‘We are not Robots!’ Nurturing the Enumerator-Principal Investigator Relationship for Improved Data Quality

The treatment arm includes three core components delivered during training and active data collection:

1. PI code of conduct for interaction during training
2. Enhanced ice-breaking session.
3. FO Champions


1. PI code of conduct for interaction during training

The PI code of conduct outlines particular ways in which PIs will promote flattened hierarchy while demonstrating professionalism during the training. These should be easily replicable.
The PI will not be present throughout every part of the two day training sessions, but will be an active presence during specific sessions at which they will feature.


2. Enhanced ice-breaking session.


Training normally involves a general icebreaker and a top-down networking event followed by going over the instrument multiple time. The intervention involves additional ice-breaker activities at training designed to reduce internal hierarchy, build trust among FOs, and improve communication between FOs and PIs.


3. FO champions.

Each small team (8-9 FOs) will have one FO champion. The champion’s role is to facilitate communication with PIs during training and fieldwork, surface questions and issues, and represent team concerns.

The treatment arm includes three core components delivered during training and active data collection:

1. PI code of conduct for interaction during training
2. Enhanced ice-breaking session.
3. FO Champions


1. PI code of conduct for interaction during training

The PI code of conduct outlines particular ways in which PIs will promote flattened hierarchy while demonstrating professionalism during the training. These should be easily replicable. They will:

Not using mobile phone during sessions
Take notes (laptop or on paper) when people talk
Sitting at a table with those being trained rather than being at the front of the room (and changing tables at different points)
Make eye contact with participants (as appropriate, for example they should not stare)
Be positive and empathetic in tone of voice and in responses to issues raised.
The PI will be present for tea / lunch breaks.
Help answer any “study” questions.

The PI will not be present throughout every part of the two day training sessions, but will be an active presence during specific sessions at which they will feature.


2. Enhanced ice-breaking session.


Training normally involves a general icebreaker and a top-down networking event followed by going over the instrument multiple time. The intervention involves additional ice-breaker activities at training designed to reduce internal hierarchy, build trust among FOs, and improve communication between FOs and PIs.

The ice-breaker will involve two components. Firstly, the PI will introduce themselves along with the FOs within the ice-breaker questions. Secondly, the ice-breaker prompts during the intervention will also include a “research challenge” / “strangest thing that happened to me during research” or similar.

Secondly, at the beginning of the read-throughs there will be an activity with the PI taking a central role. This will involve a role-play exercise with a “tricky participant” who consents to participate but then is hard to engage with the survey. Here the PI plays the role of the field officer, and FOs are invited to come up and be the “tricky participant” who presents problems. Prompts on the slides suggest that this might involve (grumpy/distracted/arguing). To prevent the risk of “stagefright” this will not be the first read through exercise that the FOs undertake. This will follow several read-throughs as is carried out in the control group. It is likely that the PI will not be able to circumvent all problems, and in doing so might look foolish or certainly more human. This breaks down barriers between the PI and FOs. The format will be:

PI: Hi everyone. As you are all used to, we have been running through the script together. It is going well One of the things we need to prepare for are participants who consent to take part, but then get distracted, or difficult. This can be hard! You all know about these kinds of situation. We are going to do a roleplay. But you can be the “difficult participants.” Here are some ideas of tricky participants I have seen up on the slides. But also you can act out a different example of a tricky participant if you have experienced one. I will be the FO who tries to complete my survey. Let’s see if i can do it. Feel free to be as tricky as you can!





3. FO champions.

Each small team (8-9 FOs) will have one FO champion. The champion’s role is to facilitate communication with PIs during training and fieldwork, surface questions and issues, and represent team concerns.

The idea of FO champions can be introduced during a short training session by the PI.
“We know that sometimes it can be difficult to raise challenges, problems, and concerns with PIs and with senior research personnel. So we will be working with “FO Champions” whose job it is to represent the views of fellow FOs and to raise any challenging issues directly with PIs. They will be on your side, and it is their job to discuss things with PIs that might not normally get discussed.

