Volume 11 Number 2, 2001
METAMORPHOSIS: A FIELD RESEARCH METHODOLOGY FOR STUDYING COMMUNICATION TECHNOLOGY AND COMMUNITY
University of Kentucky
Sandra J. Ball-Rokeach
University of Southern California
Mary E. Wilson
La Sierra CollegeJennifer Gibbs
Elizabeth Gutierrez Hoyt
University of Southern California
The main research questions are: What is the state of community in Los Angeles -- this prototypical 21st-century city where profound ethnic segregation by residential area is the rule? What are the boundaries of subjective sense of belonging -- neighborhoods confined to several streets, residential areas of several square miles, or larger areas (e.g., is there a Los Angeleno identity?) -- and what is the role of ethnicity in constituting subjective sense of belonging? What combination of communication flows (interpersonal, traditional media, community media, and Internet media) is most strongly associated with community cohesion or community fragmentation? When and why does cross-residential-area contact occur, and does it promote "ties that bind?" Is there evidence of regional community formation that builds upon NAFTA and Pacific Rim political economies as revealed in telephone and Internet communication flows? Do "cyber-communities" and their "high digital" actors contribute to or detract from communities of place?
Metamorphosis: Transforming the Ties that Bind
Los Angeles used to be called a collection of cities in search of a center, but it is even more diverse and challenging than that metaphor would suggest. As the one of the most culturally and linguistically diverse region in the United States (Turner & Allan, 1989). Los Angeles is composed of scores of communities, only some of which have a geographic base, but each of which has a common language, religion, culture and/or ethnicity (Fulton, 1997). Many of these communities have their own media, and many (if not most) use and respond somewhat differently to various modes of communication.
A central goal of the "Metamorphosis: Transforming the Ties that Bind" Project is to articulate and empirically unveil the communication infrastructures of these diverse residential areas in Los Angeles. Our premise is that communication resources (Ball-Rokeach et al., 2000b) are just as important as economic resources in the building of belonging communities. Belonging communities are constructed through communication and can be destroyed through communication. This may not seem surprising, but the communication infrastructure of community is neither well understood, nor well documented.
communication infrastructure consists of the everyday conversations and stories
that people, media, and grassroots organizations create and disseminate, and
the resources of residential areas that promote communication between residents
(e.g., parks, grassroots community organizations, safe streets, libraries,
etc.; for a more detailed discussion of the communication infrastructure concept
see (Kim, Matei,
& Ball-Rokeach, forthcoming). As an entrepreneur has to rely
upon access to the financial and technical features of an economic infrastructure
to build a successful business, the individual citizen or family has to have
access to a supportive communication infrastructure to build their sense and
reality of belonging. The vast landscape of communication flows produced by
people talking with one another, media producing stories, and local organizations
bringing people together, are the milieu of daily life. If this milieu is
noxious or insubstantial, belonging communities cannot form or be sustained.
If it is pregnant with "imagined community" (Anderson,
1991) then there is an infrastructure that people can use to build
In its examination of community transformation, the communication infrastructure perspective draws upon the analytical strengths of many fields of inquiry. For example, the geographer and demographer may track peopleÆs movements across space; the economist may examine market forces; the cultural analyst may focus upon issues of identity; the sociologist may inquire into fragmentation and ethnic conflict; and other urban communication scholars might study the community-building potentials of the Internet. We seek to join all of these concerns through grounded analysis of communication flows. While the macroanalyst seeks to understand global and regional formations, and the microanalyst seeks to understand how individuals are adjusting to massive change, we seek to tap both perspectives by locating individuals within the expanding environment where communities of place face the disruptions and opportunities of cyberspace. Whereas many scholars are studying either high-tech or low-tech communication forms, we seek to connect these by looking at how people and communities integrate new and traditional media into social life.
We see communities as being composed of individuals, family units, and neighborhoods that cohere to varying degrees around issues of place, ethnicity, interests, culture, and organization. We view communities as more than a sum of aggregated individuals, and treat residential areas as meaningful parts to a larger city, regional, or other system. This research program addresses community dynamics through the communication mechanisms that afford coherence and interaction within and between the residential areas that compose Los Angeles. We chose an integrative methodology, rather than an aggregate individual level of analysis, because this approach allows us to look at community cohesion, dynamics, and variations that are embedded in history.
Through our research, we seek to contribute to a larger social dialogue about communication, technology and community by addressing such questions as:
1) Have there been changes in the very definition of community û in the enduring social bonds that sustain shared purpose and trust? What is the state of community in Los Angeles, on of the prototypical 21st century community? Is it possible to have community without sharing a common place?
2) Does the fabric of Internet and traditional communication flows among recent immigrants from North American Free Trade Agreement (NAFTA) and Pacific Rim regions suggest that these political economies have the potential to become regional cultures (Ball-Rokeach, Gibbs, Jung, Kim, & Qiu, 2000a)? Are recent immigrants from these areas less likely to assimilate into the dominant culture (Sidanius, Feshbach, Levin, & Pratto, 1997; Sontag & Dugger, 1998) due to the economic and cultural viability afforded by their regional origins?
3) Predictions for early, pre-Internet communication technologies were often filled with errorùfor example, ignoring the serendipity that comes when people construct how new technologies are incorporated into their everyday community life (Ball-Rokeach, Sturken, & Thomas, 1998). Do contemporary visions of the community transformation potential of the Internet show the same kind of predictive error? To what extent are Internet technologies being employed to maintain and enrich families or neighborhoods (Romney, 1999), or urban community (Ball-Rokeach et al., 2000a; Gibbs, Ball-Rokeach, Jung, Kim, & Qiu, forthcoming)?
