Interviews for a monitoring report
- Interviews should be conducted with attorneys, prosecutors, judges, probation officers, forensic doctors, police, physicians and other health professionals, NGOs and women’s advocates, academics, media professionals, and religious leaders. Interviews with complainant/survivors should be done only in circumstances in which confidentiality and safety is the primary consideration. See section on Ethical issues in interviewing survivors of violence. All interviews should address the access which complainant/survivors have to the legal system, any barriers to access, and the effectiveness of remedies. See: United Nations Department of Economic and Social Affairs and Division for the Advancement of Women, Handbook on Legislation on Violence against Women (2010), 3.3.1.
- Interview questions should always be preceded by a short introduction which describes the monitoring project and the contact information of the monitors.
- If applicable, at the beginning of the interview, monitors should state that sources will be kept confidential and anonymous in the monitoring study. Monitors should assure the interviewee that their words will be attributed to a general descriptor such as “doctor in women’s hospital” or “assistant prosecutor.” Descriptors must be general enough so that the respondent cannot be identified in the future. For example, if only one prosecutor is interviewed, he or she should be identified as a “criminal justice system official” or in another similar way.
- Monitors should always conduct test interviews to make sure that the questions are understandable and answerable. It is important to become familiar with the interview questions, or the methodology, in advance of the first interview so that the interviewer is relaxed and can speak conversationally with the respondent. Ideally, the respondent will not feel pushed to answer the questions and will feel a rapport with the questioner. Interviewers must be careful to listen to all answers without appearing judgmental. See: Researching Violence Against Women: A Practical Guide for Researchers and Activists, 2005, Chapter Ten, for detailed techniques for qualitative research on violence against women.
- Monitors should plan how to record the information. Often it is best to interview each subject with a teammate who will write down the answers or enter them into a computer. The teammate can make note of areas that were not covered, and can ask about those at the end of the interview, and can also ask for clarification of points if necessary. After the interview, monitors should go over the questions and responses with the teammate, and correct any errors while memory is fresh. This is important to achieve a quality end product. Monitors should make a list of follow-up questions for this interviewee, if necessary. Monitors should record any problems that occurred, and, importantly, new insights which will inform subsequent interviews.
- Structured interviews are the best way to obtain the same information from every interviewee. However, the answers may lead to other questions not anticipated. If the area is relevant to the project, following the lead of the interviewee may lead to valuable information. Yet, some interviewees may obfuscate or lead the interviewer away from difficult information. Often, carefully formulated follow-up questions, asked immediately or in subsequent interviews, will be necessary in order to obtain all of the facts.
- The questions should be organized in a logical sequence and questions on similar topics should be grouped together. Monitors should consider which questions address sensitive topics, and employ those questions towards the end of the interview, when there is a level of comfort and rapport. In general, monitors should always ask the questions that can be answered easily first.
- “Skip patterns” determine the order of specific questions that will be asked according to certain answers in a questionnaire. These are important in order to avoid the impression that the interviewers are not listening to the responses and to improve the flow of an interview. However, it is also important to make certain that if some questions are skipped on the basis of an answer, others are not inadvertently excluded which may be important questions for that particular interviewee. “Skip patterns” should be viewed as a flexible suggestion for each interviewee and not as a rigid pattern. See: Researching Violence Against Women: A Practical Guide for Researchers and Activists, 2005, Chapter Eight.
Analyzing data from the interviews
- It is important to read each set of interview notes many times during the course of the project. As monitors immerse themselves in the transcripts, they will also be able to identify key recurring themes, complicating factors, and contradictory responses which may require further investigation. There are many online tools (e.g. many eyes) available for structuring and visualizing qualitative data from interview transcripts. These can be helpful in identifying themes and relationships.
- Data coding, or attaching labels to certain themes in the interview, is one way to manage the information. It enables information from different sources to be sorted and compared. Data can be coded by hand with notes in the margins or on the computer. There are no standard rules on data coding; however, coding should not impair the result. For example, attaching a code to a complex thought expressed by an interviewee should not limit the use of that thought in the report. See: Researching Violence Against Women: A Practical Guide for Researchers and Activists, 2005, Chapter Thirteen.
- There are many online tools available for structuring and visualizing qualitative data such as interview transcripts. This can be extremely helpful in identifying themes, patterns and relationships. These tools are often free and they are relatively easy to use. Users simply upload their data to the program and after that visualizations can be made with a press of a button. Tools such as Wordle and Tagcrowd demonstrate simple word groupings, or “word clouds,” whereas Many Eyes and Leximancer provide more sophisticated options for data visualization.
- As data accumulates and is studied, draft conclusions may be drawn. These should be checked against ongoing interviews and revised with further analysis.
Determining findings of fact
- In addition to paying attention to what is said, monitors should think about what was unsaid or left out.
- Monitors should identify and record themes and insights from the interviews.
- List of all of the themes discerned. Group similar themes together and prioritize them in order of importance to the research findings.
- Find connections between themes that don’t seem to be connected at first in order to see the data in a new way.
- Assemble the data for each theme in one place and look for important relationships between the data, and for relationships between the different themes. Look for the most important relationships between people, events, perceptions and behavior patterns.
- Determine what is significant for the research. Look for the most common norms and patterns in behavior, ideas, perceptions attitudes and expressions. Make note of any important differences.
- Begin the process of writing the report.
(Adapted from: Researching Violence Against Women: A Practical Guide for Researchers and Activists (2005), page 205)
For example, in Domestic Violence in Brazil: Examining Obstacles and Approaches to Promote Legislative Reform (2010), the author made the following findings:
- In Brazil, a crime committed in public by a stranger is considered more “serious” than a crime committed in private by an intimate partner. This view was held by the public and also by the authorities, as evidence by the ineffectiveness and lack of response of both the police and the judiciary. p.77
- Citizens did not respect the Maria da Penha Law (2006) because it so often was unenforced. p. 89
- These findings led to the author’s conclusion that the government of Brazil has failed in its duty to protect women as equal citizens. p. 77