Article

E D I T O R I A L

Introduction to the Special Issue: Discrepancies in Adolescent–
Parent Perceptions of the Family and Adolescent Adjustment

Andres De Los Reyes1 • Christine McCauley Ohannessian2,3

Received: 18 June 2016 / Accepted: 22 June 2016 / Published online: 6 July 2016

� Springer Science+ Media New York 2016

Abstract Researchers commonly rely on adolescents’ and

parents’ reports to assess family functioning (e.g., conflict,

parental monitoring, parenting practices, relationship

quality). Recent work indicates that these reports may vary

as to whether they converge or diverge in estimates of

family functioning. Further, patterns of converging or

diverging reports may yield important information about

adolescent adjustment and family functioning. This work is

part of a larger literature seeking to understand and inter-

pret multi-informant assessments of psychological phe-

nomena, namely mental health. In fact, recent innovations

in conceptualizing, measuring, and analyzing multi-infor-

mant mental health assessments might meaningfully

inform efforts to understand multi-informant assessments

of family functioning. Therefore, in this Special Issue we

address three aims. First, we provide a guiding framework

for using and interpreting multi-informant assessments of

family functioning, informed by recent theoretical work

focused on using and interpreting multi-informant mental

health assessments. Second, we report research on ado-

lescents’ and parents’ reports of family functioning that

leverages the latest methods for measuring and analyzing

patterns of convergence and divergence between infor-

mants’ reports. Third, we report research on measurement

invariance and its role in interpreting adolescents’ and

parents’ reports of family functioning. Research and theory

reported in this Special Issue have important implications

for improving our understanding of the links between

multi-informant assessments of family functioning and

adolescent adjustment.

Keywords Family � Informant discrepancies � Multiple
informants � Operations Triad Model � Parenting

Introduction

Adolescents lead complex lives. Relative to earlier devel-

opmental periods, adolescence can be characterized by an

expansion in exposure to social contexts that may pose risk

for or buffer against the development and maintenance of

psychosocial maladjustment (e.g., Paus et al. 2008; Smetana

et al. 2006; Steinberg 2005). One social context in which this

is most readily apparent is the family. For example, as ado-

lescents progress from the early through mid-to-late ado-

lescent periods, frequencies of conflict interactions with

parents remain stable, and yet normatively the intensity of

this conflict increases over development (Laursen et al.

1998). Of note, very high, chronic levels of such conflict

place adolescents at increased risk for a host of poor psy-

chosocial outcomes (e.g., substance use, delinquency, and

risk-taking behavior; Ary et al. 1999; Duncan et al. 1998).

During adolescence, families may display profound

variations in domains of family functioning other than ado-

lescent–parent conflict. For instance, a family may display

relatively high levels of adolescent–parent conflict, yet the

adolescent may frequently disclose their whereabouts to

parents, a characteristic that tends to buffer adolescents

against the development of poor outcomes (e.g., Smetana

2008). As another example, consider a family that displays

& Andres De Los Reyes
[email protected]

1
Department of , University of Maryland at

College Park, College Park, MD, USA

2
University of Connecticut School of Medicine, Farmington,

CT, USA

3
Connecticut Children’s Medical Center, Farmington, CT,

USA

123

J Youth Adolescence (2016) 45:1957–1972

DOI 10.1007/s10964-016-0533-z

http://crossmark.crossref.org/dialog/?doi=10.1007/s10964-016-0533-z&domain=pdf

http://crossmark.crossref.org/dialog/?doi=10.1007/s10964-016-0533-z&domain=pdf

relatively low levels of adolescent–parent conflict, and at the

same time the parent displays both inconsistent parenting

practices (e.g., variable rule-setting) and a low degree of

knowledge of the adolescents’ whereabouts and activities,

both of which tend to pose increased risk for adolescents

developing poor psychosocial outcomes (e.g., Darling and

Steinberg 1993; Racz and McMahon 2011). Stated another

way, a family may harbor an environment typified by a

collection of characteristics that pose risk for and buffer

against the development of poor psychosocial outcomes in

an adolescent within that family.