Data Collection Overview and Sampling Methods
The multi-method research design we summarize in this paper is motivated by the goals of linkage and comprehensiveness: linkage in the sense of connections between methods of observation and comprehensiveness in the sense of attention to multiple levels of analysis (i.e., individual, interpersonal, organizational and geographic). This complex methodology, which emphasizes the actor-environment interplay, is afforded by letting our study participants drive the research process as much as possible. The multi-method and multi-lingual data collection strategy we employ consists of seven interrelated components: 1) a 40-45 minute telephone survey (RDD, random digit dialing) of approximately 250 households in seven pre-defined study areas; 2) two types of in-person structured focus group discussions -- community issues and Internet-connected "high digitals" -- where participants are selected according to eligibility criteria from the set of telephone-survey respondents in each study area; 3) a series of social mapping activities -- designed to capture peopleÆs affective maps of Los Angeles and to capture their global communication patterns -- involving a combination of focus group participants and other mail-survey respondents who met eligibility criteria, but did not actually participate in focus group discussions; 4) interviews with the staffs of "grassroots" community organizations identified by survey respondents as important in their lives; 5) a census of the communication environment of each study area, including mainstream, community, and public media, public spaces and institutions, and community organizations; (6) interviews with community leaders identified through a grassroots snowballing technique -- e.g., suggested by staffs of grassroots community organizations; and (7) secondary data sets (e.g., crime statistics, travel flow data, and marketing data).
Selection of Study Areas
We begin by identifying large and historically significant residential areas that are dominated by a particular ethnicity. Los Angeles, as one of the most ethnically diverse cities in the United States, is also one of the most residentially segregated by ethnicity (Turner & Allan, 1989). Los Angeles County study areas are selected to maximize: a) residential concentration of the largest ethnic groups in Los Angeles, b) proximity - the center of area is no more than 10 miles from the Los Angeles Civic Center û maximizing connectedness to the life of the city and to other study areas, and c) historical interest ùfor example, the areaÆs role as a transition place for new immigrants or its representation of groups that have dominated the Los Angeles scene. To date, we have studied the following seven residential areas (see Figure 1).
1) South Pasadena, Caucasian-origin (largely Protestant)
2) Westside, Caucasian-origin (largely Jewish)
3) East Los Angeles, Mexican-origin
4) Koreatown, Korean-origin
5) Pico-Union, Central American-origin (mainly Salvadoran, Nicaraguan and Guatemalan)
6) Greater Crenshaw, African-American
7) Greater Monterey Park, Chinese-origin (Mainland, Hong Kong, Taiwan) origin
Figure 1. Metamorphosis Study Areas
The ethnicities represented in these samples constitute approximately 90% of the Los Angeles county population. Also, the countries of origin of some of the groups are in the new regional economies of the Pacific Rim and NAFTA. Mexican-origin persons constitute the largest segment of the Hispanic population in Los Angeles and 86% of our East Los Angeles study area - see Table 1; (Allen & Turner, 1997). The Korean-origin population is a fast growing Asian population (Allen & Turner, 1997) and our Korean-origin study area was at the center of the 1992 Los Angeles riots or rebellion (Hubler & Mitchell, 1997). The Chinese-origin group is the new emerging Asian economic and entrepreneurial "power" in our Greater Monterey Park study area. The Central American groups (largely El Salvadoran, Nicaraguan and Guatemalan) are mainly refugees of the civil wars in Central America in the 1980s and represent a recently immigrated population living in a relatively poor and dense geographic area. The Westside sample includes representatives of "old guard" Jewish power structure of Los Angeles, the group that led the development of the movie and entertainment industry. The South Pasadena/Pasadena sample loosely represents the "old guard" Protestant, English ancestry, financial and industrial leaders in California (Davis, 1990). These two White European-origin groups are also interesting because they, unlike many other Whites, have chosen to remain in areas proximate to the city of Los Angeles, even though many of them have the resources to migrate out of the area. In subsequent phases of our research program we will survey as many other communities as funds allow.
Objective criteria employed in study area selection
Our selection of initial study areas was influenced by the data contained in The Ethnic Quilt: Population Diversity in Southern California (Allen & Turner, 1997), A Human Mosaic: An Atlas of Ethnic Change in L.A. County (Heer & Herman, 1990), and the 1990 U.S. Census. Utilizing ArcView, a geographic information system (GIS) software, we were able to map demographic variables within the census tract  geography of Los Angeles. We were concerned that significant shifts in population might have occurred since the 1990 census data, especially since the 1992 "riots" precipitated an unknown amount of out-migration of Whites, generally, and of Koreans, in particular. We used more recent population data, based on statistical projections provided by market research firms (CACI, 1999). The data generally suggest that the Los Angeles population and its ethnic distribution across areas had not varied substantially since the last census (1990).
A number of factors had to be considered before making the final selection of initial study areas. Some were instrumental in nature, others more theoretical (Conzen, 1991). Among the instrumental issues, obtaining high concentrations of the desired ethnicity was of paramount importance due to the decision to interview only one ethnicity per area by random digit dialing (RDD). Random digit dialing sample selection becomes prohibitively expensive when targeted populations have low concentrations.
Theoretical considerations were also important. In selecting the study sites, it was especially important that their contours have cultural and historical coherence. Thus, we preferred to establish area boundaries that were easily recognizable by their own inhabitants, or by residents of other areas of Los Angeles. Of course, in a fluid and young city such as Los Angeles, this is only an ideal. All communities or neighborhoods are "on the move," their identity being in a continuous transformation. Yet, cultural commonality and historical continuities should not be discounted (Myers, 1999), and our sampling procedure tried to preserve and reveal them.
Briefly described, the seven areas present themselves as follows:
East Los Angeles Study Area and Mexican-origin Sample:
Boyle Heights, at the heart of this 11.4 square-mile area, is a transitional zone for new immigrants at the same time that it is a highly stable area (Allen & Turner, 1997). There are more homeowners that one might expect, and residents live there (and in Los Angeles) for long periods of time. This pattern was evident in earlier times when residents were Eastern European Jewish and Japanese immigrants (the 1920s and 30s). In 1990, Mexican-origin residents constituted 86 percent of the area population (2.3 miles from the Civic Center). They are relatively poor and uneducated. Residents are also relatively young, but have relatively high marriage rates with more people and children living in the home (see Tables 3 and 4 to compare the socio-demographic profiles of our samples with their corresponding study area target population characteristics).
Greater Crenshaw Area and African American Sample:
The center of this 9 square-mile area lies 6.0 miles from the Civic Center. African Americans constituted 80 percent of the population in 1990, but there has been considerable Latino in-migration since. With respect to age, home ownership, and more than two people residing in the home, the Greater Crenshaw study sample falls between the two Latino areas and the two Caucasian samples. It is more like the two Caucasian samples with respect to home ownership, unemployment, having some form of health insurance, first generation immigrants, use of car as primary mode of travel, and residential tenure.
