Identifying Adolescent Runaways: The Predictive Utility of the Personality Inventory for Children

Article excerpt

Nationally, estimates indicated that the number of adolescents who ran away from home each year ranged from 1 million (Walker, 1975) to 2 million (Freudenberger & Torkelson, 1984) with predictions as high as 4 million (National Network of Runaway and Youth Services (NNRYS), 1985). The NNRYS (1991) had recently reported that the current mean estimate of runaways per year was 2 million. Pragmatically, 3% of American families have an adolescent run away from home each year (Garbarino, Schellenback, & Sebes, 1986) or approximately 1 out of 9 secondary school students may have a runaway history (Rohr & James, 1994). Prior reviews of the literature on runaway behavior have focused on the relationship between runaways and their family. A current review indicated that there were additional problematic areas and behaviors - parental problems, delinquent behavior, academic problems, peer relationship difficulties, and problems symptomatic of psychopathology (Rohr, 1991).

Problematic Areas

The problematic area of Family Relationships consisted of behaviors such as parental rejection (Adler, 1980), constant downgrading of the child (Spillane-Greico, 1984), separation and divorce (Ackerman, 1980), intolerable and conflictual home conditions (Blood & D'Angelo, 1974), sibling rivalry problems (Johnson & Peck, 1978), problems in communicating with members of the family (Gullata, 1979), family members not expressing love for each other (Blood & D'Angelo, 1974), and mutual lack of care and love for each other (Spillane-Greico, 1984). The Parental Problem area included behaviors such as parents using excessive punishment (Brandon, 1975), parents with a history of drug use (Steinbock, 1977), and inadequacy in managing children's behavior (Wodarski & Ammons, 1981; Bell, 1984). Behaviors such as stealing (Edelbrock, 1980), disobedience (Blood & D'Angelo, 1974), legal difficulties (Schmidt, 1975), truancy (Nye, 1980), and being adjudicated as a delinquent (Linden, 1979) comprised the Delinquent Behavior area. The School Problem area included behaviors such as having a negative attitude toward school (Nye, 1980), poor problem-solving skills (Roberts, 1982a), and school behavior problems (Gutierres & Reich, 1981). Poor social relationships (Gilchrist, 1984) comprised the Peer Relationships area. Finally, Symptoms of Psychopathology included problems such as anxiety (Williams, 1977), suicidal tendencies (Norey & Donohue, 1985), having been physically abused (Harris, 1980), sexually abused (Hughes, 1981), and having used alcohol and drugs (Maar, 1984).

These earlier findings have been supported by more recent research. The NNRYS (1985) in a survey of approximately 51,000 youths in 210 runaway facilities found that some of the main presenting problems were depression, suicidal tendencies, alcohol and drug use, physical abuse, sexual abuse, and the generic category - severe psychological problems. The most recent survey by the NNRYS (1991) essentially replicated some of their 1985 findings; 50,000 youths in 146 runaway facilities were surveyed. Results indicated that 46% of the youths had a substance abuse problem, with 14% being addicted; 31% reported suffering physical abuse, and 21% reported having been sexually abused; 61% reported being depressed, with 21% having had suicidal ideations.

Some authors have rejected the notion that adolescent runaways were problematic and psychopathological. Chapman (1975) suggested that runaway behavior was related neither to delinquency nor emotional disturbance. Ambrosino (1971) reported that running away was a function of normal adolescent development toward independence and autonomy; it could be viewed as a positive psychological sign. Homer (1973) concluded that running away generally was not a cry for help but a search for adventure. She viewed the notion that runaways were sociopathic or psychopathological as a myth. A recent review by Burke and Burkhead (1989) concluded that none of the above authors provided empirical data to support their conclusions, and no recent research supports this perspective of adolescent runaways as being normal and healthy.

The research on former runaways helps provide clarification of these conflicting findings. Follow-up studies such as that of Robins (1958), found that former runaways had higher rates of mental illness - specifically sociopathic personality. Robins and O'Neal (1959) found that former runaways had more frequent arrests and divorces than did nonrunaways. Later studies indicated that runaways curtailed their schooling and had trouble with the law (Olson, 1977). They also required the assistance of social service agencies for nervous and emotional problems (Olson, Liebow, Mannino, & Shore, 1980). The findings from the follow-up studies indicated that adults who had a runaway history during adolescence, may have had problems that were chronic. Identification of runaways in order to provide intervention may decrease the long-term impact and difficulties youths may experience as adults.

