THE INFLUENCE OF FEELING LOVED ON THE
ASSOCIATION BETWEEN AVERSIVE INTERPERSONAL BEHAVIORS
AND RELATIONSHIP SATISFACTION
MARY B. SMITH
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
Tntr'nTnD f DPT-ITT flQPDT-TV
Mary B. Smith
I would like to express my gratitude to Dr. Martin Heesacker for his time,
patience, encouragement, belief in me, advisement, and support. I would also like to
thank my committee members, Dr. Lisa Brown, Dr. Franz Epting, and Dr. Connie
Shehan, for agreeing to serve on my committee. I thank them for their feedback and
I thank my co-workers at the University of Florida Counseling Center for their
encouragement and support. My thanks go especially to Paul Schauble, Nancy Coleman,
Toti Perez, Mike Murphy, Jaquie Resnick, Jennifer Sager, Brett Vicario, Kevin Stanley,
Dave Suchman, Carlos Hemandez,
Wayne Griffin, and Rachel Navarro. I also thank my
friends at the Alachua County Crisis Center for their friendship and encouragement.
I thank the instructors who provided access to their students and I thank the
students who participated in the studies. I thank my research assistants, Megan, James,
Kevin, and Brian, who were willing to be helpful and supportive. I thank Sarah Lee,
Susan Bragg, and Cheryl Phillips for their warmth and for all that they do.
I thank my Mom and Dad, my sisters and brothers, and the other relatives who
have been encouraging. I especially thank my cousin, Nancy Beardall, and her family for
their love and support. A thank you goes to my friends, Jane Calabrese and Maureen
Deluca for their friendship, love, and support.
TABLE OF CONTENTS
ACKNOWLEDGMENTS ................................................................ ...................................ii
LIST O F TA B LE S ........................................................................................................... viii
A B STR A C T ................... ............ ....... .................... .... .... .......... ..... ... .... ..... ............ix
1 INTRODUCTION.............. .......... ............................... ............................................
2 REVIEW OF THE LITERATURE ....................................................8
A Basic Human Need ....
To .fea nd~nmanticRPelai
The Exchange of BehaviL
Impact of Conflict and A
Gottman's Taxonomy of
Gottman's Four Horseme
Another Look at Behavic
Further Support for the D
Every Day Aversive Inte:
Support for Intervention
The Current Study..........
...9....... ...... ...... ........." .... ... ...... .. .. .. . .. .. .. .. ... ....... .9
tionships ................... .......... ........... ........... ........... . ..... ...10
Changes over Time............................................................. 1
or in Romantic Relationships........................................... 15
nger on Relationships ...................... ....................................16
Dysfunctional Romantic Relationship Patterns ..................18
n of the Apocalypse ........... ........ ........ ................................... 20
,r Exchange: What Predicts Relationship Demise?.............21
)etrimental Impact of Bad Behaviors on Relationships ......24
personal Behaviors ................... ...................................... 26
...................... ..... C. .......... ................ .. .. ... ... ......... .....2
.. ............. .......... ....................... .......... ...... .............. .... ...33
Phase 1 ................................................................... ..................35
Procedure...................................... ..................... ............................................38
METHODOLOGY .........,,,,.. ................... ................... ................... ................... ......35
Inventory of Desirable Responding in Relationships
Other Measures .........................................................
Procedure ............................................................................. C eta et...
... .. .. ............ ,. *, ..., ,,.....-*t,, .9" ** ******** "**** t*"e"' "
LJSV ,.,. 0.ea.. ., a..,a. ... ... ,,* 0 .... ate,,......*., "e"c 1'* "''....." 0-"- '"'' 0'"---
The Love Crushing Behavior Scale (LCBS)..................
The Positive and Negative Affect Scale (PANAS).........
The Relationship Assessment Scale (RAS) ....................
The Global Measure of Relationship Satisfaction (GMt
The Relationship Happiness Scale (RHS)......................
