5.5. RESEARCH METHOD
5.5.2. Sampling
Due to the large size of the population in this research, it is important to select a subset of the population to be included in the research, which is known as sampling. Sampling is defined as “the process of selecting the right individuals, objects or events as representatives for the entire population” (Sekaran and Bougie, 2009: 264). A proper sampling procedure should be adopted in order to obtain a sample representing the population. It is true and inevitable that certain persons will be included in the sample as a result of the researcher’s personal judgement, the prospective availability of respondents, and the researcher’s criteria (Bryman, 2008: 169).
5.5.2.1. Research population
The target population in this research can be divided into two groups. The first is microenterprises, the borrowers of Islamic MFIs, who must have received financing for productive purposes in order to improve their businesses. For this, the relevant regulation that defines microenterprise should be consulted, as this determines the population of this study. Based on the Government Act No. 20/2008, a ‘microenterprise is a business with productive assets, excluding land and buildings, of less than IDR 50million and annual sales of less than IDR 300 million’. For Islamic microenterprises, the type of financing used can be any type offered by Islamic MFIs: musharakah, mudharabah, murabahah, salam, istisna or qardhasan.
The second group, as regards the interview schedule population, is the bankers or officials of Islamic MFIs, which might include BMT (including Islamic cooperatives), BPRS, PNM Shari’ah, Pawnshop Shari’ah, and venture capital Shari’ah. A semi- structured interview schedule is employed to obtain in-depth information from these respondents.
The following sections provide details of the sampling issues for this study:
5.5.2.2. Sample frame
The sampling frame is “a physical representation of all elements in the population from which the sample is drawn” (Sekaran and Bougie, 2009: 267). Ideally, the samples for this research should be microenterprises that are borrowers of Islamic MFIs and officials from all Islamic MFIs. However, it was impossible to reach all of them since there were limitations in terms of data, time and funding availability. Thus, the sample frame was set based on statistical data provided by Bank Indonesia, the Statistical Bureau and samples of Islamic MFIs. However, the complete list of BPRS in Indonesia and East Java could not be accessed from Bank Indonesia, in which the official data only show the total number of BPRS in each province. Based on the information from Bank Indonesia (East Java branch), there were 30 BPRS in East Java in 2011. The exact number of BMTs in East Java is unknown, and there are no valid and comprehensive data that can be accessed due to there being different associations of BMTs. Some BMT associations (PINBUK, Puskopsyah, Inkopsyah, Absindo, Al-Kamil, etc) do not have recent valid comprehensive data; they only have lists of members. For example, PINBUK estimates that, in general, there were around 200 BMTs in East Java in 2011, which is the region covered by this study. For example, the Bank Indonesia dataset (2011) gathered through a survey includes 120 active BMTs in East Java. A sampling strategy to deal with the regional population for this study is explained below.
Data on the number of customers and borrowers of samples of BPRS and BMTs are collected through the interviews with the Directors/Managers. Samples used in the fieldwork were a shortlist of borrowers provided by BPRS and BMTs, which also included some walk-in borrowers who were present at the offices of BMTs and BPRS to
pay their instalments. Such individuals were also considered for this study with the objective of increasing the sample size.
5.5.2.3. Sampling method
Defining the most appropriate sampling method is the next important process in selecting the samples. The relevant literature indicates that there are two broad categories of sampling method: probability sampling and non-probability sampling (Sekaran and Bougie, 2009:267).
Probability sampling means that all members of the population have the same probability of being selected as a sample (Sekaran and Bougie, 2009: 268); this type is used when “the representativeness of the sample is importance in the interest of wider generalizability” (Sekaran and Bougie, 2009: 268). Non-probability sampling, on the other hand, does not give the entire population a predetermined chance of being selected as subjects, which means that the findings of the research cannot be generalised vigorously (Sekaran and Bougie, 2009: 276). This method is suitable for researchers with limited schedules, resources and finance.
There are two categories of non-probability sampling: convenience sampling and purposive sampling (Sekaran and Bougie, 2009: 268). Convenience sampling refers to “the collection of information from members of the population who are conveniently available to provide it” (Sekaran and Bougie, 2009: 276). Although this is the least reliable method compared to other sampling methods, it is suitable for exploratory research (Sekaran and Bougie, 2009: 278). Purposive sampling, on the other hand, is an approach where “the researcher purposely chooses subjects who, in their opinion, are thought to be relevant to the research topic” (Sarantakos, 1994: 138). In this approach, subjects are a specific group of people who have the required information.