Predictive Studies

Several authors have determined that identifying a constellation of behaviors that were predictive of the adolescent runaway was of clinical and social utility. Englander (1984) was able to achieve an 85.4% correct classification rate in discriminating a group of white female runaways from a white female nonrunaway group. He used three instruments (Intermediate Texas Social Behavior Inventory, Parental Rolationships Questionnaire, and Parental Attributes Questionnaire). The derived variables of SES, self-esteem, endorsement of socially undesirable self-oriented traits, and family harmony contributed approximately equally in separating runaways from nonrunaways. Gilchrist (1984) used an Adlerian lifestyle assessment instrument, the Self-Administering Life Style Inventory. A revised version of the instrument was administered to 51 subjects (sex unknown). The runaway sample was from juvenile homes, and the control group was from local secondary schools. A correct classification rate of 76.5% was achieved in discriminating runaways from nonrunaways. The discriminating variables were personal characteristics, peer relationships, paternal relationships, school experiences, maternal relationships, parental-marital relationships, parental absenteeism, and early recollections. Barber and Hosch (1984) used the Barber Inventory of Runaway Determinants (BIRD) and achieved a 72% correct classification rate in discriminating runaways from nonrunaways. The male and female runaways came from a runaway center and were matched with a control group by age, sex, race, and number of residents in the home. Of the ten scales on the test, low self-esteem, neglect, repetitive runaways, repressed libido, sibling rivalry, and bad influences had the most predictive power. Cahill (1988), using the MMPI, developed a 10-item runaway scale. He compared male and female hospitalized adolescent runaways, hospitalized nonrunaways, and a national sample of control adolescents. He correctly classified 65% of the runaways and 56% of the nonrunaways using the Runaway Scale. The items on the scale were drawn from the MMPI Scale 4 (Psychopathic Deviancy) and the Social Alienation Subscale derived from Scale 8 (Schizophrenia). These scales mostly reflected the characteristics and behaviors associated with adolescent runaways. This review indicated that respectable classification rates were achieved, but with some limitations: only two of the studies used male and female subjects, some studies did not assess all of the problematic areas of runaway behavior, and no study used just one instrument that reflected the six problematic areas of adolescent runaway behavior.

Review of Representative Instruments

The literature indicated that while the general focus has been on assessing the personality characteristics of runaways, very few studies used valid, reliable, and normed personality instruments. Those instruments that were used had serious psychometric limitations. A listing and brief summary of the Mental Measurements Yearbook critiques of the instruments used to assess the problematic areas follows.

In studying the area of Family Relationships, Steinbock (1977) used the Family Environment Scale, the Family Life Space Diagram, and the Family Life Questionnaire. Maar (1984) used the Index of Family Relation Scale, the Child's Attitude Toward Mother Scale, and the Child's Attitude Toward Father Scale. In assessing Parental Problems, Adler (1980) developed the Runaway-Prone Questionnaire. Van-Houten (1977) developed the Life Events Inventory. To investigate the area of Delinquent Behavior, Linden (1979) used the Jesness Inventory. Adler (1980) and Van-Houten (1977) investigated the problematic area of School Problems, using the self-developed instruments mentioned above. The area of Peer Relationships was also assessed by Adler (1980) using the Runaway-Prone Questionnaire. The problematic area of Symptoms of Psychopathology was investigated by Burke (1985), who used the Millon Adolescent Personality Inventory, and Phillips (1976) who used the High School Personality Questionnaire and the Rutgers Self-Descriptive Questionnaire - Parts I and II.

Of the twelve tests cited, professional reviews were found in the Mental Measurements Yearbooks (MMY) (Buros, 1972; 1978; Mitchell, 1985) only for the Family Environment Scale, the Jesness Inventory, the Millon Adolescent Personality Inventory, and the High School Personality Questionnaire. The MMY reviewers concluded that the reliability of the Family Environment Scale was relatively acceptable but no validity studies had been provided. The Jesness Inventory had reliability data for males, but none for females. No validation data were provided for the scales of the Jesness Inventory. The reliability of the Millon Adolescent Personality Inventory (MAPI) was relatively acceptable, but there was very little empirical support for 16 of the 20 subscales. Validation data were lacking for the MAPI. The reliability of the High School Personality Questionnaire was modest, and validation data were lacking.

In summary, there were several concerns regarding the above review of the runaway literature. One, the psychometric limitations of the assessment instruments used may qualify some of the prior research findings on adolescent runaways. A second concern was the issue of parsimoniousness. No one test reflected the range and degree of behaviors and personality characteristics associated with the adolescent runaway. Third, investigating the psychopathology of runaways using a symptom-free control group and a comparison group with known psychopathology would begin to address some of the conflicting research on runaways. Fourth, none of the predictive studies cross-validated their findings. Finally, there is clinical and social relevance in identifying this high-incidence population that may be emotionally disturbed and highly problematic. The purpose of the present study was to investigate whether adolescent runaways exhibited a distinct personality profile and whether this profile could be consistently predicted Using the Personality Inventory for Children (PIC).

METHOD

Hypotheses

1. The selected scales of the PIC (ADJ, ACH, D, FAM, DLQ, SSK) will not predict membership in one of two groups: a runaway sample and a control sample.

2. The selected scales of the PIC (ADJ, ACH, D, FAM, DLQ, SSK) will not predict membership in one of three groups: a runaway sample, a control sample, and a Day Treatment Program sample.

3. There is no difference on selected scales of the PIC (ADJ, ACH, D, FAM, DLQ, SSK) between a runaway sample, a control sample, and a Day Treatment Program sample.

