SAMPLE DESIGN
Two separate regions were deliberately picked, the more highly developed region Central Region near to Bangkok and the poorer semi-arid Northeast. Four provinces (changwats) were chosen in total, two in the Central region -- Lop Buri and Chachoengsao -- and two in the Northeast-- Sisaket and Buriram. These particular provinces were chosen as each one contained one county (ampheur) that had been sampled every year of the Thai SES, available from this Thai data archive, thus providing us with benchmark or comparative information. The two regions, the Central and Northeast regions, can also be ordered from relatively wealthy to relatively poor not only by wealth and income but also by soil moisture, soil chemistry, and other environmental characteristics.

Within each of the four provinces, 12 tambons were selected at random using stratification based on a Correspondence Analysis of satellite imagery; the latter are part of the Geographic Information System. See Binford, Lee and Townsend in Papers and Publications for details of the stratification and sampling. Within each tambon, four villages were selected entirely at random. Thus there are in total l92 villages in 48 tambons of the four changwats.

Details on sample selection and administration of the survey for each of the various instruments are contained in this part of the site. Briefly there are interviews with 2880 households, 262 BAAC groups, 161 village financial institutions, and l92 key informants. There are also soil samples from 1920 individual plots with an accompanying questionnaires. For each plot there are five photos. We also have air photos of each surveyed village from 50's, 70's, and 90's.

COLLECTION AND PROCESSING NOTES
Instruments were designed by Townsend, Paulson, and Lee in Chicago with the testing of theory very much in mind. Summaries of theoretical articles were prepared with a list of variable suggested by the theory appended at the end. Questionnaires from other surveys were also utilized, e.g., ICRISAT and RAND. Preliminary versions of the instruments were translated into Thai and then extensively pretested in the field. Questions that were difficult to answer or that used excessive time were dropped. Not infrequently five to six versions were necessary before convergence.

Enumerators were hired from Ratchapak, Chulalong Korn, Thamasat Universities and other Universities in Thailand. Formal training sessions were held. Residual unclear questions or ambiguous instructions led to final revisions of the instruments as well as to manuals and codebooks.

Data were collected by 12 teams, 14 persons per team. The bulk of the team consisted of 11 enumerators who did the interviews for the various instruments, household, BAAC key, village institutions. The completed questionnaires were read over by both a field editor and team supervisor. Errors or missing information triggered re-interviews. Open ended Thai language responses were translated to English, and all responses were double blind entered in Bangkok on an ISSA-LAN system designed by a consultant for RAND.

Each team also contained a soil person who dug the soil sample, took the plot photos, and completed the soil interview, occasionally with the help of other enumerators. The soil samples were shipped to Bangkok for analysis.