OR/18/002 Assessment methodology

Liddle, E and Fenner, R. 2017. Review of handpump-borehole implementation in Uganda. British Geological Survey Open Report, OR/18/002.

As stated in Section 1, this research seeks to understand the siting and D/I process in Uganda and aims to identify any factors within this process that may be adversely affecting the quality of the siting and D/I work, and the subsequent functionality of rural Ugandan HPBs. To address this aim, five weeks of fieldwork were undertaken in Uganda (two separate visits, one in February 2017 and one in May 2017). Three of these weeks were spent in Kampala interviewing key informants who were based in and around the city, while the remaining two weeks were spent travelling to DLGs in the central and eastern regions of the country.

Key informant interviews

Key informants included those within DLGs, CSOs, and the private sector who are involved in the process of planning, procuring, and overseeing the implementation of rural HPB projects. For DLGs, this included employees from the DWO and PDU, for CSOs, project managers, and for the private sector, DCs and consultants. The interviews sought to understand the role these actors play in rural HPB projects, the processes they follow, and any concerns they have with the way things are currently operating. Additionally, employees from MWE, including from DWD, DWRM, and the TSUs, as well as an employee from UDCA were interviewed to help the researcher to better understand how the planning, procurement, and implementation process is meant to work and any concerns these key informants have with the way this process is working in practice.

Eighty interviews were conducted in total:

• DLG employees (across 20 districts: 10 in both the central and eastern regions)
  • District water officers = 20
  • Procurement officers = 16
• CSO project managers = 11
• DCs = 14
• Consultants = 9
• MWE
  • DWD employees = 4
  • DWRM employees = 4
  • TSU employees = 1
• UDCA employees = 1

All interviews were conducted one-on-one with the respondent and these took place in their offices. Between 0.5–1.5 hours was spent with each respondent. All interviews were semi-structured, i.e. a set of interview questions were pre-written (see Appendix 1), although the flow of the interview and the order the questions were asked in developed as the interview advanced. Through using a semi-structured approach, the researcher could mould the interview as it progressed, exploring particular issues or topics that arose during the interview as well as through the duration of the fieldwork. The semi-structured approach was also beneficial as it allowed respondents to openly express their concerns and did not limit them to a structured set of questions. The majority of the questions the researcher asked were open-ended. Open-ended questions allowed respondents to give data-rich answers and meant that they were not forced to answer in a certain way. Close-ended questions were only asked when the researcher needed to clarify a certain point. All interviews were conducted in English by the researcher and a research assistant from Makerere University was present for most of the interviews.

Sampling strategy

Snowball sampling was used to select the MWE and UDCA respondents. Under snowball sampling the sample population is obtained through the use of a few key contacts who then recommended other participants who may have been useful to the study (Henn et al., 2006). In the case of this research, this began with an employee at DWD. From this initial interview, the researcher gained additional contacts, and so forth. Snowballing is commonly used when the researcher does not have access to a sampling frame, whether this be due to the target population being unknown or it being difficult to contact people, as was the case in this research when it came to the MWE and UDCA respondents (Tolich and Davidson, 2011).

For CSO selection, the researcher had a list of all the CSOs registered with UWASNET and those with the greatest WASH investment portfolio for the financial year 2015/16 were purposefully sampled. Purposeful sampling was used as this allowed the researcher to focus on those that are most active, i.e. those that have the widest impact on HPBs across the country. For the DCs and consultants, simple random sampling was used based on a list of all that were operating in Uganda at the time of the research.

Stratified random sampling was used to select 20 DLGs across the central and eastern regions of Uganda. To do so, the total DLG population (55) was divided into three strata based on their MWE reported combined non-functionality rate for shallow wells and deep boreholes in the financial year 2016/17. The strata were labelled low, medium, and high non-functionality. As the average non-functionality rate across the central and eastern regions in the financial year 2016/17 was 16% (range = 2%–44%), any DLG with a non-functionality rate ± 5% of this value was classified as medium non-functionality. Any DLG with a non-functionality rate below this range was classified as low non-functionality, while those above were classified as high non-functionality. Twenty DLGs were then randomly selected, with the number from each stratum being weighted based on the stratum size. Non-functionality rates among the sample population ranged from 4%–37%. Stratified random sampling ensured that a range of DLGs were selected, not only those with high, or vice versa, low, non-functionality rates.

Ethical considerations

Ethical issues were an important consideration in this work. In this, the researcher ensured that:

• Interviews were always booked in-advance via phone or email. Cold-calling at offices was avoided. In these emails/phone calls, the researcher always explained the purpose of the research, the type of questions that would be asked, and the expected time commitment.
• Before any interview began, the researcher would once again explain the purpose of the research, the type of questions that would be asked, and the expected time commitment.
• From having understood the purpose of the research and the interview process, verbal consent was gained from the respondent. At this stage, the researcher asked for specific permission to audio-record the interview. Those who declined were not recorded; hand-written notes were taken instead.
• Where necessary, the researcher obtained permission from the necessary authorities before contacting the specific person of interest. This was especially the case for the DLG interviews. Letters were sent to the respective CAOs in advance and these were stamped and signed before the researcher started any DLG interviews.
• Interview data was treated in a professional manner, ensuring that all confidential information remained confidential and that anonymity of participants was maintained throughout. Respondents were classified by their organisation type, with a number then being assigned to them within this organisation type, instead of using their unique names. Any references to their own organisation (or other organisations) were blanked out in their interview transcripts.

As a part of this process, an ethics application was submitted to and approved by the Engineering Research Ethics Committee, University of Cambridge.