Instrumentation

The PIC is a 600-item, parent-informant, multidimentional measure of child and adolescent behavior, affect cognitive ability, psychopathology, and family functioning. The original scales were constructed using either an empirical or rational/content scale construction strategy. The scales were normed on a sample of 2,582 normal children (no previous mental health contact; 192 subjects ages 3 to 5; 2,390 subjects ages 6 to 16). Norms were established for each gender, ages 3 to 5 and 6 to 16. The standard PIC profile included 3 scales that measured informant response set, Lie (L), Frequency (F), and Defensiveness (DEF); a general screening scale, Adjustment (ADJ); and 3 scales which reflected intellectual and academic functioning. Achievement (ACH), Intellectual Screening (IS), and Development (DVL; and 9 clinical scales, Delinquency (DLQ), Hyperactivity (HPR), Somatic Concern (SOM), Depression (D), Withdrawal (WDL), Anxiety (ANX), severe psychopathology, Psychosis (PSY), social skills functioning, Social Skills (SSK), and family conflict and parental emotional instability, Family Relations (FAM). Scale scores were reported in T score units (M = 50, SD = 10); high scores indicated pathological adjustment. Factor-derived broad-band and shortened profile scales were also available (Lachar, 1982; Lachar, Gdowski, & Snyder, 1982).

Lachar's (1982) shortened version of 280 items was used in this study. Estimates of internal consistancy indicated that there was no significant change between the original and shortened forms. Test-retest reliability indicated that the shortened version retained the temporal stability of its full-length counterpart. Correlations between the shortened and full-length versions of the PIC ranged from .88 to .89 (Forbes, 1986). The percentage of clinical interpretive agreement between the original and shortened version scales was from 92% to 97% (Lachar, 1982).

Validity and Reliability

The Mental Measurements Yearbook (MMY) (Mitchell, Jr., 1985; Conoley, 1989) critiques indicated that citing the numerous validity analysis conducted was inappropriate. A concluding summary statement of the PIC's validity studies indicated that the validity was excellent. Three test-retest reliability studies were cited. For a psychiatric outpatient sample, the mean reliability coefficient was .86; for a normal sample, the mean was .71; and for a different sample of normal children, the mean was .89. Internal consistency estimates had a mean alpha of .74. Mother-father interrater reliabilities had a mean of .57 for a sample of normal children, a mean of .64 with a clinical sample, and a mean of .66 for a psychiatric outpatient sample.

The PIC was chosen for several reasons. First, the scales of this single test seemed reflective of the behaviors, characteristics, and problematic areas associated with runaways as reviewed above and elsewhere (Rohr, 1991). Second, the use of a parent-informant test had several advantages. Typically, parents have been able to generate a more comprehensive description of their child's behavior than can other observers such as teachers or clinicians (Achenbach, 1978). Third, the PIC was not limited to children/adolescents who had relatively mature conceptual skills, adequate reading ability, and motivation (Lachar, 1984). Fourth, the PIC was aptly suited to an assessment that was family or systemically oriented. Finally, the PIC is associated with an actuarial interpretive system (Lachar & Gdowski, 1979) as well as classification typology (Kline, Lachar, & Gdowski, 1987). One of the benefits of the classification procedure was that it generated tentative DSM-III diagnoses.

Scale Selection

The PIC manual (Lachar, 1984) was consulted and the description of what each PIC scale measured was compared to the problematic areas and behaviors associated with each area. The scales Adjustment, Intellectual Screening, Developmental, Anxiety, Withdrawal, Achievement, Depression, Family Relations, Delinquency, and Social Skills were identified as representative of the problematic areas and behaviors associated with adolescent runaways. Examination of the selected scales indicated that some were redundant. According to the manual for the PIC-R (Lachar, 1984), Achievement correlated with the Developmental and Intellectual Screening scales - .84 and .56, respectively. Intellectual Screening correlated with Developmental - .74. Therefore, it was decided not to include the Intellectual Screening and Developmental scales in the analysis. The Withdrawal and Anxiety scales were also excluded. The Depression scale correlated with the Withdrawal and Anxiety scales - .62 and .81, respectively. The final scales to be included in the analysis were ADJ, ACH, D, FAM, DLQ, and SSK.

Subjects and Data Collection

Runaway sample (RUN). The target group was all of the active runaway residents at the Family Link/Runaway House in Memphis, Tennessee between 1986 and 1988, whose custodian/guardian (primarily female) completed the PIC prior to their first therapy session. Of the 250 eligible subjects during this period, 63 guardians consented to participate. Two subjects' protocols were invalid, reducing the sample size to 61. There were 12 (20%) males and 49 (80%) females. The age range was 13 to 17, with a mean age of 15.0. Thirty-five (57.3%) were first-time runaways and twenty-six (42.7%) had run away from home more than once.

This naturalistic sample presented with numerous emotional and psychological problems. Assessment findings and clinical interviews conducted by licensed and certified Master's degree level clinicians indicated that 73% met DSM-III or DSM-III-R criteria for either Dysthymia or Major Depression, with 59% having attempted suicide or experienced suicidal ideations. Nineteen percent had alleged physical abuse; 18% had alleged sexual abuse; 29% had prior mental health treatment of which 16% had prior psychiatric hospitalization. These findings were similar to the survey findings of the NNRYS (1985; 1991).

There were additional similarities between the runaway sample and national estimates on runaways regarding age, gender, and racial identity. Table 1 provides a comparison of the target sample to other samples (Family and Youth Services Bureau (FYSB), 1990; United States General Accounting Office (GAO), 1989; NNRYS, 1991). Racial identity and age were the two demographics that were the most similar among the groups. Gender characteristics were the most dissimilar.

