• Keine Ergebnisse gefunden

Chapter 4. Methodology of field research and results of descriptive data analysis

4.3 The Survey

Household surveys generally are the most important data source for poverty comparison and poverty measurement. They can tell directly about the distribution of living standard in a society or in a certain region, for example how many households do not attain a certain consumption level (Ravallion 1992). For the measurement of people’s living standards, traditional costly large-scale surveys were undertaken. Therefore, there is a need for more practical tools for measuring and assessing poverty (Aho et al. 1998). The background of the survey was to test a new tool for poverty assessment. This tool was developed in order to meet the requirements of the US Congress, which wanted USAID to develop and certify a low-cost and easy-to-implement poverty assessment tool with a high accuracy. In cooperation with IRIS (research and advisory Centre at the University of Maryland) different tools were developed and tested. The official project was accomplished in Uganda, Peru, Kazakhstan and Bangladesh (Zeller et al. 2004). The survey in Indonesia can be seen as additional test for two of the nine IRIS regression models with a smaller sample.

The legal text of the US Congress contains two definitions of the term very poor:

1. Living in the bottom 50% below the poverty line established by the national government or

2. Living on the equivalent of less then US$1 per day (in purchasing power parity)

4 Nowadays, Lawe belongs to kecamatan Pipikoro

5 Berdikari, due to an administrative reform is now two village namely Berdikari and Sejadra

6 Rompo actually belongs to kecamatn Lore Tengah

7 Nopu was a former hamlet (dusun) of Rahmat, in 2004 it became an independent village

38

(Povertytools 2005). The first definition refers to the national poverty lines of the countries, the second definition refers to the international poverty line.

For the international poverty line of 1US$ PPP, the headcount index is 19.4% (respectively weighted 20.6%) surveyed in the research area in 2005.

The survey done in Central Sulawesi, Indonesia was a quantitative study. In general, quantitative methods can measure key variables associated with the basic dimensions of poverty and therefore can provide information for the construction of a profile of poverty.

With quantitative data the poor can be characterised and an impact evaluation of poverty policies can be done (Ravallion 1992). The observation unit in the survey conducted in the vicinity of Lore Lindu National Park were households.

“The household can comprise one person or a group of people who meet certain criteria: for instance, living under the same roof or in the same building, recognizing the authority of the same individual (the head of the household), sharing meals, having a common source of income or pooling resources so as to satisfy the household’s essential needs” (Larivière et al.

1998a, p. 24).

The survey was a cross-sectional survey, which means that the households were only visited once. Originally, it was planned to ask first the composite questionnaire and exactly two weeks later the benchmark questionnaire, but this was not further seen as necessary after the experiences in other countries. So, when possible, both questionnaires were asked within the same interview session.

Especially in a cross-sectional survey the recall technique is used to gather information. Here, in a personal interview, “an interviewer asks household members to recall the value of variables such as quantities, prices and amounts spent on different goods consumed, bought, produced or sold by the household during a given period of time” (Larivière et al. 1998a, p.

28). Of course, short recall periods provide more reliable data.

The two questionnaires which are described below in more detail focused on indicators of poverty, considering the different dimensions of poverty, and on consumption expenditures which were aggregated on household level.

In December 2004 I selected 19 potential enumerators out of about 50 applicants. The main criterion for the selection was experience in living or working in the rural area of Sulawesi.

Applicants with former experience surveys or with other useful skills for a household survey were preferred (like knowledge of the local languages, mathematical skills, knowledge in 39

socio economics of rural development etc.). Preference was also given to people who were recommended by other Indonesian assistants working for STORMA. An Indonesian-English translator guaranteed the communication.

In January 2005 I provided a three-week ‘training’ for the enumerators. In the training, the questionnaires which were already adapted to the local situation together with an Indonesian colleague in Germany were discussed and further changes were made to meet the current situation in Central Sulawesi. The enumerators had to do several role games to simulate the interview situation. At the end of the training fifteen enumerators did the pre-test in Kalwara (kecamatan Sigi-Birumaro). The village of Kalware is located in the research area but is not included in the sample. There my team and me had extraordinary good conditions. The village head (kepala desa) provided very good help to our team. There the enumerators conduct the interviews in teams of two persons accompanied by Dr. Stefan Schwarze, Anastasia Wida, who was responsible for the data entry in Palu, Kristina A. Rissi (my research assistant) and translator and me, to supervise the first interviews and to get an impression about the interviewing skills of the trainees. After the pre-test, further adoptions in the questionnaire were made and problems with certain questions were discussed. After the pre-test 10 people (7 male and 3 female) were employed as enumerators. Moreover, two persons (both male) were employed as field supervisors. Besides of doing the household interviews, they had the task of controlling all questionnaires of their team in the field. For the data entry in Palu two persons (one male and one female) were employed as data entry operators.

In February 2005 the survey was conducted in the first sample village, namely Pandere. In the first village the whole team of 10 enumerators and 2 supervisors did the interviews. After the first village the group was split into two teams, each consisting five enumerators and one supervisor. First the survey was conducted in kecamatan Sigi Biromaru, then in kecamatan Kulawi and Pipkoro and then in kecamatan Palolo, Lore Utara and Lore Tenga. My field assistant and I visited the teams regularly in the field to solve problems and provide the interview material.

Due to the fact that various households refused to answer STORMA questionnaires any more, a number of 2818 valid interviews were carried out.

8 during the analysis two household were dropped because of outliers in the data set

40