The first of the three experiments which are the subjext of this report was described in detial in RRB Report No. 71-18. Two subsequent experiments were conducted with the identical forms and similar procedures. In addition, data from the National Edit Sample were tabulated in categories making comparison with experimental results possible.

The experiments were intended to be the first in a projected series having the objective of obtaining informaion for the improvement of self-administered questionnaires. Results have been sufficiently consistent from experiment to experiment, and between the experiments, a 1970 census study, and some observations of people filling the census-like forms, that further experimentation with the same questionnaires now seems unnecessary. Two alternative new formats for a census questionnaire are attached as exhibits 2 and 2a and are recommended for testing.