Combining academic research and community development is a core component of the SAFI Project’s objectives. The project started as a collaboration between youths from Laikipia and a researcher from Yale University; all subsequent phases of the project have been tailored to the dual goals of meeting community needs and designing robust social science research.
Pilot Study
The initial trial implementation of the SAFI project took place in 18 town centers across Laikipia North, West, and East Districts from November 2007 until March 2008. In each of these towns, we implemented a version of the SAFI Project program and collected data to evaluate the effectiveness of the intervention. We implemented the project in 1 town without collecting data so that we could train coordinators and check the validity of implementation protocols. We also collected the same measures in 18 towns in which there was no SAFI Project implementation; this set of towns served as our control group.
We are continuing to compile results, and will soon begin providing feedback in these centers. We will also take the results of the pilot study into account when planning for the next phases of implementation, which will add recycling and income generating activities to the existing anti-littering and waste removal protocol
Program Evaluation
In order to evaluate the effect of institutional design on program performance, we developed three different versions of the basic anti-littering program and randomly assigned each project town to one of these three treatment groups.
MOBILIZATION-The first treatment group follows only the basic mobilization program. In particular, there are no fines for littering and town trash committees are formed entirely of representatives of civil society groups from within the town. This corresponds to “coordination/common knowledge” theories of behavioral compliance with rules and laws, which stipulate that the problem of legal compliance is simply one of coordination.
PUNISHMENT-Towns in the second treatment group also undergo an anti-littering mobilization, but are different because the rules articulated by the local authorities will be backed by a formally articulated punishment. Individuals who are caught littering the first time will engage in community service work; subsequent violations will lead to fines. In addition, individuals who are caught stealing trash cans pay a fine equivalent to three times the price of one trash can. These punishments will be enforced through the joint efforts of town committees and local government chiefs.
ELDERS- Finally, towns assigned to the third treatment group have both mobilization and fines, but differ with respect to the composition of their town trash committee. In addition to recruiting committee members from civil society groups in the town, committee members will also be chosen from “traditional community organizations” in the community surrounding the town- namely clans, councils of elders, and warrior groups. The idea behind this is that “civil society groups” and “traditional groups” are supported by different networks and norms of social compliance- the expectation is that linking the enforcement of anti-littering rules to both kinds of social organization will generate higher rates of compliance (lower rates of littering) than enforcement based only in town networks.
Randomization
Random assignment to treatment and control groups is the central component of the research design employed in the SAFI Project pilot study. Three aspects of the program implementation were randomized: 1) whether or not a town was “treated” with an anti-littering and trash program, 2) which of the three variants of the trash program each “treatment” town received, and 3) when “treatment” towns received treatment.
In order to make logistics of program implementation easier, the 36 pilot study towns were grouped into 6 regional clusters of 6 towns each. 3 clusters are located in Laikipia East District, 2 are in Laikipia West District, and 1 is in Laikipia North District. Random assignment to treatment groups occurred within each regional cluster. In each cluster, 1 town was assigned to MOBILIZATION, 1 was assigned to PUNISHMENT, and 1 was assigned to ELDERS; the other three towns in the cluster were assigned to control (no SAFI Project program).
The timing of program implementation was also randomized between clusters; each cluster was assigned to a three-week implementation period. The effect of this randomization structure was that during each implementation period, the SAFI project was implemented three towns in one cluster: one MOBILIZATION town, one PUNISHMENT town, and one ELDERS town.
Data Collection
We utilize two different data collection activities to measure littering behavior in both treatment and control towns. These measures have been collected on a weekly basis since September 2007; regular collection is ongoing in all 18 pilot study towns and all 18 control towns.
Litter Count- Every week, a SAFI Project enumerator counts the number of pieces of trash on the ground in five plots in his town center. Each plot measures 3 meters by 2 meters, and the five plots were chosen to be representative of the amount of litter in the town as a whole.
Littering Behavior Observation- In each of the 36 towns, a hidden SAFI Project enumerator observes littering behavior for one hour every week. The unit of observation is each time an individual in the town center has a piece of waste in his or her hand; in each such situation the enumerator observes whether the individual drops the waste to the ground, holds on to it, or puts it into a trash can or pit.
Preliminary Results
Although data is still being collected and analyzed, preliminary results are available for the Litter Count measure from September 2007 to March 2008. The measure calculated is the average number of pieces of trash counted each week in each town’s five enumeration plots. The table below compares this measure between four groups: towns with no program, towns with the MOBILIZATION treatment, towns with the PUNISHMENT treatment, and towns with the ELDERS treatment.
The basic finding is that receiving a PUNISHMENT or ELDERS program cuts the average amount of litter in a town in half; receiving a MOBILIZATION program results in a slightly smaller reduction.
There are several implications of this exploratory analysis. First, towns in which the SAFI Project anti-littering and waste removal programs have been implemented are substantially cleaner, on average, than towns with no programs. Across the board there is a substantial positive effect of mobilizing collective action to clean a town and to prevent littering. Second, program design does seem to effect outcomes; in particular, towns in both the PUNISHMENT and the ELDERS groups have fewer pieces of trash on the ground than towns in the MOBILIZATION group. This implies that the problem of littering cannot necessarily be solved simply by coordinating collective action; behavior change must be backed at least in part by credible threats of punishment. Even when projects are operated by local civil society groups, the success of these efforts appears to depend on linkages to broader governance structures.
As more data are collected and analyzed, these results will be examined in greater detail and will be incorporated in future program design and implementation efforts.