Greater Monterey Park/Alhambra Area and Chinese-origins Sample:
This large study area (30.1 square miles) is 7.6 miles from the Civic Center. By 1990, Chinese-origin people constituted 27 percent of the population. In the 1970Æs, this area, and Monterey Park in particular, was developed as the "Chinese Beverly Hills." It is an ethnically diverse, young, and middle class area. Included in the study sample by design, are people originating from the Mainland China (N = 176), Taiwan (N =72), and Hong Kong (N =48). Seventy-two percent of our combined sample are first generation immigrants. Relative to other study samples, this sample has the lowest percent with health insurance, fairly high access to car travel, high employment, moderate home ownership (46 percent), moderate educational attainment (41 percent college graduates or more), and moderate levels of low-income households (57 percent with less than $35,000/year).
Greater Koreatown Area and Korean-origin Sample:
The center of this ethnically-diverse 6 square-mile area lies 3.6 miles from the Civic Center. Despite the fact that only 22 percent of 1990 residents were Korean-origin (the remainder largely Latino), Koreatown is the major service, cultural, and political center for Korean Americans in Los Angeles County. Korean residents are rich relative to the two Latino samples and more highly educated than the two Latino and African-American samples.
Pico Union Area and Central American-origin Sample:
This is the most ethnically diverse of our study areas and it is closest to the Civic Center (1.7 miles). Pico Union is contiguous to Koreatown. In 1990, 37 percent of residents originated from Central American countries. Relative to the other study areas and samples, Pico Union is the poorest, least educated, youngest, least likely to own a home, least likely to be employed, residence farthest from work, and has the highest use of public transportation as the primary mode of travel. Only 26 percent of sample members have lived in the area for more than 10 years.
South Pasadena Area and Caucasian/Predominantly Protestant Sample:
Most of this 6.0 square-mile area lies within the incorporated area of South Pasadena, but it also extends to Southerly portions of Pasadena. Since the 1870s, this general area has been home to a largely English-origin Protestant/Episcopalian elite (Allen & Turner, 1997). The center of the area is 7 miles from the Civic Center. While the area was relatively homogenous in 1990 when Non-Hispanic Whites constituted about two-thirds of the population, there is a growing population of Latino and Asian residents. Of all study areas, South Pasadena has the best reputation for being a strong community and a good place to raise children. Thirty-nine percent report having more than 2 people in the home, but only 3 percent have more than 2 children living at home. South Pasadena ranks among the more stable areas (49 and 83 percent, respectively, report living in the area and in Los Angeles for more than 10 years).
The Westside Area and Caucasian/Predominantly Jewish Sample:
Historically, this 7.6 square-mile area has been the destination point for upwardly mobile Russian and Eastern European Jews from less affluent areas of Low Angeles (e.g., Boyle Heights and the Fairfax District). Our study area includes large parts of the residential areas of Cheviot Hills, Century City, and Beverly Hills. At its center, this area is 9.7 miles from the Civic Center. As of 1990, it was the least diverse area. The Westside sample is the oldest of our samples (median age is 47 years), and most residents have lived in the area for more than 3 years (78 percent).
Subjective area identification
In order to measure the degree to which our sampling methodology succeeded in carving out neighborhoods that have real meaning for the people inhabiting them, telephone respondents were asked "How do you identify the neighborhood you live in?" Our interest concerned the size of the area identified (e.g., a few blocks, a subarea, or labels referring to areas even larger than our study areas), whether responses suggested conformity to well-defined incorporated areas or unincorporated place labels (e.g., real estate markers), and the number of different names employed. All of our area labels are used in common parlance and in the media to refer to sections in and around Los Angeles. When asked to identify the neighborhood where they live, peopleÆs responses reflect not only an attempt to tell another where they live in a geographic sense, but also how they locate themselves socially.
South Pasadena (Caucasian) and Koreatown (Korean-origin) stand out as having good consensus as to their neighborhood names. In each case, more than three fourths responded with the same name, which is also our area name. In contrast, the majority of Westside (Caucasian) respondents employed well-known and high status real estate area names to identify their neighborhood, revealing a preference for well-known and high-status subarea names. The Greater Monterey-Park/Alhambra (Chinese-origins) study sample used only 8 names to identify their neighborhood in this, geographically large study area (30 square miles). This area has many incorporated subareas, and these names are used by the vast majority of respondents. This study sample had the highest percent of respondents who refused to give any name (9%).
More and more varied responses were given by the Greater Crenshaw (African American) study sample. Most of the 34 names referred to specific real estate, but non-incorporated, areas. Most surprising of these were the percent responding with Los Angeles (15%) or South Central L.A. (13%) names. In contrast to the well-known subareas of the Westside, many non-residents would not know the middle and upper middle class residential subareas of Greater Crenshaw. It is likely that many non-residents would identify the whole area as South Central, which carries a pejorative connotation due, in part, to its association with the 1992 riot/rebellion. ResidentsÆ use of this name may reflect the positive connotations of South Central as an area that has been articulated by visible political and economic development activity.
Of the 18 neighborhood names given by the East Los Angeles (Mexican-origin) study sample, a large minority (44%) gave the East L.A. name. The use of the large East L.A. or, even larger, Los Angeles (10%) neighborhood names may reflect the fact that these respondents live in unincorporated areas that are not well-defined by real estate marking. Generally speaking, the East L.A. name will elicit negative associations in non-residents unfamiliar with the area, but who learn about it from media stories reporting gang or other undesirable events that are identified as happening in East L.A.
Finally, the names given by Pico Union (Central-American) respondents are different from any other study sample. They gave the most names (87) of any study sample. The Pico Union study area is adjacent to, and overlaps somewhat, with the Koreatown study area; yet their responses diverge dramatically. In contrast to the 77 percent who identified their neighborhood as Koreatown, only 5 percent of the Pico Union study sample used the well-known, but poor reputation, Pico Union name. A plurality (31%) use Los Angeles. Also distinctive of this study sample, is the 20 percent of respondents who gave street names to identify their neighborhood. This may reflect an orientation of new immigrants who are not used to thinking in terms of larger neighborhood areas. Another 5 percent did not respond with a name, but used either a positive (e.g., calm or quiet) or negative evaluation (e.g., dirty, getting worse). While the Koreatown and Pico Union areas share a downtown location, both of which have reputation problems, the Pico Union responses suggest a profound disassociation with the name of the area most commonly employed by non-residents.
