fl“... Eigfi’ééfii' %§€?iga2‘a estate Umvetsizy This is to certify that the dissertation entitled THE RELATIONSHIP BETWEEN POST-TRAUMATIC STRESS DISORDER SYMPTOMS AND SIX PSYCHO-SOCIAL VARIABLES presented by Monica Anne Green has been accepted towards fulfillment of the requirements for Ph.D. Education degree in ‘ 1 Major professor Date W MSU is an Affirmative Action/Equal Opportunity Insritulion 0-12771 MSU LIBRARIES .—3_. RETURNING MATERIALS: PIace in book drop to remove this checkout from your record. FINES wiII be charged if book is returned after the date stamped below. THE RELATIONSHIP BETWEEN POST-TRAUMATIC STRESS DISORDER SYMPTOMS AND SIX PSYCHO-SOCIAL VARIABLES By Monica Anne Green A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY College of Education ABSTRACT THE RELATIONSHIP BETWEEN POST-TRAUMATIC STRESS DISORDER SYMPTOMS AND SIX PSYCHO-SOCIAL VARIABLES BY Monica Anne Green The purpose of this study was to rigorously define, operationalize and predict pre-service, service and post-service psycho-social variables associated with Vietnam veterans with post-traumatic stress disorder. Sixty Vietnam veteran subjects from a Mid-western VA Medical Center were surveyed for symptoms of post-traumatic stress disorder. Characteristics of those sixty subjects were analyzed to determine their relationship to symptom outcomes. These included: nature of entry into service, intensity of combat experienced in Vietnam, current subective impact of the previously experienced stress of Vietnam experiences, current level of life stress, extent and nature 0f social support available to the veteran during the first year of return from Vietnam, and pre-service psycho-social functioning. Multivariate tests involving discriminant function and linear regression analysis were conducted. Both analyses revealed that the intensity of combat experienced and the current subjective in?“t 0f the previously experienced stressor of duty in Vietnam were most highly associated with current post-traumatic stress disorder symptoms. Univariate tests of correlations and analysis of variance also supported the above findings. In addition, current levels of life stress, especially disruption in interpersonal relationships, also were found to be significantly associated with post-traumatic stress disorder symptoms. These results support the findings of related studies of natural disaster victims and previous studies of the etiology and correlates of post-traumtic stress disorder symptoms. These data provide support for the existence of a quantifiable constellation of symptoms associated with psychological sequelae of severely stressful trauma. ACKNOWLEDGMENTS I wish to gratefully acknowledge financial support for completion of this study through a pre-doctoral research fellowship grant from the Veteran's Administration, Health Services Research and DevelOpment Division. The time and effort of dissertation Chairperson, William Farquahar, are recognized. The academic guidance, encouragement and personal support of Drs. Rebecca Henry and Rochelle Habeck were helpful and appreciated. Drs. Ken Chamberlain, Spencer Falcon and Colleen Ryan contributed time and expertise in conducting the psychiatric ratings. Dr. Chamberlain helped me develop the original research questions during his supervision and guidance of my internship work. Drs. Robert Gunn and Stanley Berent of the Ann Arbor VA Medical Center provided encouragement, ideas, and editing of the grant proposal. Dr. Dan Roarke contributed many hours of assistance as statistical consultant. Through his kind effort, I learned more about statistics and research methods than in any formal academic setting. There are many friends and relatives who provided their support, encouragement and patience as I labored in the last hurdle--the dissertation. I wish to thank my parents for the values of hard work and perseverence which they helped to instill in me. These values are reflected in the completion of this degree. Friends Jan, Betsy, Grace, Cheryl, Jim, Dick and Carolyn helped me with their kindness and thoughtful patience during the difficult and lonely task of completing this dissertation. Ray Howe provided the editorial assistance of a colleague and the support, kindness and encouragement of a fine friend. I wish to acknowledge the help provided by those Vietnam veterans who voluntarily provided time and effort in the conduct of this study. Without their generous and sometimes painful efforts, this study would not have been completed. Lastly, I wish to dedicate this dissertation to me -- who persevered through this arduous and trying process of completing this dissertation and ultimately the Doctor of Philosophy degree. I am so proud! CHAPTER I: CHAPTER II: CHAPTER III: CHAPTER IV: TABLE OF CONTENTS INTRODUCTIONIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIl PrOblemI I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I 1 need I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I 1 Purpose I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I 2 ResearCh HypotheSiSIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII3 TheorYIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII3 REVIEW OF LITEMTUREIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII12 IntrOduct ion to Rev ie' I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I 12 “View Of LiteratureI I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I13 cone lus ions to ReView I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I 25 DESIGN OFTHESTUDYIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII29 Research Hypotheses..................................29 Sample...............................................30 Method of Sampling...................................3l Selection Criteria...................................33 Exclusion Criteria...................................36 Grouping of Sample...................................37 Sample Size..........................................37 Measures.............................................38 Composition of Dependent Variable....................49 Composition of Independent Variables.................50 Design...............................................52 Analysis.............................................S3 Summary..............................................55 MALYSIS OF RESULTSIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII56 Final Sampling Procedure and Outcome.................56 Independent Variable Means and Standard Deviations...58 Sample Characteristics...............................59 Psychometric PrOperties of Scales....................67 Verification of PTSD Symptoms........................72 Assessing Response Set Bias..........................76 Assessing Accuracy of Self-Reported Data.............89 Results of Data Analysis.............................92 Summary..............................................98 CHAPTER V: SUMY AND CONCLUSIONSsoeoooososoesosssssoosssssssolOl Summary.............................................lOl Conclusions.........................................lOS Discussion of Significant Findings..................107 Discussion of Non-Signficant Findings...............llZ Choice of Independent Variable......................ll6 Sampling Procedure..................................118 Rating of PTSD Symptoms.............................lZO Efforts to Verify Collected Data....................121 Implications for Future Research....................122 “HEROESIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII126 ”PWDICESIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII131 CHAPTER ONE THE PROBLEM The "Vietnam Era" was defined by Presidential proclamation as the period of time from August 5, 1964 to May 7, 1975. It is estimated by the Veteran's Administration that over 9.1 million persons served in the armed forces during that span of time. The prevalence figures regarding how many persons served specifically in Vietnam and how many now experience psychological difficulties are unclear. Ewalt (1981) tendered an approximation of 2.8 million persons serving in Vietnam, with 222 of the survivors receiving service-connected compensation at 502 or more. One in five of these individuals is being compensated for a psychiatric disability. He further estimated that "...of the total-discharges in a year from the VAMC's (VA.medical centers) psychiatric discharges show 391 are Vietnam Era veterans." He estimated that 202 of Vietnam Era veterans are experiencing "severe (adjustment) problems." NEED Assessing the prevalence and nature of psychological problems of Vietnam veterans has proved difficult. The President's Commission on iMental Health (1978) suggested that most of the mental health problems (sf Vietnam Era veterans occur among those who served specifically in Vietnam, also estimating this population to be 2.8 million persons. The figures for the percentage of psychologically troubled out of this 2.8 million range from 202 to 502, according to various research findings (Center for Policy Research, 1979; Figley, 1978). The purpose in presenting these discrepant and confusing estimates of the problem is to demonstrate that it is at this basic level that the confusion regarding the Vietnam.veteran and his psychological difficulties begins. Current research findings on the etiology of Post-Traumatic Stress Disorder (hereafter referred to as PTSD) are varied and in conflict. Critical variables contributing to symptom formation are thought to be: 1) combat experience, 2) pre-service social adjustment, 3) voluntary or involuntary nature of service rendered, 4) a variety of post-war personality and attitudinal correlates, and 5) numerous factors related to the socio-political climate in the United States and in Vietnam.during the Vietnam.era. PURPOSE The purpose of this study was to identify pre-service, service, and post-service psycho-social variables associated with Vietnam veterans reporting Post-traumatic Stress Disorder (PTSD) symptoms. In addition to examining the associated variables of PTSD, clarification Of the interaction among these variables was explored. The results of this study are discussed within the context of Previous research findings related to those of the work presented in Chapter 2. These findings help to clarify the relationship of pre-service, service, and post-service variables to the current existence of symptoms of PTSD. Such research contributions may assist in the understanding of etiological factors of the disorder. Understanding the etiology assists in developing helpful treatment approaches to remedy symptoms of PTSD. Most importantly, understanding factors correlated with PTSD may have implications for interventions designed to prevent the development of severe, chronic, and disabiling symptoms of PTSD. RESEARCH HYPOTHESIS The generic research hypothesis of this study was: There is a positive correlation between the existence of PTSD symtoms and pre-service psycho-social functioning, nature of entry into service, combat experience, social support upon return from military service, current subjective experiencing of the earlier traumatic stressor, and current levels of life stress. THEORY The concept of war neurosis has been observed since the early laistory of armed conflict, with reference being made to it as an observable entity in the Greco-Roman wars. Although it has been known by a variety of labels, e.g., "nostalgia," "traumatic neurosis," "shell shock," the symptoms observed have been similar. The DSM-III criteria for Post-traumatic Stress Disorder are as follows: I. The existence of a significant stressor II. Re-experiencing of the traumatic event as evidence by at least one of the following: A. Instrusive memories of the event R. Recurrent dreams of the trauma C. A sudden feeling or acting as if the traumatic event were recurring, triggered by an environmental or emotional stimulus III. Numbing of experience, as evidenced by: A. Diminished interest in activities B. Feelings of alienation C. Constricted affect IV. At least two of the following symptoms that were not present before the trauma: A. Exaggerated startle response to stimuli 8. Sleep disturbance C. Survival guilt D. Difficulty concentrating E. Avoidance of activities that arouse recollection of the traumatic event ~F. Intensification of symptoms by exposure to events that resemble the traumatic situation. However, identification of a discrete symptom complex is only the first step toward effective psychological intervention. The emotional arousal which is induced by a person's perception of his environment as threatening can be an adaptive response facilitating survival. Conversely, the response itself, if overwhelming, chronic and inappropriate to environmental stimuli, can lead to psychological disintegration and disability. In the investigation of an individual's reaction to stress, attention has been paid to the nature of the stressor itself, the psychological structure of the individual, the