The Runaway Program

A runaway program is typically a state-licensed, 24-hour-a-day, crisis-oriented, residential treatment facility for adolescents, ages 13-17, who have run away from home. Runaway programs offer an array of services including individual, group, and family counseling; educational and vocational services; leisure and recreational activities; alcohol and drug counseling; health care; and information, referral and outreach services.

TABLE 1

Comparison of runaway sample and national estimates on runaway
demographics (age, sex, race).

Group        Runaway    GAO    FYSB    NNRYS

Age -14       36.1%     NA     42%      38%
Age 15-17     63.9%     NA     56%      54%
Male          18.0%     35%    43%      47%
Female        82.0%     65%    57%      53%
Black         36.1%     NA     NA       20%
White         63.9%     NA     NA       75%

The general treatment approach of runaway programs is family and systems oriented, focusing on identifying problems, deescalating the crisis, establishing communication among family members, and attempting to effect enough systemic change so that the home environment can be stabilized. Aftercare and follow-up services are also provided.

Definition of Runaway

The definition of "running away" varies in the literature (Walker, 1975). Predominately it refers to: A child under eighteen years of age who leaves home without his or her parents' permission (Cahill, 1988; United States General Accounting Office, 1989). Particular government agencies may add qualifiers such as being away overnight (United States Department of Health and Human Services, 1989) or being away for at least 24 hours (Saltonstall, 1973). For the purposes of this study, an active runaway was defined as: An adolescent who left home voluntarily, who eventually ended up at a runaway program, and who voluntarily asked for shelter at the program.

Day Treatment Program Sample (DTP)

This comparison group was taken from consecutive adolescent admissions (98) between 1986 and 1991 at a university-based day treatment program (partial hospitalization). The random sample of 60 adolescent subjects ranged in age from 13 through 17 with a mean age of 14.7. Forty-one (68.3%) were male and 19 (31.7%) were female. Forty-nine (81.7%) were white and 11 (18.3%) were black. None of these subjects had a history of running away. They had DSM-III-R (American Psychiatric Association, 1987) diagnostic classifications (primary diagnosis) based upon a complete psychological and/or a psychiatric evaluation and their cases were reviewed by a multidisciplinary team. Specifically, three (5.0%) had a Schizophrenic disorder, 7 (11.6%) had Major Depression, 1 (1.7%) had Bipolar Disorder, 1 (1.7%) had Social Phobia, 23 (38.3%) had Dysthymia, 4 (6.7%) had Separation Anxiety Disorder, 3 (5.0%) had Post-Traumatic Stress Disorder, 1 (1.7%) had Depressive Disorder - NOS, 6 (10.0%) had a Conduct Disorder, 1 (1.7%) had Intermittant Explosive Disorder, 2 (3.3%) had Overanxious Disorder, 3 (5.0%) had Oppositional Defiant Disorder, and 5 (8.3%) had Attention Deficit Hyperactivity Disorder.

Control Sample (CON)

Approximately 350 potential subjects, ages 13-17 were contacted through youth organizations in the Memphis, Tennessee area. Introductory letters and permission forms were sent to the parents of the youth asking permission for their children to participate in the study. Sixty (17.%) parents gave permission. The 60 adolescent subjects consisted of 26 (43%) males and 34 (57%) females. Fifty-six (93%) were white and four (6%) were black. The parents reported that none of the subjects had ever run away from home or had seen a mental health professional for services.

RESULTS

A MANOVA was conducted to determine if there were significant differences within the RUN and CON groups on the demographic variables, gender and race. Results indicated no significant differences (p [greater than] .05) on the 6 PIC scales within each group regarding gender and race. The age range for the PIC is 3-16.5. In this study 17-year-olds were used. Analysis to determine if the norms of the PIC could be extended upward to include-this age group was not done for this study. Prior research has indicated that PIC profiles of 17-year-olds could be included in samples (Lachar & Gdowski, 1979; Lachar, Gdowski, & Snyder, 1984).

Hypothesis 1

To test whether the selected PIC scales would predict membership in one of two groups, a Step-Wise Discriminant Analysis was conducted. Results indicated that the PIC variables ADJ, ACH, FAM, and DLQ were the most important determinants in the functional equation. SSK and D did not add specificity to the predictive power of the canonical function and were dropped from the equation. The total canonical correlation was .91 and the chi-square was significant ([[Chi].sup.2], 4 = 210.117, p [less than] .001.) These findings indicate a strong relationship between the groups and the variables. The total accuracy of prediction was 97.52% with 100% (60) of the Control group being correctly placed and 95.1% (58) of the Runaway group being correctly placed. Only 4.9% (3) were incorrectly placed. The results are summarized in Table 2. Thus, the first hypothesis is rejected.

The results of the above MANOVA also indicate that gender differences are not significant within the DTP group. There are significant differences regarding race within the DTP group on two PIC scales, ACH and DLQ. Wilks' Lambda is F(6, 53) = .71890, p [less than] .05; ACH, F(1, 58) = 4.818, p [less than] .05; DLQ, (F(1, 58) = 5.66, p [less than] .05.