* *t C C ScSI sect
,cc. cc. acet e
ItC .t C C C 5CC
. a a.. a tc* aCes
c..tt Ct C S C t
cam a etc a Sec St
IBIS) .. ...
The Perceived Relationship Quality Component Scale (PRQC)
Predications about the Future Scale ..................... .............................
The Partner-Rated Self-Disclosure Index (PRSD)......................
The Self-Rated Self-Disclosure Index (SRSD)...........................
The Mutual Psychological Development Questionnaire (MPDC
O their M measures ...........................................................................
:edure...................... .. .. ................................................... .................
2A nalysis.................. t........ ........... ...............................................
RESULTS ............... ...................... ................ ............................................................
Phase 1: Item Generation....................
Phase 2: Scale Refinement..................
Exploratoratory Factor Analysis and:
Additional Phase 2 Analyses.......
Phase 3: Assessment of Psychometric
Normative and Reliability Data ..
Construct Validity .....................
Hypotheses Testing .....................
Confirmatory Factor Analysis.....
The Love Crushing Behavior Scale ...
,,...S... a....,a ....
,,, ..,.... acte...,.
D ISCU SSIO N ........ e... .......................... .~es............... ........ .... .... ............................... 89
Key Finding/Integration with Existing Research....
Limitations and Directions/Implications for Future
Potential Clinical Applications................................
Conclusion ................. .................................
AND 3 COUPLES' COUNSELING RESOURCES ...........................110
PRELIMINARY UNDESIRABLE BEHAVIOR SCALE (PUBS) ......................... 111
DESIRED LOVING BEHAVIOR SCALE (DLBS) .................. ................... ...........122
INVENTORY OF DESIRABLE RESPONDING IN RELATIONSHIP (IDRR) ...123
DEMOGRAPHIC AND RELATIONSHIP STATUS
LOVE CRUSHING BEHAVIOR SCALE (LCBS) ........... ..................................... 127
INFORM ED CON SENT .........................................................................128
DEBRIEFING STATEM ENT .................................................................129
QUESTIONNAIRE ................................................................................. 130
POSITIVE AND NEGATIVE AFFECT SCALE (PANAS) ................................... 139
RELATIONSHIP ASSESSMENT SCALE (RAS) .................................................. 140
GLOBAL MEASURE OF RELATIONSHIP SATISFACTION SCALE
RELATIONSHIP HAPPINESS SCALE (RHS) ........ ........................................ 1 42
PERCEIVED RELATIONSHIP QUALITY COMPONENT SCALE (PRQC)......143
PREDICTIONS ABOUT THE FUTURE SCALE ............ .................... ...................144
PARTNER-RATED SELF-DISCLOSURE INDEX (PRSD) .................................. 145
SELF-RATED SELF-DISCLOSURE (SRSD) ............. ..........................................146
MUTUAL PSYCHOLOGICAL DEVELOPMENT QUESTIONNAIRE:
PHASE 3 DEMOGRAPHIC AND RELATIONSHIP STATUS
PHASE 1 INFORMED CONSENT ................... ................... ................... ......109
2 QUESTIONNAIRE ........................................ 15
QUE STIONNAIRE ITEM S ................... .. ,
LIST OF REFERENCES........................................ ..................s......... .... ...................... 153
BIOGRAPHICAL SKETCH....... ............................ ................................................ 162
LIST OF TABLES
Descriptive Statistics for the 22-Item Love Crushing Behavior Scale (LCBS).......70
Means and Standard Deviations for Phase
Questionnaire Items.............. .............72
Descriptive Statistics of the IDRR Items and Subscales.......................................... 74
Descriptive Statistics of the Phase 3 Scales .............................................................76
Item to Total Correlations for the LCBS in Phase 3 ..............................................77
Inter-item Correlations for Phase 3 Scales ...............................................................86
LCBS Item Rank Comparisons................................................................................87
Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy
THE INFLUENCE OF FEELING LOVED ON THE
ASSOCIATION BETWEEN AVERSIVE INTERPERSONAL BEHAVIORS
AND RELATIONSHIP SATISFACTION
While much research on intimate relationships tends to focus on positive aspects
of relationships, there is evidence suggesting negative behaviors have a detrimental effect
on the quality of close relationships. Therefore, a goal of this study was to identify love-
crushing behaviors, which are defined as behaviors provided by romantic partners that
evoke in the receiving partners the feeling that they are unloved. The dissertation consists
of three phases. In Phase 1, participants generated received behaviors that elicited in them
the feeling that they were unloved by romantic partners. The resulting list of items was
presented to Phase 2 participants, who rated the items regarding how unloved they would
feel if the items were received from romantic partners. Factor analysis of these rated
items resulted in a 22-item uni-factorial Love Crushing Behavior Scale (LCBS).
