CAPÍTULO III. DIAGNÓSTICO SITUACIONAL
3.1 MACRO AMBIENTE
3.1.3 Factores tecnológicos
A quantitative assessment of informal finance program that will prove useful, should have a large sample size that will ensure the effective use of control variables and indicators, be able to account for refusals and program drop outs and allow for issues relating to invalid data, while remaining small enough to fit the evaluator’s budget (Edgcomb and Garber, 1998; Sebstad, 1998; Barnes and Sebstad, 2000; Hulme, 2000). To accomplish the task of selecting an appropriate sample size, “trade-offs are required between the number of variables, the margin of error, the confidence interval and the budget “(Sebstad 1998: ii).
Since the sample size should be large, there is a need to determine the number of participants that can be referred to as large enough to make the research meaningful. As a result, Imp-act (2005: 6) suggests that a researcher “should plan to have not less than 30 respondents in each group”. Moreover a researcher should “allow for at least 30 in any sub-sample of interest in the study” (Sebstad, 1998: 13). If this principle is followed, the sample size for this study would have been 60 since there are two groups of respondents. “While 30 is often considered the minimum number of respondents for statistical significance, a good analysis often requires more than that, especially when the
Page | 99 analysis relates to subgroups of the population” (Edgcomb and Garber, 1998: 48). Onwuegbuzie and Collins (2007: 288) argue that if the “recommended sample size of 30 for both correlation and causal-comparative design is followed, it would lead to statistical tests with inadequate power because they (sample size of 30) are not based on power analyses”. To buttress their opinion, they went further to calculate the statistical power of 0.51 for a one tail test and 0.38 for a two tailed test of moderate relationship using a 5% level of statistical significance for a minimum sample size of 30. Despite Onwuegbuzie and Collins’ (2007) postulation, the determination of sample size from the findings of Sebstad (1998) (after considering the results of some impact assessment undertaken in Africa - Mali and Uganda, Asia - Bangladesh and Philippines, and Latin America - Honduras) concludes that “the studies that had survey samples of less than 200 all reported limitations in analysing the data” (Sebstad, 1998: 23). A trade off is therefore needed since Onwuegbuzie and Collins (2007: 288) recommend a minimum sample size of “82 participants for two tailed hypotheses” and Sebstad (1998) notes that a sample size of less than 200 cannot provide meaningful statistical results.
Ogun State consists of 20 local governments grouped into three senatorial districts, namely: Ogun East, Ogun West and Ogun Central. The population for the study are cooperative societies that are not registered with the Ogun State goverment. The choice of unregistered cooperative societies is based upon the fact that they are mostly found in rural areas and that they also function more like financial institutions for rural dwellers. Moreover, relatively large numbers of informal finance providers in sub-Saharan Africa operate as savings cooperatives (Chiumya, 2006). Two local governments that are more rural - based on Nigeria's poverty index - were selected in each senatorial district for study. Stratified and random sampling techniques were used as follows: Based on a preliminary survey carried out between June and July 2009, there is an average of fifteen (15) unregistered cooperative societies in each local government area with an average membership of 43 individuals. This implies the existence of 90 cooperative societies within the six local government areas
Page | 100 used for this study. The sample population is made up of 90 cooperatives multiplied by 43 individual members, which equals 3,870 individuals.
Five cooperatives were randomly selected in each local government area to make a total of 30 cooperatives out of the 90 cooperatives within the six local governments. Thereafter, a random sampling of eleven individuals from the membership list of the 30 choosen cooperatives were selected to participate in the impact survey questionnaires. For the focus group discussion (FGD), two cooperatives from the 6 local governments were selected randomly while 6 members from each cooperative were randomly selected to participate. Samples for the interview were drawn from two cooperatives from the six local government area selected with four members randomly selected from each cooperative. In all, 54 cooperative societies were used for the study. Therefore, the proposed sample prior to the field work is 330 for questionnaire, 72 for FGD and 48 for interview making a total of 450 selected individuals as sample. This sample size fulfils Sebstad’s (1998) and Onwuegbuzie and Collins’ (2007) criteria. The sample meets the statistical two tailed test hypothesis requirement and the analytical implications of obtaining reliable results since the quantitative research instrument was subjected to quantitative analysis and interpretation after the field study.
It was not possible to determine accurately the number of no-loan members prior to the field work because cooperative members who were not eligible for loans during the familiarisation period may have become loan members during the survey exercise. However, because participation in the study is voluntary, the researcher was able to receive responses to the questionnaire from only 302 people (91% of total sample). The remaining 28 people include the few who withdrew their participation and those who had to leave while administering the questionnaire because of other commitments they considered more important. Table 4.2 below show the breakdown of the 302 participants based on their senatorial districts, membership condition (Loan and No-loan) and gender. The detailed analysis of the participants which explains more of the demographic
Page | 101 information based on membership condition and period of membership is covered in sections 6.2 and 6.3 of chapter six.
Table 4.2 Impact Survey Participants Senatorial
District
Loan Members No-loan Members
Total Male Female Male Female
Ogun Central 28 42 15 13 98
Ogun East 35 40 10 15 100
Ogun West 46 32 12 14 104
Total 109 114 37 42 302
Compiled by the author: Field study (2010)