The functional significance of the statistically significant but small effect sizes is yet to be determined. These quantitative differences may not translate into relevant behavioral differences between individuals or groups. Conversely, these small effect sizes may be an accurate representation of these four biological systems, largely because these systems are subject to many physiological interactions with one another, with other biological systems, and with social and epigenetic forces that contribute to phenotypic variations in temperament. Moreover, other studies show very small size effects and suggest that the small effect sizes reported in this paper are appropriate and could be meaningful (de Moor et al., 2010).
Further, it has been argued that almost any data will be significant using a large sample. But statistically significant differences are not inevitable with large samples. They only appear if there is an effect in the population, and they indicate that the effect would still be found with replication. Large samples provide the opportunity to find small but significant effects that normally would be overwhelmed by statistical noise. In fact, small effect sizes are not unusual for studies of large populations (de Moor et al., 2010).
Another limitation is that for the analyses, random samples of the population were not used; instead, the samples were largely based on unmarried individuals who were looking for a partner, who had access to a computer, who were willing to pay to join an Internet dating site, and who felt comfortable using an online dating service.
However, the Internet population we tested represents a significant and important group. Over one–third of the adult U.S. population is single (over 100 million individuals); and with a current divorce rate exceeding 45%, almost half of Americans have been or are likely to become single at some point in their lives (Taylor et al., 2011). The populations examined in this study represent a large and growing percentage of the broad U.S. population and those of several other countries. Subjects also ranged in age from 18 to 88 years; they were from every major ethnic group (e.g., European American, African American, Asian American, and Latino); they lived in rural, suburban and urban areas; and they resided in all 50 states in the U.S., as well as in Canada and six additional cultures, both Eastern and Western.
Last, participants may have skewed their responses to enhance their social desirability. However, participants responding to any questionnaire that uses self-appraisals will approach the task with an array of subliminal and cognitive agendas that cannot be fully screened. In fact, the correlation analyses and the Eigen analysis of the FTI samples are more comprehensive than the samples used in most psychological studies that canvas the attitudes and behaviors of college populations paying a large fee for college entrance, coming largely from similar backgrounds, of the same general age, and sharing similar life styles and life goals.
To further explore the FTI measure, an investigation is underway to assess the relationship between 63 specific alleles and the four FTI temperament dimensions. The essential study of test–retest reliability of the FTI is in preparation as well. To apply these data to life situations, we examined the role of these proposed temperament dimensions in initial mate choice (Fisher, 2009; Fisher et al., 2010b); this investigation continues. Further research could also explore how these four broad proposed styles of thinking and behaving effect one’s proneness to divorce, adultery, and other social, reproductive, cognitive, affective and/or motivational processes, as well as their varying expression in different cultures, different age groups, different occupations, and among those of different sexual orientations and those with different medical conditions.