This dissertation investigates the relationship between feeling loved, love
crushing behaviors, and romantic relationship satisfaction. Over the years, romantic
relationship research has increased in both breadth and depth within the field of
counseling psychology. However, after reviewing 25 years of research focusing on the
investigation of marital conflict, Fincham and Beach (1999) wrote that the connection
between research findings and their clinical relevance for improving relationships were
controversial. As a result, Fincham and Beach challenged researchers to work toward
integrating research and theory in order to develop more unified frameworks for guiding
empirical investigation and clinical activities, rather than continuing to amass
observations in a relatively theoretical manner.
Coincidentally, a group of studies (Cawood & Heesacker, 1999; Heesacker &
Lawrence, 1994; Heesacker, Smith, & Lawrence, 1998; Mejia-Millan, 1999; Samson,
1996; Smith, 2000; Tiegs, Heesacker, Lawrence, Smith, Cawood, & Mejia-Millan, 2004)
addressed Fincham and Beach'
challenge by conducting theory driven research and
proposing efficacious strategies for improving the quality of romantic relationships. This
group of studies investigated the association between the provision of specific desired
loving behaviors, feeling loved, and relationship quality. To date, this body of research
the assessment of the psychometric properties of the DLBS (Cawood & Heesacker, 1999;
Heesacker, Smith, & Lawrence, 1998; Mejia-Millan, 1999; Samson, 1996; Smith, 2000;
Tiegs, Heesacker, Lawrence, Smith, Cawood, & Mejia-Millan, 2004), (c) the
establishment of a clear connection between the receipt of desired loving behaviors and
important relationship outcomes (Heesacker, Smith, & Lawrence, 1998; Smith, 2000;
Tiegs, Heesacker, Lawrence, Smith, Cawood, & Mejia-Millan, 2004); and (d)
recommendations for clinically relevant applications of the DLBS (Cawood &
Heesacker, 1999; Heesacker & Lawrence, 1994; Heesacker, Smith, & Lawrence, 1998;
Mejia-Millan, 1999; Samson, 1996; Smith, 2000; Tiegs, Heesacker, Lawrence, Smith,
Cawood, & Mejia-Millan, 2004).
Initially, DLBS researchers were interested in looking at the association between
emotion and behavior within the context of romantic relationships (Heesacker &
Lawrence, 1994). The goal of their research was to identify specific behaviors that
members of romantic relationship dyads desired from partners in order to feel loved.
Through a multi-phase process that included item generation and exploratory factor
analysis, 39 behavioral items were identified and labeled the Desired Loving Behavior
Scale (DLBS; Heesacker & Lawrence, 1994). Additional analyses of the scale resulted in
each of the 39 items falling into one of four factors. These factors were labeled: Caring
Actions, Relationship Support, Scripting, and
(Heesacker & Lawrence, 1994).
Analyses from subsequent studies (Cawood & Heesacker, 1999; Heesacker, Smith, &
Lawrence, 1998; Samson, 1996; Smith, 2000) investigating the psychometric properties
the DLBS possesses ample reliability and validity and that the DLBS shows promise as a
viable, efficacious clinical and research assessment instrument.
