Increasing Participation in Community Dialogue Workshops in Yemen

Intervention #1 (cluster randomization):
a) A behaviorally informed recruitment flyer that uses a reminder, pro-social motivation, the salience of safety, scarcity, and the provision of a map result for the purpose of having a higher proportion of individuals coming to register for the workshop compared to a flyer without these elements.

Intervention #2 (individual alternating assignment [quasi-experimental]):
a) A registration form with a commitment to attend and a prompt to place the form at home as a reminder.
b) SMS reminders before each of the 3 sessions.
c) Writing down their learning next to their attendance poster after each session to remind them of their public commitment to continue attendance for all 3 sessions.
d) A public commitment to continue to work on their team projects.

2018-06-26

2018-08-31

Intervention #1:
a) Average number of individuals registered or not from each block.

Intervention #2:
2a) Proportion of registrants attending each session
2b) Average number of sessions attended per participant.
2c) Proportion of participants attending the final project pitch in July.

Intervention #1:
Different city blocks (N = 41) randomized to receive different version of the recruitment flyer (cluster randomized) and they will be directed to one registration location if they receive the treatment and another if they receive the control flyer. The number of registrations from each block will then be calculated and this will be used to compare the treatment with the control blocks.

Intervention #2:
The attendance at each session will be tracked. In addition, attendance at the final project pitch in July will be collected for 2c. Attendance will be collected by project staff (not self-reported).

Phase 2, Intervention #2:
2d) Proportion of participants who present (participate in) the project pitch.

Participants may attend the event but might not be a part of a project pitch. Thus we will secondarily investigate if the intervention (#2) affected the probability that individuals participate in the cohesion project pitch.

Intervention #1: Three districts were divided into a total of 41 city blocks. Those blocks were then randomly allocated to treatment and control using the district and the location of the registration as stratification variables.

Intervention #2: Individuals who registered were then alternately assigned (quasi-experimental) at the individual level to receive either the treatment registration and session components listed above or to receive the business-as-usual sessions. Alternation will be stratified by gender and whether or not individuals came in a group or on their own.

Intervention #1: The three districts in Aden were Crater, Tawahi, and Alma'ala. The blocks were drawn on a map by locals who knew the context. This generated 14 blocks in Crater, 15 in Tawahi, and 12 in Alma'ala. Secondly 2 safe registration locations were selected for each district. The research team created a stratification variable by interacting the district with the existence of a registration location, this were used for stratified randomization of the city blocks. The city blocks were then randomized by the research team and this was reported to the local partner for distributing the treatment flyers to the treatment blocks and the control flyers to the control blocks.

Intervention #2: At each of the six registration locations (2 per district), individuals will be alternately assigned (quasi-experimental to simplify logistics) to either treatment or control conditions for the remainder of the program. Individuals will be directed to either a desk for females, males, or groups at which location they will be assigned to treatment or control using a pre-determined sequence (odd numbers to treatment, even to control). This will be stratified by gender (male or female) because the sessions will be conducted separately for gender (both treatment and control) and whether or not they came in a group that was then assigned jointly.

Randomization sequences will be generated by computer by the research team using randomizr v0.12.0 in R (seed = 20180425). However, the sequence will be followed in person by the staff of the implementing partner and (in intervention #2) will be based on the order in which individuals register.

Intervention #1: City block (cluster)
Intervention #2: Individual (except if they register as a group)

Yes

Intervention #1: 41 clusters (city blocks)
Intervention #2: 240 individuals

Intervention #1: 500 individual flyers (250 treatment, 250 control) distributed among the 41 clusters. Intervention #2: 240 individuals alternately assigned to treatment and control (120 and 120) but stratified by gender and if they came as a group.

Intervention #1: 20 treatment city blocks and 21 control city blocks.
Intervention #2: 120 treatment individuals and 120 control individuals.

Intervention #1 will be able to detect a standardized mean difference of 0.33 assuming an ICC of 0.10 (the ICC is unknown in this case) and an alpha of 0.10 and a beta of 0.20 (power of 80%). Intervention #2 will be able to detect a standardized mean difference of 0.32 standard deviations, assuming an alpha of 0.10 and a beta of 0.20 (power of 80%). For the proportional outcome, assuming that 70% of the participants in the control group show up for the final presentation, then the trial will be able to detect an increase of 15 ppt (from 70% to 85%).