physiological mechanisms which are activated in reaction to the experience of stress, and the social context in which the individual exists. Initial formal theoretical speculation regarding the etiology and dynamics of "traumatic neurosis" are frequently attributed to Breuer and Freud (1895). Freud observed We apply the word 'traumatic' to an experience which within a short period of time presents the mind with an increase in stimulus too powerful to be dealt with or worked off in a normal way, and thus must result in permanent disturbances of the manner in which the energy operates. He emphasized the repetitive recollection and intrusion of the traumatic experience as an attempt at mastery of the traumatic CXPerience . Grinker and Spiegel (1963) in observing "war neurosis" of World War II veterans paralled Freud's analytic approach stating: The reaction to the stimuli of combat depends upon the meaning given to these stimuli in terms of recognizing them as a threat and of feeling confident of the ability to neutralize the threat.......Porces, located in the superego or in the external environment, demand a continued attempt to master the threat, and at their behest the ego attempts to inhibit or control the anxiety. (If) the ego loses its inhibitory power, anxiety dominates more and more, becoming an ungovernable force which swamps the ego and disintegrates its function. Taking a macroscopic view which incorporated more than dynamic theory, Crinker postulated that the etiology of the condition lay in the interplay among the soldier's personality and character structure, his training and confidence, his personal and combat unit morale, and other impinging situational stressors. He thought that any person, under sufficiently intolerable stress, had the potential to develop "war neurosis“ symptoms. Rado (1942), distilling psychological thought on the "traumatic syndrome" seen in combatants of World War II , drew from Freud's eXp10rations of trauma in formulating his theory. He emphasized ' a view which is echoed by "neurosis as a form of adaptation,‘ contemporary theorists in PTSD research- (Horowitz, 1976; Lifton, 1973). According to Redo, symptom formation was a by-product of the intra-psychic experiences of "ego contraction" which he defined as the "shrinking of the organism's inner resources" (in the face of overwhelming trauma) and "(ego) disorganization rather than disintegration." Lifton (1982), in an extension of Rado's views, perceives the development of PTSD as inevitable for anyone exposed to a sufficiently catastrophic trauma. He stated that "...predisposition is only a matter of degree." He elaborated upon the symptom complex by emphasizing the development of the "death imprint...the radical intrusion of an image-feeling of threat or end to life." Further, he stated that there is a tendency to cling to the "death imprint" through imagery and feeling in an attempt to master and assimilate the traumatic threat to life, as Freud also noted. Even though much of Horowitz' work has been directed toward treatment, he has also explored those factors related to etiology and the personal meaning of the traumatic stressor. His theory has been an integration of object relations theory and developmental understanding of character structure. The state of stress imposed by a particular life event may impose a general regression in which developmentally primitive adaptive patterns will be noted, latent conflicts will be activated and more apparent, and increased demand for parental objects will affect all interpersonal relationships. Horowitz suggests that an understanding of the nature of the stress response syndrome is dependent upon an exploration of pre-stressor psycho-social functioning. Further, he suggested that information regarding developmental stage achieved pre-morbidly be gathered in order to understand the individuals' adaptive and maladaptive style of coping with the activities of everyday existence consequent to the stressor. In summary, Horowitz states: Some persons are unable to complete trains of thought and affect instigated by the event because pre-existing problems or conflicts block the processing of these themes....This can lead to a prolonged post-traumatic reaction that may disrupt work and social functioning. Wilson (1978). a Vietnam veteran and social psychologist, evaluated the impact of war-related trauma on Vietnam.veterans within the context of psycho-social stages of development. The mean age of combatant in the Vietnam War was 20 years, which was the youngest age for.American men to have seen combat in any American-fought war. Using Erikaonian stage theory, attention was drawn to the fifth Psycho-social developmental crisis, Identity vs. Role Confusion. According to Erickson, the "task" of this developmental period is to develop a more integrated and enduring sense of self and personality ‘Cmcture. This stage of psychological growth and development is Characterized as a time of: increasing autonomy from parental guidance, exploration of mutual heterosexual intimacy, initial career decision-making, committment to ideological principles, values and morals, and the discovery and acceptance of one's strengths and weaknesses. Experiencing and observing oneself in all of these contexts provides further self-definition and clarification. Wilson suggested that young adults who might otherwise proceed through this developmental stage may experience an interuption when exposed to combat. In summary, Wilson stated: Under the best circumstances one would hope for good role models, a clear sense of purpose or mission, a moral and political cause worthy of committment, the opportunity to broaden one's geographical-historical world image, the opportunity to believe in the trustworthiness of authority and leaders, collectively shared experiences with age-mates such that a more positive sense of self emerges and finally, to come to a more profound and complex understanding of cultural processes and prevailing technologies. It is not within the purview of this research to address the moral-laden issues provoked by Wilson's comment. However, his thoughts are offered in an effort to evoke thoughtful consideration of the psychological vulnerability of the young military recruit and appreciation for the impact of experiences, dilems, and threat to Physical and psychological integrity experienced by many of these ll"Susliarticularly given the morally conflict-laden context of the Vi'P-tznam War. 10 Understanding the environmental context in which the stressor occured and was experienced by the individual is a critical aspect of appreciating psychological sequelae. For example, Goodwin (1980) suggested that the unique qualities of the Vietnam War may have contributed to the development of a delayed stress reaction. Goodwin studied the DEROS system (date of expected return from overseas). which was intended to minimize psychological casualties. He found that the DEROS system proved disadvantageous in that it undermined unit morale, cohesion, and identification. He also noted that in previous wars, soldiers experienced emotional closeness and support for the working through of the experience of combat with each other during combat and the long trip home from overseas. However, it was common during the Vietnam War for a soldier to be in the midst of combat one day and back in his home town the next. Other factors which Goodwin noted were the confusing ideological and political conflict over the war effort, the lack of clear identification of the enemy, the unclear demarcation of enemy territory, difficulty in coping with women and children combatants, and the high rate of drug usage. Consequently, environmental factors external to the combatant My contribute to the development of PTSD symptoms. ‘The groundwork was laid for this research project-its in? lementation, the presentation and discussion of results—by reviewing the literature. The development of PTSD symptoms are understood from a review of the nature of the stressor, including it's 11 environmental context. An intra-psychic or dynamic appreciation of the individual from the analytic perspective helps to understand the formation of PTSD symptoms. The interface of psycho-social developmental tasks and the environment in military and combat experiences provide an understanding of the impact of military service upon the young military combatant. In Chapter 2, research efforts specific to the psycho-social variables cited previously in the generic research hypothesis and their relationship to PTSD are reviewed and critiqued. In Chapter 3, the method by which the research hypotheses were operationally tested are described. The previous review of theories germane to the develOpment of the symptom complex of PTSD serves as the foundation for the following discussion of specific variables assumed to be related to the existence of PTSD. CHAPTER TWO INTRODUCTION TO REVIEW OF LITERATURE Much has been written in the last fifteen years about variables thought to be related to the existence and etiology of "maladjustment" of the Vietnam veteran. As with other issues warranting empirical study, research efforts on PTSD have been problematic. These problems include research endeavors which incorporate errors in method and those which are limited in scope of exploration. For example, investigators have explored one or more independent variables ranging across the time intervals which include pre-service, service and post-service experience. In some cases, the dependent or criterion variable has not been accurately measured or routinely defined as Post-traumatic Stress Disorder symptoms but has been imprecisely labeled "maladjustment," "depressive syndrome," and "attitude problems." Since the Vietnam War, there has been an evolving recognition of a "delayed stress response" or PTSD symptoms. There has been empirical OXPloration of variables thought to be related to the development of 'Ylptoms of PTSD. This evolution of awareness of and exploration of PTSD is reflected in the chronology of research endeavor and theoretical postulation. This review is designed to critically e"Cuisine those research efforts directly related to the work reported here- For a comprehensive review of the development of psychiatric “03° logy related to combat stress and a history of treatment for the condition the reader is referred to Figley (1973. Chaps 1 5 ‘0' 13 REVIEW OF THE LITERATURE In an early study evaluating Vietnam veterans, Huffman (1970) looked at the relationship of pre-service functioning and nature of entry into the service to "emotional difficulties precipitated by combat fatigue." His findings were based on interviewes of 610 "psychiatric casualties" of the Vietnam War. He reported that 111 of this sample were drafted into service. Of the 610 interviewed, only 48 experienced "emotional difficulties precipitated by combat." " the author determined that Through "brief psychiatric interviewing, 281 of his sample had pre-service legal problems and 621 had not completed high school. He concluded that men with unsuccessful social adjustment prior to military service were prime candidates for emotional difficulties in combat. However, the author did not describe how "pre-service social adjustment" was measured. Furthermore, efforts to standardize the interviewing procedure were not reported. The author acknowledged that interviewing depth and length varied among "subjects." Also, no explanation was given for the conclusion that only 48 of the 610 psychiatric casualties were luffering emotional difficulties precipitated by combat. Sampling Procedures which would assist in determining the generalizability of his results to a broader population of Vietnam veterans were not do8<:1:ibed. Since interviews in Huffman's study were conducted in field hospital settings in Vietnam during 1970, he may not have 14 identified those soldiers who later deve10ped a "delayed" post-traumatic stress disorder. Strayer and Ellenhorn (1975) studied the relationship of a variety of service and post-service factors to post-service adjustment. Their sample of 40 was randomly selected from a pool of 1000 recently discharged army veterans whose names were obtained through the Veterans of Foreign Wars. Data were collected through taped, structured interviews and the administration of the California F Scale (an instrument to measure authoritarian personality charcteristics), Rotter's Locus of Control Scale (1966), and a sentence completion form developed by the authors. Transcriptions of the taped interviews were rated independently by three judges on the following post-service factors: goal orientation, hostility-aggression, attitude toward United States participation in the War, level of guilt feelings, .adjustment to civilian life, and attitude toward college war protesters. Level of combat intensity experienced by subjects was Incasured by self-report on a l to 4 Lickert-type scale. Specific details of subjects' combat experiences were not identified or considered. In reporting their findings, Strayer and Ellenhorn (1975) stated :that introspective veterans who were involved in heavy combat tended t:<>* be apathetic, unemployed, and opposed to the war effort. 