The efficacy of the DLBS has been demonstrated in several ways. First, the
DLBS identifies specific behaviors that are desired from romantic partners in order for
them to feel loved (Heesacker & Lawrence, 1994; Heesacker, Smith, & Lawrence, 1998;
Smith, 2000). Second, the DLBS efficiently assesses how much each item is desired from
romantic partners (Smith, 2000).
Third, the scale can be utilized by participants to rate
subjective impressions of how much or how often the desired loving behaviors are being
received from romantic partners (Smith, 2000).
When the discrepancy between what is
desired (want) and what is received (get) is calculated, therapeutic interventions can be
developed and targeted specifically and directly toward the areas where the discrepancy
between want and get is the greatest. Thus, the possibility for impacting and improving
the quality of a given relationship is increased.
The positive consequences for relationship partners whose want-get discrepancies
have been identified and addressed through clinical interventions can be objectively
evaluated (Smith, 2000). Research has shown that there is a significant inverse
relationship between the want-get discrepancy and relationship satisfaction. That is, as
the discrepancy between what is desired from a romantic partner and what is received
from that partner increased, relationship satisfaction decreased (Smith, 2000). In contrast,
when the want-get discrepancy is decreased, relationship satisfaction increases. In
addition, research to date has demonstrated that intact romantic relationships have
Decreasing want-get discrepancies through the provision and receipt of desired
loving behaviors has been significantly linked to increased relationship satisfaction.
Nonetheless, the magnitude of the increases in satisfaction and the feeling that one is
loved varies based on the desirability of the specific behaviors that are being provided to
and received by romantic partners (Smith, 2000). Just as receiving desired loving
behaviors from romantic partners has a positive impact on how loved recipients feel and
influences how satisfied they are with their relationships, a corresponding inference may
be made that receiving negative behaviors will have a negative impact on the affective
state of relationship partners and on how satisfied they are with their relationships.
The empirical evidence suggests that negative behaviors may be even stronger
predictors of relationship satisfaction than positive behaviors. After exhaustively
reviewing the literature across multiple domains, Baumeister, Bratslavsky, Finkenauer,
and Vohs (2001) found abundant, convincing empirical evidence supporting the greater
pervasiveness and power of the effect of negative or aversive stimuli compared to
positive or rewarding stimuli. In fact, according to these authors the concept, bad is
stronger than good is one of psychology's most basic and far reaching principles.
Baumeister and colleagues (2001) stated that bad events may be stronger than good
because people tend to treat bad experiences as if they are isolated events (Baumeister et
al., 2001). In contrast, good experiences tend to be seen collectively and they become
integrated into an ongoing general perception of goodness (Baumeister et al., 2001). As
most people have experienced, the impact of an entire day's worth of kindness and
inconsiderate act is attenuated only by the accumulation of an extensive array of positive
acts within which the negative act is embedded.
Others have demonstrated that focusing on negative behaviors is as important as
investigating positive behavior and potentially more so because of the impact that
negative behaviors can have on relationships. For example, (a) while both positive and
negative behaviors influence relationship satisfaction, the negative impact of negative
behaviors is much greater than the positive impact of positive behaviors (Baumeister et
al., 2001; Gottman, 1994a, 1999; Gottman & Krokoff, 1989; Rozin & Royzman, 2001);
(b) a difference between couples who are content and couples who experience marital
misery is a healthy balance between positive and negative feelings (Gottman, 1994a,
1999); (c) in order for marriages to remain stable over time, the minimum ratio of
positivity to negativity is five to one and when a relationship contains more negative
interactions than positive interactions, the relationship will very likely terminate
(Gottman, 1994a; 1994b; 1999).