Beyond family-level variations in displays of risk and

protective factors for adolescents’ psychosocial function-

ing, displays of family functioning vary in their observ-

ability. For example, intense adolescent–parent conflict

may occur frequently and in public view. As such, the

adolescents and parents involved as well as outside

observers (e.g., adolescents’ peers and other family mem-

bers) may have frequent opportunities to observe displays

of such conflict, even within short time windows. In con-

trast, inconsistent parenting practices within a family (e.g.,

rule-setting sometimes and not other times; presence/ab-

sence of weekend curfew) may only be observable by

people who have both an intimate perspective on the

family’s functioning, and a long time window within which

to observe displays of inconsistent parenting practices.

Collectively, concerns about both family-level variations

in displays of family functioning and the observability of

these displays necessitate the use of comprehensive

approaches to assessment. The most commonly imple-

mented approach involves taking multiple informants’

reports of family functioning domains (see also Hunsley and

Mash 2007). Using this approach, researchers gather reports

from those involved in family interactions (e.g., parents and

adolescents). Multiple informants’ reports may also be

augmented by data from other sources, such as independent

observers’ ratings of family interactions (e.g., level of

warmth or hostility displayed within a laboratory-based

family discussion task; De Los Reyes et al. 2015b), or direct

assessments of physiological processes as they manifest

within relevant contexts (e.g., elevations in arousal or

decreased physiological flexibility displayed during com-

puter-based tasks, unstructured home observations, periods

of social stress, or a resting period; Aldao and De Los Reyes

2015; De Los Reyes et al. 2015a; De Los Reyes and Aldao

2015; Cohen et al. 2015; Franklin et al. 2015; Leitzke et al.

2015; McLaughlin et al. 2015; Youngstrom and De Los

Reyes 2015). Further, a key focus of this approach involves

collecting assessments of psychosocial outcomes commonly

linked to family functioning, such as adolescent psychoso-

cial functioning, which may also leverage multi-informant,

multi-method measurement approaches (e.g., reports of

adolescents’ mental health from adolescents, parents,

teachers, clinicians, and independent observers).

Ubiquity of Adolescent–Parent Reporting

Discrepancies

The value of multi-informant approaches to assessment lies

in the unique views that information sources have about the

constructs for which they provide reports (Achenbach et al.

1987). In particular, adolescents and parents may vary in

the domains of family functioning about which they have

robust capacities to observe (e.g., parents and perceived

levels of knowledge about adolescents’ whereabouts and

activities vs. adolescents and perceived levels of disclosure

about their whereabouts and activities; Kraemer et al.

2003). Consequently, adolescents and parents may provide

reports that provide incrementally valuable information

about family functioning, relative to each other (i.e., each

report contributes non-overlapping information that is not

contributed by the other report). Yet, researchers often

encounter challenges with using and interpreting adoles-

cents’ and parents’ reports of family functioning, because

their reports commonly result in discrepant estimates of

family functioning (for a review, see De Los Reyes 2013).

Much of our knowledge about these informant dis-

crepancies comes from research that examines magnitudes

and moderators of correspondence (i.e., relations between

two or more reports of the same person) among informants’

reports of psychosocial functioning. To assess children and

adolescents, these informants may include parents, teach-

ers, peers, and the children/adolescents themselves (Hun-

sley and Mash 2007; Rescorla et al. 2014). For adults, self-

reports and clinician ratings may be the most often used

sources, although over the last decade, researchers have

increasingly leveraged reports from collateral informants,

such as significant others of the adults being assessed (e.g.,

spouses, caregivers in the case of elderly adults; Achen-

bach 2006). Over 50 years of work across hundreds of

investigations of informants’ reports of children, adoles-

cents, and adults indicates that mean cross-informant cor-

respondence hovers in the low-to-moderate range (e.g.,

Pearson r’s in the .20 s–.40 s; Achenbach et al. 1987, 2005;

De Los Reyes et al. 2015b). However, correspondence does

not remain uniform across informants. That is, informants’

reports tend to exhibit relatively higher correspondence

levels when they (a) come from informants who observe

behavior in the same context (e.g., pairs of teachers; pairs

of parents), (b) estimate levels of behaviors that are rela-

tively easier to observe (e.g., externalizing behaviors such

as aggression/hyperactivity vs. internalizing behaviors such

as anxiety/mood), and (c) come from continuous versus

discrete scales (De Los Reyes et al. 2013e, 2015b).