Overall, it would seem that incorporation, clear definition of real estate subareas, or a business/cultural center label contribute to more consensual naming of a neighborhood. However, different study samples, living in different residential area conditions, vary considerably with respect to the size of geographic referent of neighborhood. Thus, lower name consensus, and a tendency to name larger neighborhood units (e.g., Los Angeles) is more characteristic of poorer unincorporated areas than do not have a Koreatown-type commercial/service area name, and have fewer well-defined real estate subareas.
Stages in the Household Sampling Process
Having selected our initial broadly-defined residential areas, we proceeded in stages to randomly select phone numbers to reach potential survey respondents. The general procedure entails selection of census tracts that fall within a cluster of phone exchanges (smallest geographic unit for phone service defined by the three digits following the area code). Unfortunately, census and phone boundaries do not correspond with one another. Moreover, the geographic range of each exchange is not discrete; any one neighborhood or even street may be served by one or more phone exchanges. This problem has worsened in recent years with the proliferation of phone numbers due to the increased use of new communication technologies (Piekarski, Kaplan, & Prestegaard, 1999).
We began the actual selection process with mapping 1990 Census socio-demographic data (race, ethnicity, ancestry, and income) at census tract levels, using the Geographic Information System (GIS) software, ArcView 3. Then, we selected those tracts with satisfactory population densities for each of our ethnic groups, followed by a geographic statistical analysis by which we determined what phone exchanges serve the selected areas. To insure an optimal "hit rate" and because exchanges straddle each other, and their contours do not correspond to those of the census tracts, we had to select phone numbers from exchanges that fell as much as possible within a census tract defined area, but with as few as possible spilling over into the outside area. The hit rate is the probability that a respondent reached will be of the targeted ethnicity and within the desired geographic area. More specifically,
The unfolding of the process and its theoretical hit rate outcomes are presented in Table 2. Due largely to the lack of convergence among census tracts and phone exchanges, and also to the fact that only one specific ethnicity was selected per area, our obtained average hit rate was 42 percent. The individual area rates vary between an anticipated low level of 13 percent in Koreatown and 16 percent in Pico Union (due to low concentration of desired ethnicities), to 76 percent in Westside.
The complicated nature of this sampling procedure raises the question of why spatially-focused random digit dialing (RDD) is worth the complexity. The well-known RDD sample advantages of bias reduction are particularly evident in the Los Angeles area. It is estimated that 50 percent of Los Angeles household phone numbers are unlisted. Thus, the financial and procedural advantages of employing phone (or other) directories, while considerable, are outweighed by the advantages of the RDD procedure. This is particularly the case when the research design objective is to gain access to representative samples of geographically-located area residents.
ethnicity and location (by zip code) filters were built into the telephone
survey (see next section for a description of the survey contents). The target respondent was an adult 18 years
or older who reported being of the desired ethnicity for their target area:
Introductory remarks included: "May I speak with an adult who is 18 years of
age or older?" To select on ethnicity,
respondents were asked: "How do you
usually describe your ethnicity?" The
response categories read to the respondent were: Latino or Hispanic, Caucasian or White, African American or Black,
Asian/Pacific Islander, Native American, or Other. Multiple responses were accepted. If one of the responses fit the target
ethnicity, the respondent was included in the sample. In South Pasadena, East
Los Angeles and Westside study areas respondents were also screened by location.
Only those living in a selected cluster of zip-codes in each area were included
in the sample. .
The percentage of phone calls made that reached eligible respondents who completed the survey was low: 40.71 percent in East Los Angeles, 40.33, percent in Pico Union, 29.53 percent in South Pasadena, 27.73 percent in Greater Koreatown, 27.22 percent in Greater Monterey Park, 23.99 percent in the Greater Crenshaw sample, and 19.95 percent in the Westside. Our final response rate was calculated by dividing the number of completed interviews by the number of eligible units in the sample (Frankel, 1982). This most conservative way of calculating response rate produces an overall rate of 31 percent (see Table 5). Once the respondents agreed to participate in the study, 62 percent of them finished the interview, providing a good completion rate
There is much discussion within professional survey research circles about declining response rates in studies like ours employing RDD techniques -- e.g., the listserv of the American Association of Public Opinion Research (AAPOR, n.d). A search of the literature, however, produced little systematic study of the RDD response rates obtained in the last decade (Brady and Orren, 1992; Keeter, Kohut, Groves and Presser, 2000). This is surprising, given the frequently-noted problems created by the confounding effects of new technologies (e.g., fax machines, cell phones, caller id, modems, and pagers) that make it difficult to know if a phone number is associated with a household or if the household has more than one telephone number. Moreover, there is a great deal of discussion about increasing resistance of people, generally, to responding to telephone interviews (e.g., due to computerized marketing calls and decreasing time availability). While it is well-established that urban areas generally have lower response, there is no contemporary study that gives us a baseline against which to compare our response rates.
There are several reasons why the response rate is low. Two of the most important are the large percentage of numbers for which eligibility could not be determined due to no answer or answering machines (39 percent of the sampling frame) and the large number of business, modem, fax, or other non-residential numbers (34 percent of the sampling frame; see Table 5). Also, the length of the survey announced to the prospective respondents at the beginning of the interview could be expected to increase rates of refusal or termination (15 percent of the sampling frame).
Several other factors may have lowered response rates including the sensitivity of the ethnicity screener and questions asking: What country or countries did you or your family originally come from? and Who in your family first came to the United States? These questions may have been off-putting especially to people who fear immigration authorities and might be responsible for the 7.4 percent of the prospective respondents who refused the interview at this point in the process.
With these low response rates, we need to more specifically examine how well our obtained sample households match important demographic characteristics of their respective ethnic-specific populations. Keeter et al. (2000) suggest that low response rate should not necessarily translate into significant response bias. Keeter compared two different survey administration protocols, one employing much effort to reach respondents by phone that resulted in 60 percent response rate, and the other employing less effort to reach respondents that resulted in a 36 percent response rate. He then compared differences along 91 variables and found that the average difference was approximately 2 percentage points. More important, few significant differences were found with respect to variables that parallel our major variables (e.g., attention to media, social trust, and connectedness).