There are also affective consequences for partners who are recipients of negative
behaviors (Kowalski, 1997). For example, recipients of negative or aversive behaviors
may feel aggravation, annoyance, anxiety, boredom, contempt, depression, disgust,
frustration, hostility, hurt, jealousy, and loneliness (Baumeister & Leary, 1995; Kowalski,
Kowalski, 2001; Leary, 1990). These consequences occur in all types of
interpersonal relationships but tend to be the greatest in relationships between close,
intimate partners as compared to individuals who are distant or less invested (Kowalski,
Achieving a healthy balance between good and bad behaviors appears to be
necessary given that close relationships are essential, vital parts of most people's lives
and that relationships are often the source of people's most meaningful rewards
(Kowalski, 2001). Despite the fact that research conducted to date has unequivocally
demonstrated that negative behaviors are more influential on relationships than positive
behaviors (Baumeister et al., 2001; Baumeister & Leary, 1995; Gottman & Krokoff,
1989; Gottman, 1994a; Gottman 1994b; Holmes, 1991), much of the research on close
relationships has been directed at positive relationship qualities (Duck, 1994; Kowalski,
1997). In addition, little of this relationship research has been theory driven. Moreover,
clinical interventions have not consistently been guided by research findings (Fincham &
In response to these circumstances, the purpose of the present study was to build
upon an emerging line of research whose goal has been the improvement of romantic
relationship quality. Findings to date have demonstrated that the impact of the exchange
of negative behaviors is detrimental to relationship quality and that the impact negative
behaviors have on relationships is greater than the impact of positive behaviors.
Therefore, this dissertation's primary goal has been to construct a valid and reliable scale
of love crushing behavioral items. The secondary goal has been to develop a theory
driven, empirically based instrument that will have utility for clinicians and their clients
in order to obtain a clearer understanding of and appreciation for the types of behaviors
that are most detrimental in their relationships.
The result of this investigation was the Love Crushing Behavior Scale. This scale was
developed over a series of three phases. In the first phase, items were generated by a
group of participants. In the second phase, the non-redundant items were rated by a
second group of participants. Results were analyzed to determine scale items and factor
structure. In the third phase, the psychometric properties of the resulting scale were
REVIEW OF THE LITERATURE
The literature review that follows includes research supporting the current study.
A search of the literature was conducted as such: Initially, keywords and phrases (e.g.,
"behavior," "negative," aversivee," "relationship,
" "close relationship,
" "romantic relationship,
" "romantic dyads," dyadss,
"love," "feeling loved,
" "unloved," "feeling unloved,
and "relationship satisfaction") were used individually and in
two- and three-word combinations. Numerous keyword searches were conducted using
these search terms with the PsycINFO and PsycARTICLES data bases. The study's
author conducted subjective analysis of the resulting articles and chapters. Decisions
were made by the study's
author regarding the relevance of resulting articles and chapters
by reviewing abstracts and/or various portions of the articles or chapters. Studies and
chapters that contained information on aversive behaviors in romantic relationships and
relationship outcomes were retained whereas studies and chapters that did not were
discarded. Article and book chapter reference lists were reviewed and any articles
included on those lists that appeared to be relevant were also reviewed. The topics that
are covered in this review include a brief overview of the exchange of interpersonal
(1994) "Four Horsemen of the Apocalypse," information on the impact of good and bad
behaviors on romantic relationships, an overview of aversive interpersonal behaviors, and
A Basic Human Need
The need to belong is a very basic human need. Most would agree that "feeling
loved" by another likely contributes toward filling this basic need. Many would also
agree that feeling loved by a partner is a positive experience especially when this is
compared to the reciprocal alternative, which is to "feel unloved." However, relationships
are not always positive or satisfying experiences for all; participants in romantic
relationships do not always feel loved by their partners. At times partners may even feel
as if they want out of their relationships or that they do not belong in the relationship.
Sometimes there are even adverse consequences for people who are in relationships
(Baumeister & Leary, 1995).
One of the reasons why relationships are sometimes negative and unsatisfying is
because relationship partners are sometimes the providers or the recipients of negative,
aversive behaviors to or from their partners.
