Date of Award


Document Type


Degree Name

Doctor of Philosophy (PhD)



First Advisor

Donald Williamson


This study describes the development of a self-report inventory to measure bereavement. The inventory was called the Hayes Bereavement Inventory (HBI). Items for the inventory were provided by four grief therapists with at least three years of experience in bereavement counseling. Items were written in a format similar to the Beck Depression Inventory. Four hundred and forty adults bereaved in the past five years completed the bereavement inventory, three widely used self-report measures of depression, anxiety, and anger, and the Eysenck Personality Questionnaire. Respondents lost loved ones to death from a variety of causes which included medical diseases, car accidents, cancer, suicide, and homicide. Relationship loss included immediate and extended family members and close friends. Odd or even numbers were alternately assigned to each respondent during the data collection period. An exploratory factor analysis of the HBI was computed on the odd-data set. This produced a Three Factor Model of the HBI that was tested on the even-data set using the EQS confirmatory factor analysis program. Based on structural equations derived from the odd-data set, the Three Factor Model was a poor fit based on the obtained covariance matrix of the even-data set. Three additional methods for comparing two factor pattern matrices indicated that Factor One, called Bereavement Pain, was a strong factor replicated in the even-data set. Factors Two and Three, called Religious Doubt and Conflicted Relationship, were not replicated as strongly. Nevertheless, item loadings for these two factors were still significant in the even-data set and in the total sample. The HBI and Bereavement Pain had Cronbach alphas of.94. The HBI had a test-retest reliability coefficient of.93 over a period of two weeks. The HBI and Bereavement Pain each had a correlation of.59 with the Beck Depression Inventory. A research agenda addressing reliability, replication, and construct validity was discussed.