We believe that while there are some biases in our methodology, they are within the normal ranges for a survey of this complexity. In Tables 3 and 4, we compare the income, education, age, and gender of our samples with the respective ethnic population characteristics (based upon the 1990 Census) for each area. Briefly, these differences can be summarized as follows:
1. East Los Angeles: study sample includes a higher percentage of women.
2. Greater Crenshaw: study sample is better educated and has higher income.
3. Greater Monterey Park: study sample is relatively older.
4. Pico-Union: study sample includes a higher percentage of women.
5. Koreatown: study sample is considerably better educated and slightly younger and female.
6. South Pasadena: study sample is better educated, older and includes a higher percentage of women.
7. Westside: study sample is better educated and has higher income.
All in all, our best estimates of sample bias suggest an outcome that is not substantially different from outcomes obtained in more conventional and less complex sampling designs. Its biases are in the direction of females and higher income, education and age. With these cautions in mind, we note the advantages of having relatively large numbers of ethnically and residentially-homogeneous study samples that include Non-English speaking persons.
Data Collection Methods
As briefly noted earlier in this paper, our data collection strategy is somewhat distinctive for its multiple and interrelated methods of observation. In sequential order, we start with a 40-45 minute telephone survey that contains closed-and open-ended questions, gathering data from approximately 250 respondents in each of seven residential study areas. In the process of administering the survey, we obtain agreement from appropriate respondents to participate several months later in one or another of two types of focus group discussions that include mapping activities (see below). A key section of the telephone survey concerns respondentsÆ participation in four types of community organizations: political/educational, religious/cultural, neighborhood/homeowner, and sports/recreational. These responses are analyzed to identify those community organizations that are most implicated in the lives of our respondents in each area. Structured interviews are conducted with the administrative staffs of these organizations. Community organizations interview data, supplemented by our ethnographic, historical, and media monitoring activities in each study area, are employed to identify key issues extant in each community and the persons who serve as community leaders. Interviews are conducted with these community leaders about those key issues and related matters. In addition, a comprehensive media census and secondary data collection is conducted in each of the seven areas. This "tissued" mode of data collection enhances our ability to grasp and communicate the dynamics of community life in each study area, particularly as revealed in communication flow patterns. In what follows we are focusing on the most thoroughly deployed to date: the phone survey, the focus groups, socio-spatial mapping, and community organization interviews. The media census and secondary data collection are still works in progress. They will be touched upon briefly in the next section.
The Telephone Survey
An extremely ambitious, but essential aspect of our methodology is inclusiveness; inclusive in the sense that all eligible respondents can participate. This means that the telephone survey had to be accessible to the many people who either do not speak or do not feel comfortable speaking in English. Accordingly, the survey was translated (and back translated) and administered for the initial study areas in Spanish (with Mexican and Central American colloquialisms), Chinese (both Mandarin and Cantonese) and Korean. The interviews were conducted with the help of a subcontractor with whom we worked to train bilingual interviewers. The survey was programmed for Computer Assisted Telephone Interview (CATI) administration. Each linguistic version of the survey was pretested (N = 10) on a separate sample in its corresponding study area. Pretesting objectives included maximizing question clarity, and culling down the total administration time to 40-45 minutes. Constant random monitoring of actual interview process occurred for quality control. Open-ended questions were recorded in the language spoken during the interview and translated into English as needed. The major content sections of the survey are:
Focus group studies were designed to give texture and depth to the data obtained from the telephone interview; that is, they supplement and expand upon issues touched on in the survey and get at more latent attitudes and perceptions. Two types of focus groups were or are to be conducted in each study area: one dedicated to community issues and one to integration of Internet in community life (what we called the "high digital group"). Focus group participants were recruited at the end of the phone interview.
To maximize inclusiveness the approximately two-hour focus groups were conducted in the language of the participantÆs choice (English, Spanish, Korean or Chinese). Two sessions of each type were conducted with 8-10 participants per session. While ethnicity is held constant, we selected from our eligible pool of participants a good mix of gender, age, socio-economic status, and residential subareas. The protocol or sequence and style of questions were pretested on a trial group. Most focus group discussions took place in libraries or in other convenient public facility. Adult participants received $50 for their participation and children received a University of Southern California T-Shirt. Each session was videotaped for subsequent analysis.
Extant community issues focus group:
There were two basic eligibility criteria for recruiting participants for these focus groups:
The key issues probed in the structured focus group discussions were:
À Psychological sense of belonging: spontaneous associations with the words community, neighborhood, and differences between the two, images and feelings about their residential area
À Community issues: participant identification of key issues facing the area and in-depth discussion of education, racial/ethnic relations, transportation, and safety issues
À Most and least liked aspects of living in oneÆs area
À Socio-spatial mapping activities: participants were asked to color in a map of LA, and another of their specific residential area, according to their familiarity and comfort levels with different areas
À Images of other areas: spontaneous associations with the other study areas
À Changes and future predictions: participantsÆ observations on how the residential area has changed, direction they foresee it going, what one thing they would change about their area
À New technologies and their perceived impact on community: participantsÆ connections to new technologies, extent to which they help/hinder sense of community and interactions with neighbors
High digital focus groups
The eligibility requirement for the high-digital adult focus group was that they had completed the telephone interview and were connected to the Internet. For the low rate of Internet-connectedness Latino study samples, eligibility was expanded to computer owners. Eligibility criteria for the child high-digital participants were: 1). a parent had completed the telephone survey and when possible were participants in the adult groups 2) they were between 9 and 14 years old, and 3) they were Internet-connected. The adult and child groups were held separately, but at the same time. All high-digital groups were held at our University site in order to advantage access to a large-screen web access facility where the children could roam sites of their choosing.
The focus groups aimed at capturing how the Internet is integrated in the life of each family and community. We also sought to assess if and how new technologies create an environment where ordinary people are empowered or not in the process of globalization, or if there is a "globalization of everyday life" at work in Los Angeles (Ball-Rokeach et al., 2000a). Another concern, especially among the Asian and Latino groups, was whether "high digital" children are a form of "social capital" for their families helping their parents to reach to their relatives located abroad through new technologies of communication and thus contributing to the creation of new regional communities.
Other concrete topics explored in discussion and in supplemental paper and pencil forms:
Spontaneous associations with the words "computer", "Internet", "globalization"
À The number of "high-tech" gadgets owned
À The breadth and depth of their Internet connectedness
À Semantic scale associations with globalization
À Their everyday life ways of connecting to the Internet, including family and work related activities.