1958 J Youth Adolescence (2016) 45:1957–1972

123

The low-to-moderate correspondence levels among

informants’ reports seen in mental health assessments

generalize to correspondence between adolescents’ and

parents’ reports of family functioning domains. Four

important observations about these effects warrant com-

ment. First, as mentioned previously, one observes the

largest magnitudes of cross-informant correspondence in

reports of mental health from informants who observe

behavior within the same context (e.g., reports about a

child’s behavior from a pair of teachers at the child’s

school). Interestingly, adolescents and parents often pro-

vide reports of family functioning domains that, by defi-

nition, occur within the same context of observation (i.e.,

the family unit). Based on this, one might presume that

correspondence between adolescents’ and parents’ reports

of family functioning should resemble the relatively high

levels of correspondence between the mental health reports

of informants from the same context (i.e., Pearson r’s in

.50 s; see Achenbach et al. 1987; De Los Reyes et al.

2015b). Yet, this is often not the case: Correspondence

levels between adolescents’ and parents’ reports of family

functioning routinely hover in the low-to-moderate range.

Indeed, this pattern manifests in assessments of a host of

domains including adolescent–parent conflict (e.g., Gon-

zales et al. 1996), inter-parental conflict (e.g., Epstein et al.

2004), parenting behaviors (e.g., Guion et al. 2009; Otter-

pohl and Wild 2015), parental monitoring (e.g., parental

knowledge, adolescent disclosure, parental solicitation and

control; Kerr and Stattin 2000), and adolescent–parent

relationship quality (e.g., Pelton and Forehand 2001).

Second, underlying the low-to-moderate levels of cor-

respondence between adolescents’ and parents’ reports of

family functioning, there exist substantial dyad-level vari-

ations in patterns of adolescents’ and parents’ reports about

the family. That is, not all adolescents and parents diverge

in their reports about the family. In fact, within samples of

adolescent–parent dyads, some provide reports that con-

verge quite highly with each other whereas other dyads do

not (e.g., De Los Reyes et al. 2010; Lippold et al. 2013).

Further, among those adolescent–parent dyads who evi-

dence divergence between their reports, sometimes it is

because the parent views family functioning more favorably

than the adolescent, and sometimes the reverse is the case

and the adolescent views the family more favorably than the

parent (e.g., Lippold et al. 2011, 2014; Yaban et al. 2014).

Third, on the surface, discrepancies between adoles-

cents’ and parents’ reports about family functioning may

have the ‘‘look and feel’’ of other family functioning

domains, namely conflict. Indeed, prior work indicates that

disagreements arising from daily life occurrences (e.g.,

doing chores and homework) give rise to conflict between

adolescents and parents (e.g., Smetana 1989). One question

may be, to what extent are discrepant views between

adolescents and parents about the family distinguishable

from behavioral conflict? Importantly, adolescents’ and

parents’ discrepant views of the family can be empirically

distinguished from their levels of behavioral conflict. For

instance, research indicates that Pearson correlations

between indices of adolescent–parent discrepant views and

adolescent–parent conflict hover in the .10 s–.60 s range

depending on the measurement method and informant (De

Los Reyes et al. 2012). Further, whereas indices of ado-

lescent–parent discrepant views uniquely predict scores on

performance-based measures of interpersonal perception

(i.e., emotion recognition), indices of adolescent–parent

conflict do not (De Los Reyes et al. 2013c). Taken together,

these findings indicate that adolescent–parent discrepancies

and conflict, though related, provide distinct information

about family functioning and interpersonal perception.