TABLE 2

Classification Results for Cases Selected for Use in the First
Discriminant Analysis

                No. of       Group Membership
Actual Group    Cases     Predicted 1    Predicted 2

Group 1           60         60               0
Control                     100.0%            0.0%
Group 2           61          3              58
Runaway                       4.9%           95.1%

Percent of "Grouped" Cases Correctly Classified: 97.52%

Hypothesis 2

To test whether the selected PIC scales would predict membership in one of three groups, a Step-Wise Discriminant Analysis was conducted. The Results indicated that the PIC variables AIM, FAM, DLQ, and D were the most important determinants in the functional equation. ACH and SSK did not add specificity to the predictive power of the canonical function and were dropped. The total canonical correlation for the first function was .85 and the chi-square was significant ([[Chi].sup.2], 8 = 258.638, p [less than] .001). The total canonical correlation for the second function was .39 and the chi-square was significant ([[Chi].sup.2], 3 = 30.231, p [less than] .001).

These findings indicate a relatively strong relationship between the groups and the variables. The total accuracy of prediction was 73.48%; 96% (58) of the Control group was accurately placed, with 3.3% (2) being misclassified; 50% (30) of the DTP group was accurately placed with 50.0% (30) being misclassified to the other two groups; 66% (53) of. the Runaway group was accurately placed with 33.8% (27) being misclassified to the other two groups. The results are summarized in Table 3.

The high rate of misclassification may indicate that the samples are not significantly different from each other. The second hypothesis is not rejected.

Hypothesis 3

To test whether there is a difference on selected scales between a Control sample, a Runaway sample, and a Day Treatment Program sample, a ONEWAY ANOVA was conducted. If significance were found, a Scheffe' post-hoc analysis would be conducted to determine which specific groups were significantly different from each other and on which variable. Significant differences were found on the following variables.

TABLE 3

Classification Results for Cases Selected for Use in the Second
Discriminant Analysis

                No. of        Group Membership
Actual Group    Cases     Pred.1    Pred.2    Pred.3

Group 1          60        30         9         21
DTP                        50.0%     15.0%      35.0%
Group 2          60         2        58          0
Control                     3.3%     96.7%       0.0%
Group 3          61        15         1         45
Runaway                    24.6%      1.6%      73.8%

Percent of "Grouped" Cases Correctly Classified: 73.48%

Adjustment. For the Adjustment scale there is a significant difference between the samples with F(2), 178) = 164.85, p [less than] .001. The critical value of Scheffe's test is 3.49, and based on that value, the following significant differences were found. The Control sample mean score, (M = 49.96), is significantly different (p [less than] .05) from the Runaway sample mean score (M = 89.14) and the DTP sample mean score (M = 88.71). The Runaway sample is not significantly different (p [greater than] .05) from the DTP sample.

Achievement. For the Achievement scale there was a significant difference between the samples with F(2, 178) = 37.20, p [less than] .001) The critical value of Scheffe's test is 3.49, and based on that value, the following significant differences were found. The Control sample mean score (M = 45.90) is significantly different (p [less than] .05) from the Runaway sample mean score (M = 60.39) and the DTP sample mean score (M = 60.85). The Runaway sample is not significantly different (p [greater than] .05) from the DTP sample.

Depression. For the Depression scale there is a significant difference between the samples with F(2), 178) = 65.37, p [less than] .001. The critical value of Scheffe's test is 3.49 and based on that value the following significant differences were found. The Control sample mean score (M = 51.45) is significantly different (p [less than] .05) from the Runaway sample mean score (M = 72.36) and the DTP sample mean score (M = 78.71). The Runaway sample and the DTP samples are not significantly different (p [greater than] .05) from each other.

Family relations. For the Family Relations scale there is a significant difference between the samples F(2), 178) = 39.46, p [less than] .001. The Scheffe's test critical value is 3.49 and based on that value, the following significant differences were found. The Control sample mean score (M = 48.85) was significantly different (p [less than] .05) from the Runaway sample mean score (M = 62.09) and the DTP sample mean score (M = 64.56). The Runaway sample is not significantly different (p [greater than] .05) from the DTP sample.

Delinquency. For the Delinquency scale there is a significant difference between the samples F(2, 178) = 184.24, p [less than] .001. The Scheffe's test critical value is 3.49 and based on that value, the following significant differences were found. The Control sample mean score (M = 49.91) is significantly different (p [less than] .05) from the Runaway sample mean score (M = 102.13) and the DTP sample mean score (M = 89.00). There is also a significant difference (p [less than] .05) between the Runaway sample (M = 102.13) and DTP sample (M = 89.00).

Social Skills. For the Social Skills scale there is a significant difference between the samples F(2, 178) = 52.86, p [less than] .001. The critical value of Scheffe's test is 3.49 and based on that value, the following differences were found. The Control sample mean score (M = 43.66) is significantly different (p [less than] .05) from the Runaway sample mean score (M = 64.63) and the DTP sample mean score (M = 65.05). The Runaway sample is not significantly different (p [greater than] .05) from the DTP sample.

DISCUSSION

Edelbrock (1980) concluded that research into significant differences between runaways and nonrunaways was not very fruitful and should be discontinued. That position may have been premature because in 1980 it had not been concluded, empirically, that runaways were not significantly different from nonrunaways on measures of psychopathology. In fact, it was only recently that psychopathology had been empirically investigated. But even that research had significant methodological problems.