À Degree of connectedness (via travel, telephone, fax or Internet connections) to other countries.
Socio-Spatial Mapping Activities
The GIS methodology was used not only for sampling purposes but also for data analysis. Concretely, we used geographic data for creating and analyzing maps pertinent to local and global processes of social perception and communication. We present some general findings in the process of discussing this method. We do so because it is a method we are adapting to communication research purposes and because the findings should assist the reader in understanding a relatively novel technique.
Los Angeles Comfort Maps
From the pool of telephone respondents who either participated in the focus groups or were eligible to do so, we gathered mapping data from approximately 30 respondents in each area. These data were collected in-person for the focus group participants and through a mail survey for eligible non-participants. All participants were given a black and white map of Los Angeles  including major freeways and community names, and were instructed to color:
a) in black: their neighborhood area
b) in green: the areas where they feel comfortable
c) in yellow: the areas they feel somewhat but not completely comfortable
d) in red: the areas they fell uncomfortable (or they fear)
e) in blue: the areas they do not know (areas left blank were assumed to be unknown and were assigned the color blue).
Socio-affective maps maps are designed to capture both the areas that people imagine as being "good," i.e., places they consider secure and desirable, as well as those that are perceived as insecure or threatening, generative of fear. The resulting mental maps (see Figure 2) are indicative of the way in which people perceive Los Angeles urban space, rather than the actual social and economic health of various areas of the city. In fact, in designing this methodology we started from the premise that while some perceptions might be accurate, a good number of them will be distorted by stereotypes and clichÚs. We also hypothesized that communication media will be at least partly responsible for this distortion. In the "Pulling it together" section that concludes this paper, we present in some detail the methodology used for analyzing socio-spatial data from a communication perspective.
Figure 2. Metamorphosis affective map (N=215). The darker the red, the higher the fear (discomfort), the more intense the green the higher the comfort. The map is oriented North. Blue lines represent highways.
Global Connectedness Maps
Another thirty respondents from each area who are Internet connectors  were asked to provide a list of 10 countries where they know people with whom they communicate by letter, phone, fax, email/chat or in person, by traveling there. The data obtained were standardized in terms of contact frequency per year and then summated into a raw global connectedness index. Using a GIS contour of the world we then generated two maps of world connectedness, one for the new immigrant (Chinese, Korean, Central American and Mexican) and one for the old immigrant (Caucasian and African-American) respondents. The maps, displayed in Figures 2 and 3 indicate that the countries with the most contact closely mirror the countries of origin of new versus old immigrants. New immigrants tend to have the most contact with NAFTA countries (the US, Mexico, Central America) and some Pacific Rim countries (Korea, China, Taiwan, Thailand) û Figure 3. Old immigrants, on the other hand, have the most contact with countries in Europe (the UK, Ireland, Spain, Germany, France, Finland) and the Middle East (Turkey, Saudi Arabia), as well as Liberia, the US, and Canada û Figure 4. This suggests that peopleÆs global connections follow the lines of ethnic affiliation of both new and old immigrants.
Figure 3. New immigrant (Korean, Chinese, Latin-American) respondentsÆ connectedness map
Figure 4. Old immigrant (White and African-American) respondentsÆ connectedness map.
Interviews with Community Organizations
On-site interviews with the staffs of community organizations identified by our survey respondents as important in their everyday lives affords critical information to enrich our assessment of the state of community within and between study areas. In addition to enumerating their respective resources and community-oriented activities, we ask the staffs to give us the benefit of their observations with respect to community cohesion, the social, historical, cultural and economic bases of conflict, and the degree of interarea contact fostered by their organization. At least two of each type of community organization addressed in the survey (political/educational, religious/cultural/ethnic, sports/recreational, and neighborhood/homeowner) are interviewed in each study area. We interview a similarly diverse range of community leaders with respect to their community concerns, perceptions, and activities. In both the community organization and the leader interviews, we incorporate relevant findings from the survey and focus group research to stimulate discussion and to gain their reactions to their constituentsÆ beliefs and behaviors. Finally, an important part of our research action plan is to establish good working relations with these important community figures so that their actions can be informed by our findings and they, in turn, may become disseminators of those findings.
Survey respondents (1812) from the seven areas named 1322 organizations. However, community organizations identified by the respondents from the Greater Monterey Park study sample have not yet been interviewed. Respondents in the remaining six study areas (1491) mentioned 1199 organizations, this group composes the data set from which the organizations were selected for interviews. The process of identifying the organizations began with a review of the raw data, and some preliminary work: correcting spelling errors, standardizing the names of organizations, recording the number of mentions, and deleting multiple listings of the same organizations. This process was repeated for each study area. Korean and Spanish names of the organizations were translated. The next step was to find the organizationsÆ addresses, phone numbers, contact people, and get information about the organizations. Very few organizations were mentioned by multiple respondents; there was not a clear indication of which organizations to interview, by virtue of having a plurality respondents identifying a core group of organizations. However, several organizations were mentioned in more than one category, for instance, some churches provide cultural/religious, political/educational, and sports/recreation opportunities. Organizations that were mentioned by three or more respondents, within a community, or across communities were identified. Then through a combination of snowballing and purposive selection, organizations were selected to be interviewed.
A total of 64 organization interviews in six study areas were conducted. Interviews were conducted between September 14, 1999 and March 9, 2000. The interviews focused on finding out the major issues with which the organizations are involved, learning how this information is communicated to their members and potential members; assessing the amount of discussion and activity these issues generate among the organizationsÆ members; and looking at the organization in the context of their community. This information reveals how organizations position themselves within the communication infrastructure of their communities.
These semi-structured interviews were loosely focused around specific topics, but tailored to suit the particular organization being interviewed. The topics were designed to gather information about the organizations:
À Goals, mission, purposeùpast and present
À Scope of activities
À Avenues used to communicate with members and the public
À Membership eligibility requirements and composition
À Population targeted
À Support and funding
À Issues of governance and leadership
À Self-Identification of its role in the community
À Self-Identification of involvement in any organizational network
Quite probably because of the very concrete cultural and other inherent differences between the study areas, the organizations in each area differ considerably from one another. Even the organizations that share common roots and missions (YMCA, PTA, Neighborhood Watch, Senior Centers) are distinctively different between the areas. There are vast differences in the amount of resources available for the organizations on the Westside and South Pasadena (areas mainly populated by White, higher socio-economic residents), in contrast to the organizations in Pico Union and East Los Angeles (with residents of primarily Mexican and Central American origin). The needs of the respective area residents also differ significantly. The organizationsÆ facilities, services offered, staff, and programs provide a study in contrasts that will be examined in Wilson (forthcoming).
