4.5 Control variablesIn light of current literature on the topic of CSRD, it is considerednecessary to incorporate a number of control variables in the research design.
Company size is considered to have a significant positive effect on CSRD,given that larger companies are more visible and tend to be under stronger stakeholderpressure (Branco and Rodrigues, 2008; Khan et al., 2013). Given the scopeof the study, the methodology should control for company size when testing theoutlined hypotheses. Previous literature confirms that stakeholder pressure for CSRdisclosure is influenced by the industry to which a company belongs (Gamerschlag,Möller and Verbeeten, 2011). Thus, I controlfor industry profile. Consistent with Chan, Watson and Woodliff (2014), and Brammer andPavelin (2006), the researchdesign will control for industry differences by measuring industry profile.
Chanand colleagues (2014) classify thefinancial sector as a low profile industry, the materials sector as a highprofile industry, and all other sectors as medium profile industries. Previous research on the relationship between corporate socialperformance and financial performance is largely inconclusive (Margolisand Walsh, 2003). However, inline with Majeed and colleagues (2015) the regressionanalysis will control for financial performance, given that recent research suggestsa positive relationship (Ehsanand Kaleem, 2012). 5. Proposed Research DesignThefinal part of the research proposal will outline the proposed researchapproach. The outline will include a discussion of the population and sample,the method of data collection, the method of data analysis and the limitationsassociated with the design of the proposed research. Finally, a time scheduleof the major milestones is proposed, under the assumption that the research isto be conducted within the time limits of a bachelor’s thesis. 5.
1 Population and Sample The population, to which the results of the proposed research is meantto apply, will be multinational corporations in the Nordics. Regardless of theresults obtained, the research is important to legislators and investors, amongothers. In order to draw inferences about the total population, a sample of the60 most-traded publicly listed companies from Denmark, Finland, Norway andSweden will be selected. This sample size is congruent with similar studies (Majeed et al., 2015; Dias et al., 2017).
Thechosen sample is a convenience sample. Similar studies have chosen to use asimilar sample type, expecting that these companies are the most visible andhave the highest levels of disclosure (e.g., Dias et al., 2017).
Given thevariables assessed and the methods used, it is necessary to select companiesthat disclose information in English, which limits the sample severely. Theselected 60 companies currently all publish annual reports in English. Any ofproblems of access should, therefore, be avoided by using the selected sample.My sample will focus on the reports published for the accounting year 2017.5.2 Data CollectionThe methodsof data collection for each of the selected variables are outlined in thefollowing section. 5.
2.1 Dependent VariableCSRdisclosure is the chosen dependent variable of the study. Similar to previousstudies, CSRD is examined using content analysis. Content analysis is a method usedto codify a piece of text based on an established criterion (Weber, 1990, p. 9). Contentanalysis is regularly used when analyzing annual reports and is consideredempirically suitable within CSR research (Guthrie and Parker, 1990).5.2.
1.1Unit of observationOne ofthe main issues in content analysis is the unit of observation. Previousstudies have used various units for analysis including word count, number ofsentences and number of pages (Aliet al., 2017).Another method of analysis is to used advances software applications, for exampleLeximancer. The study will use words as the unit of observation for tworeasons.
The use of words removes the obligation of the coder to makesubjective judgements about the topic. Beyond that, this method allows foreasily repeated and replicated trials that yield the same results (Gamerschlaget al., 2011). Giventhe simplistic approach to the content analysis, the collection of data fromthe annual reports is to be conducted with the word finder function installedin Adobe’s PDF reader or similar tools.
2Identification of keywordsInline with previous research (Holder-Webbet al., 2009; Gamerschlag et al., 2011) the keywords forthe content analysis are based on the framework of the Global ReportingInitiative (GRI). Within the area of CSRD, GRI is considered the most relevantinstitution (Moneva,Archel and Correa, 2006) and the organizationrefers to its guidelines as the global standard. Based on the GRI guidelines,and in line with previous research (e.g.,Gamerschlag et al.
, 2011), a list of 32keywords for the content analysis is outlined in Table 1. When conducting thecontent analysis, the researcher should consider both singular and pluralforms, as well as both British and American spelling.