4 Phase one: ethnographic case study Uganda
4.3 Phase one objectives and methods
4.3.1 Method 1: Document analysis
As documents can provide background information and historical insight (Bowen, 2009), document analysis was selected to best respond to the first research objective. This provided historical overview and contextualising information to aid the interpretation of the interview results (from both insiders and outsiders) and participant observation (Bowen, 2009). Although document analysis is a method commonly used in qualitative research, there is often a lack of transparency and variability in the descriptions of the procedures used, and no systematic rules to the approach (Altheide et al., 2008; Elo and Kyngäs, 2008). As such, both Elo and Kyngäs’ (2008) and Hsieh and Shannon’s (2005) procedures for content analysis were used as a guide to inform the researcher’s procedure for the analysis of documents.
Objective Data collection
1 To explore and describe the process and history of development of services for PWCD in Uganda.
• Document analysis • Interviews
2 To describe and explain outsiders’ perceptions and experiences of their contribution to the development of services for PWCD in Uganda.
• Insider interviews
3 To describe and explain outsiders’ perceptions and experiences of their contribution to the development of services for PWCD in Uganda.
• Outsider interviews
4 To analyse and interpret insider/outsider collaboration and working practices.
• Participant observation
Procedure
Relevance of documents was guided by the first research objective: to explore and describe the process and history of the development of services for PWCD in Uganda. Both manual and electronic searches were used to identify
potentially relevant documents which included published and accessible documents and grey literature (unpublished or published documents that have limited distribution, McAuley et al., 2000) in the public domain. Some
documents had also been identified as being potentially relevant during the literature review stage (and from the identification of potential participants for interviews): these were kept for their potential inclusion in the analysis.
Searches were repeated over a 15 month period from May 2016 to August 2017. Search terms were limited to English only. To address the first research objective, specific key words were used to allow for the focused identification of potentially relevant documents. Using truncation, the key words used in
combination as search terms were: Uganda; speech and language therap* and synonyms such as speech therapy, SLT, speech and language pathol* SLP;
communication disability and similar terms such as speech and language delay/ disorder/disability. The following databases were selected for use: ERIC,
CINAHL, PubMed, Google Scholar and Google.
Wright et al. (2015) noted a lack of guidance on the optimum selection for search databases. As such, combination of databases were used to search for specific types of documents. ERIC, CINAHL and PubMed were used to identify peer-reviewed articles in the field of health and education. Google scholar and Google were used to access more general “grey literature” documents such as conference reports and NGO annual reviews. Published and peer-reviewed articles were searched for on academic databases. The RCSLT’s Bulletin was also searched from the online available editions (2003-2017). Titles and abstracts were examined for relevance. If the contents of documents met the criteria of describing Uganda plus either speech and language therapy (and synonyms), and communication disability (and synonyms), they were included for analysis. Grey literature was searched using the search engine Google. The first 30 pages of each Google search were explored to locate potentially
relevant sources to allow for the focused search of documents. As with the academic search, titles and introductory content of search results were read and included for analysis if they included Uganda and either speech and
language therapy, and synonyms, or communication disability, and synonyms.
The manual literature search was conducted on a personal collection of documents regarding PWCD in Majority World countries belonging to the Director of Studies. This consisted of published and peer-reviewed articles and grey literature. These documents were read for the same key terms and
inclusion criteria described above. Inclusion of both manual document searching and general web searching aimed to reduce publication bias (McAuley et al., 2000).
As the literature review revealed low numbers of documents relating to the topic area of outsider support for services for PWCD in Uganda, the inclusion worked on the basis of relevance to the topic. As such, all types of documents were considered, regardless of date. This was because searches were not expected to yield results prior to 1980, and earlier documentation was unexpected yet would have been new information based on the existing narrative of services for PWCD in Uganda. Only English-language documents were included. Duplicate or near-duplicate documents were excluded. Where documents were near duplicates, the first published or produced document would be included. Where documents had been published but had become inaccessible, authors and publishers were written to ask for permission to access documents for use in the research.
