Professional Documents
Culture Documents
The Adolescent Problems Inventory (API) and the Problem Inventory for
Adolescent Girls (PIAG) are commonly employed in studies of antisocial and
delinquent behaviour. This paper offers a review and critique of these inventories.
The development and construction of the inventories are detailed, followed by dis-
cussion of their psychometric properties. The weight of evidence suggests that the
inventories are related to behavioural indices associated with antisocial and delin-
quent behaviour. More recent studies have considered the factor structure of the
API, suggesting several dimensions of social functioning. It is concluded that the
API and PIAG are robust instruments that can yield valuable clinical and research
data. 1996 The Association for Professionals in Services for Adolescents
Introduction
There are many uses for a valid and reliable measure of adolescent functioning and ability.
Those engaged in clinical work with adolescents may require such a measure as a screening
tool or as part of a full assessment; clinical evaluation can be aided, in terms of both process
and outcome, by using such a measure; and empirical research obviously needs valid and
reliable measures. In terms of measures of adolescent social functioning, two scales have
been widely used in the literature: the Adolescent Problems Inventory (API) for adolescent
males (Freedman et al., 1978) and the Problem Inventory for Adolescent Girls (PIAG;
Gaffney and McFall, 1981). The purpose of the present paper is to offer practitioners and
researchers an overview of the API and PIAG.
The original theoretical background to the API lay in the social skills model of adolescent
delinquent behaviour: this model proposed that delinquents will show situation-specific
skill deficits as compared to nondelinquents. As McFall (1982) suggests, skill level
determines an individual’s perceived social competence in a given situation. The API was
designed to provide a profile of a young person’s strengths and weaknesses in responding to
the demands of a range of social situations. Thus, an API score is not in itself predictive of
delinquency, but offers an assessment of deficits that are potentially related to delinquent
behaviour. The Problem Inventory for Adolescent Girls (PIAG) was later developed by
Gaffney and McFall (1981) to provide a parallel measure for female adolescents. In the
long-term, these assessments were seen as providing a foundation on which to structure
Reprint requests and correspondence should be addressed to Dr C. Hollin, School of Psychology, University of
Birmingham, Edgbaston, Birmingham B15 2TT, U.K.
social skills training programmes for delinquent adolescents by indicating areas for training
(Henderson and Hollin, 1983, 1986; Hollin, 1990).
Inventory construction
The API and PIAG were both developed using the same methodology, involving five
sequential steps: (i) situational analysis; (ii) item development; (iii) response enumeration;
(iv) response evaluation; and (v) construction of the inventory and the rater’s manual.
Item development. The set of general problem situations was then translated into
specific items, setting the scene, characters, history and goal of a given social situation.
There were two types of item that differed according to the question posed to respondents:
(1) “What would you say or do now?”; and (2) “What can you do to go about solving this
problem?” Thus from the API, an example of the former:
You’re visiting your aunt in another part of town and you don’t know any of the guys your age
there. You’re walking along her street and some guy is walking toward you. As he is about to
pass you, he bumps into you and you nearly lose your balance.
What do you say and do now?
While for the latter, again from API:
It is Saturday morning and you have nothing planned for the whole day. There’s nothing to
look forward to all day. You feel bored already just thinking about it. You need some kicks.
What can you do about solving this problem?
At this stage there were 90 potential items for the API, and 78 for the PIAG.
Construction of the inventory and rater’s manual. The criteria for retention of
Adolescent problem inventory 349
an item in the inventory were two-fold: first, an interjudge agreement of response levels of
competency of 75% or more; secondly, at least 25% of the responses to an item judged as
incompetent by at least 75% of the judges. This procedure gave an API of 44 items and a
PIAG of 52 items. The rater’s manual was formulated to score the items on a five-point
Likert scale: 0 (very incompetent), 2 (incompetent), 4 (neither competent nor
incompetent), 6 (competent), 8 (very competent). Explicit criteria from the response
evaluation stage are given in the manual, along with sample responses at each level for each
item.