Fourth, the work reviewed previously indicates that

(a) low-to-moderate adolescent–parent correspondence is

the norm; (b) dyads vary considerably as to patterns of

convergence and divergence between reports; (c) when

dyads’ reports diverge, the direction of this divergence may

vary between dyads (adolescent [ parent vs. par-
ent [ adolescent); and (d) discrepancies between adoles-
cents’ and parents’ reports contain information about family

functioning that relates to, but is distinct from, other domains

of family functioning. In light of this work, it is important to

consider the importance of understanding and interpreting

points of convergence and divergence between adolescents’

and parents’ reports of family functioning. To begin, con-

sider that when adolescents and parents provide researchers

with reports about family functioning domains (e.g., conflict,

parenting, relationship quality), they are providing their

impressions of features of their lives that may matter a great

deal to them (see also De Los Reyes 2011; De Los Reyes and

Kazdin 2006a). Thus, patterns of convergence and diver-

gence between adolescents’ and parents’ reports of such

functioning may reflect important aspects of their interac-

tions and how they relate to one another (De Los Reyes et al.

2013c; Goodman et al. 2010). In line with this view, recent

work indicates that both the convergence between adoles-

cents’ and parents’ reports, as well as the divergence between

these reports, longitudinally predicts psychosocial outcomes

among adolescents (e.g., De Los Reyes 2011; Laird and De

Los Reyes 2013; Lippold et al. 2013; Ohannessian and De

Los Reyes 2014). Consequently, understanding patterns of

adolescents’ and parents’ reports of family functioning may

result in tools for predicting adolescent adjustment.

Importance of Sound Approaches to Modeling

Informant Discrepancies

Overall, prior work in adolescent development greatly

informs our understanding of discrepancies between

J Youth Adolescence (2016) 45:1957–1972 1959

123

adolescents’ and parents’ reports of family functioning.

Further, work on these informant discrepancies and how

they operate in other assessment literatures (e.g., multi-

informant assessments of mental health) might augment

research in adolescent development, and provide

researchers with avenues for hypothesis generation and

theoretical development. Yet, three key issues need to be

addressed to further improve our understanding of multi-

informant assessments of family functioning and their links

to adolescent adjustment.

The first involves improving our approaches to mea-

suring informant discrepancies. As in other literatures (e.g.,

mental health; organizational behavior; neuroscience; De

Los Reyes and Kazdin 2004; Edwards 1994; Meyer et al.

2016), researchers examining discrepancies between ado-

lescents’ and parents’ reports of family functioning often

seek to measure the distance or ‘‘gaps’’ between these

reports, and test whether variations in these gaps relate to

variations in scores from criterion variables (e.g., adoles-

cent psychosocial functioning). Historically, these mea-

surements often consisted of taking the difference between

one informant’s report of a family functioning domain

(e.g., adolescent report of family conflict) from another

informant’s report on that same domain (e.g., parent report

of family conflict). Researchers subsequently treated this

difference score as an individual differences variable but

more importantly, as a new construct that exists over-and-

above the construct(s) reflected in the individual reports of

the informants (i.e., discrepancies between adolescents’

and parents’ views of family conflict vs. adolescents’ and

parents’ unique views of family conflict).

We have learned a great deal about these difference scores

and what they are capable of providing in the way of mea-

suring informant discrepancies. In short, they provide very

little information. Specifically, work from organizational

behavior research finds that difference scores are often inca-

pable of contributing incremental or unique information about

psychological constructs (e.g., discrepancy between

employee’s attributes and fit with an organization), over-and-

above the scores used to create them (Edwards 2002). Stated

another way, difference scores are statistically redundant with

the scores contained in the difference scores. Further, these

inherent limitations to difference scores generalize to assess-

ments of informant discrepancies in assessments conducted in

clinical research and developmental psychopathology gener-

ally (Laird and Weems 2011). In fact, recent work provides a

set of analytic tools that one can use to test whether a specific

use of difference scores can meaningfully inform prediction of

scores from criterion variables, over-and-above its component

scores (Laird and De Los Reyes 2013). In many cases,

researchers may encounter disappointment with what a dif-

ference score can offer in the way of incremental prediction of

scores from criterion variables.