Pulling It All Together
We combine the telephone interview, focus group, and community organization/leaders data to create a contextualized understanding of the dynamic interplay of social and cultural forces at work in the transformation of community in Los Angeles. Each research method already has added value to the others: for example, the selection of focus group participants from the pool of telephone interview respondents enriches the telephone interview data, and vice versa; selecting community organizations from telephone respondentsÆ identification of their most important organizational affiliations establishes a textured link between individuals, families, and community organizations. All of these interrelated data sets were brought together into our two main tools: multidimensional communication maps and the Total Communication Environment Diagnostic methodology (TCED).
For the study to have maximum applicability for area researchers, journalists, industry, and policymakers, we seek to visualize the findings in a form that heightens comprehensibility and adds value to the data sets. Specifically, we employ the latest mapping technology to create three-dimensional social topographies that afford insights into community dynamics that might not otherwise be gained from more conventional data analytic tools. We create maps that let people see the complexity of interaction between communication flows, community cohesion and sense of belonging, community participation and links to organizations, transportation flows, ethnic identity, and demographic characteristics. Our investigations suggest that we can take advantage of the latest geographic mapping developments that offer something like an animated or moving virtual reality of a social topography. For example, the animated Figure 5 depicts a three-dimensional image of the study areas where the amount of time that respondents spend weekly on-line is represented by the elevation of the landscape.
Figure 5. Weekly time spent on-line projected in three dimensional space for the 7 Metamorphosis neighborhoods.
Note: The lower "mounds" in the center represent East Los Angles, Koreatown and Pico-Union; the relatively greater "heights" in the middle represent Crenshaw. The higher "mountains" on the extremities represent Westside (to the left) and South Pasadena/Greater Monterey Park to the right.
Based on a special geographic methodology of projection, the steepest mountains represent areas where respondents spend the longest time on-line: Greater Monterey Park, the Westside, and South Pasadena. In contrast, the East Los Angeles and Pico Union cyberlandscape is characterized by low mounds, indicating weak Internet connectivity.
Using the same geographical-statistical package (ArcView) we have processed the comfort maps digitally and generated a series of two and three dimensional maps of fear / comfort. For example, in Figure 5 elevation represents census tract income levels (the higher the terrain, the higher the 1990 median income of the inhabitants in each census tract). Over this income landscape, we have projected the areas that are most feared (in red) or perceived as being comfortable (in green) by out study respondents. The immediate association made is that relatively wealthy areas are perceived as comfortable and relatively poor areas are feared.
These maps go beyond the heuristic value of communicative description. Employing spatially oriented statistical packages (S-Plus) we can generate comfort maps that vary in terms of media connectedness and then measure their goodness of fit with socio-demographic variables. For example, Figure 7a-b depicts 2 maps: one (a), generated from people who connect to television to attain understanding, orientation, and play goals (Ball-Rokeach, 1985; Ball-Rokeach, 1988), and the other (b) from people who do not connect to television for these purposes. Comparison of these two maps Figure 6a and 6b) readily reveals a pattern whereby television connections are associated with weaker comfort (less intense green) and stronger fear (more intense red).
Figures 6 a-b. Spatial Comfort levels for Los Angels county areas for (a) television connectors and (b) non-television connectors.
Spatially oriented statistical analysis can also provide exact assessments of the degree to which comfort is significantly associated or not, from a statistical point of view, with any number of area characteristics (crime, population diversity, etc.). In this case fear is associated with presence of minority populations but not with area crime levels. For more detailed analysis and discussion see Matei, Qiu and Ball-Rokeach (forthcoming).
Total Communication Environment Diagnosis (TECD)
We are generally concerned that our research findings have practical utility for policy analysts, urban journalists, and the many social action organizations that deal with one or another urban problem. To further this aim, we report our findings in both academic and policy venues (i.e., journal article and White Papers). One illustration of substantive outcomes is a White Paper entitled "The challenge of belonging in the 21st Century: The case of Los Angeles" which we released on our web page in May 2000 (Ball-Rokeach et al., 2000b). We report a multivariate regression / path-analysis model of the interaction between a large array of communication (multiple media, interpersonal, and organizational), location-specific variables, and an index of belonging to urban spaces that we developed on the basis of the rich literature on community integration and attachment (Chavis & Wandersman, 1990; Hui, 1988; McLeod et al., 1996). The end product is a model in which we can observe the force of major elements in the communication infrastructure of belonging.
How can this model be used in day-to-day life? For example, an agency seeking to reach diverse populations with public safety materials can readily grasp a detailed picture of unique combination of important communication actors in each residential environment. Or community builders can identify the weakness in the communication infrastructure that undermine belonging and design tailored strategies of intervention to promote belonging communities. Thus, the TECD model affords knowledge of residentsÆ communication patterns that can be applied in a direct way to understand why, how and how strong they "belong." This affords a unique insight into the relative degrees of community cohesion evident in areas with different ethnic populations and different communication infrastructures.
Relevance for Explorations of the Effects of the Internet on Social Life
Of the many features of the Metamorphosis Project that bear upon frequently-asked questions about the consequences of the Internet, several are of particular importance to the research team. The first is how the Internet is being incorporated into the communication infrastructure of everyday life. Because we include all major forms of contemporary communication -- interpersonal, traditional media, community media, and Internet media -- we will be able to explore how Internet media are changing the functions of other communication forms and their centrality to the conduct of personal and community life.
A potentially important tool we have developed to this end is the Internet Connectedness Index (Ball-Rokeach et al., 2000a). This measure is designed to capture the scope, depth, and centrality of Internet incorporation into the conduct of everyday life. It includes historical indicators (e.g., time since initial Internet connection), indicators of goal scope or the number of personal goals people seek to attain through Internet connections, evaluative indicators (e.g., assessment of whether the Internet has had a generally positive/negative effect on the respondents life), and other dimensions. This ICI index is particularly useful in investigating the question of the digital divide, i.e. unequal access to computers and the Internet. Most discussions approach the issue in terms of who gets access and who does not, and how long people stay online. These approaches reduce all discussions about social inequality to computer / network access. The unstated assumption is that simple access to the Internet will solve the problem of social inequality, in general. However, what do people do once they get connected to the Internet? How do they incorporate new media into their everyday lives? Are the concrete uses of the Internet helping social advancement, improving job and social skills, or will they replicate the "couch potato" effect of television?
