Our supervisors keep everything together, ensuring data collection is as smooth as possible.
Our field workers are the backbone of what we do.
Outline India believes in ensuring data quality at every step of the research cycle, right from study design, through data collection, to dissemination.
Our focus is on data quality improvement, not just assurance. Standard data quality assurance protocols focus on the obvious gaps, incomplete information, the inconsistencies and the missing data points. However, our emphasis is on ensuring the reliability, accuracy, and relevance of data.
We ensure data quality by the following:
We qualitatively pre-test survey tools in a location with a similar socio-demographic profile to the study site. This helps us check consistency, appropriateness of translation, relevance, and context, as well as uncovering inconsistencies.
Through our learnings from the pre-test, we prepare detailed training manuals for each study in consultation with the client, to ensure clarity and standardization in data collection.
Field workers are trained on the background of the evaluation, a question-by-question overview of the survey tool, cultural sensitivity, and ethical research considerations, technical training, obtaining informed consent, data security and data transfer. Training includes field mocks and debriefing sessions so that we can resolve any outstanding issues before data collection begins.
Post the training period, our researchers stay on the field for initial days to monitor each field worker, clarify doubts, address linguistic and comprehension inconsistencies and implement the sampling strategy. The Field Manager continues to stay on the field for additional days to provide ongoing support.
We make sure to regularly obtain ethical approvals for our projects. We ensure the privacy of our respondents, and thereby take all the necessary precautions to keep their responses confidential.
We use digital data collection devices like Computer-Assisted Personal Interviewing (CAPI) platforms to conduct face-to-face interviews. Through CAPI we incorporate hints and instructions for field workers, record and visualize the location of interviews, take pictures, switch languages and monitor data in real-time.
Before each project, we put together project-specific field teams who have cultural and linguistic familiarity along with relevant domain experience. The finalized teams are established from their performance post-training by our researchers, based on standardized performance assessment protocols.
There is a ceiling limit to the number of interviews that a surveyor can conduct in a day, to ensure quality. The duration of each interview is recorded and included in the final data set.
Each survey form and data point goes through three levels of approvals and quality checks – Field Supervisor, Field Manager, and Data Manager. We can troubleshoot inconsistencies and errors, and re-train/debrief individual field workers in real-time. Additionally, we conduct telephonic and in-person back-checks.
For qualitative data collection, we re-visit recordings and field notes to ensure that transcriptions are accurate and context is provided to make sense of them.
After each study we put together a field report: detailing data collection context and progress, documenting field definitions and assumptions and providing recommendations.
We work at the confluence of theoretical research and the field with all its messy realities. Things often don’t go as planned.
This is our attempt to document the strange, sometimes funny and sometimes disturbing experiences we have had in the field and the hard-won lessons we have learned from them.
“How do you clean your hands?”
“With mud”, he answered.
“Could you show us?”
He
walked over to the tap, diligently scrubbed his hands with water, applied soap
and washed it off. No mud ever entered the process.
It
was a study to understand hand hygiene practices, and we were using a survey
tool, enumerators’ observations and spot checks to understand how people washed
their hands. The rationale for using multiple methods was to check how people
washed their hands, without trying to impress the enumerator or changing their
behavior because they were being watched.
As
this example shows, different tools found different answers. Yet it was
difficult to reconcile the findings during analysis as there was no
accompanying qualitative data to explain why the respondent would say one thing
and do another. Luckily, during the pre-test, our researchers had noticed this.
We found that respondents, who had interacted with peer educators telling them
to wash their hands with soap and water, thought they were being asked to show
the ideal way to wash hands and not how they normally do it.
“What's your age didi?”
“17”, she giggles, her grey hair
glinting in the sunlight.
“Didi are you sure you are 17?”
“Yes, Yes”
Concepts
of time and space in some rural areas aren’t the same as Western ones. Yet
almost every survey you come across asks the question - ‘what’s your age?’ Over
the years, we have developed strategies to answer this question.