0n the <=‘c’t1trary, authoritarian veterans tended to be employed, goal-oriented, c"’l>¢:>sed to anti-war protesters and in favor of the Vietnam War effort. 15 Strayer and Ellenhorn (1975) reported that adjustment to civilian life, feelings of hostility, depression, guilt and a negative post-service attitude toward the Vietnam War were all significantly associated with intensity of combat experience. Then they proceeded to state that "severity of adjustment problems were highly correlated with amount of combat experienced, low goal orientation, high hostility, positive attitudes toward the war protest movement, severe guilt feelings, poor self-concept, high intraception and severe depression." Interpretation of Strayer and Ellenhorn's study was problematic. A dependent variable was not identified and statistical methods such as specific tests and levels of significance used were unreported. Causal statements made in the study were apparently based upon correlation coefficients. Yager (1976) compared and contrasted post-service violent and non-violent veterans of active-duty combat. Thirty-one subjects were re-service psycho-social functioning was attempted. 4. The impact of social support upon return from Vietnam was 88 ain explored 28 5. The relationship between PTSD symptoms and current life stress was explored, as this variable has not been previously examined in the literature. The rationale for the exploration of this variable was to identify potential areas of life events that may have been especially troublesome for veterans with more PTSD symptoms. Clinical experience with this population indicated that this hypothesis may have been verified through empirical exploration. 6. The current impact of the previously-exerienced stressor upon current cognitive and affective experience was explored, as this factor has not been examined in previous reports on PTSD research. The rationale for examining this variable was that to identify the quality of the impact of a previously-experienced stressor on cognition and effect may shed helpful information on the disabling hmpact of the disorder. 7. The criterion variable of PTSD symptoms was operationalized and measured in a standardized fashion according to replicable procedures, using the DSHPIII criteria. CHAPTER 3 DESIGN OF THE STUDY In this chapter, research hypotheses are specified. Sampling procedures, selection criteria and exclusion criteria are discussed. Hethods for Operationalizing the measurement of each variable, composition of the independent and dependent variables, research design and subsequent data analysis are also described. RESEARCH HYPOTHESES The primary question which stimulates the following hypotheses is: What psycho-social variables distinguish those Vietnam.veterans who develop more symptoms of PTSD from those Vietnam,veterans who develop less symptoms? The specific hypotheses to be tested are the following: 1. The reported frequency of PTSD symptoms will be higher for drafted veterans than for enlisted veterans. 2. Those veterans who report more symptoms of PTSD will have experienced more intense combat and will score higher on the Combat Rating Scale than those veterans who are less symptomatic for PTSD. 3. Those veterans with more reported symptoms of PTSD will be experiencing greater current subjective stress related specifically to the experiences of Vietnam.conf1ict and will score higher on the meact of Events Scale than their less symptomatic counterparts. 30 4. Those veterans with more symptoms of PTSD will have more recent life stress, as measured by the Life Events Inventory than Vietnam veterans less symptomatic for PTSD. 5. Those veterans who report more symptoms of PTSD will have had less social supports available to them within their first year of return from Vietnam and will score lower on the Social Support Questionnaire than those veterans who report less symptoms of PTSD. 6. Those veterans with more symptoms of PTSD will have more problematic pre-service psycho-social functioning as evidenced by higher scores on the Pre-Service Social History Questionnaire than less symptomatic Vietnam veterans. SAMPLE The sample was drawn from a population of Vietnam veterans served by the Ann Arbor VA Medical Center. It is estimated that in Washtenaw County alone (only a portion of the Ann Arbor VA Medical Center's catchment area) there are approximately 11,660 Vietnam Era veterans, according to Health Services Research and Development Division of the Veteran's Administration. Based on nation-wide percentages, it could be postulated that there are approximately 3,380 veterans who served specifically in Vietnam in the Washtenaw County area. A figure of 202 31 to 501 of psychologically disturbed Washtenaw County veterans could be extrapolated from this figure. During fiscal year 1981, the Ann Arbor VA Medical Center served 939 Vietnam.Era veteran in-patients, 282 (302) of which were discharged from the psychiatric in-patient service. In fiscal year 1981, 2150 Vietnam era veterans were seen for out-patient care and, again using a percentage of 30, it may be approximated that 645 potential Vietnam Era veterans received psychiatric out-patient care during that fiscal year at the Ann Arbor VA Medical Center. Therefore, it was determined that a sample for study could be drawn from Vietnam veteran patients seen at the Ann Arbor, Veterans Administration Medical Center. METHOD OF SAMPLING Initially, sampling was attempted in a systematic fashion in the following manner. As part of any treatment visit, a form called a "router" is filled out on each patient at the hospital each time he/she receives any inepatient or out-patient care. The routers were obtained on all 694 patients served within the months of March and April, 1983. Each router (containing demographic information on a patient) was then reviewed for information regarding era of service or Vietnam veteran status. Each patient identified as a Vietnam veteran or a veteran of that era was then contacted by phone (where phone 32 numbers were available from patient information or telephone directory information) or by mail where no phone number was available. The purpose in drawing a systematic and sequential sample using this router system was to minimize biases resulting in differential selection of respondents for each group. Two hundred and fifty-six subject candidates were contacted by phone and 338 were contacted by mail. The results obtained from this method of sampling are described in Table 3.1. TABLE 3.1 RESULTS OF INITIAL SYSTEMATIC SAMPLING PROCEDURE Outcome Total Did not serve in Vietnam. 104 (17.52) Lived too far away to participate 19 (3.11) Refused to participate 32 (5.52) Met other exclusion criteria 67 (11.32) Did not respond to phone messages or letters 341 (57.4%) Met inclusion criteria 31 (5.22) Total: 594 (1001) 33 It took approximately 9 months to recruit 51.42 of the final sample of 60. Because of time and monetary constraints, it was necessary to sacrifice this systematic and sequential method of sampling. A description of the method by which the remainder of the sample was drawn may be found in Chapter 4, pps. 56-58. SELECTION CRITERIA Each subject candidate was initially screened by phone to determine if they met any exclusionary criteria. Where phone numbers were unavailable for subject candidates, they were sent a letter requesting that they contact the investigator by mail or telephone. When the potential subject was contacted, a brief description of their military involvement and emotional sequelae was elicited. Demands of study participation were then discussed with each potential participant. If the subject candidate indicated interest in voluntarily participating in the project an appointment was scheduled. At this appointment, the subject candidate was asked to read and sign the informed consent form (appendix B). The informed consent form also served as a means of assuring reading level and comprehension by having each subject read aloud and paraphrase aloud the first paragraph of the form. The subject was then interviewed by 34 one of two psychiatrists involved in the project to determine the extent of PTSD symptoms by use of the Figley Rating Scale (appendix C). The scale was administered in a structured interview format (refer to Chapter 4, pps. 71-74 for details regarding the Figley Rating Scale and it's application). Also, each psychiatrist was to further assess the existence of any exclusion criteria that may have been overlooked in the initial screening process (see Chapter 3, pg. 36 for exclusion criteia). After this psychiatric interview, if the subject was accepted into the study, data collection was begun, under the direct supervision of the investigator. It took approximately two to four hours for completion of all instruments, depending upon reading skill. All paper-and-pencil questionnaires were completed at the hospital. Each questionnaire was then reviewed by the investigator with the subject to assure that all items were responded to where appropriate and all inconsistencies were rectified, where possible. Each subject completed a Minnesota Multiphasic Personality Inventory (MMPI) as part of the research project. Initially, the MMPI was to be used as a means of identifying and excluding those subject candidates who appeared to meet psychotic criteria (through application of Henrich's Rules for scale configuration). It became apparent after running the first six subjects through this screening that all met psychotic criteria or were "indeterminant" as the clinical scales were quite elevated for those subjects who were 35 experiencing greater degrees of PTSD. Those subject who appeared psychotic or indeterminant in their MMPI profiles were not identified in the psychiatric interviews, nor by the researcher, as exhibiting any clinical indications of psychosis. Keane, a nationally-known researcher in the field of PTSD, was consulted (personal communication, 1983). He has extensive experience in the research and clinical use of the MMPI with persons experiencing PTSD. His observations confirmed the above findings (Malloy et a1., 1983). It was his opinion that many Vietnam veterans with PTSD symptomatology appear psychotic according to the MMPI scale e1evations--with a predominant pattern of 2-8 or 8-2, but do not present as psychotic in clinical evaluation (personal communication, February, 1983). His observations have been supported elsewhere (Foy et. al, 1984). It was Keane's opinion that the phenomenon was a clinical aberration to be explored through further research. He recommnded that MMPI score configurations not be used as exclusion criteria for this study. His rec o-aendations were imp lamented. At the point of data collection, each subject was assigned a code number. This number only was placed on questionnaires, survey instruments and demographic data forms to assure subject confidentiality. 36 EXCLUSION CRITERIA Subject candidates were excluded if they were referred from the Alcoholism Treatment Unit but had not been sober for at least one month. Also, there were two instances where subject candidates came for evaluation for the study and were obviously intoxicated. In these two cases, questionning revealed a history of alcoholism treatment admissions as well. Because of the nature of the study, requiring detailed memory of certain experiences and feelings these individuals were excluded from the study. Subjects were screened for the nature and extent of drug use/abuse since discharge from the service. Based on self-report, no individual appeared to be habituated to narcotic or narcotic-like substances. However, there were subjects who were chronic users of marijuana. This information was elicited and quantified, although no subject was excluded for marijuana use. To do so would have excluded a large number of the sample drawn. Clinical experience and empirical observation suggests that a significant minority of Vietnam veterans continue to be regular marijuana users. For this reason, no subject candidate was excluded because of marijuana use. No person was accepted into the study who had been served on the Neurology in-patient or out-patient service because of the same concerns for acuity of memory and affective experience. 