The issues raised by difference scores have led

researchers to develop new techniques for modeling dif-

ferences and/or similarities between reports. For instance,

researchers may study interactions between informants’

reports within a polynomial regression framework to

examine discrepant perceptions in dyads. This approach

allows for the direct examination of whether differences

between reports contribute to predicting scores on criterion

variables, beyond the main effects of individual reports

(Laird and De Los Reyes 2013). Moreover, polynomial

regression methods can be modified for use in examinations

of discrepant views as either predictors, outcomes, or both

(De Los Reyes et al. 2016b; Laird and LaFleur 2016). Thus,

the polynomial regression approach can generalize to

examining changes in informant discrepancies over time.

Further, the polynomial regression approach has been

extended to understanding and interpreting informant dis-

crepancies in other areas, including neuroscience, person-

ality, and treatment (Fjermestad et al. 2016; Meyer et al.

2016; Tackett et al. 2013). Other approaches possess similar

capabilities and have been successfully implemented in the

study of informant discrepancies. These include meta-

analysis of cross-informant correspondence (e.g., Achen-

bach et al. 1987; De Los Reyes et al. 2015b), direct

assessment of discrepant views (i.e., via structured inter-

view: De Los Reyes et al. 2012, 2013d), nested repeated

measures analytic models (e.g., generalized estimating

equations; Alfano et al. 2015; Augenstein et al. 2016; De

Los Reyes et al. 2013b); and person-centered models (e.g.,

latent class analysis; De Los Reyes et al.

2009, 2011, 2016a, 2013a; Lippold et al. 2011, 2013, 2014).

In line with this recent work, a key aim of this Special Issue

is to report recent work on discrepancies between adoles-

cents’ and parents’ reports of family functioning, using

these emerging measurement and analytic models.

An Increased Focus on Measurement Invariance

A second issue related to the first is that of the inter-

pretability of discrepancies between adolescents’ and par-

ents’ reports of family functioning. Specifically, when

interpreting differences between informants’ reports,

informants ought to provide such reports on identical or

parallel measures. Indeed, to do otherwise would present a

confound: Informants might provide discrepant reports

because the item content or response options differed

between the measures they completed (see Schwarz 1999).

Thus, methodological differences in measurement might

parsimoniously account for the discrepancies between two

informants’ reports, rather than any meaningful difference

in how the two informants perceived the psychological

phenomena about which they were tasked to provide

reports. Thus, one prerequisite of research on informant

1960 J Youth Adolescence (2016) 45:1957–1972

123

discrepancies involves use of parallel instruments (De Los

Reyes et al. 2013e).

Yet, even when informants do provide reports on par-

allel measures, their reports might be based on measures

for which scores taken from them differ in their psycho-

metric properties. If informants’ reports do not come from

measures for which their scores carry the same properties,

then differences between reports might be parsimoniously

explained by measurement error. Consequently, in recent

years research on multi-informant assessment has focused

on the measurement invariance of parallel forms admin-

istered to multiple informants (e.g., Dirks et al. 2014), or

whether scores from reports of multiple informants can be

meaningfully interpreted as carrying the same or similar

psychometric properties. However, only recently have

these methods begun to be applied to understanding the

measurement invariance of adolescents’ and parents’

reports of family functioning (e.g., Gross et al. 2016;

Janssens et al. 2015). Therefore, a second key aim of our

Special Issue involves reporting research on the measure-

ment invariance of adolescents’ and parents’ reports of

family functioning.

Need for Theoretical Modeling to Unify Research

Efforts

A key condition underlying current problems with mea-

surement and analytic modeling of multi-informant data is

the lack of a unifying framework to guide research on

multi-informant assessments of family functioning. That is,

a few theoretical models exist that seek to explain or

improve interpretability of informant discrepancies (e.g.,

De Los Reyes and Kazdin 2005, 2006b; Goodman et al.

2010; Kraemer et al. 2003). However, these models focus

on domains other than family functioning, such as mental

health, treatment outcome, and youth victimization.

