Analyzing Existing Statistics
Secondary analysis may involve obtaining raw data collected by other researchers and undertaking a statistical analysis of the data, or it may involve the use of other researchers' existing statistical analyses. In analysis of existing statistics. the unit of analysis is often not the individual. Most existing statistics are aggregated: They describe a group. Durkheim wanted to determine whether Protestants or Catholics were more likely to commit suicide; however, none of the available records indicated the religion of those who committed suicide. Although Durkheirn suggested that Protestants were more likely to commit suicide than Catholics, it was impossible for him to determine that from the existing data.
In a contemporary study of suicide, K. D. Breault (1986) analyzed secondary data collected by government agencies to test Durkheims hypothesis that religion and social integration provide prctection from suicide. Using suicide as the dependent variable and church membership, divorce. unemployment, and female labor force participation as several of his independent variables. Breault performed a series of sophisticated statistical analyses and concluded that the data supported Durkheirns views on social integration and his theory of egoistic suicide. He also found support for Durkheirns proposition that Catholics are less likely to commit suicide than are Protestants. However, it should be noted that Durkheim did not attribute lower rates of suicide among Catholics to the role of church beliefs as much as to the tendency of Catholicism to promote social integration through rituals and regulation of standards of faith and moral conduct (Ellison, Burr, and McCall, 1997). Numerous other studies have used secondary data to examine the relationship between religious factors and rates of suicide. For example, in a recent study researchers used data from sources including the J ational Center for Health Statistics and a large survej of religious denominations to examine the extent to which religious homogeneity-how well community residents adhere to a single religion or a small number of faiths-is associated with lower suicide rates (Ellison, Burr, and McCall, 1997). Using larger categories such as conservative Protestant, moderate Protestant. Catholic, Mormon, Cohort and Jewish, the researchers concluded that religious homogeneity
was linked with lower suicide rates, particularly in the northeastern and southern United States (Ellison, Burr, and McCall, 1997).
Analyzing Content Content analysis is the systematic examination of cultural artifacts or various forms of communication to extract thematic data and draw conclusions about social Iife, Cultural artifacts are products of individual activity. social organizations. technology. and cultural patterns (Reinharz, 1992). Among the materials studied are written records, such as diaries. love letters, poems. books. and graffiti. and narratives and visual texts, such as movies. television programs. advertisements. and greeting cards. Also studied are material culture, such as music, art. and even garbage. and behavioral residues; such as patterns of wear and tear on the floors in front of various exhibits at museums to determine which exhibits are the most popular (see Webbetal., 1966). Harriet' Martinet stated that more could be learned about a societ;y, in a day by studying "things- than by talking with Individuals for a year (Martineau, 1988/1838). Researchers may look for regular patterns. such as frequency of suicide as a topi~ on television talk shows. They may also examine subject matter to determine how it has been handled. such as how the mass media handle the subject of suicide (see Box 2.3 for an
example). Content analysis provides objective coding procedures for analyzing written material (see Berg. 1998; Manning and Cullum Swan, 1994). It also avows for the counting and arranging of data into clearly identifiable categories (manifest coding) and provides for the creation of analytically developed categories (latent or open coding). Using latent or open coding. it is possible to identify general themes. create generalizations. and develop "grounded theoretical- explanations (Glaser and Strauss. 1967)_ As this explanation suggests. researchers use both qualitative and quantitative procedures in content analysis.
How might a social scientist use content analysis in research on why people commit suicide? Suicide notes and diaries are useful forms of cultural artifacts. Suicide notes have been subjected 19 extensive analysis because they are "ultra personal documents" t.\at are not solicited by others and frequently are written just before the person's death (Leenaars, 1988: 34). Many notes provide new levels of meaning regarding the individuality of the person who committed or attempted suicide. Suicide notes and diaries often reveal that people committing suicide consider their death as a "passing on to another world" or simply "escaping this world:' Some notes indicate that people may want to get revenge and make other people feel guilty or responsible for their suicide: "Now you'll be sorry for what you did" or Alt's all your fault!" Thus, suicide notes may be a valuable starting point for finding patterns of suicidal behavior and determining the characteristics of people who are most likely to commit suicide (Leenaars, 1988). Today, researchers analyze the suicide notes of both women and men. However, earlier studies of suicide notes primarily focused only on those written by men, e~en though women have been found to leave notes more often than do men (Lester, 1988, 1992), Strengths and Weaknesses of Secondary Analysis One strength of secondary analysis is that data are readily available and inexpensive. Another is that, because the researcher often does not collect the data personally, the chances of bias may be reduced. In addition, the use of existing sources makes it possible to analyze longitudinal data to provide a historical context within which to locate original research. However, secondary analysis has inherent problems. For one thing, the may be incomplete, unauthentic, or inaccurate. A second issue is that the various data from which content analysis is done may not be strictly comparable with one another (Reinhart, 1992), and coding this data sorting, categorizing, and organizing them into conceptual categories-may be difficult (Babble, 2004