Purposive sampling was employed as the interview method for interviewing the Directors/Managers of BMTs and BPRS. There is no exact number of samples required for qualitative research: it mostly involves continuing the sampling process until no new insights are obtained; thus, it depends on the “heterogeneity of the population” (Sekaran and Bougie, 2009: 298). In other words, based on the information requirements, in this
case the impact of Islamic micro financing on developing microenterprises that are specifically owned by Islamic microfinance borrowers, non-probability sampling is employed. Due to the limitations described above, this research employs both convenience and purposive sampling as part of non-probability sampling. Purposive sampling is applied in selecting the Islamic MFIs; in this case the selected institutions are BMTs and BPRS. These two types are selected since they fulfil these criteria: (i) they are well-established institutions serving microenterprises, and will thus provide sufficient evidence to measure the impacts of Islamic micro financing; (ii) they represent different natures and characteristics - while BMTs are cooperative in nature, less regulated and member-based, BPRS are banks, well-regulated and monitored by Bank Indonesia.
In quantitative data collection, the questionnaire method, hence convenience sampling, is then applied to select the borrowers of selected BMTs and BPRS. The selected respondents are either walk-in borrowers, who come to the offices to repay their loans, or those drawn from the list of borrowers provided by BMTs and BPRS, who repay their loans via the institutions’ collection services.
In relation to geographical delamination and sampling, it should be stated that the large population size and limited resources implied that it was impossible to undertake nationwide sampling. Therefore, this research purposively limits the sample scale to certain cities in East Java province, such as Malang regency, Malang city, Batu, Surabaya, Pasuruan, Pandaan, Bangil, Lumajang, Tulungagung, Probolinggo, Jombang, and Gresik. The reasons for choosing East Java as a case-study are as follows:
(i) East Java is the second most populous province in Indonesia (Biro Pusat Statistik, 2012: 78);
(ii) Based on Inkopsyah (Centre of Shari’ah Cooperative and BMT), there are 32 BMTs in East Java, which makes it the third largest in Indonesia in terms of the number of BMTs; the largest is West Java with 92 institutions, followed by Central Java with 75 institutions (Inkopsyah, 2012);
(iii) Based on Bank Indonesia’s data (2011), there were 30 BPRS in East Java in 2011, which made it the second largest in Indonesia in terms of the number of BPRS (Bank Indonesia, 2011: 3);
(iv) The number of microenterprises in Indonesia was 50,697,659 (in 2008), around 23% of which are in East Java. Thus, it is considered the largest compared to other areas/provinces in terms of the number of microenterprises (Ministry of Cooperative and Micro Small Medium Enterprises, 2009: 24);
(v) The number of people working in microenterprises in Indonesia was 83,647,771 in 2009, and 20% of them are in East Java, which is the second largest in Indonesia in terms of microenterprise workforce (Ministry of Cooperative and Micro Small Medium Enterprises, 2009: 24);
(vi) The total number of borrowers in East Java is unknown;
(vii) In East Java, among other cities Malang and Surabaya have the highest level of potency in terms of Shari’ah banking development (Ascarya, 2007).
In addition to the above-mentioned reasons, there is a very limited amount of research on BMTs in East Java, as such available studies are mostly concentrated on Central Java and West Java. In addition, the research on BPRS in Indonesia is even more limited, while no comparative studies of BMTs and BPRS have yet been conducted.
5.5.2.4. Sample size
Determining a proper sample size is necessary in order to obtain a reliable and valid sample that is representative for data analysis. A sufficient sample size should fulfil two requirements: precision and confidence (Sekaran and Bougie, 2009: 288). Precision refers to “how close our estimate is to the true population characteristics” (Sekaran and Bougie, 2009: 287), while confidence indicates that the certainty level of researcher’s estimation “…will really hold true for the population” (Sekaran and Bougie, 2009: 288). The larger the sample size, the greater the precision and confidence. However, a large sample will increase the cost of data collection.
Bryman (2008: 179) argues that there is no fixed size or number required for a sample, since it depends on a number of factors including time limitation, funding availability and the need for precision. He summarises that a large sample size is likely to provide more precision and fewer sampling errors. As a straightforward guidance, a comprehensive table on the sample size of a given population is prepared by Krejie and Morgan (1970, cited by (Sekaran and Bougie, 2009: 295-296). Based on the information presented in the table, for a population of one million or more, a sample size of 384 is required (Sekaran and Bougie, 2009: 296). On the other hand, Roscoe (1975, cited by (Sekaran and Bougie, 2009: 296) suggests that, for most research, the proper sample size is between 30 and 500. Thus, due to cost, time and other resource limitations, this research used the above guidance stated in Sekaran and Bougie (2009) to determine the sample size.