Recent work on theoretical modeling of multi-informant

mental health assessments seeks to guide research on

interpreting the outcomes of these assessments (De Los

Reyes et al. 2013e). With some modification, this frame-

work may improve the study and interpretability of multi-

informant assessments of family functioning. Specifically,

researchers designed the Operations Triad Model to

understand and interpret multi-informant assessments of

mental health. In Fig. 1 we present a graphical depiction of

the Operations Triad Model. In these assessments, multiple

informants provide reports about a target person’s mental

health (e.g., parent and teacher report about a child’s

behavior problems). As with assessments of family func-

tioning, these reports may provide unique information

about mental health that converge on estimates of such

mental health (i.e., Converging Operations; Fig. 1a). This

convergence may result in the informants’ reports pointing

to a common conclusion. Additionally, this convergence

might reflect meaningful consistencies in assessed behav-

iors across contexts. For example, if a parent and teacher

both report that a child displays relatively high behavior

problems, then this convergence in reports may signal that

the child displays these problems consistently across home

and school contexts.

Multiple informants’ reports may also diverge in their

estimates of mental health. To continue with the problem

behavior example, a parent and teacher may differ in their

reports such that the teacher’s report indicates relatively

high levels of problem behavior, whereas the parent’s

report indicates relatively low levels of such behavior.

Such divergence in reports, for instance, may reflect that

the child displays problem behavior to a far greater degree

at school than at home. If so, then the reasons for the

divergence may reflect meaningful variations in the child’s

problem behavior across relevant contexts (i.e., Diverging

Operations; Fig. 1b). Conversely, the reports may not

reflect any meaningful divergence, and instead could reflect

methodological differences between the informants’

reports (e.g., item content, response options, psychometric

properties). These methodological factors could parsimo-

niously explain the divergence between reports (i.e.,

Compensating Operations; Fig. 1c), and as a result could

provide justification for procedures to integrate multi-in-

formant data that assume the lack of convergence among

reports reflects measurement error (e.g., structural equa-

tions modeling; AND/OR rules; selection of primary out-

come measures; De Los Reyes et al. 2015b).

We have developed versions of the Operations Triad

Model to understand multi-informant assessments of

mental health in relation to contextual variations in

observed behavior (De Los Reyes et al. 2013e), and more

recently in relation to variations in physiological processes

(De Los Reyes and Aldao 2015). In line with prior work, a

third aim of this Special Issue is to advance a version of the

Operations Triad Model that is modified for use in inter-

preting multi-informant assessments of family functioning.

Such a framework might not only guide hypothesis testing

with multi-informant assessments in this area, but also

inform the selection of measurement and analytic models.

The Present Special Issue

Overall, innovative theoretical, measurement, and analytic

developments in multi-informant assessments of psycho-

logical constructs may inform advancements in using and

interpreting adolescents’ and parents’ reports of assess-

ments of family functioning. Researchers in this area would

benefit from a collection of articles that leverage these

advancements. To this end, in this Special Issue we address

J Youth Adolescence (2016) 45:1957–1972 1961

123

four aims. First, we provide a guiding conceptual frame-

work for using and interpreting multi-informant assess-

ments of family functioning. Second, we report state-of-

the-art scholarship on adolescents’ and parents’ reports of

family functioning using the latest methods for measuring

and analyzing patterns of convergence and divergence

between these reports. Third, we report research on the

measurement invariance of adolescents’ and parents’

reports of family functioning. Collectively, this research

includes diverse areas of study. Fourth, commentaries by

Lerner and Rescorla outline directions for future research

on using and interpreting multi-informant assessments of

family functioning and their links to adolescent adjustment.

Applying the Operations Triad Model

to Adolescent–Parent Reports of Family

Functioning

As described previously, we originally designed the

Operations Triad Model to ‘‘make sense’’ of or understand

patterns of convergence and divergence between infor-

mants’ reports of mental health. In particular, we focused

Informant 1 Informant 2

Informant 1 Informant 2

Research
Conclusion

Research
Conclusion

1

Research
Conclusion

2

Informant 1 Informant 2

Research
Conclusion

1

Research …