Methodological problems underscore the conclusions of Adams & Monroe, (1979): that psychological profiles were not adequate predictors of adolescent runaways. This was primarily due to: (1) not knowing whether the psychopathological behavior associated with runaways was a precipitant to running away or a consequence of the episode itself; and (2) were runaways significantly different from non-runaways regarding psychopathology. These and other methodological problems made synthesizing and summarizing the multidisciplinary literature on runaways extremely difficult.

The findings from the first hypothesis indicate that: (1) there was significant empirical support for conceptualizing runaway behavior into the problematic areas; (2) by using a valid and reliable measure of psychopathology, representative of the problematic areas, a distinct psychological profile of the adolescent runaway was identified, which was highly predictive and significantly different from the psychological profile of youth who had not run away or required mental health services; and (3) the PIC is extremely sensitive to the behaviors associated with runaways and their family, and its use is encouraged whenever professionals work with runaways.

The findings for the investigation of the second and third hypotheses lend empirical support to mental health professionals' clinical impressions that runaways exhibit psychopathological behavior. The distinct psychological profile of the adolescent runaway became blurred when the runaway group was compared to a moderate to severely emotionally disturbed clinical group. Runaways were essentially not significantly different from the DTP group and, in fact, on one scale, DLQ, the runaways were more psychopathological than the clinical group. These findings provide empirical support for the NNRYS 1985 and 1991 survey findings that runaways have psychological problems.

However, the questions is: Were the psychological problems a consequence of the runaway incident or a precipitant (Adams & Monroe, 1979)? An additional advantage of the use of the PIC with runaways was that if the PIC was administered to the parent prior to the first therapy session at the runaway program, the items the parent endorsed about their child's behavior occurred prior to the runaway episode. In this study the PIC was administered prior to the first therapy session. Additional support comes from the type of experiences of runaways while "on the run." Brennan, Huizinga, & Elliot (1978) found that approximately 25% reported that they experienced a good time while away from home; 20% reported unpleasant experiences such as hunger, cold, fear, boredom, and lack of any positive experiences. These figures leave over 50% of runaways reporting neither type of experience. Only about 3% reported traumatic experiences - being beaten, robbed, raped, or jailed (Brennan et al., 1978). These findings lend some credence to the view that psychological problems were present prior to the runaway incident.

In summary, adolescent runaways do exhibit a distinct personality profile that is predictable. However, the profile loses its distinction when compared to a known psychopathologically disturbed group. This is due to the statistical similarity between the two groups. Several methodological concerns, as stated in the multidisciplined literature, were addressed in an effort to minimize threats to validity.

Even so, some limitations were present: Not all of the groups were randomly selected; there may be time constraints when obtaining a representative, randomly selected runaway sample; this runaway sample took two years to collect. Further, very few runaway programs have a psychological assessment component, much less a staff member qualified and licensed to administer the tests. Another limitation was that when using discriminant analysis, the overall correct classification rate may be inflated due to the fact that the functional equation was generated from the very sample to which it is applied. To counter this inflation, cross-validation procedures are necessary. Due to the above-noted time constraints, a cross-validation sample was not collected. It could be argued that with a significant chi-square, a .91 canonical correlation, and a total classification rate of 97.52%, the cross-validation procedure would probably only provide further support for the results rather than yield a significant detraction from the findings. There was a demographic difference within the DTP comparison sample on the variable of race. While the PIC manual (Lachar, 1984) stated that race does not bias PIC profiles, it did occur on two scales (ACH, DLQ) within this DTP sample. There may be a sampling bias due to the fact that the Day Treatment Program provided mental health services to a higher percentage of black adolescents whose range of PIC scale scores was greater than those of the white subsample. Research that uses comparison groups may need to consider whether random stratification procedures should be employed.

Future research with adolescent runaways should continue to use the PIC. Besides the actuarial interpretive system, the PIC also has an associated typology or classification system which has an extremely high correct classification rate (over 90%) (Kline, Lachar, & Gdowski, 1987). This typology also generates tentative DSM-III diagnosis. Suggested avenues for investigation should include determining a descriptive personality profile of the adolescent runaway. Based on this study's findings, determining such a profile would be helpful to mental health professionals who work with runaways. Since psychopathology seems to be associated with the adolescent runaway, determining its severity and chronicity would also be useful. Additionally, the runaway sample included youth who had run away only once and those with multiple runaway episodes. There may be a psychological uniqueness associated with the multiple runner. Also, more emphasis should be placed on conceptualizing the runaway as emotionally disturbed rather than not. Investigation in this direction would focus research efforts on understanding the adolescent runaway and providing intervention and treatment services.

Specific recommendations for runaway programs support some of the NNRYS (1985) survey findings. The survey concluded that the runaways at that time were exhibiting more psychological problems than were manifested 5 to 6 years ago, and that qualified staff were needed to provide the mental health services necessary. The findings from this study provide support for three relevant recommendations. First, since runaways seem to have psychological problems, upgrading the existing assessment component is warranted; second, employing personnel qualified to treat these mental health problems is also appropriate; and third, use of the PIC in runaway programs is advocated.

