ORANGE COUNTY SURVEY

METHODS

The sample for the 1984 Orange County Survey consists of 1,003 randomly selected residents who were 
interviewed by telephone. The sample is stratified geographically, with half of the sample selected from 
north of the Santa Ana River and half from the south. For data analyses, the sample is statistically 
weighted to represent the actual distribution of the Orange County population. 

The sample in each area was chosen using a computer program which randomly generates telephone numbers 
from among working blocks of telephone exchanges. A working block is one that contains numbers in use. 
The total of telephone numbers generated within an exchange was in proportion to the number of 
residential phones represented by that exchange in the northern part of the county or the southern part 
of the county. Using this procedure, approximately 2,200 telephone numbers from the south and 
approximately 2,2,00 telephone numbers from the north were drawn. This procedure of random digit dialing 
ensures that unlisted as well as listed numbers are included in the sample, Also, since over 95 percent 
of the households in Orange County have telephones, random dialing yields a sample representative of the 
population of Orange County.

The Troldahl-Carter Method was used in randomly selecting which adult member of the household was to be 
interviewed. This method consists of enumerating the total number of adults in the household and the 
total number of men in the household. Then, using a prearranged grid, the interviewer selects the 
individual in the household for interviewing.

As further evidence of the representativeness of the sample chosen by the above methods, characteristics 
of the sample were compared to characteristics of the total Orange County population using the 1980 
census. On the basis of age, income, sex, marital status, household size, and home ownership, the sample 
is representative of the population of Orange County. Characteristics of the 1982 and 1983 Orange County 
Survey samples were also contrasted with the characteristics of the 1984 Orange County Survey sample. 
Marital status, ethnicity, age, sex, and education were closely comparable in the three surveys.
The sampling error for this survey is plus or minus three percentage points. This means that if this 
survey were to be repeated 100 times, in 95 out the 100 times the answers obtained for a particular 
question would match those we obtained in this survey within three points. The sampling error for any 
particular sub-group would be greater. These calculations assume that the data were collected under ideal 
circumstances. Since there are a° large number of practical problems in conducting social surveys, the 
actual sampling error for any particular result might be slightly higher.

Data Collection and Survey Instrument

As noted above, the interviewing for the Orange County Survey was done by telephone. Cost considerations 
and methodological improvements have led to telephone surveys' increased adoption in the social sciences. 
In addition, several studies show similar quality in telephone and face-to-face interviews. Interviewers 
were closely supervised during the data collection. Interviewers participated in a two-hour training 
session on the Orange County Survey instrument. Supervisors were available during the telephone interviewing to answer 
questions of interviewers or respondents. The telephone system used also allowed supervisors to monitor 
interviews to correct for errors in administering the questionnaire.

The interviewing was done between September 5th and September 22nd, 1984. On weekdays, interviewing 
occurred between the hours of 5:30 and 10:00 .m., and on Saturday between 10:00 a.m. and 4:00 p.m. For 
each number in the sample, at least four call back attempts were made. The resulting sample of 1,003 
represents 25 percent of the numbers dialed (4,021 in all). For 13 percent, a refusal to cooperate was 
received. The majority of calls not completed (56%) was due either to nonworking residential phone 
numbers or to o answers, which may indicate nonresidential phone numbers. The refusal rate for the survey 
was 34 percent, that is, 25 percent completions plus 13 percent refusals divided into 13 percent. This is 
consistent with the general refusal rate in surveys, which varies between 25 percent and 40 percent.

The Orange County Survey, as shown in Appendix D, includes 86 questions. There were also a few open-ended 
questions. Individuals were asked to name their residential zip code. In addition, there is a question on 
the respondent's occupation. Categories for coding were developed prior to interviewing, and reliability 
tests indicated that the coding of open-ended questions was conducted accurately.