Administration
The text of the problem situations is presented on audiotape to respondents (although items
can be administered verbally or in written form). After each individual item, the
respondents are asked to say aloud what they would do in such a situation. This response is
audiotaped for subsequent evaluation, according to the criteria given in the manual, by
trained judges. Transcription of responses is desirable to eliminate coding biases that may
arise from the tone or accent of voices. (The problem of bias does not arise in the multiple-
choice formats where responses are simply circled on the form.) The judges’ ratings may be
averaged to give a mean competency score, or added to give a total score. The mean score is
preferred, as total scores of inventories are not very informative (McFall, 1982). The modal
score is a useful alternative, indicating the most commonly used level of rated competency.
Multiple-choice version
A multiple-choice format was developed for both the API (Freedman et al., 1978) and the
PIAG (Gaffney, 1984) as a quicker alternative to the open-ended format that would also
facilitate group administration. Five response alternatives were written for each item,
derived from the scoring criteria in the manual, corresponding to the five points of the
manual’s rating scale. Scores on the multiple-choice versions are generally higher than
those found with the free response format (Freedman et al., 1978; Gaffney, 1984).
Short form
Gibbs et al. (1987) developed the Inventory of Adolescent Problems—Short Form (IAP-
SF) as a shorter open-ended format alternative to the API. A panel of expert judges (e.g.
clinical psychologists and juvenile detention centre officials) independently reviewed the
original items, using two criteria: (1) their relevance to current adolescents; and (2) how
generic they were with reference to gender. In all, 22 such situations were identified, while
still retaining the situational diversity of the original inventory. The IAP-SF is scored on
the same five-point scale as the API. A revised version of the IAP-SF has been compiled for
use with both male and female adolescents (Gibbs et al., 1994).
al., 1984; Hunter and Kelley, 1986; Ward and McFall, 1986; Leadbeater et al., 1989;
Simonian et al., 1991; Leeman et al., 1993; Oyserman and Saltz, 1993).
To estimate rater consistency, two studies asked raters to repeat their ratings of selected
items: the mean r for the API was 0.90 (Freedman et al., 1978) and 0.91 for the PIAG
(Gaffney and McFall, 1981). There are no test–retest data on the reliability of the API or
PIAG.
Reliability analyses were conducted during the validation studies. However, as the items
were designed to be situation-specific and hence not overlap, the inter-item and item-total
correlations would be expected to be low. Also, the internal consistency values are probably
slightly inflated due to the extreme samples used in the validation studies, but are high
enough for this not to be of concern (see Table 1).
Validity
There are several comparisons of the performance of delinquents and nondelinquents on
the API and PIAG: all report highly significant differences between these two groups
(Freedman et al., 1978; Gaffney and McFall, 1981; Dishion et al., 1984; Gaffney, 1984;
Ward and McFall, 1986; Oyserman and Saltz, 1993). Delinquent adolescents consistently
perform at a lower level, both on the multiple-choice and free response formats. Also, when
instructed to respond with their “best” response, although their scores improve delinquents
still give significantly less competent responses than nondelinquents (Freedman et al.,
1978).
Discriminant analyses performed as part of the API validation study (Freeman et al.,
1978) gave a validity coefficient of r=0.60, with an 89% correct classification of
respondents as delinquents or nondelinquents. A parallel analysis of the PIAG (Gaffney
and McFall, 1981) produced a phi coefficient of 0.70, and an 85% correct classification of
the respondents.
Most of the studies using the API relate the scores to some kind of behavioural index of
conduct, using both official and self-report sources. Self-report delinquency checklists are
most common, relating both type and frequency of behaviours to inventory scores.
Significant correlations between inventory and behavioural index scores, showing that
greater involvement in misconduct and delinquency is associated with less competent
responses, have been reported by Dishion et al. (1984), Gaffney (1984), Leadbeater et al.
(1989), Leeman et al. (1993), Oyserman and Saltz (1993), Simonian et al. (1991) and Ward
and McFall (1986).
