One acre, one farmer, 1,000 problems

Editor’s note: Nick Easen is the specialist agricultural project manager of Asia Market Intelligence, a Hong Kong-based research firm.

The vast majority of Southeast Asians are still toiling in the paddy fields, filling the region’s rice bowls. In many cases agricultural research in the region is a headache for multinational companies in the agriculture industry, especially if the research is being done at a distance with limited control.

More often than not both clients and market research companies may have a limited understanding of the local situation. For example, it is a tall order to expect one company to understand the attitudes of corn farmers in Heilongjiang Province, China; or the effects of pest infestation in Mindanao, Philippines; or how El Niño influences local grain prices in Sumatra, Indonesia.

To account for all the variables in the field, researchers have to be willing to innovate. The keys to effective agricultural research in Asia are a willingness to learn and the ability to adapt research techniques to the requirements of the specific job.

To successfully complete a recent research project, which examined Indonesian farmers’ usage of and attitudes towards planting and harvesting corn, we relied on investigation, education, understanding and innovation.

Issues were complex

Over one million small farm holdings exist in Indonesia. Early on, we recognized that organizing a sample survey of 800 farmers in this country was equivalent to interviewing 10 farmers in the U.S. and saying that their usage and attitudes reflected the entire North American continent’s grain farmers!

The issues involved were understandably complex. How could the client expect the results to represent or mean anything? How could we really expect to understand all the spheres of influence affecting Indonesian farmers? On the agency side there was the question of coverage and logistics - how to have widely dispersed interviews in rural areas, yet still conduct the research within the constraints of a fixed budget.

After initial investigation, many more questions were raised as even more variables surfaced. These fell into five categories, which are associated with any agricultural research project:

  • Climate. Indonesia straddles three time zones, and its climate varies significantly across the archipelago, from Java to Lampung to Sumatra. This can widely affect growing conditions, timing of planting and harvest, as well as yield. Last year raging forest fires coupled with the unusual weather effects of El Niño also had a major impact on agriculture.
  • Agronomy. Semantics added to the confusion. There is one name in Bahasa (Indonesia’s national language) for three types of insects. There are also remote Bahasa terms for various diseases. Up to three crops may be grown, and hand-weeding may be more prevalent than pesticide control.
  • Geography. Irrigation, transport to and from market, availability of seed and quality of the soil differed from holding to holding. There are six different units of measurement for land area: ru, bumi, patok, tumbak, lobang and the bahu. Each is roughly about a third of an acre. However, again, sizes were not standard. The average size of a plot of land is about an acre.
  • Socioeconomic situation. The farmers’ decisions are based on how much they could afford for seed, fertilizers and pesticides. A low level of education and illiteracy are common, as well as a lack of understanding of farming practices.
  • Culture. Agriculture in Indonesia is considered a livelihood rather than a profession. Decisions are based on habit and tradition rather than monetary gain. Farmers are influenced more by the village chief and their next-door neighbor rather than advertisements or promotions. Each area has its own dialect, with few farmers speaking Bahasa.

All these variables had to be accounted for within the framework of the research program, so that the differences in responses concerning usage and attitude could be attributed to one of the categories and reasoning listed previously. Cause and effect were crucial to understanding any response.

The key was to use a combination of research techniques. Initially, secondary research was conducted in order to decide where to sample. Government statistics on area and yields of corn by geographical region allowed us to home in on the major corn areas at a regent level (equivalent to a group of counties in the U.S.).

A combination of secondary research and trade interviews then allowed us to adapt a U.S.-formulated questionnaire to the local situation. Common English names for diseases and pests had to be translated into their Latin equivalent and then into Bahasa to preserve their meaning. This ensured that there was no disparity between languages whatsoever, as interpretation of the common English name in most cases would not have specified the correct disease. Latin acts as the best medium, for instance:

English
Southern Corn Leaf blight

Latin
Heminthosprium maydis

Bahasa
Bercak daun

The key to doing a good agricultural survey is to have a questionnaire that encompasses every eventuality likely to occur in the field, yet still have the flexibility to adapt the survey tool to the local situation when the interview is in process.

Well-researched questionnaires and good interviewer briefings are critical. This particular questionnaire spent eight weeks in development with back translations (English into Bahasa and then back into English independently), consultations with in-country agricultural experts and questionnaire drafting in Indonesia, Hong Kong and the U.S.

Thorny point

In many ways it was realized that once in the field, the number of widely dispersed interviews would not allow any to be called back or repeated. One thorny point was the concept of area and application, i.e., application of seed, fertilizer and agrichemicals per unit area. Corn yield also fell into this category.

Basically, there was no specific formulation in the questionnaire that could encompass these concepts. Only through a rigorous series of interviewer briefings could we relay the importance of trying to communicate area and how it relates to application rates. In many cases we ended up with bags per batu rather than pounds per acre!

In total, five areas and 10 regions were chosen that spanned an area the size of the European Community. Due to the fact that local dialects were necessary to communicate effectively with the respondent farmers, local agricultural students teamed up with regular interviewers and field controllers. These people knew the local agricultural situation and interviews were conducted in more than five dialects. Local farmers were also less likely to be hostile and more open to discussing the research more freely.

In this case, the qualitative phase of research ran alongside the quantitative survey. This was necessary to color in the whole picture and provide an understanding of the underlying factors affecting each farmer in his decision-making process.

Each quantitative section had a qualitative side to it. For instance the farmer may have had an incidence of leaf blight, but why was it this bad? Could he have applied the seed at a high rate? Why? In each case the reasons could easily fall under one of the five categories stated previously.

It would have been easier to have run the survey and analyzed the results without doing the qualitative, trade interviews and secondary research. However that would have only given half the picture.

For instance, in one area farmers plant two seeds in one planting hole - a practice born out of tradition and low germination levels in local soil. On the quantitative side, when the tables are run it looked as if the seed application rate was double and they are growing twice the density of corn plants. This, of course, was not the case.

In addition, there was not one question that encompassed the level of importance of the village chief and neighboring farmers on product choice. This was only expressed by qualitative coverage at a local level.

Understood in context

In analyzing the quantitative data it became clear that the data could only be understood in the context of each qualitative account of the local situation. We were fortunate in selecting 10 areas that reflected very different situations with respect to the five categories. They ranged from dry areas with a poor socioeconomy to irrigated, large corn farms run as true commercial enterprises.

The complexity of the data itself was also immense, with multiple crops of different brands of corn. Different times of planting and multiple plantings likewise overlap with harvesting! This was coupled with yield and application rates in numerous local units.

In many ways the survey brought to light how complex each local situation was in terms of molding the farmers’ usage, attitudes and decision making. The client was also surprised at just how much information there was. More importantly, the client now understood the Indonesian customer more easily, focusing on the real underlying processes applicable to each farmer, and the unique relationship he has with his crop and brand selection.