When aversive behaviors are provided by
one partner and experienced by the other, not only do these behaviors interfere with the
basic human need to belong, but these behaviors in fact produce several negative
affective consequences for the recipients of the behaviors (Baumeister & Leary, 1995;
2001). Resulting emotional consequences include anger, anxiety,
betrayal, boredom, depression, disgust, hostility, jealousy, loneliness, and sadness
Love and Romantic Relationships
For centuries, poets and philosophers have written about love, attempting to
describe it. More recently, scientists have attempted to define and conceptualize love. For
most people in Western cultures love is believed to be an essential element for romantic
relationships and "being in love"
is what makes a couple a couple (Simpson, Campbell,
& Berscheid, 1986).
To better understand the construct love, Sternberg (1986) used a comprehensive,
systematic approach in his theorizing, partitioning love into three components. Sternberg
identified several different types of love based on the absolute strength of these three
components and their relationships to one another. The three components making up
Sternberg's theory of love are intimacy, passion, and commitment.
The intimacy component is synonymous with warmth and is characterized by an
emotional investment in the relationship. Intimacy is defined as connectedness,
bondedness, and a feeling of closeness to another (Sternberg, 1986). Passion is often
referred to as the hot component of love. Passion is derived in part from the motivation to
be involved with another. Passion describes the drives that direct one to physical
attraction, romance, and sexual consummation. Passion includes forms of arousal and
sources of motivation that lead to passionate experiences. Commitment includes the
cognitive elements that are involved in decision making about committing to a
relationship. Sternberg identified two different types of commitment. These types are
dependent on whether a relationship is a new, short-term relationship or whether the
deciding whether or not to work toward maintaining that love. Because commitment
involves cognition, it is often viewed as the cold component relative to passion and
intimacy, the other two components (Sternberg, 1986).
Using these three components (intimacy, passion, and commitment) in various
combinations, Sternberg identified eight different types of love. For Sternberg,
is made up of high levels of all three of the components-intimacy, passion and
commitment. However, couples' experiences of perfect love tend to be ephemeral and/or
elusive and according to Sternberg, most people who describe themselves as being in love
either experience romantic love or companionate love. Romantic love generally occurs in
the beginning of relationships and companionate love tends to be experienced when
relationships are more enduring.
Romantic Relationships: Changes over Time
According to Sternberg (1986), romantic love is composed of intimacy and
passion and produces feelings of physical arousal in the presence of the beloved. The
early stages of romantic relationships tend to be made up of happy and blissful times as
partners are optimistic about their relationships. Couples experiencing romantic love are
characterized by a lack of objectivity and often have an idealized view of each other
(Sternberg, 1986). Partners are motivated to attend to and to notice each other'
qualities more than they notice each other's
negative qualities (Huston, Caughlin, Houts,
Smith, & George, 2001; Huston, McHale & Crouter, 1986; Miller, 1997
Waller, 1938). During this period of time partners' expectations of each
however, after dating someone for some time, receiving only one phone call within the
span of two or three days might be quite irritating (Love, 2001).
Most people expect that the experiences they have with their romantic partners
will remain pleasurable. They do not anticipate that conflict or pain will be a part of their
experience (Attridge & Berscheid, 1994). Despite people's high hopes and expectations
at the beginning of relationships, as they progress over time, there may be decreases in
feelings of love for or feelings of being loved by one's partner, which can be
disheartening (Sprecher & Regan, 1998).
When this occurs romantic partners may begin
to wonder whether the relationship they thought is right for them was right after all and,
as a result, they may begin to feel threatened (Simpson, Campbell, & Berscheid, 1986).
In contrast to romantic love, the experience of love for couples who have been
together for a long time and describe themselves as happy tends to be transformed into
what Sternberg calls companionate love (1986). Companionate love is a combination of
intimacy and commitment and tends to be experienced by partners whose lives have
become intertwined with each other and who possess a deep friendship (Sternberg, 1986).