37 GROUPING OF SAMPLE Initially, two groups were to be drawn, a symptomatic group and an asymptomatic group. This procedure would have allowed the dependent variable of PTSD symptoms to be treated dichotomously, thereby simplifying data analysis. However, this was not done for two reasons. First, it was apparent after examining 30 subjects that there were no asymptomatic Vietnam veterans in the sample. That is, all subjects appeared to experience some symptoms of PTSD. These symptoms ranged in severity from mild to severe. Second, even though it may have been possible to identify asymptomatic Vietnam veterans outside of the VA health care system, this procedure was beyond the scope of this study. SAMPLE SIZE The method for determination of sample size was that based on formulas available in Beyer (1966). These formulas were based on the method of analysis to be used using one-way analysis of variance. According to Glass and Stanley (1970), the same assumptions hold true for multiple analysis of variance and therefore, were considered appropriate for this study. For those measures using scales with range of scores from 1 to 5, with an alpha of 0.05 and beta of 0.80, 38 one would need a sample of 17 in order to detect a difference of at least two points, if a significant difference exists. For those measures using scales with range of scores from 1 to 4, with an alpha of 0.05 and beta of 0.80, one would need a sample size of 11 in order to detect a difference of at least two points, if such significance exists. The total sample drawn was 60. Glass and Stanley further assert that the normal distribution prevails for an N of 30. The sample of 60 which was studied and reported here appears more than adequate for the analysis of the research hypotheses. MEASURES The previously identified hypotheses were Operationally tested through the administration and statistical analysis of the measures identified in Table 3.2: 39 TABLE 3.2 MEASURES USED IN THE QUANTIFICATION OF THE INDEPENDENT VARIABLES Hypothesis Variable Studied Means of Measurement 1 Draft or volunteer status Military History Form 2 Combat intensity Combat Rating Scale 3 Current subjective impact of Vietnam experiences Impact of Events Scale 4 Recent life stress Life Events Inventory 5 Post-service social support system Social Support Questionnaire 6 Pre-service psycho-social functioning Pre-Service Social History Questionnaire Explanations of the instruments used to test the hypotheses mentioned above follow below. Social Desirability Scale The Social Desirability Scale (Crowne & Marlowe, 1961) is a 33-item self-report questionnaire designed to assess propensity or "set" to present oneself in a socially desirable or undesirable way (see appendix D). Social desirability was defined 40 by its originators as "...behaviors which are culturally sanctioned and approved but which are improbable in occurrence." The 33 items included in the final Marlowe-Crowne revision were found to discriminate at the .05 level or better between high and low scores. Internal consistency using Kuder-Richardson's split-half formula was estimated to be .88 by Crowne and Marlowe (1961). The use of the Social Desirability Scale for this study was to aide in the identification of those subjects who were unreliable historians because of the "set" to present their behavior in a socially acceptable fashion. Impact of Events Scale The Impact of Events Scale (see appendix E) is a 15-item self-report scale developed by Horowitz et a1. (1979). The purpose of the scale is to assess the form and quality of conscious experiences (related to a previously experienced traumatic event) during a recent (past seven days) period, with the event specific to the subject inserted on the form as referrent for response to the list of questions. Examples of items from the Impact of Events Scale are: 1. (During the last seven days) I thought about Vietnam when I did not mean to; 2. (During the last seven days) I avoided letting myself get upset when I thought about Vietnam or was reminded of it. The meact of Events Scale was considered appropriate for testing hypothesis three: "Those veterans with PTSD will be experiencing 41 greater current subjective stress related specifically to the experiences of Vietnam conflict and will score higher on the Impact of Events Scale than those veterans with less symptoms of PTSD." According to Zilberg et a1., (1982), the past seven days (the time referrent for assessing the subjective on-going impact of a previous stressor) was found to be the best time unit for clinically valid reports of current subjective distress and state of mind related to the stressful life event. The scale yields two sub-scales, that of intrusion and avoidance. The items were found to be both logically and empirically consistent. According to Zilberg et a1. (1982), the reliability of internal consistency of these subscales was estimated to be 0.78 for intrusion and 0.82 for avoidance in the original research. In a replication study of the scales' psychometic properties (Zilberg, et a1., 1982), alpha coefficients ranged from 0.86 to 0.89 for three groups over a period of time. Combat Rating Scale The Combat Rating Scale (see appendix F) is a lO-item self-report scale quantifying combat intensity. Six items receive a single rating of 225 and four items receive a double rating of £32. The scale ranges from a score of 0 (no combat) to a maximum of 14 (most intense combat). This scale is a revision of the original scale devised by the Center for Policy Research (1979). The revision by Gallup, Laufer, and Yager is an 42 unpublished version which correlates highly with the old scale (Cronbach's alpha of 0.84) but is simpler and shorter for administration. It is a statistically reliable measure of combat experience. The Combat Rating Scale was used to identify the level of combat intensity experienced by each subject. An example of items taken from the scale are as follows: 1. In an artillery unit which fired on the enemy; 2. Flew in an aircraft over Vietnam; 3. Stationed at a forward observation post. This scale was used to test hypothesis two: "Those veterans who report more symptoms of PTSD will have experienced more intense combat and will score higher on the Combat Rating Scale than those veterans who are less asymptomatic for PTSD." Life Events Inventory The Life Events Inventory (Cochrane and Robertson, 1972) is a 55-item self-report questionnaire designed to measure current levels of life stress (see appendix C). It is a revision of the Schedule of Recent Events (Holmes and Rahe, 1967), attempting to remedy the following deficiencies noted in the Schedule of Recent Events: inappropriate, irrelevant or ambiguous items, a lack of systematic inquiry into common stressful events in peoples' lives, and lack of weights assigned to the life events being meausured to increase the accuracy of the assessment of the impact of the stressful life event. The authors of the Life Events Inventory concluded that "...(the scale) is a comprehensive measure of recent life stresses equally suitable for use with all sections of the population." 43 The Life Events Inventory was used to test hypothesis four: "Those veterans with more symptoms of PTSD will have more recent life stress, as measured by the Life Event Inventory than Vietnamlveterans less symptomatic for PTSD." An example of items considered to be potential life stressors and taken from the Life Events Inventory are: l) Unemployment (of head of household), 2) Trouble with superiors at work, 3) New job in the same line of work. Each of the 55 items is weighted, with a theoretical range of scores from 0 to 2,879. Three groups of judges (psychiatrist/psych- ologists, students, and psychiatric in-patients) were asked to weight each item according to face validity for amount of stress. Coefficients of concordance for all three groups was 0.89. Therefore, the weights attached to each stressful life event were deemed to be evidence of face validity of that events' degree of impact on a person. The Life Events Inventory was specifically designed to quantify the amount of "turmoil, disturbance and upheaval" that people are subjected to, rather than simply pleasant or unpleasant life events, which is the index of life stress that would assist in the exploration of Hypothesis 4. Social Support Questionnaire The Social Support Questionnaire was an instrument developed for the purposes of this study (see appendix H). A review of the literature indicates that there has been no instrument developed to measure the nature of social support available in a person's past history. It was developed to measure sub-factors of the general concept of social support available to the veteran during the first year of his return from Vietnam. Sarason et a1. (1982) have identified the number of persons available to the subject and the degree of satisfaction with the available support as the two basic elements imbedded in the concept of social support. Therefore, sub-factors measured were the number of individuals comprising the social network of a subject and the satisfaction inherent in the subject's relationship with each person comprising the social network. Three items were included in the measurement of social support which pertain to the subject's ability to use the social support network to assist in coping and adaptation during that first year of return from Vietnam. These items were: "talking with people about Vietnam.experiences," "feeling close to anyone" during the first year of return from Vietnam, "spending time with anyone during the first year of return" from Vietnam. For the purposes of standardization, if a subject asked what was meant by "talking about Vietnam experiences" and/or "spending time with..." the reply was "whatever that means to you." If a subject asked for a definition of "feeling close," the standardized response was "feeling emotionally close to someone." The measurement of quantity of people comprising the subject's 45 social support network was operationalized by simply asking the veteran to indicate the first name and relationship of each person that he considered important to him during his first year of return from Vietnam. To insure standardization, if a subject asked for the definition of "important" the standard response was "whatever important means to you." The instrument gives an index of satisfaction with the social supports available to him when he first returned from Vietnam. This sub-scale index was derived by asking the subject to rate the relationship that he had with anyone he indicated as important to him during his first year of return from Vietnam on the qualities of " "sharing," and "degree of contact," "how important," "helpfulness, "good or bad feeling." These concepts were used by Pattison in the development of the Pattison Psycho-Social Inventory (1981). This instrument was used to test hypothesis five: "Those veterans who report more symptoms of PTSD will have had less social supports available to them within their first year of return from Vietnam and will score lower on the Social Support Questionnaire than those veterans who report less symptomm of PTSD." The hypothesis was derived from the observations that social support contributes to positive adjustment and personal development and provides a buffer against the effects of stress (Bowlby, 1969; Hirsch, 1980; Bronfenbrenner, 1961). 46 Military_History Questionnaire The Military History Questionnaire (see appendix I) is a 29-item self-report instrument designed to elicit information regarding military experiences while in Vietnam. It was developed for use in this study. The questionnaire is a derivation and abbreviation of the Figley Vietnam Era Veterans Survey (1977). In the modification of the Figley Vietnam Era Veterans Survey, items were deleted that seemed redundant or more detailed than deemed necessary for this study's purpose. A revision was piloted on four hospitalized Vietnam veterans with Post-traumatic Stress Disorder. The final version used in this study was based on input from the pilot cohorts. Examples of the kind of information elicited from the questionnaire are draftee or volunteer status, length of tour in Vietnam. drug and alcohol usage while in Vietnam, and nature of job or military occupational status while in Vietnam. Only one item from the questionnaire contributed to the testing of the research hypothesis. The question was "Were you drafted or did you volunteer for military service?" Therefore, a dichotomous variable was created for the testing of the following hypothesis: "The reported frequency of PTSD symptoms will be higher for drafted veterans than for enlisted veterans." No information regarding psychometric properties of this instrument is available. It's use was considered exploratory in nature. 