Regarding the sampling size for the interviews, the number of officials from institutions targeted for interview was 20 institutions, consisting of 10 BMTs and 10 BPRS. For the questionnaire, a target of 300 borrowers from both BMTs and BPRS were considered sufficient.
It should be noted that the final sample of BPRS and BMTs were based on the following criteria:
(i) A total of 10 BPRS were selected from 30 BPRS in East Java based on their willingness to be involved in the study as respondents;
(ii) A total of 12 BMTs were selected based on their willingness to be involved in the study and their financing products, which support the development of MEs.
As indicated in Table 5.1, 348 questionnaires were completed by 12 BMTs and 10 BPRS .(see: Table 6.8).
5.5.2.5. Data collection process
The field research took about three months from mid-August to mid-November 2011. As for the interviews, in the sample there are two groups of respondents: the officers of MFIs and the borrowers who use the funds to develop their microenterprises. The list of
interview questions was sent to the institutions providing funds for productive purposes; then, when the respondents were ready, the interviews were conducted in their offices. For the data collection process using questionnaires, three part-time enumerators were employed to help respondents answer the questions. The respondents were selected from among the borrowers who had been using the funds for productive purposes for more than 6 months. Some respondents were approached as ‘walk-in’ customers, but most of them were approached at their businesses premises in traditional markets or at home; their contact details were obtained with the assistance of the Account Officers of BMTs and BPRS who normally visit the borrowers to collect the repayments.
Overall, it should be noted that the response rate from both types of respondents is considered to be sufficient. Initially, 13 BPRS were approached by sending interview questions by post, followed by direct contact via phone calls for confirmation. As a result, 10 of them were able to participate in the survey, which means the response rate in relation to the target reached 77%, although it took quite a long time to secure their consent. The response rate from BMTs was higher than that of BPRS, since only 1 of the 13 contacted BMTs could not participate, which resulted in a 92% response rate in relation to the target. Thus, in total the institutional response rate was 84.61%. BMT officers probably provided better responses and were more willing to be involved in the research and survey due to their nature as community-based organisations which served the members of the local community, while BPRS were more concerned about confidentiality issues. Table 5.1., thus, summarises the general characteristics of the sampled institutions based on the initial descriptive results.
Table 5.1a: General Characteristics of Respondents (Institutions)
No Description BPRS BMT
1. Number of institutions 10 institutions 12 institutions 2 Legal type PT (Perseroan Terbatas/
Limited Corporation (Bank)
Cooperative
3 Association Asbisindo Absindo, Pinbuk, Dekopin,
Puskopsyah, Puskopsyah Al- Kamil,
4 Number of branches 1-2 branches Varied (1-2 branches, MMU has 44 branches and UGT has 138 branches)
Table 5.1b: General Characteristics of Respondents (Institutions)
No Description BPRS BMT
6 Most popular financing Murabahah Murabahah 7 Average financing
duration
18 months 16 months
8 Time for financing evaluation
3 days 3,5 days
9 Collection of repayment Monthly Montly, Weekly, daily
10 Collateral Required Required
11 Number of borrowers Varied : 250 – 2600 Varied : 100-1000, MMU and UGT has more than 25,000 borrowers.
12 Maximum financing Varied: IDR 50.000.000,- to IDR 200.000.000,-
Varied : max IDR 50.000.000,- 13 Minimum financing Varied: IDR 500.000,- to
IDR 3.000.000,-
Varied : IDR 100.000,- to IDR 500.000,-
14 Average financing Varied: IDR 1.500.000,- to IDR 50.000.000,-
Varied: IDR 500.000,- to IDR30.000.000,-
15 Margin per year Varied : 18% - 24% per year Varied: 12%-20% per year 16 Number of staff Varied: 14 to 80, average 24
staff.
Varied: 3-20 staff, MMU has 285 staff and UGT has 670 staff
The response rate from borrowers was also satisfying. Almost all the respondents approached were willing to answer the questionnaires. This is because they have direct and close relationships with the Account Officers; thus, when the enumerators approached them through the Account Officers, all of them were willing to participate. Table 5.2 depicts the general characteristics of the questionnaire respondents, namely the borrowers, based on the initial descriptive findings.