Initially, when partners fall in love, according to Brehm (1992), it is often "with
their own imagined constructions rather than with the concrete reality of another human
(p. 103). However, after a couple has been together for an extended period of
time, the rewarding behaviors that may once have characterized courtship begin to
diminish as people's best behaviors become difficult to maintain over time. Best
behaviors tend to diminish and behaviors that are more reflective of and congruent with
behaviors begin to erode. The positive images partners had of their beloveds are replaced
with images, which are based on reality (Swann, De La Ronde, & Hixon, 1994; Huston &
Vangelisti, 1991). The accumulated effect of exchanged behaviors (positive and negative)
has an impact on the overall quality of relationships (Kamey & Bradbury, 1995).
relationships progress over time, partners learn through their mutual experiences and on
the basis of their interactions with each other and their evaluations of these interactions,
whether or not their relationships are rewarding and satisfying (Bradbury & Fincham,
Kayser (1993) used the terms "disillusionment" and "disaffection" when
describing the processes involved in the loss of love and the loss of a sense of closeness
between partners. Kayser (1993) explained disaffection as the replacement of positive
affect with neutral affect and added that for disaffection to occur positive feelings had to
exist at the beginning of the relationship. Kayser (1993) interviewed spouses and results
demonstrated that disillusionment typically occurs for spouses when partners' behavior
after marriage falls short of the expectations, dreams, and fantasies that were held prior to
marriage. Many researchers have found support for a disillusionment model (Brehm,
1992; Huston et al., 2001; Huston, McHale, & Crouter, 1986; Miller, 1997
Holmes, 1997), and have suggested that in the early stages of romantic relationships, new
partners tend to behave in ways that are consistent with their idealized view of the
relationship. For example, in the beginning of relationships, displays of affection are
embellished, conflict between partners is avoided, and negative feelings are muted. The
attractive, and responsive traits and lacking negative, unattractive and disagreeable traits.
In addition, because newlyweds tend to accentuate positive psychological and behavioral
tendencies, they should feel deeply in love with each other and they should experience
little ambivalence about the relationship and about being together. That is, until
disillusionment sets in (Brehm, 1992; Huston, McHale, & Crouter, 1986; Huston et al.,
2001; Miller, 1997
Murray & Holmes, 1997).
In another study looking at disillusionment, Huston et al. (2001) followed 168
newlywed couples over 14 years. Participant data were collected two months after the
wedding and twice again over the next two years. Researchers' contact with the
relationship partners included face-to-face interviews, completion of questionnaires, and
telephone interviews, which were done to assess partners' interactions with each other.
During the phone interviews, spouses were read descriptions of affectionate and non-
affectionate behaviors, and asked to rate how often their partners had done each of the
behaviors in the preceding 24 hours. After couples had been married for 13 to 14 years,
164 of the original 168 couples were re-contacted and additional data were collected. The
additional data included an assessment of marital status. One hundred and five of the
original participants who were re-contacted were still married, 56 were divorced, and 3
were widowed. Results of this study demonstrated that the couples who divorced quickly
had the weakest initial romantic bonds, showed the least amount of affection toward each
other, and were the most antagonistic and negative toward each other compared to any
other category of participants. Even after only two months of marriage, those who
researchers also found that the average daily ratio of positive to negative behaviors for
the couples who divorced quickly was one to four for husbands and one to three for
wives, indicating that these couples were deeply distressed. This study lends support to
the disillusionment model, which is reflective of an abatement of love, a decline in
affection, an increase in ambivalence, and changes in the way spouses view each other.
The Exchange of Behavior in Romantic Relationships
Many studies have attempted to identify which factors distinguish happy couples
from unhappy ones and to answer a closely-related question,
"Why do some marriages
become more satisfying over time while others become less satisfying?" Several
approaches have been utilized to investigate these questions. One such approach
addresses the exchange of both positive and negative interpersonal behaviors and was
conducted mainly from a behavioral theoretical perspective (Karney & Bradbury, 1995).
