One of the critical issues in this type of self-administered questionnaire is assessing the responses of the administrators. Are they accurate? When they report that in-migration flows are substantial, are they correct in their assessment? When they report that transportation and air pollution are serious problems, are these merely their own personal judgments, or do they in fact reflect a real problem in traffic flows and air quality? We have been able to address this problem in part, through two internal reliability checks, and through the use of external objective indicators to asses the validity of the responses for Korea and Japan. We can also make an observation from a question on family planning policies and estimates of contraceptive use from the second round of the survey.

Reliability 1. For all cities in the first and second round of the Inquiries, we asked for data on the city's population for the current and past two census periods. From this we could calculate the rate of growth of the city in the past decade. We could then ask if there was any relationship between the calculated rate of growth, and an administrator's perception of the size of the migrant flow. This is only an indirect test, to be sure, but we did find a very close association between the actual growth rate and the administrators' perceptions. In cities that were growing rapidly, the administrators tended to see a great inflow of migrants; for those that were stagnant or declining, the administrators tended to see no in migration flows, or an outflow. The relationship was in the predicted direction and was close and statistically significant.

Reliability 2. In both the first two rounds, administrators were asked for actual data on migration flows; either the proportion of total growth that came from migration, or the actual numbers of migrants. Since these are usually more difficult data to obtain, not all cities were able to report them. For those who could report the figures, however, we could examine the direct relationship between the reported size of the migrant flow, and the administrators' perceptions of that flow. Here again, there was a very close, and statistically significant, relationship.

Validity. For Korea and Japan we have annually published data on cities and countries, which give a vast amount of information on conditions comparable to those in our Inquiry. For South Korea we have city data on vehicles per kilometer of road, on numbers of high school teachers per 100 high school students, on per capita city revenues, and on the extent of sewage facilities. These objective measures were found to be closely related to city administrators' perceptions of problems of traffic volume, secondary education, financial resources, and sewage.

For Japan we had city data on sewage coverage, the ratio (or numbers) of city personnel to population, and on the number of doctors per population. These were found to be closely related (statistically significant) to city administrators' judgments on the sewage problem, on the quantity of personnel, and on the quality of health care.

A Minor Discrepancy: views on family planning. In the second round of the survey, administrators were asked to estimate the contraceptive prevalence rate in their cities. They were also asked what efforts had been taken to control fertility. Examining the estimates, we found that ninety-five out of the one hundred responses accorded well with na tional statistical reports. Pakistan reported very low use; China, Korea and Indonesia very high levels, for example. But five administrators from Indonesia and Malaysia reported levels were far below the national levels. On the reports of policy we find the same thing: general agreement with some discrepancies. For the most part administrators in the poorer countries knew about their national policies, while most of the Japanese administrators left this response blank. All of this fits well with official policy in the various countries. But there were 14 Indonesian and 4 Philippines administrators who said there was no government policy, although family planning services were in place. Thus there are slight discrepancies on the facts of government policies, and on the levels of contraceptive use. We cannot explain these, but they should lead to some caution in taking administrators' responses as completely valid.

In reviewing these tests of validity and reliability, the AUICK report suggested a specific research strategy. When we discover city administrators whose views or reports appear discrepant, it would be useful to return to those cities to see if there indeed might be some specific condition in their environments that produce the discrepant perception, or whether the discrepancy is the result of misunderstanding of the question, or actual administrator's error.

Impact of Position. The fourth round of the Inquiry experimented with questions that would show how different positions might affect the responses. Administrators were classified by level - top and middle levels; by gender - male and female; and by years of service - more or less than 40 years. Then responses could be analyzed by these categories to see if the administrator's level, gender, or years of service affected the responses. It is difficult to interpret this analysis, because the samples were not randomly taken, and the numbers in any category are quite small, leading to large sampling errors. Still, some of the findings appear interesting and can be summarized here, though it is important to keep in mind these technical statistical problems. It is possible that the results will not stand up in larger samples.

1. Perception of High Population Growth Rates. Middle managers were more aware of high population growth rates than were top level managers in Indonesia, South Korea and Pakistan. Middle managers were also more likely to see fertility as too high in South Korea and Pakistan; this was reversed in Indonesia, however, where top managers were more likely to perceive fertility as high.

2. The necessity of empowering women. Women were more likely than men to agree with the necessity of empowering women in India, Japan and South Korea. All administrators in Indonesia agreed, and there were no women respondents in Pakistan.

3. Economic participation and improved household status. Women were more likely than men to agree that economic participation would improve their status at home in Indonesia and Japan. In the other countries there was no dif ference by gender.

4. Awareness of ICPD. Older and top level managers were more likely to be aware of the ICPD in Indonesia, Japan and South Korea. This was reversed in Pakistan, where the younger, middle managers were more aware, and there was no difference in India.