A Study of the North American Hispanic Adventist Church - Adult Survey, 1994

Data Archive > U.S. Surveys > Religious Groups > Members or Leaders > Adventist > Summary

The purpose of this study was to examine the unique needs and challenges facing the Latino Adventist community in the North American Division of the Seventh-day Adventist Church, which includes the United States, Hawaii, Canada, and Bermuda. "The major focus was on illuminating the nature, current trends, perspectives, and trends within the Adventist Latino community" (Hernandez, 1995, p.29). AVANCE was conducted as a follow-up study to Valuegenesis.

Data File
Cases: 2,143
Variables: 367
Weight Variable: None
Data Collection
Date Collected: April, 1993 - March, 1994
Funded By
A major grant from the North American Division of the Seventh-day Adventist Church and additional support from Adventist educational institutions including: Andrews, La Sierra, and Walla Walla Universities, and Atlantic Union and Pacific Union Colleges.
Collection Procedures
Self-administered surveys were distributed at congregational "youth society meetings." All Hispanic Adventist churches regularly hold "youth society meetings" where youth and adults are both in attendance. The pastor of each church was given a short survey to determine whether the group that was present at the "youth society meetings" was representative of the church membership.
Sampling Procedures
In order to draw a random sample of Hispanic Seventh-day Adventist churches, all of the churches were listed by conference within union. Within each conference, the churches were listed by size. For the first conference, they were listed from largest to smallest, and then in the next conference from smallest to largest, etc. This "serpentine" pattern preserved the similarity of size of churches adjacent on the list. If it was necessary to replace a church that had been chosen on the list by its "shadow" (the next one down on the list), this pattern avoided a sudden jump in size from large to small. A single running cumulative total column was calculated alongside the entire list of churches. The last entry in this column totaled 56,974.

To do the actual sampling, the grand cumulative total was divided by the total number of churches desired for the sample (in this case, 60). This produced an "interval size" of 56,974 /60 = 949. A random number between 1 and 949 was then generated; it happened to be 636. The actual churches were drawn from the sample by going down the "cumulative sum" column until a number equal to or greater than 636 was found. The church associated with that cumulative sum was the first church in the sample. Then the interval (949) was added to the random start (636 + 949 = 1585). The church associated with a "cumulative sum" total of 1585 was chosen as the second church. All remaining churches were chosen by successively adding the interval to the running total, and picking the church associated with that cumulative sum: 2534, 3483, 4432, 5381, etc. The resulting sample is stratified by union, by conference within union, and by size.
Principal Investigators
Edwin I. Hernandez
Sergio Hernandez
Mario Negrete
Ramona Perez-Greek
Johnny Ramirez
Caleb Rosado
Saul Torres
Alfonso Valenzuela
Daisy Machado.

Note: The senior PI, Edwin I. Hernandez, is interested in cooperative publications with this data.
Related Publications
Hernandez, Edwin I. (1995). The Browning of American Adventism. Spectrum, 25(2), 29-50.

Leege, David and Michael R. Welch, "Catholics in Context: Theoretical and Methodological Issues in Studying American Catholic Parishioners." Review of Religious Research, V61. 31, No. 2 (December, 1989).
Cleaning Data Procedures for the AVANCE Research Project
The AVANCE research project employed basically the same procedures to check the data for quality and accuracy (a process called "cleaning the data") as those used during the Valuegenesis research project. Given the differences in questions and procedures in the AVANCE project, some additional steps were taken to ensure data quality, which are noted here.

One of the most important issues in survey research is the reliability and validity of the survey responses. While the accuracy and honesty of survey answers can never be guaranteed, there are procedures available to increase the confidence in the data obtained. In the present study, the following procedures were employed:

1. Examining frequencies of missing items: The first method is to determine whether large numbers of questions have been left blank. Blank questions toward the beginning of a survey may indicate reading difficulties; further on, blanks may indicate inappropriate time allotments, fatigue, or inattentive responding, calling into question the validity of responses. In order to address this problem, questionnaires with 10 or more missing items in the first 75 questions were omitted from the analysis.

2. Nonresponse: As was the case in the first 100 items of the core, there was concern that youth or adults whose surveys were retained in the final data set not have undue numbers of omissions in the items specifically relevant to youth and adults. Thus youth and adults who omitted more than 4 items out of the first 10 in their respective sections, beginning with FAITINGD, were omitted from the analysis.

3. Deviance Index: A frequent concern is whether the respondents honestly answer when asked about certain norm-breaking behaviors, such as drug use and stealing. While there is virtually no way to ensure the accuracy of the measurement of such behaviors, it is reasonable to believe that the numbers obtained here are comparable to those obtained in other national surveys, particularly the Valuegenesis survey. Following the procedures in such surveys, respondents who indicate abnormally high levels of deviant behaviors will be deleted.

There are 8 major measures of extreme norm-breaking behavior in the common core section of the surveys. Questions OFTNDRNK, OFTNSMOK, OFTNDRUG, BINGEDRK, and GOPARTY-PREMASEX assess the use of alcohol, smoking, illegal drug use (marijuana, cocaine, etc), binge drinking, going to parties where others are drinking, shoplifting, viewing pornography, and engaging in premarital sex. In each case, respondents were asked to report how many times, during the last 12 months, they did each of these things, on a scale ranging from 1 = "never" to 8 = "more than once a day." The total range is 8 to 64. While it is possible that a given individual may engage in a number of these behaviors "more than once a day," we viewed it as highly unlikely that any one individual would have done all eight of them "more than once a day," or at least unlikely that such an individual would be attending a "youth society meeting" at church to complete the survey. In order to address this issue, all respondents who answered "more than once a day" for each of the eight items for a total of 64 will be omitted from the analysis.

4. Inconsistent Responding: The following procedures for addressing inconsistent responding have to meet the minimum requirement of missing data as discussed above.

Youth: When a respondent filled out a youth survey, but his/her age was 26 and above and/or the respondent was married (indicated by item MARITAL option 3), the responses to core items HLPQUEST to TELENMBR were kept and the rest coded as missing. Responses to core items HLPQUEST to TELENMBR from unmarried youth from ages 13 to 25 who filled out the adult survey were kept and the rest coded as missing. If a youth was not in school (indicated by marking 1 on SCHOOL), then responses to TCHRQUAL to FORGTSPN were coded as missing.

Adults: When adults who were 26 and older and/or who claimed to be married filled out a youth survey, their responses were included to core items HLPQUEST to TELENMBR and the rest of the questions (KIDSMALE to BIGFAMIL) were coded as missing.

5. Substituting missing data: For those respondents who had less than 10 missing items in the first 100 questions (see above) and who had four or less missing items in the first 10 questions starting with FAITINGD of the respective youth and adult surveys, the following procedure was used to deal with other missing data:

The total sample was subdivided by congregation, and parish-specific means of at least 60% of the adults or youth within the congregation were imputed in the missing cases. In other words, if 60% of sampled adult members answered the item, then their mean score was imputed to the remaining 40% who had missing data; otherwise, the missing data remained as missing. This procedure minimizes problems associated with conventional "mean plugging," which reduces sample variance and is viewed as being appropriate when a sample has distinct clusters and subdivisions as is the case with congregational studies (See Leege and Welch, 1989).