47 Preservice Social History Questionnaire The Preservice Social History Questionnaire (see appendix J) is a 43-item self-report instrument designed specifically for this study to elicit information indicative of the quality or nature of psycho-social adjustment prior to entering the service. The questionnaire is composed of sub-scales of items for six areas of psycho-social functioning. These areas of functioning were determined through consultation with social science researchers at the Institute for Social Research (personal conversations with Amiram.Vinakour and Melvin Mania; November, 1982) at the University of Michigan. Furthermore, a review of the literature regarding social functioning (Moos, 1974; Horowitz, 1979; Weissman, et a1., 1981) indicates that the areas addressed in the subscales are critical components of the predictor variable of pre-morbid psycho-social functioning. The following is a description of each subscale and the information assessed. Scores are derived for each subscale with lower scores indicating least problematic functioning and higher scores indicating more problematic pre-morbid functioning. The seven subscales are: 1. Family History. This subscale includes 11 questions regarding the stability of the environment of the home of origin, mental health history of immediate family members, household moves and unemployment of the head of household. The subscale yields a score ranging from 0 to 11. 48 2. School History. This subscale includes six questions regarding academic performance, disciplinary problems and last grade completed. The subscale yields a score ranging from 0 to 6. 3. Legal History. This subscale contains three items and assesses information regarding juvenile pranks, involvement with juvenile authorities, or arrests prior to entering the service. The subscale yields a score ranging from O to 3. 4. Relationship History. This subscale contains eight items and includes questions about the nature and depth of friendships, supportive relationships with an adult, and significant intimate relationships and their outcome. The subscale yields a score ranging from 0 to 7, as the responses to item 26 from the scale were not used in the ultimate subscale and scale scores. 5. Drug and Alcohol History. This subscale contains nine items and assesses information regarding drug and alcohol intake habits prior to entering the service. The subscale yields a score ranging from 0 to 9. 6. Mental Health History. This subscale contains six scorable items which measure information regarding emotional difficulties and help sought by the subject prior to entering the service. The subscale yields a score from 0 to 6. A subscale of employment history was originally incorporated into the scale. There were only two items comprising the subscale, with 49 only one item yielding a score contributing to a measure of pre-service psycho-social functioning. Therefore, this item was used ultimately in analysis incorporating a total scale score. No psychometric data regarding the use of the subscale was computed, however. This instrument was piloted on four psychiatric in-patients to assess the instruments clarity. Furthermore, the pilot cohorts were asked if there were other areas of functioning or other questions that they thought would be helpful to ask in assessing a person's pre-service psycho-social functioning. Some revisions and rewriting of items were conducted based upon this input. This instrument was used to test hypothesis six: "Those veterans with more symptoms of PTSD will have more problematic pre-service psycho-social functioning as evidenced by higher scores on the Pre-Service Social History Questionnaire than less symptomatic Vietnam veterans." ' COMPOSITION OF DEPENDENT/CRITERION VARIABLE The dependent or criterion variable was treated as either categorical or continuous, depending upon the demands of the analysis being performed. The categorical variable was formed in the following manner. The Figley Rating Scale, with scores ranging from 0 to 61, was the source for forming the categorical variable. Three groupings 50 ' and "severe." Efforts were were formed, titled "mild," "moderate,' ‘made to balance the number of cases in each group. The final group formation was: the mild group included 21 cases, the moderate group included 20 cases, and the severe group included 19 cases. There were uneven numbers in each group because of ties between individuals at the extremes of each group. COMPOSITION OF INDEPENDENT VARIABLES There were eight independent or discriminating variables used for the study of the previously-presented hypotheses. In considering the correctness of the main analyses, the following generic or global hypothesis was tendered: There is a statistically significant and measurable relationship between any one of the eight independent variables and the dependent variable of post-traumatic stress disorder symptoms. The description and composition of each independent variable is as follows. Nature of Entry Into Service This variable is assessed in the Military History Questionnaire, question number 2, and is scored as a dichotomous variable (draftee or enlistee). 51 Combat Intensity This variable is a composite score of all item ratings from the total score on the Combat Rating Scale. It is scored as a continuous variable, ranging from 0 to 14. Current Impact of Previous Stressor This independent variable was composed of the total score on the Impact of Events Scale. It was a continuous variable with theoretically ranging scores from 0 to 75. Recent Life Stress This independent variable was continuous. It was derived as the total score from the Life Events Inventory. Theoretically, scores ranged from 0 to 2,879. Pre-Service Psycho-Social Functioning This independent variable was composed of the total score on the Pre-Service Social History Questionnaire. It was a continuous variable with theoretically ranging scores from O to 43. Social Support Upon Return from Vietnam. The means of quantifying this variable and the testing of the related hypothesis were more complex. Three variables were derived to measure certain aspects of the social support system. One variable was composed of the total number of people in the veteran's social network during the first year of return from Vietnam. The second variable used in the testing of this hypothesis referred to the quality of the relationship with each 52 person indicated as important by the veteran during the first year of return from Vietnam (see appendix H, page 2 for a definition of each criteria composing the quality index ). The third variable used in the testing of the relationship between PTSD symptoms and social support within the first year of return from Vietnam was the sum of positive responses to three dichotomously scored questions taken from the Social Support Questionnaire. The variable yielded a score ranging from 0 to 3. These questions were: 1) During you first year of return from Vietnam, did you talk with anyone about your experiences in Vietnam? 2) Did you spend time with anyone during the first year that you returned from Vietnam2 and 3) Did you feel close to anyone during the first year that you returned from Vietnam? Thus, three variables comprised the operationalization of the social support network of the veteran upon return from.Vietnam. DESIGN The study design was retrospective and exploratory in nature. Data were collected retrospectively to evaluate the relationship between PTSD symptoms and six psycho-social variables. There was no experimental treatment to be applied and measured. As noted by Campbell and Stanley (1963), there are potential sources of invalidity in this design. 53 The lapse of time since the traumatic event introduces possible bias in memory, a confound that can not be identified, nor kept standardized across all subjects. Furthermore, the nature of data collection relied heavily on self-report of past history (see Chapter 5, pps. 115-116 for detailed discussion of verification of data collected). In the gathering of specific facts of self-reported history, attempts were made to elicit data which was concrete and specific. Data was collected through a self-administered questionnaire format to eliminate the potentially confounding variable of interviewer bias or interviewer effect on subject response. There is also the potential for sampling bias in the design in that sampling was not conducted in a random fashion. Futhermore, since no subject was accepted with a recent history of drug and/or alcohol abuse or neurological impairment, this introduced a selection bias in the sample drawn (see Chapter 5, pps. 112-113 for additional discussion of potential sampling bias). ANALYSIS The appropriate statistical model for data analysis is derived from the previously stated research question: What psycho-social variables distinguish those Vietnam veterans who develop more symptoms of PTSD from those Vietnam veterans who develop less symptoms? 54 The initial plan was to conduct a discriminant function analysis, treating the criterion variable of PTSD symptoms as categorical. The purpose of the discriminant function analysis was to evaluate the non-linear relationship between each of the independent variables and the dependent variable while controlling for the influence of variance of the other independent variables (Klecka, 1980). Through the course of data collection, it became apparent that to treat the dependent variable as categorical would result in a spurious depiction of the relationship of the dependent variable to the independent variables. That is, in the clinical state, PTSD symptoms do not present as mild, moderate or severe. There appears to be a continuum from non-existent to very severe symptomatology. For this reason, data analysis was conducted differently from that initially proposed (see Chapter 4, pps. 93-97, for detailed description of final data analysis). Two univariate analyses were performed to facilitate the interpretation of the main multivariate analyses. Correlations between each independent variable and the dependent variable were computed to measure the degree of linear relationship between each independent variable and the dependent variable without reference to the influence of the other independent variables. One-way analyses of variance was then computed between each independent variable and the dependent variable to test for non-linear relationships. The dependent variable was treated as trichotomous-rather than dichotomous-to allow for assessment of 55 non-linearity in the relationship between each independent variable and the dependent variable, independent of the influence of the other independent varibles. SUMMARY As stated in the procedures section, all subjects appeared to experience some symptoms of PTSD, with a wide range in the degree of symptomatology across the sample. It was determined that to force subjects into two groupings based on symptomatology would be to spuriously and inaccurately characterize the nature of the criterion variable. Therefore, in the main analysis, the criterion variable was treated as both categorical and continuous. Interpretation of the two main forms of analysis were facilitated by univariate analyses using analyses of variance and Pearson's product moment correlation coefficients. CHAPTER 4 ANALYSIS OF RESULTS In this chapter, the final sampling procedure and outcome is discussed. Descriptive information of the sample drawn is presented. Descriptive statistics concerning the independent variables are presented. Psychometric prOperties of the instruments used for the testing of the hypotheses are described. The influence of response set bias is examined. Results of the data analyses for the hypotheses are presented, which include multivariate and univariate analysis. FINAL SAMPLING PROCEDURE AND OUTCOME Initially, sampling was attempted in a systematic fashion using the router method (as discussed in Chapter 3, pps. 31-32). After nine months using the router method, only 31 (51.21) appropriate subjects were identified and examined. Because of time constraints, financial considerations, and lack of personnel, adjustments were implemented in the original sampling procedure to expedite sampling. Therefore, potential subject candidates were recruited also from in-patient and out-patient medical and psychiatric treatment staff. This comprised 29 of the total sample of 60. Since no financial incentive for participation in the study was available, it proved expedient and practical to schedule a subject candidate for the study when their appointments coincided with other hospital-related appointments. 57 Twenty-five subjects were referred for the study from in-patient and out-patient primary therapists (usually psychiatrists, psychiatric residents, psychology interns, psychologists, or social workers). Four subjects were referred for study from medical treatment staff. In Table 4.1, the source of referral by severity of PTSD symptoms is described. TABLE 4.1 REFERRAL SOURCE* Referral Source Mild Moderate Severe Total Router 18 10 3 31 (51.71) In-Patient Psych. 0 2 7 9 (15.01) Out-Patient Psych. 3 5 8 16 (26.7%) Other 0 3 l 4 (6.71) chi-s quare , *p- . 