Economic Condition
4.6 The Nature and Sources of Data
Copestake et al. (2005) believe that when assessment is to be carried out, the cost of such impact assessments can be reduced by relying on public data. Studies in Nigeria (Anyanwu, 2004; Anyanwu, 2005) have focused on the use of secondary data made available by formal finance providers to the regulatory authorities. These data do not represent the opinion of the program beneficiaries. Moreover, the focus of this work is on cooperative societies which are neither recognised nor controlled by the government; hence they are not required to submit any report to the government. The majority of these cooperatives hardly prepare financial statements that could be relied upon as a basis on which to make meaningful decisions. This research cannot rely on such secondary data for analysis and this is one of the reasons why the
Page | 95 researcher decided to use primary data by sourcing data directly from the rural dwellers using a cross sectional study. Primary data are valuable because of the richness of the data, the directness of information from participants and the opportunity of accessing the silent but salient reactions during interviews which are not present in secondary data collection approach. The distinctive nature of this study which focuses on people that otherwise are not taken care of by overall government provision and systems justifies the use of primary data. The review of literature in chapter three shows that many studies (Ghosh and Maharjan, 2001; Lohlein and Wehrheim, 2003; Calkins and Ngo, 2005; Wanyama et al., 2008; Ramotra and Kanase, 2009; Allahdadi, 2011) used primary data since it is better to obtain such information from users and beneficiaries of the program rather than relying on secondary sources.
The researcher has to make do with cross sectional data derived at one point in time directly from cooperative members since this is the only ideal way to collect the data needed in view of the characteristics of the population – poor and rural based – for the study. It has been observed by researchers (Sebstad, 1998; Hulme, 2000; Nelson, 2000) that it is not possible in all cases to use longitudinal design. Moreover, “the problem of response increases significantly if longitudinal data are collected, as second and third interviews have much less amusement value” (Hulme, 2000: 90). For example, Eisenhauer (1995) that used longitudinal study was able to have 246 respondents at the second visit instead of 302 that took part in the initial visit. Sebstad (1998: ii) suggested that “assessment should concentrate on variables for which recall data is easily obtainable and generally reliable”. This approach was also recommended by Hulme (2000) and Nelson (2000) that recall methodology should be used where baseline data or studies are not available or possible. Apart from the advantage of collecting data firsthand from respondents, the choice of the cross sectional method becomes imperative since it may not be economically justifiable to conduct a longitudinal study due to lack of baseline data. Cross sectional design will also make the study more relevant with rapid analysis of responses, while
Page | 96 timely reporting of results, and data collected will serve as baseline data for future use.
4.6.1 Control Group
The determination of the role of cooperative societies in rural finance using a cross sectional study requires the use of a control group as identified in the literature in section 4.3.1 above. The main reason for using a comparison or control group is to find out whether members who have participated in the cooperative have been able to use it to improve their standard of living compared to those who have not taken part. The responses of the control group will be used to compare with those of program participants because the “meaningful positivist requires a critical minimum sample size, as well as inclusion of a control group”. (Copestake et al., 2002: 14). A worthwhile research on the role of cooperatives should be able to consider members and non-members or loan and no-loan members in order to determine the impact of such programs on the participants. This brought to light the possible weaknesses of sample selection in Eisenhauer (1995), Larocque et al. (2002), Adedayo and Yusuf (2004) and Adebayo et al. (2010) that used only program members without any control group. The control group is necessary in order to trace changes to participation in the program. However, the control group should be similar to program beneficiaries on key variables (Sebstad, 1998; Hulme, 2000; Imp-act, 2005).
The ability to establish a control group with the same socio-economic conditions with cooperative members may not be possible. As an alternative, members of the same program – new clients or incoming clients - who are yet to benefit from the program loan have been argued by Edgcomb and Garber (1998), Sebstad (1998), Hulme (2000) and Nelson (2000) to be an effective control group. Moreover, using new or “incoming clients as the comparison group helps to minimise the self selection bias since they also elected to join the program” (Nelson 2000: 4A-6). In this case, cooperative members that have not received loans are chosen as the comparison group for both qualitative and quantitative
Page | 97 methods. In other words, the two groups are members of the same cooperative societies, those who have received loans and those who have not taken loans. Using a control group in a qualitative study has the potential to help the researcher to maximise his understanding of phenomena (Onwuegbuzie and Leech, 2007). Moreover, since the control group (no-loan members) is included in the quantitative aspect of the study, it is therefore consistent that the control group should be included in the qualitative aspects, if the findings are to be comparable and credible.