationalism a nominal variable. Table 2 shows the central position within the dataset – the mode – of Levels of Nationalism is 1. This indicates more participants are identified with Britishness than Scottishness, Welshness, or Englishness. In addition, the position of mode is a positively skewed distribution, as can be seen from Appendix 5.