Documents
The resulting documents eligible for analysis were textual documents that described services for PWCD in Uganda, and outsiders’ contributions to these. The aim was to focus on the topic of outsiders supporting services for PWCD in Uganda. Data were included from a number of different types of secondary sources in order to represent the range of perspectives. Secondary documents were predominately from outsider perspectives and efforts were made to
include insider authored documents where possible. Data sources consisted of: academic journal articles; annual reports; university policy documents;
newsletters; national policy documents; project reports; university recruitment materials; volunteer adverts; NGO documents; webpages; blogposts.
Ethical considerations
Documents were in the public domain. Both webpages and blogs were assessed to consider the privacy level (Eastham, 2011). Furthermore, the names of specific individuals mentioned in these documents were not disclosed in the write-up or discussion, nor were individuals identifiable.
Analysis
All documents were analysed using content analysis. Content analysis is the most typically used approach for analysis of documents and is consistent with an ethnographic methodology (Bryman, 2012; Altheide, 1987). Content analysis can be either quantitative or qualitative. Hsieh and Shannon (2005) identified three main types of qualitative content analyses: conventional, directed, and summative. According to Hsieh and Shannon’s (2005) descriptions:
Conventional Content Analysis is useful where there is limited literature or
theory. It is an inductive approach, in which the researcher is immersed in the data. It avoids a preconceived coding framework or schedule. Instead, the researcher familiarises themselves with the text, making note of their initial thoughts. This process continues and the thoughts are used to develop a coding scheme. These codes are then arranged into categories of related and linked meanings. Finally, categories group codes into “clusters.”
Directed Content Analysis is best used for describing a phenomenon where
there is existing literature or theory, and allows for the validation or extension of a theory. It uses a coding framework which is developed from the existing theory and applied to the text. Where the codes from the framework cannot be applied to text, a new code is then assigned.
Summative Content Analysis uses an initial word/phrase search and count to
calculate the frequency of the words/phrases in a text, as well as interpret the context and ways in which they are used.
Conventional content analysis was identified as being the most appropriate analysis approach for these data due to the relatively small amount of literature relating to Ugandan services and the lack of a clear existing theory about these services. Conventional content analysis has been used by a number of
researchers (Blundell, 2012; Benner, 2013) for the analysis of documents, and allowed for a richer understanding of the phenomenon Hsieh and Shannon (2005).
Coding strategy
Documents were first categorised into types and numbered. Using an inductive approach, all texts were read to make note of first impressions and ideas. After all texts had been read, notes were used to create a coding schedule
developed from key words and concepts from the texts and the researcher’s interpretation and reflection on the documents (Hsieh and Shannon, 2005). This coding schedule was applied to all the texts. The texts and documents were analysed critically and skeptically (considering what was not mentioned in the texts, what was implied, and potential underlying themes or subtext (Rapley, 2007)). This approach was important as Rapley (2007:115) noted: ‘descriptions are never neutral but produce a specific version or understanding of the world.’ Once all texts were coded, codes were arranged into subcategories and
categories organising by meaning.
4.3.2 Method 2: Interviews
As outlined in 3.2.4, semi structured Interviews were selected as the most appropriate method of data collection to respond to the second and third research objectives.
Participants
Inclusion criteria was guided by the second and third research objectives: to interpret insiders’ perceptions and experiences of the contributions of outsiders and outsider support to the development of services for PWCD in Uganda; and to interpret outsiders’ perceptions and experiences of their contribution to the development of services for PWCD in Uganda. Therefore inclusion criteria of participants was people who had experience of services for PWCD in Uganda, both insiders and outsiders. Insiders were defined as Ugandan nationals,
outsiders were defined as non-Ugandan nationals. Insider participants needed either to have worked or trained as part of services for PWCD with outsiders, or be parents or carers to have been a recipient of such services. Outsiders
needed to have worked in services for PWCD in Uganda. Furthermore,
participants needed to be competent enough in English to speak and take part in an interview (as a first or other language), and able to give voluntary fully informed consent.