Those studies using official delinquency as a behavioural criterion overwhelmingly find
significant correlations between conduct (both within and outside institutions) and API
scores (Dishion et al., 1984; Simonian et al., 1991; Leeman et al., 1993). Only one study
failed to find this relationship: Hunter and Kelley (1986) correlated nine official criteria of
institutional antisocial and disruptive behaviour with API scores and found no significant
associations. However, this may be partly explained by the short period, 3 weeks, that
assessment of the young people lasted. This compares with monitoring of behaviour over a
6-month period in the Freedman et al. (1978) study that reported significantly different API
scores between high and low disruptive incarcerated delinquents. In addition, with an
incarcerated delinquent population, Hains and Herman (1989) found a significant
difference between high and low behavioural functioning individuals, as defined by their
level on a token economy system. However, further analysis revealed that this was only the
case for nonaggressive offenders (as defined by DSM-III criteria).
Veneziano and Veneziano (1988) divided a sample of institutionalized juvenile
delinquents according to their scores on the API. Those delinquents in the least competent
band showed the highest levels of disturbance on several measures, including impulsivity,
dangerousness, aggression, moral values, anger and family difficulties. Hollin and Swaffer
(1993) found no differences in API scores between adolescent sex and non-sex offenders,
replicating work with adult sex offenders. They also looked at the relationship of API scores
with indices of social perception and social performance. They found that while social
perception and social performance were significantly correlated with each other, neither
correlated significantly with API scores. Parenthetically, Hollin and Swaffer slightly
modified the American form of the API for use with an English population. They reported
similar scores, both by item and overall mean, to Freedman et al.’s delinquent norms.
Mediating variables
With regard to potential mediating variables, neither Freedman et al. (1978) or Gaffney and
McFall (1981) found a significant relationship between API scores and socio-economic
status (SES). Gaffney (1984) did find a relationship between SES and PIAG scores:
however, when SES was controlled, the difference between delinquent and nondelinquent
groups on the PIAG was maintained. There are similar mixed findings regarding the
relationship between API scores and intelligence. Some studies find no relationship
(Gaffney and McFall, 1981; Dishion et al., 1984), others highly significant correlations
(Freedman et al., 1978; Gaffney, 1984). This discrepancy may be a function of variation in
IQ measurement and matching between delinquent and nondelinquent groups.
Alternatively, the relationship may be due to some other mediating variable: studies that
control for IQ do find a difference between delinquent and nondelinquent samples on the
measure of social competence (Gaffney, 1984).
Factor structure
The validation studies for the API and PIAG explored the factor structure of the inventory
using statistical clustering techniques (Freedman et al., 1978; Gaffney and McFall, 1981). A
lack of clear identifiable clusters suggested a minimum of overlap between items. This is not
a surprising finding given the wide range of items included in the API. However, the
shorter version of the inventory (IAP-SF) has proved more amenable to factor analysis.
With IAP-SF scores from a sample of 100 incarcerated juvenile delinquents, Simonian et al.
(1991) conducted a principle components factor analysis that produced three factors,
accounting for 41% of the variance. The first factor was composed of items labelled
352 E. J. Palmer and C. R. Hollin
Discussion
Studies comparing the performance of delinquents and nondelinquents on the API (and
PIAG) consistently show that delinquent young people perform significantly less well than
nondelinquents. Those studies concerned with delinquents suggest strongly that
behavioural indices (both official and self-report) are related to scores on the API. These
correlational relationships are significant, although not always of a high magnitude.
However, low correlations are to be expected as the scores used in these correlations are
either summary scores or mean scores, which simplify the picture. The API is an inventory,
and so in practice it is the overall pattern or skills profile that should be related to
behaviour.
Looking across studies at the scores generally obtained for delinquents and
nondelinquents, there is consistency in the level of scores obtained. As yet, these
inventories have been mostly used with American Caucasian populations, although some
studies show no differences in the scores attained by different cultural backgounds (Ward
and McFall, 1986; Veneziano and Veneziano, 1988). API situations have also been used in
studies requiring interpersonal vignettes for other reasons. Fondacaro and Heller (1990)
employed 12 items to elicit the causal attributions of blame in adolescents. Leadbeater et al.
(1989) used API items to elicit the interpersonal negotiation strategies of high-risk youth.
Clearly the API and PIAG are potentially useful measures for both research and practice.