REFERENCES

Achenbach, T. M. (1978). The Child Behavior Profile: I. Boys Aged 6-11. Journal of Consulting and Clinical Psychology, 46, 478-488.

Ackerman, R. J. (1980). The effects of parent death, long-term illness, and divorce on children running away from home. (Doctoral dissertation, Western Michigan University, 1980.) Dissertation Abstracts International, 41, 8231A.

Adams, G. R., & Munro, G. (1979). Portrait of the North American runaway: A critical review. Journal of Youth and Adolescence, 8(3), 359-373.

Adler, I. S. (1980). A profile of the runaway-prone adolescent girl and a psycho-diagnostic instrument for her detection. (Doctoral dissertation, University of California, Los Angeles, 1979). Dissertation Abstracts International, 40, 3910.

Ambrosino, L. (1971). Runaways. Boston: Beacon.

American Psychiatric Association (1987). Diagnostic and Statistical Manual of Mental Disorders (3rd ed. rev). Washington, DC: Author.

Barber, A. V., & Hosch, H. M. Development and evaluation of the Barber inventory of runaway determinants. Unpublished manuscript, 1983.

Bell, R. O. (1984). Parent-adolescent interaction in runaway females. (Doctoral dissertation, University of Minnesota, 1984.) Dissertation Abstracts International, 44, 3507A.

Blood, L., & D'Angelo, R. (1974). A progress research report on value issues in conflict between runaways and their parents. Journal of Marriage and the Family, 36(3), 486-490.

Brandon, J. S. (1975). The relationship of runaway behavior in adolescents to the individual's perceptions of self, the environment, and parental antecedents. (Doctoral dissertation, University of Maryland, 1975.) Dissertation Abstracts International, 36, 646B.

Brennan, T., Huizinga, D., & Elliott, D. S. (1978). The social psychology of runaways. Lexington, MA: D.C. Heath.

Burke, W. H. (1985). Runaway youth: An analysis of personality and psychological characteristics. (Doctoral dissertation, The Florida State University, 1985.) Dissertation Abstracts International, 46, 2923A.

Burke, W. H., & Burkhead, E. J. (1989). Runaway children in America: A review of the literature. Education and Treatment of Children, 12(1), 73-81.

Buros, O. (Ed.). (1972). Seventh Mental Measurements Yearbook. Lincoln: The University of Nebraska Press.

Buros (Ed.). (1978). Eighth Mental Measurements Yearbook. Lincoln: The University of Nebraska Press.

Cahill, K. R. (1988). The prediction of adolescent runaway: Psychometric and demographic approaches. (Doctoral dissertation, University of Nebraska, Lincoln, 1988). Dissertation Abstracts International, 50, 1149B.

Chapman, C. (1975). America's runaways. New York: Morrow.

Conoley, J. C. (Ed.). (1989). Tenth Mental Measurements Yearbook. Lincoln: University of Nebraska Press.

Edelbrock, C. (1980). Running away from home: Incidence and correlates among children and youth referred for mental health services. Journal of Family Issues, 1(2), 210-228.

Englander, S. W. (1984). Some self-reported correlates of runaway behavior in adolescent females. Journal of Consulting and Clinical Psychology, 52(3), 484-485.

Family and Youth Services Bureau. (1990). Annual report to the Congress on the Runaway and Homeless Youth Program fiscal year 1989. Washington, DC.

Forbes, G. B. (1986). Comparison of the short and original scales of the Personality Inventory for Children. Journal of Clinical Psychology, 42, 129-132.

Freudenberger, H. J., & Torkelsen, S. E. (1984). Beyond the interpersonal: A systems model of therapeutic care for homeless children and youth. Psychotherapy, 21(1), 132-140.

Garbarino, J., Schellenback, C., & Sebes, J. (Eds.). (1986). Troubled youth, Troubled families. New York: McGraw-Hill.

Gilchrist, L. (1984). Use of the Revised Self-Administering Lifestyle Inventory in distinguishing between runaway and nonrunaway adolescents. (Doctoral dissertation, Idaho State University, 1984.) Dissertation Abstracts International, 45, 3071A.

Gullotta, T. P. (1979). Leaving home: Family relationships of the runaway child. Social Casework, 60, 111-114.

Gutierres, S. E., & Reich, J. W. (1981). A developmental perspective on runaway behavior: Its relationship to child abuse. Child Welfare, 60(2), 89-94.

Harris, L. (1980). Middle-class high school dropouts: Incidence of physical abuse, incest, sexual assault, symstomatic behavior, and emotional disturbance. (Doctoral dissertation, University of Minnesota, 1980.) Dissertation Abstracts International, 41, 2058A.

Homer, L. (1973). Community-based resource for runaway girls. Social Case-work, 54, 473-479.

Hughes, K. A. (1981). The reported incidence of incest among runaway female adolescents. (Doctoral dissertation, California School of Professional Psychology, Berkeley, 1980.) Dissertation Abstracts International, 41, 4638B.

Johnson, N. S., & Peck, R. (1978). Sibship composition and the adolescent runaway phenomenon. Journal of Youth and Adolescence, 7, 301-305.