For
young children, we check their government-issued MCP card. If they don’t have
one, we ask if they were born before or after the most recent local natural
disasters and then make an educated guess. For adolescent girls, we ask how
long ago they started menstruating, and use the average age of menstruation to
calculate. For men, we ask family members and neighbors.
And
for the didi we met in Bihar, we asked if she was alive when India became
independent? Turns out she was.
Respondents don’t just stay at the field
site – they move around over time. Especially in rural areas, members of the
household often migrate to nearby cities or urban areas in search of work. But
what happens when we are required to conduct a baseline, midline as well as an
end-line study in the same location with the same respondents? You may not find
your respondents in their home all year round. And this is exactly what
happened to us!
We
were conducting an end-line study in Rajasthan in November, but just couldn’t
find the baseline respondents we had surveyed earlier. We knocked on their
homes only to find from the family members that the original respondent had migrated
either to Gujarat to harvest cotton or to Punjab to sow wheat.
Lesson learned: Look at migration patterns and cropping cycles before
deciding on study phases!
A loo isn’t supposed to be pretty but
this one was. A bright yellow building, newly painted, it gleamed in the sun.
The teacher who was guiding us pointed it out proudly – “that’s the girls’
toilets” she said.
“Can
we go in?” we asked. “It’s locked” she stammered. “Don’t you have the key?” we
asked, wondering why anyone would lock a toilet. By this time the School Principal had joined us. We told him it was a beautiful building and could we
look inside? Again he didn’t seem keen, but he unlocked the building.
And
inside was a bare floor, no latrine, no tap, not even a hole.
It was during an interview with a
school principal, a well-respected man in the community. The respondent was all
too happy to cooperate, answering our researcher’s questions at length. But in
the middle of the interview, he paused and grabbed her hand, explaining that he
holds the hands of disobedient children. But he didn’t let go.
She
finished the interview, extracted her hand and left with her colleagues. But it
shook us all up, an unnerving and unwelcome incident. For the remaining
interviews, we asked a male field worker to accompany our female staff, and
there were no more incidents. But as an organization, we do vehemently defend
and uphold our independence and work towards making the development sector
gender-neutral. This was a sad setback.
She had been following us for hours.
Survey after survey, a steady shadow that dogged our steps despite the mid-day
heat, at the height of the Delhi summer. “Didi, why are you here?” we finally
asked. “Survey me as well” she answered.
She
wasn’t one of our randomly selected respondents. The survey was long, almost
two hours, and we normally had to beg respondents to take it, not fend them
off. “Didi, why do you want to be surveyed?” we asked. Her answer was garbled
with her passion but with some help from her neighbors, we finally got the
story.
A
couple of years ago, another set of researchers had visited and administered
surveys. Like us, they were randomly choosing respondents. But unlike us, it was
for the baseline of Randomized Control Trial where the selected respondents
were given monetary and technical help to construct houses. She hadn’t been
selected, but her neighbors had. And she had watched over the years, as they
built their fancy homes while she was forced to live in her shack. Determined
not to be omitted from a survey again, she now makes sure that surveyors
include her.
“Aapke ghar mein kitne purush rehte hain?”
“Purush?”
“Kitne aadmi rehte hain?”
She glared at us.
Did we make a mistake? Yes, we did. In Rajasthan, the
word aadmi denotes “husbands”, not men. For our respondent,
the question translated to “How many husbands do you have?” No wonder she was
offended.
We
wanted to know the number of male members living in the households. Finally,
after apologies, some subtle probing we got the number.
Lesson
learned: There are more linguistic variations than what we might be aware of.
So be wary while translating!
The village was,
as is often the case, remote – 3 to 5 kilometers from the nearest road. Coming
in we were met with stares instantly, our guide, a local ASHA, told us that
outsiders were rare. This wasn’t unique but odd to us. The villagers followed
us around, tense, listening carefully to what we were saying in our unfamiliar
accents. Our study – on the topic of sexual
and reproductive health – was a sensitive one and we were to talk to young
adolescents in groups by themselves. The village members were clearly not happy
with this, and although the village’s ASHA and Mukhiya supported us, we were
not welcome. We ended up having a small group discussion with the few
adolescents whose parents were comfortable. We left quickly, the villagers
followed us to ensure that we were gone.