We think that FO Champions will be good communicators, will represent your views, and will be honest. They might have a great deal of experience. Or they might be quite new to work as an FO. Either way, they need to represent you if you notice and problems or concerns during data collection.”


After selection, FO Champions will receive a short additional meeting with the PI and research manager about their role and responsibilities. This will be based on the responsibilities outlined below. It could include example illustrations. It will address the incentives that FO Champions have to “stay quiet” about problems by letting FO Champions know that the PIs want to hear about problems big or small. This meeting will also address protection they will receive from saying difficult things to PIs.

FO Champions will not be paid anything extra.



Selection process:

1. Small teams of 8-9 will be selected by the Research Team.
2. Each group has a huddle where they get together and introduce themselves in smaller groups.
3. The introduction will recommend specific qualities to all the FOs, as above.
4. After this, everyone can make nominations for FO champions (including self-nomination and nominations from the research team).
5. Each group of 9 then votes for its FO champion.
6. Nominations and votes will happen on paper.
7. The FO Champions will receive an additional session on their responsibilities.

For those who are not selected, we will use the following script

“Don’t worry if you were nominated as an FO champion but not ultimately selected. We can only have a small number, and you can still have an important role in what they are doing by telling them any problems.”


Champion responsibilities:

The role of the FO Champion will be to communicate any issues from the perspective of their team of 9 FOs with PIs. This means they will have a daily check-in report/brief. This could be a one line assessment of “things have gone well.” They will not comment on the performance of individual members (X has done well, Y has not fulfilled their target). They will only comment on any challenges that the team has faced. “People are getting bored of the survey and it is difficult”, “we ask people about AI and no one is interested.”

There will be a brief prompt that both FO Champions and PIs have. The FO Champion should lead any discussion (text or call) but the PI can prompt if required.

1. How is the work going?
2. Are there any challenges with data collection?
3. How are respondents finding the survey?
4. Are your team concerned about any aspect of the research?
5. Does anyone have anything they want to bring up with the PI about the research?
6. Is there anything that needs to be escalated?


Control Arm

Control teams will receive the standard enumerator training and follow the usual fieldwork protocols used in the area. They will not take part in the enhanced icebreaker or have FO champions as part of the intervention.

2025-12-01

2025-12-22

H1. Nurturing the enumerator-PI relationship will improve overall data quality.
H1a. Nurturing the enumerator-PI relationship will improve accuracy.
H1b. Nurturing the enumerator-PI relationship will improve completeness.
H1c. Nurturing the enumerator-PI relationship will improve consistency.
H1d. Nurturing the enumerator-PI relationship will improve validity.
H1e. Nurturing the enumerator-PI relationship will improve question and survey duration.
H1f. Nurturing the enumerator-PI relationship will improve uniqueness.

The first primary outcome relates to accuracy, which refers to the extent to which data is correct and free from errors. Inaccurate data includes anomalous data points that are unusual or implausible. Having extremely high or low values, or numerous spikes might signal improbable observations. One way of assessing anomalous data points is by using standard deviation as a benchmark. The mean of the variable of interest can be calculated over time, and observations that are two standard deviations higher or lower flagged. Other measures of accuracy include checking whether variables such as the date of birth and the age in years align.
The second primary outcome is completeness. It relates to having all the expected data points in the dataset. It involves checking data is complete across time, space, and all units of observations. Furthermore, location data including Global Positioning System (GPS) coordinates is useful for assessing whether enumerators are adhering to protocols requiring them to take location coordinates. Higher than average item non response including refusals or don’t know responses from individual enumerators might indicate poor interview practices. Complete data is essential for deriving meaningful insights.
As the third primary outcome, consistency includes having variable values aligned with expectations. This means that responses must align across different modules, using the same format and uniform units of measurement. For example, dates, phone numbers and emails take should take the prescribed format. Inconsistent data such as dates entered in different formats can cause errors when data is merged and analyzed.
Validity as the fourth primary outcome refers to the extent to which data adheres to the defined rules and constraints. While bench mark value ranges might be pre-defined, checks include whether variable values like age fall within the acceptable range, prices have positive values among others. Invalid data leads to incorrect analysis and thus flawed decision making.
The fifth primary outcome includes question and survey duration. This outcome is based on the average time a question or survey takes taken from survey timestamps. Questions and surveys that take a relatively short time as compared to the average time might indicate that enumerators are skipping questions e.g. not collecting data on household rosters or fabricating data rather than conducting interviews.
Uniqueness, the sixth primary outcome, refers to the extent to which observations are unique i.e., there are no duplicate observations, which cause redundancy. Duplicate observations lead to incorrect analysis and conclusions. Ensuring data uniqueness is important for maintaining data reliability.
This study shall also examine secondary outcomes of interest including productivity, integrity, and work satisfaction. Productivity relates to the number of surveys completed by an enumerator. Integrity encompasses enumerator honesty and transparency in upholding the best data collection practices. Work satisfaction refers to the degree of happiness enumerator feel in their role. It is influenced by factors like the work environment and relationships with other enumerators, supervisors and PIs.