Table 5.2a: General Characteristics of Respondents (Borrowers)
No Description
1. Microenterprises (productive assets other than land and building less than IDR 50.000.000,-)
81% respondents are microenterprises
2 Marital status Married (91%)
3 Gender Male (54%)
4 Age Average : 42.41 years
5 Education Mostly having less than Senior High School education (87%)
6 Religion education Formal education (23%), Informal education (20%), Nothing (56.6%)
7 Business area Urban (66%), Rural (34%) 8 Age of business In average 13 years business age.
Table 5.2b: General Characteristics of Respondents (Borrowers)
No Description
9 Financing sources BMT: % , BPRS: % 10 Number of financing Average : 3.66 times 11 Original purposes Working capital (93%)
12 Financing used Working capital only (68.4%), Consumption and working capital (23.6%)
13 Repayment Monthly (84.2%)
14 Decision maker in using the fund Self (77.3%)
15 Arrear No arrear (79.5%)
16 Sales before IDR 8.500.000,-/per month 17 Sales after IDR 10.000.000,- / per month 18 Income before IDR 1.500.000,- / per month 19 Income after IDR 2.250.000,- / per month
5.5.2.6. Reliability and Validity
In social research, to ensure the quality of the research three issues are raised: reliability, replication and validity (Bryman, 2008: 31-35). This implies that good research should be repeatable and consistent, meaning that when it is conducted in other areas or on other occasions, the measurements should still be reliable and repeatable to produce good outcomes. Replication refers to the capacity of the measurements or procedures to be replicated by other researchers (Bryman, 2008: 32).
In technical terms, reliability refers to “consistency of measures” (Bryman, 2008: 149), which can be measured using ‘Cronbach’s alpha’; this measurement is also valuable for ensuring internal consistency of the data (Bryman, 2008: 161). Moreover, the reliability of a scale indicates the extent to which it is free from random errors, for which indicators are used to test-retest reliability and internal consistency (Pallant, 2007: 6). First, test- retest reliability administers the survey to the same people at two different times and calculates the correlation between the two scores obtained. Accordingly, the higher the test-retest correlation, the higher the reliability. Since the survey measures enterprises’ profiles which are quite stable, the test-retest will have a high correlation. The impact aspect will also be expected to be stable. Second, internal consistency is “the degree to which the items that make up the scale are all measuring the same underlying attribute”, in other words, the extent to which the items “hang together”. Similar to Bryman’s
suggestion, the internal consistency measured by Cronbach’s coefficient alpha should be above 0.7 (Pallant, 2007: 95).
In order to measure the reliability of the data in this study, the Cronbach’s alpha was measured and found to be 0.925 for all items in the questionnaires; thus, it is considered reliable as it was above 0.7. In particular regard to interview data, reliability was ensured by audio-taping the interviews and taking notes, which were utilised as a source of data for transcription and interview analysis using coding.
Validity relates to the ‘integrity’ of the research outcome (Bryman, 2008: 32; Bryman and Bell, 2007: 42). In other words, as Pallant (2007: 7) argues, validity refers to “the degree to which it measures what it is supposed to measure”. Bryman (2008) discusses four aspects of validity: measurement validity, internal validity, external validity and ecological validity. In quantitative methods, measurement validity mostly relates to reliability of measurement (Bryman and Bell, 2007: 41).
Internal validity means that the conclusion derived has a ‘causal relationship’ between two or more variables (Bryman, 2008: 32). External validity is concerned “with the question of whether the results of the study can be generalized beyond the specific research context” (Bryman and Bell, 2007: 42). Ecological validity is about ensuring that the environment is neutral, as when respondents respond to questionnaires, the environment should enable them to react naturally without any inference.
Pallant (2007: 7) states that the validation of a scale involves the collection of evidence concerning its use, which technically includes: (i) content validity, which refers to “the adequacy with which a measure of scale has sampled from the intended universe or domain of content”; (ii) criterion validity, which is defined as “the relationship between scale scores and some specified measurable criterion”; (iii) construct validity, which is related to “testing a scale not against a single criterion but in terms of theoretically derived hypotheses concerning the nature of the underlying variables or construct” (Pallant, 2007: 7). It is measured by investigating its relationship with other constructs.
To ensure the validity of this study, the researcher relied on the empirical evidence developed in previous research and carefully chose the wording applied in the questions. It should be noted that no straightforward, simple calculation method is provided in the statistical software. Therefore, an attempt to ensure the measurement and content validity has been conducted by adopting variables and scales from previous related studies (see section 5.5.1.1, Chapters 3 and 4). The pre-tests, both in English and Bahasa Indonesia, were also beneficial in providing feedback on whether the wording applied in the questions was clear and not misleading; thus, based on the feedback, corrections were