Implications for Poverty Reduction in Rural Households in Ghana: Sampling Technique and Attribution

Implications for Poverty Reduction in Rural Households in Ghana: Sampling Technique and AttributionThe study employed a quasi-experimental survey. Thus the data for the study was obtained from both beneficiaries (treatment group) and non-beneficiaries (control group) of MFI loans in 2011 through a random survey of 500 women engaged in agro-processing in the Upper East Region of Ghana of whom 250 were beneficiaries of microfinance while 250 were non-beneficiaries. Questionnaires were administered to the randomly selected respondents in a face-to-face interview. The questions included in the interview relate to access to microfinance, initial savings, consumption expenditure on basic needs, the number of business activities the woman engages in at the moment, the location of the business, and several other socio-demographic characteristics.
The study’s sampling procedure for reaching the treatment and control groups was done in a manner to minimise biases that characterise non-experimental impact research. The rationale was to mimic a randomized control trial. The following highlights some of the strategies employed to minimize spill-over effects, confounding problems, and contamination and selection biases. First, to deal with spill-over effects, the control and treatment groups were selected from different communities. The choice of communities was preceded by a focused group discussion in all the communities in the district. The rationale was to ascertain information on the extent of interaction among communities and gain insight on issues such as the similarity between communities and interventions related to poverty and finance that have been received by communities. Placement bias has been associated with selecting treatment and control groups from different communities. In this study, this is less of a concern as MFIs are situated mainly in the District capitals. Thus, the likelihood of the control group indirectly receiving some benefits from the treatment group in view of their access to credit is minimized. Buyer seller

Second, selection bias; as indicated by Duvendack et al. and Hulme(n.d.) occurs when there is no randomization in the assignment of subjects under study into either treatment or control group. This therefore creates a pre-existing difference between the treatment and the control groups. When this happens it leads to an inconsistent or bias estimate of the impact of the programme intervention. Thus to minimize the problem of selection bias, the study selected respondents with similar characteristics, such engagement in agro-processing business, respondents resident in rural communities and other household characteristics. The entrepreneurial drive and ability which is an invisible attribute was therefore effectively taken care of as well as other economic, physical and social environment.
Thirdly, Contamination; this is said to occur when there is communication about the experiment between groups of participants. That is subjects under study are aware of the study and communicate among themselves about the study. There are three possible outcomes of contamination. Some participants’ performance may worsen because they resent being in a less desirable condition; also participants in a less desirable condition may boost their performance so they don’t look bad; and diffusion of treatments: control participants learn about a treatment and apply it to themselves. This issue of contamination was taken care of in the study by interviewing individual respondents in each group in their respective homes, so that no one knows of the other in the study. Again the control and treatment groups were selected from different communities.

Tags: , , , ,