Another avenue of inquiry is the extent to which Internet connectedness detracts from, or contributes to, communities of place. Of particular interest is the question "how pre-existing social connections to family, community organizations or churches contribute to social participation on-line?" Are the two types of connections, to real and virtual space, engaged in a zero-sum game, participation in on-line groups growing at the expense of involvement with real neighborhoods? Also, are there variations in this relationship according to the kind of community of place (e.g., dominant ethnicity, SES variations, or degree of community cohesion)?
Finally, what are the utopian and dystopian visions of Internet technologies held by everyday people, and do these visions vary by class, ethnicity, and gender? Technological visions give us accounts of the hopes, fears, and cultural receptivity to the cyber mode of human contact engendered by the Internet. Having open-ended data on the technological visions of our study groups will allow us to extend this discussion beyond the elite discourse that tends to dominate scholarly analyses, to introduce everyday peopleÆs points of view.
Our long-term vision is to create a contextualized understanding of how Los Angeles is evolving into the 21st century that will directly contribute to academic inquiry and also to social action efforts designed to reduce the costs of change for people and their communities. While the financial costs and range of requisite staff linguistic and research skills to implement this research design are considerable, they seem warranted by the prospect of theory-driven communication research whose very comprehensiveness increases policy and action potential.
Table 1. Target Population and Metamorphosis Sample Percentages of Los Angeles Population
* Target population: Non-Hispanic White, Mexican, African American, Korean, Chinese, Central American origin people
** Specific Ethnic Target population: Mexican in East Los Angeles, African American in Greater Crenshaw, etc.
Table 2. Statistical data identified in the selection of phone exchanges for target communities
Table 3. Profiles of Residential Areas from which Study Samples were drawn
Table 4. Profiles of Study Samples
Table 4. Profiles of Study Samples (continued)
Table 5. Response rates
Table 5. Response rates (continued)
Ball-Rokeach, Gibbs, J., Jung, J.-Y., Kim, Y.-C., & Qiu, J. (2000a). The globalization of everyday life: visions and reality (White Paper 2). Los Angeles: Annenberg School for Communication. Metamorphosis Project. [On-line]. Available: http://www.metamorph.org/vault/globalization.html.
Ball-Rokeach, S., Gibbs, J., Gutierrez Hoyt, E., Jung, J.-Y., Kim, Y.-C., Matei, S., Wilson, M., Yuan, Y., & Zhang, L. (2000b). The challenge of belonging in the 21st century: the case of Los Angeles (White Paper 1). Los Angeles: Annenberg School for Communication, Metamorphosis Project. [On-line]. Available: http://www.metamorph.org/vault/belonging.html
Ball-Rokeach, S. J., Sturken, M., & Thomas, D. (1998). Technological visions: utopian and dystopian perspectives conference website, [Website]. Annenberg Center and Annenberg Schools for Communication. Available: http://www.metamorph.org/confer/index.html [2000, July 1].
Conzen, K. N. (1991). Immigrants, immigrant neighborhoods, and ethnic identity: Historical issues. In G. E. Pozzetta (Ed.), American immigration & ethnicity--ethnic communities formation and transformation (Vol. 3, pp. 97-109). New York: Garland Publishing.
Gibbs, J., Ball-Rokeach, S., Jung, J., Kim. Y., & Qiu, J. (forthcoming). The globalization of every day life: visions and reality. In M. Sturken, D. Thomas, & S. Ball-Rokeach (Eds.), Technological Visions: Utopian and Dystopian Narratives .
Keeter, S., Kohut, A., Groves, R. M., & Presser, S. (2000, January). Consequences of reducing nonresponse in a national telephone survey. Keeter, Scott. Available: http://mason.gmu.edu/~skeeter/nonresponse.zip [2000, July].
McLeod, J. M., Daily, K., Guo, Z. S., Eveland Jr., W. P., Bayer, J., Yang, S. C., & Wang, H. (1996). Community integration, local media use, and democratic-processes. Communication Research, 23(2), 179-209
Piekarski, L., Kaplan, G., & Prestegaard, J. (1999, May, 15). Telephony and telephone sampling. The dynamics of change. Paper presented at the Annual Conference for Public Opinion Research, St. Petersburg, Florida.
Sontag, D., & Dugger, C. W. (1998, July 19). New immigrant tide: Shuttle between worlds. New York Times, pp. Available on-line: http://search.nytimes.com/search/daily/bin/fastweb?getdoc+site +site+15827+2+wAAA+immigrant%7Etechnology
 Census tracts are geographic
units, defined by the United States Census Bureau, usually determined by
major roads whose area contours vary considerably from less than one to
more than ten square miles. The population contained in a census tract varies
between several hundreds to several thousands.
 To verify the accuracy of our spatial targeting technique all respondents were asked to indicate the closest cross streets to their home. The same GIS software used for sampling was then employed for locating each respondent on the map and for counting how many respondents actually live in the selected areas. Generally, the accuracy proved to be very high, around 90% of the respondents (regardless if they were screened for location or not) living in the target census tracts.
 Completion rate defined as (the Completed Interviews) divided by (the Completed Interviews + the Suspended Interviews + the Terminated Interviews + Qualified but Refused).
 The mail respondents were contacted by phone, after the survey, and asked to help out with a second part of the study, involving maps for which they were promised two free movie tickets. A mapping kit including the same black and white map of Los Angeles, detailed coloring instructions and crayons were sent through mail. The completed maps were sent back by the respondents in a self-addressed envelope included in the mapping kit.
 This mapping activity was carried during focus groups and through mail-outs, the recruiting methodology being similar to the one used for the affective/comfort maps.
Copyright 2001 Communication Institute for Online Scholarship, Inc.
This file may not be publicly distributed or reproduced without written
permission of the Communication Institute for Online Scholarship, P.O.
Box 57, Rotterdam Jct., NY 12150 USA (phone: 518-887-2443).