Quantitative Methods in Relationship Research

Of the various ways that scholars study relationships, the most common involve quantitative methods. The quantitative approach uses statistical procedures to identify trends or regularities in data, which are then examined to yield insights about the subject matter under investigation. A fundamental underpinning of the quantitative approach is that systematic statistical methods, when applied according to conventional standards, provide unbiased tests of the phenomenon or hypothesis in question. Quantitative researchers believe, “When in doubt, count.” In this respect, relationship research is built on the same empirical foundation as other quantitative sciences, spanning nearly all of the biomedical, behavioral, and social sciences.

Relationship researchers use quantitative methods in two general ways. Some studies are primarily descriptive—that is, data are collected from a reasonably large and presumably representative group of individuals and then aggregated, yielding summary statistics. For example, the statements that the divorce rate among Americans married in the 1970s was approximately 50 percent or that women tend to be more emotionally expressive than men are intended to concisely describe general trends within these groups. Other quantitative methods, called inferential methods, are used to test hypotheses. This means that the researchers specify in advance the predicted nature of the association between two (or more) variables and then collect and evaluate data to determine whether the predicted association is correct. Usually these hypotheses are based on theory so that these tests amount to tests of the validity and usefulness of the relevant theories. For example, sociometer theory predicts that when individuals are rejected by members of a valued social group, their self-esteem will suffer. This hypothesis might be tested by comparing the mean level of self-esteem among rejected individuals with that of individuals who had not been rejected. If this test results in a so-called statistically significant difference between groups, the theory is said to be supported. A nonsignificant result implies that the theory was not supported, at least in this particular instance.

Quantitative methods are typically used with data collected from experiments, surveys, or archives, although in principle any data that can be quantified can be analyzed with these methods. A large majority of relationship researchers favor quantitative methods over other methods, such as qualitative methods, because quantitative methods are believed to be more objective. This is because quantitative methods rely on standardized procedures that in principle prevent, or at least substantially minimize, the likelihood of the researcher biasing the findings of research. Another reason that researchers favor these methods is their precision and elegance. With the right method, massive data sets assessing many variables from thousands of research participants can be represented in a reasonably efficient, clear, and informative manner. Indeed, for many researchers, the most fulfilling moment in the entire research enterprise occurs when years of thought and effort culminate in a statistical result. The major disadvantage of quantitative methods is that they do not allow research participants to describe the phenomena in question in rich detail from their own perspective (something that qualitative methods do well).

Quantitative statistical methods began to be developed in the 1870s, with the pioneering work of Sir Francis Galton, who sought a mathematical way to represent the influence of heredity. Others who were highly influential during this early period were Galton's protégé, Karl Pearson, who introduced the correlation coefficient; W. S. Gosset (known then as “Student” to mask his job at the Guinness Brewery), who introduced the t test; and Ronald Fisher, who further developed analysis of variance techniques. Since then, statistical methods have become vastly more complex and sophisticated, so much so that virtually all graduate programs involve extensive training in quantitative methods. Even then, many research projects require consultation with statistical specialists. There is little doubt that the contemporary growth of statistical theory, along with advances in the computational hardware necessary to carry out these tests, will continue and likely escalate these trends.

...