2. Methodology

2.1 The Sample

The content analysis sample was chosen by the BCI and the NDA in advance of the research team's involvement. The programme sample consisted of a mix of national, commercial and public service radio and television broadcasters, local radio from an urban and rural background and community radio from both an urban and rural background. The sample also included Ros FM, a community station in Roscommon Town with a particular emphasis on people with disabilities. The 408 sample hours were made up of 167 hours of television and 361 hours of radio. These were sourced as follows:

The sample was drawn from the period between February 2007 and July 2007. In the case of the television stations sampled, it was drawn from material broadcast between 5pm and 10pm over the course of composite week^[3]. Similarly, in the case of radio stations, 50% of the material was drawn from weekdays and 50% from Sundays. Thus Saturday is not represented on either medium. The radio broadcast hours in the sample are drawn from the hours between 7am and 7pm. In effect then, the sample is based upon those hours of the day when the respective media of television and radio are attracting their largest audiences. Radio listenership peaks in the morning and slowly declines until 7pm. Meanwhile television only begins to pick up significant viewership as it approaches the primetime hours after 6pm. In sum, this research reflects representation of people with disabilities during those hours when the consumption of the media examined are at their respective peaks rather than the sum of all broadcasting in Ireland.

Although the research looked at 408 hours of material, advertising was not examined as part of the project. Thus 29.5 hours consisting of advertising were excluded from the sample and final analysis was conducted on the basis of a sample size of 378.5 hours.

2.2 The Coding Process

The first stage of the research involved "coding" the 408 hours of programme material. This was conducted by five researchers using the questions laid out in the coding questionnaire (the questionnaire is included in the Appendix 1 to this report). For the most part, the coders only applied the coding questionnaire up to the point where it was established whether or not the programme actually featured a person with a disability, referred to such a person at length, or referred to disability as a discourse (i.e. featured content which was relevant to the content analysis).

Since the subsequent questions in the coding questionnaire examined the nature of this representation and because these representations occurred in a minority of the programmes examined during the research, the majority of programmes coded only recorded the following information:

• Title
• Channel
• Time and duration of broadcast
• Programme Genre
• Programme nationality
• Number of characters/individuals featured

For those programmes which either featured persons with a disability, made extensive reference to a person with a disability or which discussed disability, the coders were required to address a number of subsequent questions contained in the coding questionnaire relating to:

• Characteristics of those represented
• Type of disability
• Level and type of representation offered
• Relevance of disability to the representation
• Occupation of those represented
• Role played by featured person with a disability
• Subject of relevant person's contribution

Their findings were directly inputted into SPSS, a social statistics software package. The analysis of this SPSS data was used to answer the research questions above.

The work of the coders was informed by reference to a coding guidebook, drafted by DCU and approved by the BCI and the NDA, which provided definitions and outlines of what was meant by each of the categories included in the questionnaire.

For further quality control purposes, inter-coder reliability (i.e. consistency amongst coders in the manner in which they categorised programme content) was tested at a number of stages by having all five coders separately code the same pieces of programme content and then comparing their results for agreement. Inter-coder reliability was comfortably within acceptable standards.

The coding process began in early August 2007 and was completed by the beginning of the second week of October 2007. In the interim the project leader conducted some mid-term analyses of the incomplete data and work began on examining the range of discourses thrown up by the content analysis. This work of data and discourse analysis was completed by the end of October 2007.

2.3 Some Definitions - Appearance vs. Reference

Throughout this document we refer both to "appearances" by people with disabilities and "references" to disability. It is important to clarify at the outset what we mean by these terms.

The word "appearance" implies the visual manifestation of a person. However this understanding was considered too narrow for two reasons:

1. First on radio, for obvious reasons, a visual appearance is simply impossible.
2. However even on television, it is possible for an individual to be referred to - possibly at some length - without ever actually physically appearing. Nonetheless such references contribute to how the audience comes to understand those individuals.

Since the purpose of the research was to explore the nature of all broadcast representations of people with disabilities in the sample, (not just those actually on-screen or in-studio), it was decided to define "appearances" as referring both to individuals with disabilities actually present on-screen/in-studio and individuals with disabilities who were discussed in television and radio programming (i.e. "represented") even if they were not actually present.

"References" then referred exclusively to those cases in which disability was referred to (or "represented") without referring to a particular individual with a disability.