Some Problems of Inference from Chain Data

Means of collecting & analyzing chain data (eg, rumor transmission, serial interpersonal influences) are discussed. Distinctions between chain methods & related samples such as networks are reviewed, & four primary uses of chain data are described (inferences about individual social actors, intrinsic properties such as chain length, or chaining processes & networks). Four techniques for chain sampling are outlined: (1) snowball sampling, seen as useful chiefly for inferences about networking; (2) the small-world method, often providing shorter chains & lower completion rates; (3) natural chains sampled at endpoints (ie, chains not delineated at the researcher's request); & (4) natural chains sampled at any locus. Different limits of each method & their appropriateness to different survey types are briefly discussed. 2 Tables, 3 Figures. L. Whittemore.