Non-financial Incentives of Community Health Workers in Guinea-Bissau

1. Intrinsic Motivation:

The treatment on intrinsic motivation has as objective to increase the performance of the agents through an alteration in what we claim is their level of intrinsic motivation, i.e. doing something because it is inherently interesting or enjoyable. We will intervene through a manipulation of task significance level using an first-person recorded interactive video making salient the Social Impact of the CHW task (Grant 2008). Social Impact understood as the extent to which employees feel that their own actions improve the welfare of others. The video replicates daily activities of an agent, and offers different stories that materialize as the CHW plays the video. During the video, the agent faces a single central-interactive decision on how much effort to elicit virtually Depending on the choice, different consequences will come up. Within this treatment, a group of agents also visualize a short video in which a traditional healer endorse the CHW role in the community. In addition, another group of agents visualize a placebo-static video in which they do not have to make any decision.

2. Social Status:

This intervention aim to test the effect of social recognition in the general effort of agents. All agents participating in this intervention who reach an established performance threshold will receive an award in a public ceremony. The prize is symbolic and has no monetary value. On top of the prize and the ceremony, all the households assigned to a winner CHW receive a text message to inform them that such an agent outperforms and won an award.

3. Information Campaign:

All the households assigned to an agent in this treatment receive 3 distinct messages with information about the CHW work. This intervention aims to facilitate the introduction of the agents to the household, and at the same time increase households’ interest and understanding about CHW work.

We implemented 3 rounds of each intervention.

2017-12-01

2018-10-13

1. Baseline and endline surveys of all 1,015 agents working in SAB at the time of the start of the intervention in August 2017.
- CHW’s self-reported measure of motivation
- CHW’s perceptions about social impact and task significance of agents’ activities
- CHW’s self-reported time allocation to CHW activities (visit households, training, monthly meetings) and performance of households visits
- CHW’s involvement in other community activities
- CHW’s labor and educational aspirations

2. Baseline and endline surveys of a random subsample of 2 households per agent.
- Number of CHW visits
- Household perceptions about quality of the CHW, relevance of the information received
- Household trust in the CHW
- Household’s self-reported implementation of the health practices disseminated by the CHW
- Household’s self-reported visits to the health center of kids under 5 and pregnant women
- Household’s self-reported health outcomes on kids under 5 and pregnant women

3. Five minutes phone-calls survey of a random sub-sample of roughly 8-10 households per agent.
- Number of CHW visits
- Household perceptions about quality of the CHW, relevance of the information received
- Household trust in the CHW
- Household’s self-reported implementation of the health practices disseminated by the CHW

4. Administrative data collected by the local counterpart implementing the project. Data collected from this source includes:
- Monthly reports of agents’ activity (self-reported): Number and quality of CHWs’ monthly reports about household visits
- Supervision reports on agents’ performance
- Agents’ Pre-tests and Post-tests scores performance at every monthly meeting before and after CHW training sessions.
- Number of months that each CHW was active
- Drop outs

5. Registry books of patients collected by the research team in every health center, clinic and hospital in Autonomous District of Bissau. The final use of this data in the analysis would depend on the quality of the matching process between patient administrative records and household baseline data.
- Households’ number of visits to health centers
- Health outcomes

4. Registry books of patients collected by the research team in every health center, clinic and hospital in SAB. The final use of this data in the analysis would depend on the quality of the matching process between patient administrative records and household baseline data.

The identification of the causal effects of the incentive schemes and the information campaign is based on random assignment of the treatments at the individual level (intrinsic motivation and information campaign) and the neighborhood level (social status). By comparing different groups of individuals, we can isolate the effects of each intervention and shed light on the fundamental questions posed above, without being plagued by methodological problems related to endogeneity of incentives and information.

The interventions in this project are as follow:

1. Neighborhood-level treatment assignment allowing the identification of the effects of Social Status Incentives on CHWs’ performance (all SAB-76 neighborhoods):
(a) Control Group (C) (38 neighborhoods)
(b) Social Status Treatment (TSS) (38 neighborhoods)

2. First level of agent-level treatment assignment within neighborhoods in groups C and TSS: We will randomly assign all agents in the 76 neighborhoods to one of the following 2 interventions:

(a) Individual level control group (Ca). These agents will have no intervention related to information campaign (508 CHWs).
(b) Information Campaign (TIC). Households assigned to CHW in this group receive information about CHW program through a set of text messages (507 CHWs).

3. Second level of agent-level treatment assignment within neighborhoods and Information Campaign Treatment groups.
(a) Intrinsic Motivation control group (CIMa): These agents will have no intervention related to intrinsic motivation incentives (254 CHWs)
(b) Intrinsic Motivation Incentives (TIM)-Placebo: Agents are exposed to a static version of the interactive video in which they do not have to make any decision (254 CHWs)
(c) Intrinsic Motivation Incentives (TIM): Agents are exposed to an interactive video (254 CHWs)
(d) Intrinsic Motivation Incentives (TIM)-Traditional Healer. Agents are exposed to an interactive video and to a message from a traditional healer supporting CHWs activities (254 CHWs)

The randomization was done by a computer using Stata.

The first unit of randomization is the Neighborhood, and the second unit of randomization is the CHW (see details above).

Yes

76 Neighborhoods and 1,015 CHWs.

1,015 CHWs, ~ 2,030 households interviewed in person and ~ 3,000 households interviewed by phone

Pure Control group: 63 CHWs
Only TSS: 64 CHWs
Only TIC: 67 CHWs
Only TIM-Placebo : 60 CHWs
Only TIM: 68 CHWs
Only TIM-Traditional healer: 64 CHWs
TSS+ TIC: 60 CHWs
TSS+ TIM-Placebo: 67 CHWs
TSS+ TIM: 59 CHWs
TSS+ TIM- Traditional Healer: 63 CHWs
TIC+ TIM-Placebo: 64 CHWs
TIC+ TIM: 63 CHWs
TIC+ TIM-Traditional Healer: 65 CHWs
TSS+TIC+ TIM-Placebo: 63 CHWs
TSS+TIC+ TIM: 64 CHWs
TSS+TIC+ TIM-Traditional Healer: 61 CHWs

Pre-analysis plan