Sampling
Purposive sampling was used to sample participants who were representative of the target population. The planned sample size was 30 (15 insiders, 15 outsiders).
Within this sample number, there were specific target groups of desired participants: aiming to capture the diversity of different roles in the target
population. Table 4.2 shows the target populations and planned sample size for both insiders and outsiders.
Table 4.2 Participant interview sample
These groups were chosen to reflect the different individuals involved within the development or reception of services for PWCD in Uganda, and the relative sizes of each groups’ population. More outsider SLTs were planned for the sample as their roles included clinically practicing and lecturing on the SLT programme. The total planned participant size was appropriate due to small
Insiders Sample size Outsiders Sample size
SLTs 3 SLTs 9 Student SLTs 3 SLT Mentors of SLT graduates 3 University SLT programme staff
3 NGO coordinators involved in SLT programme
3
NGO coordinators involved in SLT programme
3
Carers of PWCD who had received SLT
target population sizes and the fact that this research was for a PhD (for which a medium-sized sample was likely to suit the time constraints, while still
providing enough data to analyse effectively (Baker et al., 2012). Flexibility was allowed for, as the sample size was to be responsive to emerging data and the possibility of previously unknown participants coming to light. For example: although visiting practitioners were not part of the planned target groups, this group emerged following the beginning of the interview process and was later included.
In keeping with the ethnographic case study methodology, ethnographic principles were used to determine sample size. Ethnographic research often aims to reach a point of data saturation: a point at which sufficient participants have been observed or interviews carried out, that no new themes emerge (Guest et al., 2006). Sample size is often a factor in achieving saturation, as is the diversity of the population (Mason, 2010). Although the small sample size could have appeared to preclude saturation, the sample formed a relatively large proportion of the population of insiders and outsiders, considering the small number of potential participants involved in SLT services in Uganda.
Recruitment
Participants were recruited for interviews in four ways. All participants received participant information sheets (PIS) and an interview topic guide (ITG) which described interview topic areas (see appendices 4 and 5).
1. Direct contact. An outsider-led CPD training event for SLTs and allied professionals was hosted in a Ugandan University. Here, the research was introduced and interviews were arranged and scheduled. Participants had the opportunity to ask questions about the research.
2. Direct contact. Email was used to contact potential participants who were identifiable (and contactable) online due to their public roles in relation to services for PWCD in Uganda.
3. Snowballing. Potential participants were asked to suggest and identify other potential participants who met the inclusion criteria. PIS and ITG were given to
the initial participants to pass on to potential participants. Following contact with new potential participants, interviews where then offered and arranged.
4. Via gatekeeper. A participant from the CPD training acted as a gatekeeper. A PIS and ITG appropriate for parents or carers (see appendices 4 and 5) were given to the gatekeeper who identified and accessed potential participants who were parents and carers of PWCD. The PIS and ITG were developed for participants with limited literacy skills. For these individuals, the PIS was also read out to participants.
After providing PIS to potential participants, all participants were given a minimum of 24 hours before they were contacted again by phone or in person to ask if they wanted to continue. If, after this period of consideration,
participants still wanted to take part in the research, consent forms were signed (see appendix 6) prior to interviews (either in person for face to face interviews, or digitally for phone or Skype interviews).
Ethical considerations
Issues of power were important to consider as an outsider. Although there are potential difficulties in using gatekeepers to access participants (Sixsmith et al., 2003), this measure helped to minimise potential coercion to take part in the research as a result of power dynamics between themselves and an outsider researcher. One participant from the CPD training session acted as a
gatekeeper to recruit and gain consent of parents or carers of PWCD.
Furthermore, the researcher’s Director of Studies (DoS) knew a number of people involved in supporting services for PWCD in Uganda through her role as the project lead on the mentoring project described in 4.2. It was important to ensure participants did not feel obliged to take part in the research. Participants were given PIS and reminded of their right to not take part in the research.
Informed consent was achieved by providing details of the research, how data was to be collected, and how it was to be used in a clear and comprehensible way to potential participants (Sanjari et al., 2014).