H2. Nurturing the enumerator-PI relationship will improve productivity by increasing the number of surveys completed and submitted to the server.
H3. Nurturing the enumerator-PI relationship will improve integrity in terms of honesty and transparency in data collection practices.
H4. Nurturing the enumerator-PI relationship will improve enumerator work satisfaction.
H5. Nurturing the enumerator-PI relationship will improve enumerators trust in PIs.
H6. Nurturing the enumerator-PI relationship will improve enumerators interpersonal skills including assertive communication and teamwork.
H7. Nurturing the enumerator-PI relationship will improve enumerators intrapersonal skills including self-motivation, self-discipline, self-reflection, and sense of well-being.

Other secondary outcomes comprise potential mechanisms including trust, interpersonal skills, and intrapersonal skills. Trust refers to the feeling that PIs are on the enumerators’ side, and that enumerators feel safe providing feedback. Interpersonal skills focus on how enumerators interact and communicate with other enumerators and PIs and generally contribute to teamwork. Intrapersonal skills involve a more introspective and self-reflective approach. They contribute to self-confidence, adaptability, emotional resilience, and effective self-management. Intrapersonal skills are crucial for maintaining overall well-being in a work environment.

The intervention this study tests involves randomly assigning enumerators to receive direct channels of communication with PIs coupled with field office (FO) champions as an additional component to the standard enumerator training.

The intervention this study tests involves randomly assigning enumerators to receive direct channels of communication with PIs coupled with field office (FO) champions as an additional component to the standard enumerator training.

In each country, all enumerators in both the treatment and the control groups will be on the field for 5 days of data collection. Each FO will target 4-5 participant surveys in a day.


Data collection is divided into different stages for enumerators and tracked with scheduled surveys:

1. Baseline (before training): A Google Form baseline survey will be sent to all eligible field officers from Kenya and Uganda from the database. This will happen before field officer recruitment and will act as the baseline enumerator survey.
2. Daily pulse surveys are sent to the 150 FOs per country throughout the data collection period of 5 days to capture quick day-to-day changes in trust, wellbeing, and other key indicators.
3. Midline: After the first 3 days of data collection, all FOs will complete a programmed midline survey on SurveyCTO.
4. Endline: After the 5th day of data collection, all FOs will complete an endline survey.
5. Follow-up: A follow-up survey will be administered 25 days after the data collection exercise. All FOs will be expected to complete this.

Randomization done in office by a computer using R.

Individual.

No

300 individulas

300 individuals

150 FO's control and 150 FO's treatment.

Sample size calculation based on binary outcome Minimum detectable effect: 0.17, 17% Sample size: 292 Level of significance: 0.05 Power: 0.8 Take up in treatment: 100% Take up in control: 1% Attrition: 5% Outcome proportion at baseline: 0.5 Proportion in treatment: 0.5