However, the conceptual tangle highlighted by measures such as the API and PIAG, lies in
the distinction between terms such as social skill and social competence (Rubin and Rose-
Krasnor, 1992). While such global terms refer to social functioning generally, there is the
issue of what exactly the API and PIAG measure: the literature variously refers to social
skills, social competence and social problem-solving. Given that these inventories ask
young people to make judgements about their likely actions in the various situations, then
social decision-making is probably nearest the mark. As such, social decision-making is a
Adolescent problem inventory 353
part of social functioning, clearly associated with both social skill and social competence.
The factor analytic studies have highlighted the social arenas in which adolescents make
decisions related to eventual antisocial behaviour.
The practical implications of these points are two-fold. First, it seems likely that the API
and PIAG actually measure specific aspects of social cognition and therefore cannot be
taken as global measures of social competence. Thus, to achieve a rounded assessment, the
API and PIAG should be supplemented by more general measures of social competence and
social problem-solving (Rubin and Rose-Krasnor, 1992). Second, as the measures tap
aspects of social cognition, it follows that they must be augmented by measures of social
behaviour, for example by observation.
In summary, the logical next step in the genesis of these measures, which would
potentially answer many theoretical questions, is a large scale study looking at the
relationship of API and PIAG scores with other measures of adolescent functioning. This
level of knowledge would, in turn, lead to more comprehensive and sensitive assessments to
inform research and practice.
References
Dishion, T.J., Loeber, R., Stouthamer-Loeber, M. and Patterson, G.R. (1984). Skill deficits and male
adolescent delinquency. Journal of Abnormal Child Psychology, 12, 37–54
Fondacaro, M.R. and Heller, K. (1990). Attributional style in aggressive adolescent boys. Journal of
Abnormal Child Psychology, 18, 75–89.
Freedman, B.J., Rosenthal, L., Donahoe, C.P., Schlundt, D.G. and McFall, R.M. (1978). A
social–behavioral analysis of skill deficits in delinquent and nondelinquent adolescent boys.
Journal of Consulting and Clinical Psychology, 46, 1448–1462.
Gaffney, L.R. (1984). A multiple-choice test to measure social skills in delinquent and nondelinquent
adolescent girls. Journal of Consulting and Clinical Psychology, 52, 911–912.
Gaffney, L.R. and McFall, R.M. (1981). A comparison of social skills in delinquent and
nondelinquent adolescent girls using a behavioral role-playing inventory. Journal of Consulting
and Clinical Psychology, 49, 959–967.
Gibbs, J.C., Swillinger A., Simonian, S.J., Shockley, K.C. and Paradissis, E.J. (1987). Inventory of
Adolescent Problems—Short Form and Scoring Manual. Unpublished manuscript, Ohio State
University.
Gibbs, J.C., Swillinger, A., Simonian, S.J., Shockley, K.C., Leeman, L.W., Rowland S.J. and Jaycox,
C. (1994). Inventory of Adolescent Problems—Short Form (IAP-SF). Unpublished manuscript,
Ohio State University.
Hains, A.A. and Herman, L.P. (1989). Social cognitive skills and behavioural adjustment of
delinquent adolescents in treatment. Journal of Adolescence, 12, 323–328.
Henderson, M. and Hollin, C. (1983). A critical review of social skills training with young offenders.
Criminal Justice and Behavior, 10, 316–341.
Henderson, M. and Hollin, C.R. (1986). Social skills training and delinquency. In Handbook of social
skills training, Vol. 1, Applications across the life span, Hollin, C.R. and Thrower, P. (Eds). Oxford:
Pergamon Press.
Hollin, C.R. (1990). Social skills training with delinquents: a look at the evidence and some
recommendations for practice. British Journal of Social Work, 20, 483–493.
Hollin, C.R. and Swaffer, T. (1993). Social functioning and delinquency: a return to basics. Journal of
Adolescence, 16, 205–210.
Hunter, N. and Kelley, C.K. (1986). Examination of the validity of the Adolescent Problems
Inventory among incarcerated juvenile delinquents. Journal of Consulting and Clinical Psychology,
54, 301–302.
Leadbeater, B.J., Hellner, I., Allen, J.P. and Aber, J.L. (1989). Assessment of interpersonal
354 E. J. Palmer and C. R. Hollin