The provision of ITG gave participants an overview of the topic areas to be explored and aimed to reduce the power discrepancy between the roles of interviewer — participant (Allmark et al., 2009). As described in 3.3, measures were undertaken to ensure participant confidentiality. This included
anonymisation of all interview data.
Data collection procedure
Interviews were scheduled to take place in venues which were convenient and safe for both the researcher and participant. MMU lone worker policy was followed (MMU, 2006). Interviews took place across a number of settings, including a University campus, participants’ homes and participants’ children’s school. The majority of the interviews with insider participants took place in Uganda, with one interview with a Ugandan participant conducted from
overseas over the telephone. Two insider participants wanted to be interviewed in person together and this wish was respected. Outsider interviews took place face-to-face in Uganda or in the UK, or over Skype. Semi-structured interview data were collected over a 15 month period from March 2014 to June 2015.
Both insider and outsider interviews were scheduled to last approximately an hour and followed the same procedure. The interviewer used topic guides (see appendix 5) to allow for focused and structured questions to lead to purposeful conversations about outsiders’ contributions to sustainable services for PWCD in Majority World countries. Interviews were structured to begin by thanking the participant for their time, and then reading through the purpose of the interview and interview facts. Participants were again reminded of their right to withdraw at any time. Following this initial information, recording began using a
dictaphone (Olympus Vn-731pc). Brief notes were made to follow up
interviewees’ comments in further detail, so as to not interrupt the flow of the interviews. On completion of each interview, participants were asked if they had any further comments or questions for discussion then thanked again for their participation.
Audio files of the interviews were uploaded onto the researcher’s password- protected laptop then deleted. Audio files were played using Quicktime (version v10x) and transcribed verbatim into a Pages document (version 5.2). Where
speech was unintelligible after 5-10 listens it was not transcribed and marked by “XXX” in the transcript. Paralinguistic features such as body language, gestures and tone of voice were added (in brackets) to accompany the transcription for face to face interviews. Participant names and identifying information were not transcribed to maintain confidentiality. Where participants wanted to speak off the record, conversations were not transcribed or used as data (so as to respect participants’ wishes). This happened in two interviews.
Data analysis
Thematic network analysis (TNA) was identified as the most appropriate method of analysis. The method of data analysis needed to be consistent with the overall philosophical approach of the research. Qualitative analysis is used to interpret data and themes to aid the understanding of the phenomena being explored (Sargeant, 2012). A number of analysis protocols were considered.
Thematic analysis is an umbrella term for method for the identification, analysis and reporting of patterns in data (Braun and Clarke, 2006). There are different types of thematic analysis: inductive, theoretical, experimental or constructive (Braun and Clarke, 2013). Although thematic analysis is commonly used in qualitative research, there is a lack of information and agreement about what it is and how it is done (Vaismoradi et al., 2013). Thematic network analysis (TNA) is a method for analysing data to arrive at themes ordered in a hierarchy of web like constructions called networks (Attride-Stirling, 2001). This method of analysis was developed as an interpretive tool, with clear procedures which may be lacking in other types of thematic analysis (Attride-Stirling, 2001). TNA was selected as an appropriate approach because, as a novice researcher, TNA offered a transparent approach for the analysis of data, which would in turn enhance replicability of the research.
Following the complete transcription of interview data, the Pages document was exported into NVivo (version 10.2.2). Each transcript was then was read and re- read to allow for the familiarisation of the text, then analysed according to the 6 stages of TNA. The process of the TNA approach used is summarised below using the stages described by Attride-Stirling (2001):
1. Coding
Separate coding frameworks were devised for both insider and outsider interview data. These were created using recurrent topics from the data following familiarisation with the texts. Codes from the framework were then applied to sections of the transcript: sometimes passages, sometimes just a few words, depending on the content. Insider data were coded first, following the first visit to Uganda.
Subsequently, the remaining insider interview transcripts were collected during the second visit to Uganda, and were coded following their collection. The outsider interview transcripts were coded following the collection of the outsider interview data set. Analysis of insider and outsider transcripts followed the same