Kline, R. B., Lachar, D., & Gdowski, C. L. (1987). A Personality Inventory for Children profile typology of children and adolescents: II. Classification rules and specific behavior correlates. Journal of Clinical Child Psychology, 16(3), 225-234.

Lachar, D., & Gdowski, C. L. (1979). Actuarial assessment of child and adolescent personality: An interpretive guide for the Personality Inventory for Children profile. Los Angeles: Western Psychological Services.

Lachar, D. (1982). Personality Inventory for Children (PIC): Revised Format Manual Supplement. Los Angeles: Western Psychological Services.

Lachar, D., Gdowski, C. L., & Snyder, D. K. (1982). Broad-band dimensions of psychopathology: Factor scales for the Personality Inventory for Children. Journal of Consulting and Clinical Psychology, 50, 634-642.

Lachar, D. (1984). Multidimensional Description of Child Personality: A manual for the Personality Inventory for Children. Los Angeles: Western Psychological Services.

Linden, B. W. (1979). Personality characteristics, family variables, and ego development in runaway girls. (Doctoral dissertation, University of Wyoming, 1979.) Dissertation Abstracts International, 41, 694B.

Maar, J. E. (1984). Critical factors related to adolescent runaway behavior in the state of Florida. (Doctoral dissertation, The Florida State University, 1984.) Dissertation Abstracts International, 45, 2993A.

Mitchell, J. V., Jr., (Ed.). (1985). Ninth Mental Measurements Yearbook. Lincoln: The University of Nebraska Press.

National Network of Runaway and Youth Services, Inc. (1985). A profile of America's runaway and homeless youth and the programs that help them. Washington, DC: Author.

National Network of Runaway and Youth Services, Inc. (1991). To whom do they belong? Washington, DC: Author.

Novy, D. M., & Donohue, S. (1985). The relationship between adolescent life stress events and delinquent conduct including conduct indicating a need for supervision. Adolescence, 20, 313-321.

Nye, F. I. Runaways: Some critical issues for professionals in society. Cooperative Extension Bulletin - 0744, Washington State University, 1980.

Olson, L. R. (1977). Suburban runaways: A follow-up study. (Doctoral dissertation, University of Maryland, 1977). Dissertation Abstracts International, 38, 2379B.

Olson, L., Liebow, E., Mannino, F. V., & Shore, M. F. (1980). Runaway children twelve years later: A follow-up. Journal of Family Issues, 1(2), 165-188.

Phillips, S. B. (1976). The runaway girl; A differentiating profile of personality, family school and social factors. (Doctoral dissertation, Rutgers University, 1978.) Dissertation Abstracts International, 39, 394B.

Roberts, A. R. (1982a). Stress and coping patterns among adolescent runaways. Journal of Family Issues, 5(1-2), 15-27.

Robins, L. N. (1958). Mental illness of the runaway: A 30-year follow-up study. Human Organization, 16, 1-15.

Robins, L. N., & O'Neal, P. (1959). The adult prognosis for runaway children. American Journal of Orthopsychiatry, 29, 752-761.

Rohr, M. E. (1991). Personality profile of adolescent runaways. (Doctoral dissertation, Memphis State University, 1991.) Dissertation Abstracts International, 52, 4, 1215A.

Rohr, M. E., & James, R. K. (1994). Runaways: Some suggestions for prevention, coordinating services, and expediting the reentry process. The School Counselor, 42(1), 40-47.

Saltonstall, M. B. (1973). Runaways and street children in Massachusetts. Boston: Massachusetts Committee on Children and Youth.

Schmidt, W. A. (1975). Runaway children. Journal of the American Medical Association, 232(6), 651-652.

Spillane-Greico, E. (1984). Characteristics of a helpful relationship: A study of empathic understanding and positive regard between runaways and their parents. Adolescence, 19, 63-75.

Steinbock, L. A. (1977). Nest leaving: Family systems of runaway adolescents. (Doctoral dissertation, 1978.) Dissertation Abstracts International, 38, 4544B.

United States Department of Health and Human Services (1989). Annual report to the Congress on the runaway and homeless youth program, fiscal year 1989. Washington, DC: Office of Human Development services; Administration of Children, Youth and Families; Family Services Bureau.

United States General Accounting Office (1989). Homelessness and runaway youth receiving services at federally funded shelters (GAO/HRD-90-45). Washington, DC: U. S. Government Printing Office.

Van-Houten, T. A. (1977). Life stress: A predictor of adolescent running away. (Doctoral dissertation, The Catholic University of America, 1977.) Dissertation Abstracts International, 38, 5719.

Walker, D. K. (1975). Runaway youth: An annotated bibliography and literature review. ERIC Document Reproduction Service No. ED138907.

Williams, A.D. (1977). A comparison study of family dynamics and personality characteristics of runaways and nonrunaways. (Doctoral dissertation, United States International University, 1977). Dissertation Abstracts International, 39, 5096B.

Wodarski, J. S., Ammons, P. W. (1981). Comprehensive treatment of runaway children and their parents. Family Therapy, 8(3), 229-240.

Reprint requests to Michael E. Rohr, Ed. D., University of Tennessee, Division of Child and Adolescent Psychiatry Day Treatment Program, 711 Jefferson 607W, Memphis, Tennessee 38105.

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.