“Why were we met with so much hostility?”
We asked the ASHA, the Mukhiya, and the few friendly respondents. Soon the
story came out – there were rumored cases of outsiders luring children away
from a neighboring village, and harvesting their organs. Us, strangers coming
in and wanting to speak privately with children had unknowingly triggered the
villagers’ fears.
For institutional review boards, the
ethics are clear – you go to a village, you get informed consent, tell the
respondent the risks and benefits, talk to the respondent privately so that
they are not ostracized for their views and then you leave. We had followed
these best practices and more, going ahead and talking to the ASHA and Mukhiya,
explaining our study and gaining their support. But the backlash still
happened. How do you get data from a place that doesn’t trust you?
A question on the labor status of an
individual seems pretty easy to explain when looked at from the outside. But as
you delve deeper into it, you discover the multiple layers. In a training
session that we conducted, while giving examples on how to categorize an
individual's current labor status,
one field enumerator asked, "What is the kind of employment of a
priest?"
Another
enumerator answered, "Self-employed".
This
led to a long drawn discussion among the field enumerators on how it is neither
of the above and the ambiguity had to be resolved with the following
explanation, given the context of the assignment. A priest employed by a temple
trust and paid a monthly salary will be categorized as a "salaried
employee". On the other hand, a priest who conducts ceremonies as and when
required at houses of people or other places will be marked as
"other" in the options list and be specifically listed as a "daily
wage earner”. This simple question of a field enumerator led to a very
important probe being incorporated in implementing the survey instrument as and
when such a response is received.
Lesson
learned - Even the simplest of questions come with its own set of connotations
and may require in-depth probing to arrive at the desired answer.
We are taught to
expect the unexpected when on the field. However, little did we expect what befell
us in Araria, Bihar. Since our guide contained sensitive questions and the
respondents were 15-19-year-old girls, we requested privacy to conduct our
focus group discussions in the local primary school. However, the next day when
we returned to the school, we were surrounded by a crowd of agitated villagers
who refused to let us continue for fear that we may kidnap their children.
Eventually, after we managed to calm them
down, we thoroughly explained the purpose of the study to them. However, the
situation had spiraled to such an extent that we had to exit that village and
drop it from our sample. It was only later that we found out that the reason
for such animosity towards outsiders stemmed from an incident that had occurred
a couple of months ago. Apparently, two young girls of the village were taken
to Mumbai on the pretext of getting employment. Instead, their organs were
harvested and they were sent back to the village on a train.
Lesson learned: We should have entered the
village a day prior to our interviews and explained our study to the Pradhan.
This would also have helped us gain the confidence of our respondents.
The first step to
smooth data collection is to develop a well thought out and sound field plan.
However, even meticulous planning cannot always prepare you for the curve-balls
that are thrown your way when you are on the field. In such situations,
thinking on your feet and improvising are the only ways to ensure that data
collection progresses unhindered. Improvisation was our greatest tool in
completing our fieldwork in North 24 Parganas district of West Bengal.
However, we did learn a valuable lesson.
The greatest take away from our field
experience in West Bengal was to never enter the field without all prerequisite
permissions and clearances. The study was to understand the role played by
elected representatives of Gram Panchayats in local governance and development.
Due to the politically sensitive nature of this study in a politically volatile
environment characterized by the polarization of parties, elected representatives
in many of the Gram Panchayats expressed distrust, and suspicion towards the
motives of such an exercise, refusing to consent to be surveyed in the
absence of formal permission from local authorities. As such, we faced great
difficulty on the field in meeting the initial sample requirements and had to
modify our sampling strategy mid-way which had many operational and budgetary
implications.
Lesson Learned: It is imperative to get
requisite permissions from local bodies, as far as possible, before entering
the field for data collection. While this may not be necessary for household
studies, it is absolutely vital for studies that are politically sensitive.
Often while
doing fieldwork, we come across respondents who don’t understand the
questionnaire and tend to interrogate more about the same. They usually find questions bizarre and challenging.
Subscribe to our newsletter