Originally, many methods for sex estimation of unidentified human remains were developed with bio-archaeological uses in mind. Bio-archaeologists use these methods more often to identify and get demographic information for historic populations, while in forensic anthropology they are used to identify specific individuals, which requires more accuracy. Over the years, more methods have been developed with a focus on the purpose of identifying specific individuals with higher precision and accuracy.
The most accurate methods for sex estimation used in forensic anthropology include analysis of postcranial long bones, the os coxa and cranium.
In 1963, Eugene Giles and Orville Elliot, published a study on the use discriminant function analysis for estimating sex using the crania. They created 21 different discriminant functions using different combinations of 9 of 11 measurements. When they tested their functions, 82-89% were accurately sexed. This method was developed and tested on a sample of crania of American Black and American White descent. If these functions were to be used for individuals of other ancestry groups, Giles and Elliot recommend adjusting the sectioning point for each group.
In 2005, Tracy Rogers did a study on the accuracy of sex estimation from visual assessment of cranial morphology. Rogers selected 17 traits to be assessed for accuracy. 89.1% of the individuals were accurately sexed using all 17 traits in the assessment. Many of the traits had a high percentage of indeterminate ratings. The analysis suggested traits to be used for sex estimation as follows:
In humans, sexual dimorphism is most present in the pelvis. This is due to the structuring of the female pelvis to accommodate childbirth. The methods most used for the os coxae are primarily visual or non-metric.
In 1969, Terrell Phenice developed a new visual method using the os coxae. The method uses three criteria on the os pubis: the ventral arc, the subpubic concavity, and the medial aspect of ischio-pubic ramus. When tested on a test sample, 95% were accurately sexed.
Since its introduction in 1969, the "Phenice Method" as it is known, grew rapidly in popularity as a viable option, likely due to its simplicity and quickness. The method has been tested for accuracy over the years.
In 1989, Nancy Lovell tested the accuracy of the Phenice method among observers with greatly variant levels of osteological training. In Phenice's original study he claimed the method did not require prior experience. Lovell had observers that included undergraduates, graduate students, and a professional physical anthropologist. She found that prior osteology knowledge had no effect on the accuracy. In the study, around 83% were accurately sexed. The individuals in the sample were 52 years and older. A slight negative correlation between age at death and estimation accuracy was found.