001 DF-6 This compromise in the original systematic and chronologically sequential selection of subjects introduced potential selection bias. Note that there was a significant relationship between source of referral and severity of PTSD symptoms. Therefore, The sample drawn 58 was not necessarily representative of the VA population of Vietnam veterans. This conclusion must be considered in generalizing the results of this study beyond the sample examined. INDEPENDENT VARIABLE MEANS AND STANDARD DEVIATIONS In tables 4.2 and 4.3 the mean scores and standard deviation scores among the three groups for the eight independent variables are presented. The reader is referred to Chapter 3, pps.49-50 for the method by which the dependent variable of PTSD symptoms was categorized into mild, moderate and severe groupings. TABLE 4.2 MEAN SCORES BY PTSD GROUP ON INDEPENDENT VARIABLES VARIABLE MILD MODERATE SEVERE ALL GROUPS Draftee vs. volunteer 1.71 1.85 1.89 1.82 Combat intensity 7.29 12.20 13.42 10.87 Impact of Events Scale 13.29 34.10 56.95 34.05 Life Events Inventory 325.33 336.30 693.74 445.75 Number in social network 4.90 5.30 5.32 5.17 Quality of network 2.99 3.04 2.85 2.96 Relations with others .55 .56 .38 .50 Preservice functioning 7.67 5.60 7.21 6.83 59 TABLE 4.3 STANDARD DEVIATIONS BY PTSD GROUPS ON INDEPENDENT VARIABLES Variable Mild Moderate Severe All Groups Draftee vs. volunteer .56 .49 .32 .47 Combat intensity 4.62 1.96 1.50 3.08 Impact of Events Scale 13.34 17.20 8.61 13.58 Life Events Inventory 166.66 209.99 365.53 258.14 Number in social network 2.59 3.18 3.02 2.93 Quality of network .81 .84 .90 .85 Relations with others .24 .32 .30 .29 Preservice functioning 4.29 4.23 3.95 4.17 SAMPLE CHARACTERISTICS The majority of veterans who served in Vietnam.were male. males saw combat conditions. males were included as subjects. Only For the purpose of this study, only 60 In Tables 4.4 through 4.9, current demographic and descriptive information of the sample drawn is presented. In Tables 4.10 through 4.14 significant pre-service and military descriptive information is provided. TABLE 4.4 AGE OF RESPONDENTS* BY DIAGNOSTIC GROUPING Agg__ Mild Moderate Severe Total Mean 35.57 35.25 34.80 35.57 Standard Deviation 3.33 2.97 2.59 3.33 Range 31.0 - 43.0 *chi-square, non-significant 61 TABLE 4.5 MARITAL STATUS OF RESPONDENTS* Marital Status Mild Moderate Severe Total Never Married 2 0 1 3 (5.0%) In First Marriage 11 7 3 21 (35.01) In Second Marriage 3 3 3 9 (15.02) In Third Marriage 1 4 2 7 (11.71) Divorced Once 2 5 3 10 (16.71) Divorced Twice 2 0 2 4 (6.71) Separated-First Marriage 0 0 2 2 (3.31) Separated-Second Marriage 0 0 3 3 (5.02) Living Together 0 1 0 1 (1.71) *chi-square, non-significant 62 TABLE 4.6 RAGE OF RESPONDENTS* Race Mild Moderate Severe Total White 20 l6 18 54 (90.01) Black 1 3 l 5 (8.32) Chicano 0 l 0 1 (1.71) Other 0 O 0 0 (0.01) *chi-square, non-significant TABLE 4.7 CURRENT EMPLOYMENT STATUS OF RESPONDENTS* Employment Status Mild Moderate Severe Total Employed 16 12 6 34 (56.7%) Unemployed 5 8 13 26 (43.32) *chi-square, p-.02 DF'Z 63 TABLE 4.8 INCOME OF RESPONDENTS* Income Mild Moderate Severe Total Under $10,000 5 1 6 12 (20.7%) $10,000 - $20,000 7 7 6 20 (34.51) 321,000 - $30,000 a 6 3 13 (22.2%) $31,000 - $40,000 a 3 1 8 (13.31) 841,000 - $50,000 0 2 o 2 (3.41) Over $51,000 1 1 1 3 (5.11) *chi-square, non-significant TABLE 4.9 RESPONDENTS CURRENTLY IN PSYCHOLOGICAL TREATMENT* Treaument Mild Moderate Severe Total Yes 3 8 14 25 (41.72) No 18 12 5 35 (58.31) *chi-square, p-.001 DF-2 64 TABLE 4.10 RESPONDENTS' AGE AT ENTRY INTO SERVICE* Age 8t Entry Mild Moderate Severe Total 18 years or less 9 12 11 32 (53.32) 19 - 21 years 10 8 7 25 (41.72) 22 - 25 years 2 0 1 3 (5.02) *chi-aquare, non-significant TABLE 4.11 NATURE OF ENTRY INTO SERVICE OF RESPONDENTS* Entry Status Mild Moderate Severe Total Draftee 7 4 2 13 (21.71) Volunteer 14 16 17 47 (78.31) *chi-aquare, non-significant 65 TABLE 4.12 EDUCATION OF RESPONDENT AT TIME OF ENTRY INTO SERVICE* Education Mild Moderate Severe Total Leas than Grade 12 5 6 11 22 (36.62) Completed High School l3 l3 7 33 (55.02) Sane College 3 l l 5 (8.41) *chi-aquare, non-significant TABLE 4.13 RANK 0? RESPONDENT AT TIME OF DISCHARGE FROM SERVICE* Rank Mild Moderate Severe Total E4 or less 7 12 10 29 (48.31) 85 - £6 14 7 8 29 (48.31) E7 or more 0 l 1 2 (3.42) *chi-aquare, non-significant 66 TABLE 4.14 DISCIPLINARY ACTION AGAINST RESPONDENT WHILE IN SERVICE* Discipline Mild Moderate Severe Total None 10 13 9 32 (53.31) Article 15 7 6 6 19 (31.72) Court Martial 1 1 2 4 (6.72) Both 0 0 2 2 (3.3%) Other 3 0 0 3 (5.02)) *chi-square, non-significant In analyzing the significance of the relationship between the degree of PTSD symptoms and the demographic variables, the dependent variable of PTSD symptoms could have been treated as continuous or dichotomous. The following assumptions pertain to the use of one-way analysis of variance (Hays, 1973): 1. normal distribution of variance 2. homogeneity of variance 3. independence among source of error 4. equal distribution of cases per cell 67 From observation, it is readily apparent that the assumption of equal cases per cell is violated. From this violation, homogeneity of variance is suspect. Further, there is no evidence of normal distribution of variance available. Therefore, it was determined that the most appropriate test of signficance for assessing the degree of relationship between the descriptive variables and PTSD symptoms was chi-square. Recognizing that chi-square is a less powerful non-parametric test of signficance, analyses of variance were also conducted on the relationship between the descriptive variables and PTSD symptoms. The results of the ANOVAs supported the chi-square tests reported above. PSYCHOMETRIC PROPERTIES OF SCALES The following section describes psychometric prOperties of scales used for the operationalization of the independent variables, where available. Social DesirabiIity Scale The use of the Social Desirability Scale for this study was to aide in the identification of those subjects who were unreliable historians because of the "set" to present behavior in a socially acceptable fashion. Cronbach's alpha was computed as an estimate of internal consistency for the data collected from the Social Desirability Scale, which was .82. This instrument is considered a reliable measure of response set bias for this sample (the reader is referred to Chapter 3, pp. 39-40 for further information on the Social Desirability Scale). 68 Impact of Events Scale The Impact of Events Scale was used a measure of the impact of a previously-experienced stressor on current affect and cognition. Estimates of internal reliability were computed for the Impact of Events Scale based on data collected in this study. On the seven-item ' internal reliability using Cronbach's alpha was subscale, "intrusion,' estimated to be .94. On the subscale measuring "avoidance" the alpha coefficient of internal reliability was estimated to be .90. The internal reliability of the total scale was estimated to be .95. The estimates were a higher meausure of internal consistency than those estimates of internal reliablity based on the normative sample (the reader is referred to Chapter 3, pp 40-41 for further details on the Impact of Events Scale). These estimates of reliablity support the scale's use in this study as a stable measure of the impact of previous trauma on current affect and cognition. Combat Rating Scale The Combat Rating Scale provided a measure of the intensity of combat experienced in Vietnam. An estimate of internal consistency reliability, was derived on the scale from the data collected for this study. Cronbach's alpha was computed to be .83. Therefore, the Combat Rating Scale was determined to be a sufficiently reliable measure of the intensity of combat experience (the reader is referred to Chapter 3, pp. 41-42 for further information on the Combat Rating Scale). 69 Life Events Inventory The Life Events Inventory was used to measure the current level of life stress in a subject's life. The inventory was concluded to be a reliable measure of life stress for this study's purpose. An estimate of internal consistency was derived from the data collected for this study, using Cronbach's alpha. The estimate was computed to be .83 (the reader is referred to Chapter 3, pp. 42-43 for further details on the Life Events Inventory). Social Support Qpestionnaire This scale was developed.specifically for this study to provide a measure of the social support network available to the veteran within the first year of return from Vietnam. No psychometric data are available on the scale. The scale was pilot tested with four persons typical of the subjects used in this study and revisions were made accordingly, prior to it's use in the study. As stated in Chapter 3, pps. 43-46, because three variables constituted the measurement of social support, the scoring method does not lend itself to computation of internal consistency. Pre-Service Social History Questionnaire The Pre-Service Social History Questionnaire was also designed specifically for this study to elicit that information indicative of the nature of psycho-social functioning prior to entering the service (the reader is referred to Chapter 3, pp. 47-49 for details regarding the development and composition of the scale). 70 The six subscales contained in the Pre-Service Social History Questionnaire were analyzed for further refinement after completion of data collection and prior to the main data analysis. Each item was evaluated for degree of correlation with other items comprising it's subscale. The SPSS subprogram of Reliability enables one to identify the change in alpha level of a subscale if an item is deleted from the subscale. After identifying those items that could be deleted to increase the estimate of internal reliability, it appeared unfeasible. The alpha level could have been increase by .015 by deleting four scored items. It was determined that the possible modest increase in alpha did not warrant revising the subscales items and total scale. In Table 4.15, information regarding the nature of the subscale, number of items comprising each subscale, Cronbach's alpha for the subscale, number of items comprising the total subscale, and Cronbach's alpha estimating the internal consistency of the overall scale of Pre-Service Social History are described. 71 TABLE 4.15 INTERNAL CONSISTENCY ESTIMATES OF PRE-SERVICE SOCIAL FUNCTIONING SCALE Subscale Number of Items Cronbach's Alpha Family History 11 .67 School 6 .63 Legal 3 .67 Relationship History 7 .24 Drugs/Alcohol Use History 9 .66 Mental Health History 5 .67 Total Scale 43 .72 A subscale of employment history was originally incorporated into the scale. There were only two iems comprising the subscale, with only one item yielding a score contributing to a measure of pre-service psycho-social functioning. Therefore, this item was used ultimately in analysis incorporating a total scale score. No psychometric data regarding the subscale was computed, however. 72 Military History The only item from the Military History Questionnaire which was used in data analysis was one dichotomously scored item. Therefore, it was not appropriate or necessary to derive psychometric prOperities of this questionnaire for this study. VERIFICATION OF PTSD SYMPTOMS The verification of reliability of assessing symptoms of PTSD was done in the following way. Each subject was interviewed separately by a psychiatrist and the investigator, using a semi-structured interview format. Each interviewer blindly rated each subject on the type and frequency of symptoms of PTSD. The Figley Rating Scale (Figley, 1978) was used for this purpose (see appendix C). The symptoms elicited in the interviews were those used in the DSM-III criteria for post-traumatic stress disorder (see appendix A). There was no criterion-based validity check. Criterion-based validation was not possible as there is no accepted criterionebased measure of PTSD. The diagnositic description available in the DSM-III is in common clinical use and is the most-commonly agreed-upon clinical description of the symptoms constituting PTSD. The Figley Rating Scale is based on the DSHPIII classification of 73 PTSD. The Figley Rating Scale was developed by Charles Figley (1978) for his pioneering work in the identification of PTSD as a diagnosable and treatable psychological syndrome. From personal communication with Figley (November, 1983), it was learned that no psychometric data has ever been reported on this scale. Therefore, a measure of internal consistency using Cronbach's alpha was computed from the data collected on the described sample. The scale is broken into four subscales, corresponding to items A through D of the DSM-III criteria for PTSD. In Table 4.16, the nature of the subscale, number of items comprising the subscale, reliability estimates for the subscales and total scale are described. TABLE 4.16 INTERNAL CONSISTENCY ESTIMATES ON FIGLEY RATING SCALE Subscale Number of Items Alpha Stressor 9 .93 Reexperiencing of stressor 4 .75 Social withdrawl 5 .79 Symptoms 7 .77 Total Scale 25 .92 74 The Figley Rating Scale yields a score ranging from 0 - 61. As stated above, two