Context

In recent years, about 5.6 million Venezuelans have fled economic hardship and social unrest, fleeing to neighbouring countries such as Brazil, Colombia, Ecuador, and Peru. In the context of this migration journey, women and girls face particular vulnerabilities such as intimate partner violence (IPV), early unions, sexual assault and harassment, as well as forced sex work, and survival sex to access basic goods and services. Additionally, the COVID-19 pandemic is thought to have exacerbated the risks for women and girls for a variety of reasons, including increased financial strain, decreased security, living in closed quarters with potentially abusive partners, and limited mobility due to public health restrictions.

Anecdotal evidence suggests that many affected women and girls are often not able to access medical care and support services along their migration route. In many cases, this means gender-based violence (GBV) survivors face significant barriers to accessing sexual and reproductive health services such as pre-natal care, testing for HIV and sexually transmitted infections, as well as clinical management of rape. Further, COVID-19 has reduced accessibility to many other services.

Project Objectives

Our research collaboration between the International Organisation for Migration (IOM) and Queen’s University is supported by Elrha’s Humanitarian Innovation Fund (HIF). Our objective was to provide more efficient mixed-method data regarding gendered migration experiences, including GBV risks, survival sex, and trafficking so that threats could be more promptly addressed, and the needs of survivors better met. The ultimate goal of this work is to improve the safety, well-being, and sexual and reproductive health of refugee and migrant women and girls from Venezuela through more efficient data collection and analysis allowing for more responsive programming.

Methodology

The Sensemaking Approach

To achieve our objective, we used an innovative ‘sensemaking’ (SM) approach with Spryng.io software. SM is based on the recognition that storytelling is a natural way to convey complex information and is used by individuals to make sense of their experiences. Using SM, participants audio-record a story in response to an open-ended prompt (in this case about the migration experiences of women &; girls), thus generating rich qualitative data. After the recording, participants then interpret their own experiences by ‘plotting’ their perspectives. SM quantifies each of the plotted points, providing statistical data linked to the accompanying narratives. Multiple-choice questions collect demographic information and help to contextualize the shared story.

By collecting many self-interpreted stories, SM leverages the ‘wisdom of the crowds,’ and collectively, the participants' responses create a nuanced picture in the same way pixels come together to produce a clear image.

Advantages to Sensemaking

SM offers several unique advantages:

• First, it provides a more comprehensive understanding of complex issues by using indirect prompting questions to elicit more revealing responses. By avoiding asking direct questions, SM allows stories to emerge from the broader landscape of experiences, thus situating them in the everyday lives of participants.
• Second, in contrast to quantitative surveys, which ask participants to choose between several discrete options, SM provides more nuanced data because it allows for a much more extensive range of possible responses.
• Third, SM reduces social desirability bias because, within a given question, the possible responses are either positive, negative, or neutral, with no one response being more socially acceptable than others.
• Fourth, interpretation bias is also reduced because participants interpret their own experiences.
• Finally, because SM data is collected digitally on hand-held tablets or smartphones and the narratives are audio-recorded, it is a more efficient method of conducting mixed-methods research compared to more traditional approaches.

Implementation Challenges

COVID-19-Related Delays and Implications

The project’s start was initially delayed by a year because of the global COVID-19 pandemic. Originally, the project had a commitment from five countries wishing to participate (Brazil, Colombia, Dominican Republic, Ecuador, and Peru) and with a goal of reaching 6,000 participants in total. However, after the first year of the pandemic, both Colombia and the Dominican Republic were no longer able to participate and so the scope was reduced to three countries with a target of reaching 3,000 participants. During the pandemic-induced delay, many costs had increased particularly those involving fuel and travel. Additionally, the research teams now also required masks, hand sanitizer and sanitizing wipes to clean the tablets between participants in order to prevent transmission of COVID-19. Therefore, the budget originally allotted for the Dominican Republic and Colombia was reallocated for higher transportation costs and for COVID-19 personal protective equipment in the remaining project countries.

An in-person training with the research teams from Brazil, Ecuador and Peru was planned for the autumn of 2021. However, with ongoing COVID-19 travel restrictions and the global impact of the Omicron wave, the training was subsequently postponed to January 2022 and switched to a blended format.

Audio Files

Approximately 80% of all the collected narratives are audio recorded, while the remainder was typed onto tablets. We realized early in the data collection that the quality of the audio files was quite poor due to significant background noise. Most of the data is collected in outdoor public spaces such as border crossings, markets, transportation hubs, and points of service delivery. Therefore, it’s not surprising that there would be lots of ambient sound. We provided feedback to the enumerator team recommending that whenever possible they move away from noisy areas and encourage the participant to hold the microphone closer to their mouth while recording the narrative. Over time, the background noise on the audio recordings did diminish but is has continued to be somewhat of a challenge.We also faced a challenge with missing audio files. Depending on the week, the percentage of missing audio files ranged from 2% to almost 30%, and the working theory is that longer audio files are failing to upload when the internet connection is unstable. To mitigate the risk of losing more audio files, in the later phases of data collection, the enumeration teams tried to upload the data from the IOM office or other location where the internet connection tended to be more reliable. In the last weeks of data collection, the percentage of missing audio files had stabilized in the 2-7% range.

Emotional Toll on Enumerators

The initial training for enumerators had two sessions on psychological first aid including self-care. However, the experiences shared by Venezuelan refugees and migrants who participated in the study were often traumatic and disturbing, and understandably enumerator wellbeing was affected. This was particularly true for enumerators from Venezuela who were themselves refugees and migrants. To mitigate the impact on the mental health of the enumerators, several measures were implemented. These included hosting a second psychological first aid session for the team, encouraging enumerators to reduce the number of interviews that they were doing per day, recommending taking a break between interviews and taking time off as needed. These measures helped and the enumeration team were better able to care for themselves in the second half of the data collection despite the disturbing nature of many of the experiences shared by Venezuelan refugees and migrants.

Transcription and Translation

We had aimed to collect approximately 3000 narratives over 10 weeks and had budgeted for their transcription and translation. However, we far exceeded our original data collection goal and have just over 9,300 narratives. This means there is not enough budget to cover the transcription and translation of all the data. We tried several different routes of transcription and translation to identify a solution and turned to artificial intelligence (AI) options to process the thousands of transcriptions and translations more efficiently and to do so within our budget. The AI transcriptions and translations will be used in our initial mixed-method analysis but any narratives being included in final results presentation will first be verified for accuracy by a professional human translator.

By: Susan Bartels - Queens University, and Monica Noriega - IOM