psychiatrists interviewed subjects to determine existence and degree of PTSD symptoms. In addition, all subjects were interviewed and diagnosed by the investigator. Not all subjects were interviewed and rated by the same psychiatrist. Twenty subjects were interviewed and rated by psychiatrist "A" and forty subjects were rated by psychiatrist "B." Each subject was interviewed twice, once by one of two psychiatrists and then the investigator. All three raters were experienced in the diagnosis and psychotherapeutic treatment of PTSD and had extensive experience providing psychological and psychiatric services to Vietnam veterans. To assure a high level of inter-rater reliability the following procedure was followed. Prior to beginning the study, the two psychiatrists and the investigator conducted three group interviews with three separate Vietnam veterans who were not included in the study. Each interviewer elicited information from the veteran through the course of the group interview. Each psychiatrist and the investigator then rated the existence and frequency of symptoms based on information elicited during the interview without collaboration. Ratings were compared, contrasted, and discussed where rating were discrepant. In this way, efforts were made to enhance consensual perception of symptoms and symptom severity. The success of this initial "pilot" effort is demonstrated by the following Pearson's 75 product moment correlation coefficients of inter-rater reliability, described in Table 4.17: TABLE 4.17 INTER-RATER RELIABILITY COEFFICIENTS ON THE FIGLEY RATING SCALE SUBSCALES RATERS (n-3) *R- Stressor psychiatrist investigator .89 psychiatrist investigator .93 Reexperiencing of psychiatrist investigator .88 stressor psychiatrist investigator .89 Social withdrawl psychiatrist investigator .65 psychiatrist investigator .81 Symptoms psychiatrist investigator .82 psychiatrist investigator .91 Total scale psychiatrist investigator .95 psychiatrist investigator .95 *R- Pearson's product moment correlation coefficients 76 ASSESSING RESPONSE SET BIAS Another effort to assess reliability of data collected was done by using the Social Desirabilty Scale (Crowne and Marlowe, 1961), to identify the influence of response set bias on data collected. Each item from each scale used in the quantification of the independent variables was correlated with the total score on the Social Desirability Scale. Each item was correlated with the PTSD Index (the overall rating of severity of symptoms of PTSD). Each item was also correlated with ‘he total score for the scale from which the item was drawn. The purpose of this correlational analysis was to determine if each item's correlation with response set bias was stronger than each item's correlation with the construct which the item was intended to measure. If any item correlated significantly with the Social Desirability Scale, and less so with the PTSD index and it's own scale total. this item was noted. The pattern of correlation with the PTSD index and it's own scale was then observed. If the item correlated .08 or higher with the Social Desirability Scale total than the other two factors and at a significant level, this item was then judged to be unacceptablely contaminated by response set bias. Pre-Service Psycho-Social History Questionnaire Three items from the Pre-Service Social History Questionnaire were identified as signficantly correlated with social desirabilty. Each of these 77 items' correlations with social desirabilty, PTSD symptoms, and the Pre-Service Social History total score is described in Table 4.18. TABLE 4.18 INFLUENCE OF SOCIAL DESIRABILTY ON RESPONSES TO PRE-SERVICE SOCIAL HISTORY QUESTIONNAIRE Item SDS CORR PTSD CORR SCALE CORR 21. Arrested prior to service -.28* .07 .46** 30. Relationship broke up within two years return from Vietnam -.31* .24 .19 32. Fights related to alcohol prior to Vietnam duty .33* -.Ol .30* *p-.05 **p-.Ol These three items from the Pre-Service Social History Questionnaire were significantly correlated with social desirability. Item 21 was more highly correlated with it's scale total than with social desirability, yet proved to have no relationship with the criterion variable. Item 30 appeared to have even less merit in 78 addressing the relationship between pre-service social functioning and PTSD symptoms in that it was significantly correlated with social desirability but neither with the criteiron variable nor it's own scale subtotal. Item 32 also proved to have no value in addressing the research hypothesis, being significantly loaded with response set bias, a poor predictor of the criterion variables of PTSD symptoms and less correlated with it's own scale total than social desirabilty. Out of 43 items comprising the total Pre-Service Social History Questionnaire, only the three items identified in Table 3.17 appeared to be significantly biased by perceived response demand. Therefore, because of the large number of uncontaminated items, the data provided in response to the questionnaire appear to have been minimally uncontaminated by the influence of response set bias. Impact of Events Scale Of the 15 items comprising the Impact of Events Scale, only three were significantly correlated with social desirability. In Table 4.19, each item and it's correlation with social desirability, the criterion variable and it's scale total are displayed. 79 Table 4.19 INFLUENCE OF SOCIAL DESIRABILTY ON RESPONSES TO THE IMPACT OF EVENTS SCALE ITEM SDS CORR PTSD CORR SCALE CORR 6. I had dreams of Vietnam -.27* .75** .82** 11. Other things making me think about Vietnam -.29* .75** .84** Intrustion subscale total -.26* .83** .95** *p-.05 **p-.Ol Every item from the Impact of Events Scale correlated signi- ficantly with the criterion of PTSD symptoms. Furthermore, although correlated with social desirabilty at the .05 level, each item was increasingly correlated with PTSD symptoms and it's own scale total at the .01 level. Responses to these three items are assumed to be largely unbiased by social desirabilty in their measure of the influence of a previous stressor on current cognition and affect. 80 Life Events Inventory, Ten of the 55 items from the Life Events Inventory were found to be significantly influenced by social desirability. Each item's correlation with the criterion variable and with the Life Events Inventory total is depicted in Table 4.20. TABLE 4.20 INFLUENCE OF SOCIAL DESIRABILITY ON ITEMS FROM THE LIFE EVENTS INVENTORY ITEM SDS CORR PTSD CORR SCALE CORR 2. Trouble with superiors at work -.31* .00 .35* 4. New Job in new line of work -.32* -.O6 .14 8. Moving to a new residence -.30* .24 .49** 18. Involvement in a fight -.49** .40** .53** 19. Immediate family member starts drinking heavily -.32* .ll .24 25. Gain of a new family member -.33** .17 .25 26. Problems with drugs/alcohol -.36** .26 .48** 38. Increase in number of arguments with spouse -.27* .12 .49** 46. Marital separation -.35** .39** .74** 47. Extra-marital sexual affair -.35** .37** .44* *p-.05 **p-.01 81 The correlational pattern between each item and it's correlation with social desirabilty, PTSD symptoms, and each item's correlation with it's own scale appeared to be more eradic than that seen with the Impact of Events Scale. Of all ten items mentioned which appeared influenced by social desirability, only items 46 and 47 were more highly correlated with PTSD symptoms and the Life Events Inventory total scale score than with social desirability. This finding would argue that these items are less influenced by social desirability and are valuable in the overall measurement of the relationship of life stressors to PTSD symptoms. However, items 2, 4, 19 and 25 appear distorted by response set bias. Therefore, their validity as items contributing to the measurement of the relationship between life stressors and PTSD symptoms is compromised by response set bias. Items 8, 18, 38 and 26 appear questionnable in their usefulness in the testing of the research hypothesis in that these three items all appear to be correlated with social desirability and with their own scale total but significantly less so with PTSD symptoms. Essentially, from.the 55-item Life Events Inventory, the data elicited on eight items appears contaminated by the influences of response set bias. In conclusion, because of the correlational findings, there appeared to be sufficient response set bias. Interpretation of analyses based on the use of the Life Events Inventory must be considered cautiously in light of these findings. 82 Combat Rating Scale Correlational analyses were conducted on the 10 items comprising the Combat Rating Scale and the Social Desirability Scale, PTSD symptoms and the Combat Rating Scale total. Unlike the Pre-Service Social History Questionnaire, the Impact of Events Scale and the Life Events Inventory, no item from the Combat Rating Scale correlated significantly with social desirabilty. Rather, all but one item from the Scale was significantly correlated with PTSD symptoms. The responses given to items from the Combat Rating Scale appear to be uneffected by response set bias to any measurable degree. Post-Service Social Support The three independent variables used to address the hypothesis regarding social support within the first year of return from Vietnam were each correlated with the Social Desirability Scale total and PTSD symptoms. There was no total scale score for Social Support Questionnaire. Therefore it was not possible to correlate each variable with an overall scale total of social support. In Table 4.21, the correlations among the three independent variables and the Social Desirability Scale and PTSD symptoms is reported. 83 TABLE 4.21 INFLUENCE OF SOCIAL DESIRABILITY ON THE REPORTING OF SOCIAL SUPPORT UPON RETURN FROM VIETNAM Item SDS CORR PTSD CORR Quality of relationships in the social support network .11 -.26* (mean score for quality of all relationships in network) Extent of relating upon return from Vietnam ' .21 -.17 (items 2, 5, and 6, appendix 1) Number of people comprising the social support network .33** .05 (total number indicated in section B, (appendix I) *p I .05 **p - .Ol 84 The only variable of the three signficantly influenced by response set bias appears to be the number of people reported to comprise the social support network upon return from Vietnam. That is, those subjects scoring in the positive direction on the Social Desirability Scale also tended to report more individuals in their social support network upon return from Vietnam. This finding may be interpreted in a different fashion than the previous items analyzed. Those persons with a higher number of peOple in their social support network were also high in the need for social approval by others. The variable composed of the number of people constituting the social support network may be less contaminated by the perceived response demands of the testing situation than by the personal need of approval by others. The item is uncorrelated with PTSD symptoms. The findings on the item may argue less for the impact of number of people constituting a social support network as a measure of psychological pathology and more for the item's function as a measure of style of adaptation or coping. The other two items used to operationalize the concept of post-service social support appear relatively uncontaminated by response set bias. Figley Rating Scales The assessment of the influence of social desirabilty on the measurement of the criterion variable, PTSD symptoms, was also critical to assess. To do so, each item from the 85 investigator-rated and psychiatrist-rated Figley Rating Scales was correlated with the Social Desirability Scale total, with it's own scale total, and with the criterion variable. ( N.B.: the correlation of each item with PTSD symptoms and it's own scale are over-inflated in that the PTSD symptom index was formulated by averaging the scores from the Figley Rating Scales obtained by a psychiatric rater and the investigator.) In Table 4.22, the significant correlations between the psychiatrist-obtained PTSD symptom rating, social desirability, and the Figley Rating Scale total are described. 86 TABLE 4.22 INFLUENCE OF SOCIAL DESIRABILTY ON PSYCHIATRIST-RATED SYMPTOMS OF PTSD Item SDS CORR PTSD CORR SCALE CORR 12. occasionally think of Vietnam unexpectedly -.31* .41** .58** 15. have a tough time completing anything you start -.50** .47** .46** 25. worry about losing your temper and hurting someone -.49** .37* .45** Subtotal for symptoms of social withdrawal -.39** .64** .57** Subtotal for variety of symptoms of PTSD ' -.36* .67** .74** Total Scale -.24 *p - .05 **p ' .01 87 In Table 4.23, the significant correlations between items from the investigator-rated Figley Rating Scale, social desirability scale total, PTSD symptoms, and Figley Rating Scale total score are described. TABLE 4.23 INFLUENCE OF SOCIAL DESIRABILITY ON INVESTIGATOR-RATED SYMPTOMS OF PTSD ITEM SDS CORR PTSD CORR SCALE CORR 12. occasionally think of Vietnam.unexpectedly -.36* .33** .48** 16. feel that the older you get the less you need people -.38* .20 .23** 21. guilty about surviving the war when others did not -.35* .57** .41** 25. worry about losing your temper and hurting someone -.43** .60** .53** Subtotal for symptoms of social withdrawal -.37* .62** .57** Subtotal for variety of symptoms of PTSD ' -.33* .73** .77* Total scale -.29 *p - .05 **p - .01 88 It appears that social desirability may have exerted more influence on the data collected from subjects by the investigator than on that data collected by the psychiatric raters. Three items in Table 4.22 proved signficantly correlated with social desirability whereas four items in Table 4.23 were influenced by social desirability. Furthermore, the overall scale total of the investigator rating correlated at -.29 with social desirability; the overall scale total for data-collected by the psychiatrists correlated at -.24 with social desirability. Although these variations are not pronounced they are noteworthy. Variations may be explained by the influence of sex of the interviewer and the subsequent perception by subjects of response expectation. Both psychiatric raters were male. The investigator was female. The Figley Rating Scale is composed of 25 items. As shown in Table 4.22, only two items (numbers 15 and 25) appeared to be more highly influenced by response set bias than by PTSD symptoms or the items' correlations with their own scales of origin. In Table 4.23, only two items (numbers 12 and 16) were more highly correlated with social desirability than with PTSD symptoms or their own scale correlations. Given the modest influence of social desirability in both psychiatrist-rated and investigator-rated scales, it appears that response set bias did not presented major distortion in the data collected regarding PTSD symptoms. 89 From clinical experience, item 25, "worry about losing your temper and hurting someone," is an especially sensitive and painful issue for Vietnam veterans who saw combat. It is noteworthy (but not surprising) that this affect-ladened issue was influenced by response set bias. Summary of Response Set Bias Results In summary, substantial efforts were exerted to assess the influence of response set bias and perceived demands of the testing situation on the veracity of data collected. Such bias may have exerted a significant influence on the Life Events Inventory. On all of the other instruments used to operationalize the testing of the criterion and independent variables, it appears that response set bias was not influential. ASSESSING ACCURACY OF SELF-REPORTED DATA The Significant Other Questionnaire, designed for this study, was also used to identify response set or lack of accuracy in information elicited (see appendiij’. The Significant Other Questionnaire, composed of eight items, was sent to a person that each subject identified as someone who knew him well prior to going to Vietnam and immediately upon his return. Two of the eight items elicited demographic information from the signficant other. The content of six questions concerned life events during the veteran's adolescence. These six items were used in the analysis as an operational measure of the veteran's status 90 prior to service. Subjects were asked to sign a release of information form which was sent to the ”signficant other" permitting the investigator to seek this information by mail (see appendix M). Forty-seven Significant Other Questionnaires were returned, from a total of sixty. If questionnaires were not returned within two weeks of initial mailing, follow-up letters were sent. Also, one subsequent phone call was made encouraging return of the questionnaires followed by a second mailing of the questionnaire with cover letter. A percentage of agreement between the "signficant other's" reponse and that of the veteran's response to the same question was computed using Kendall's coefficient of concordance (Hays, 1973). The coefficients are a measure of agreement between the responses of the significant others and each subject. Data in Table 4.24 indicates content of the six questions asked of the subject and his designated "significant other." The coefficients of concordance are also displayed: 91 TABLE 4.24 COEFFICIENTS OF CONCORDANCE BETWEEN SIGNIFICANT OTHERS AND SUBJECTS QUESTION COEFFICIENT* Who spent the most time raising him from birth to age 18? 651 Grade of school completed prior to entering the service? 942 Did he have at least one close friend during teenage years? 89% Was he involved in outside-of-school activities? 781 Has he ever arrested prior to entering the service? 72% Was he ever in a car accident, physical fight or legal trouble while under the influence or drinking alcohol? 832 *coefficients of concordance or agreement There are no standards for acceptable levels of agreement since the Significant Other Questionnaire was developed specifically for this study. Therefore, it was decided that the percentage of agreement between respondents and their significant others was acceptable. The last two items in Table 4.24 were somewhat influenced by social desirability or response set bias (correlation of each item with social desirability was -.28 and .33 respectively) but still maintained a high level of response concordance. This analysis was conducted in an effort to assess the potential bias introduced by distortion of memory over time (Campbell and Stanley, 1963) 92 RESULTS OF DATA ANALYSIS In order to test the relationship among the independent variables, a correlation matrix was constructed. The intercorrelation among the independent variables and their correlation with the dependent variable of PTSD is presented in Table 4.25 oowxnr>flmoz undtunz HZUHVMZUNZH <>~H>urnm >25 Ham DNHMZUMZH <>wn>urn fi>urn a.uu meannss ca. near-n nae-an on rpms m .oc p.90 ccspnne 1 I 1 1 1 s on .522... . S .8 .5 . 8 .5 .8 r8 meanness. u s 1 I 1 s awn: onrsna .uo .Nm .Hu .wus .cw .nn .uo ~.cc mnemonepns s I 1 mesonpoopom .om .cm .00 .Ou. .uu .pp .Op .po v.0: s.o~ es.cu ss.co~ ul.c~ asses.oo~ 94 Correlational Analysis Of the eight independent variables, three proved to be positively correlated with the dependent variable of PTSD symptoms at a statistically significant level. These variables were: combat intensity, the current subjective impact of a previously-experienced stressor on affect and cognition, and the current level of life stress. There appeared to be a statistically significant negative correlation between PTSD symptoms and one of the three variables comprising the quantification of social support available to the veteran during the first year of return from Vietnam. This variable measured the extent of relating with peOple that the veteran engaged in during that first year of return from Vietnam. The finding was in the opposite direction from that originally hypothesized. There were also some intercorrelations of statistical significance among the independent variables, most notably between the Impact of Events Scale and the Life Events Inventory and between the Combat Rating Scale and the Impact of Events Scale. No other intercorrelations were found to be statistically significant. Regression Analysis A regression analysis was conducted, treating the criterion variable as continuous. The purpose of the multiple linear regression analysis was to evaluate the linear relationship between each of the independent variables and the dependent variable, and again, controlling for the influence of variance introduced by the other independent variables in the analysis. 95 Table 4.26 documents the values for each independent variable calculated from the multiple regression analysis. TABLE 4.26 RESULTS OF THE MULTIPLE REGRESSION ANALYSIS ENTRY ORDER R R R CHANGE BETA T Impact of Events Scale .825 .681 .681 .599 10.1*** Combat Scale .921 .848 .167 .466 7.89*** *p - .05 **p I .01 ***p - .001 Examination of the multiple regression revealed that variables 1) intensity of combat experienced and 2) the current subjective distress experienced from exposure to a previous trauma are most likely to predict existence of PTSD symptoms from the eight predictor variables evaluated. Status at entry into military service, current levels of life stress, magnitude and nature of social support network upon return from Vietnam and pre-service psycho-social functioning do not contribute significantly to the prediction of the existence of PTSD symptoms. 96 In summary, the multiple regression analysis showed that the existence of PTSD symptoms was significantly predicted by the two predictor variables cited in Table 4.26, F(2, 57)'158.69, p ..001. Discriminant Function To clarify and substantiate the findings of the regression analysis, a stepwise discriminant function (Nie, et al.; 1970) was conducted using a categorical variable derived from the previously continuous dependent variable of PTSD symptoms (see Chapter 3, pg. 50 for explanation of grouping). A variable composed of three categories of mild, moderate, and severe PTSD symptoms was created and used as the dependent variable. The independent variables remained the same as those described in Chapter 3, pp. 50-52. Three independent variables were entered into the stepwise discriminant function which produced a highly significant function, Hilk's Lambda-.232; Rao's approximate F-statistic, F(6, llO)‘ 19.73, p e .001. This discriminant function was associated with an overall correct classification rate of 752 of total cases (n-60). Independent variables in the order of entry into the discriminant function were: combat intensity, Impact of Events scale total, and Life Events Inventory total. Jackknifed classification (Lachenbruch and Mickey, 1968) was performed as a test of the accuracy of the prediction. Results of the jackknife verification are displayed in Table 4.27. 97 TABLE 4.27 JACKKNIFE VERIFICATION OF THE DISCRIMINANT FUNCTION RESULTS GROUP PERCENT CORRECT NUMBER OF CASES CLASSIFIED INTO GROUPS Mild Moderate Severe Mild 71.4 15 5 1 Moderate 70.0 2 l4 4 Severe 84.2 0 3 16 Total 75.0 17 22 21 Analysis of Variance Univariate one-way analysis of variance tests were performed as an adjunct to the main multivariate tests. In Table 4.28, the results of the one-way ANOVA's are displayed. 98 TABLE 4.28 ONE-WAY ANOVAS OF INDEPENDENT VARIABLES WITH THE DEPENDENT VARIABLE VARIABLE DF F SIGNIFICANCE Impact of Events Scale 2, 57 51.53 .001 Combat intensity 2, 57 22.63 .001 Life Events Inventory 2, 57 12.84 .01 Relations with others 2, 57 2.37 non-significant Preservice functioning 2, 57 1.37 non-significant Draftee vs. volunteer 2, 57 .80 non-significant Quality of network 2, 57 .25 non-significant Number in social network 2, 57 .13 non-significant SUMMARY In general, both the univariate and multivariate tests appear to support one another. The Impact of Events Scale proved to be most highly correlated with the existence of PTSD symptoms in the regression analysis, the correlational analysis and the analyses of variance. In conjunction with the Combat Scale, the Impact of Events Scale yielded an accuracy of prediction in the discriminant function significant at the .001 level. Combat intensity proved to correlating with the existence the discriminant function. In rank second as an independent variable of PTSD symptoms, in all analyses but the discriminant function, combat 99 intensity was the first independent variable identified as most predictive of the existence of PTSD symptoms. In all but the regression analysis, the current level of life stress (as measured by the Life Events Inventory) proved to be significantly associated with the existence of PTSD symptoms. In the discriminant function, the correlation and the analysis of variance, the Life Events Inventory was identified as third in it's degree of relationship to the existence of PTSD symptoms. Surprisingly, the independent variable quantifying the extent to which the veteran related to others within the first year of return from Vietnam proved to be negatively correlated with the dependent variable in the correlational analysis at the .05 level. The reason for this finding is unclear, nor did the other univariate nor the multivariate tests yield the same finding regarding this independent variable. The other independent variables of number of persons comprising the social support network during the first year of return from Vietnam, quality of those relationships, and pre-service psycho-social functioning proved not be significantly associated with the existence of PTSD symptoms. In Table 4.29, the independent variables identified as significant in each analysis, are presented in the order of their rank of significance or correlation in each of the tests conducted. H>urw b.nw mcxx>wk om MHoszHn>za HzcmwMZUNZH <>w~>urmm UHmowH: mczneHoz >z>hw~>zom noaoon mnmpm Havmnn om mcmonm rwmm mcmono menus wmnwmmmHoz >z>hmmwm oowwmr>HHoz Hanson om m