Several like this have focused on the exchanges of behavior between romantic partners
during problem-solving discussions. These studies tend to test the hypothesis that when
positive behaviors are received by relationship partners, the impact on both relationship
satisfaction and relationship stability will be positive, whereas when negative behaviors
are received, the impact on relationship satisfaction and stability will be negative (Karney
& Bradbury, 1995). Results of these studies (Karney & Bradbury, 1995; Markman, 1981;
Stuart, 1969; Wills,
Weiss, & Patterson, 1974) have generally supported the association
between the exchange of positive behaviors and enhanced relationships and the exchange
of negative behaviors and diminished relationship outcomes.
in relationship satisfaction (MacDermid, Huston, & McHale, 1990). Huston and Chorost
(1994) showed this when they compared couples' behavior shortly after marriage with
the behavior they exchanged two years later. The findings revealed that affection between
partners decreased by approximately half after newlyweds had been married for two
Several other researchers (Gottman, 1979; Hahlweg, Revenstorf, & Schindler,
1984; Levenson & Gottman, 1983; Raush, Barry, Hertel, & Swain, 1974; Schaap, 1982;
Ting-Toomey, 1983) have looked at the association between positive and negative
behavioral interactions and relationship satisfaction. Their research has provided a great
deal of additional empirical evidence showing that negative interactions are much more
common among unhappily married couples than happily married couples. For example,
partners who indicated that they were more satisfied with their relationships tended to
express more positive verbal and nonverbal behavior toward each other. These behaviors
included more head nodding, increased expressions of caring behaviors, increased
concern for one's partner, and more smiling. Recipients of these positive behaviors also
had higher ratings of relationship satisfaction. In contrast, unhappy or unsatisfied partners
express more negative verbal and nonverbal behaviors toward each other. These negative
behaviors included more threatening, critical, and insulting comments, more frowning,
and harsher voice tones. In addition, partners receiving these negative behaviors rated
themselves as being less satisfied with their relationships (Brehm, Miller, Perlman, &
Campbell, 2002; Carton, Kessler, & Pape, 1999; Patterson, 1988).
reciprocity of negative behaviors, and decreased relationship satisfaction (Huston et al.,
1986, Huston & Vangelisti, 1991; Jacobson & Margolin, 1979; Jacobson,
Some researchers have implicated anger (Greenberg & Johnson, 1988;
Hendrix, 1988; Parrott & Parrott, 1995) and conflict (Raush et al., 1974) as the culprits
that lead to negative relationship outcomes. However, this assertion has been challenged.
Gottman and Krokoff (1989) were interested in identifying the components that make up
negative interactions and in finding out whether anger was a key determinant of marital
distress. To investigate this question, initial relationship satisfaction was controlled by
dividing participants into two groups, couples who (1) rated marital satisfaction lowest
and (2) rated marital satisfaction highest. Both groups were videotaped while discussing
an issue about which they had a high degree of disagreement. Randomly selected
segments of the video taped segments were later viewed by trained raters, who used a
reliable and valid coding system to categorize the interactions. Interactions were rated as
being either effectively neutral, effectively positive (e.g., affection, anticipation,
excitement, humor, .interest, and joy), or effectively negative (e.g., anger, contempt,
disgust, fear, sadness, or whining).
When groups were compared, although there was a
short-term, inverse relationship between expressions of anger and relationship
satisfaction, in the long term, anger was positively associated with relationship
satisfaction. In addition, anger was not predicative of divorce (Gottman, 1994b; Gottman
& Levenson, 1992). However, with regard to the expression of non-angry negativity and
relationship satisfaction over time, Gottman found a strong inverse relationship.
concurrent marital satisfaction and positive interactions. Another interesting finding from
this study was that engaging in conflict and expressing anger were actually quite
functional for marriages in the long run; but when conflict between partners included
whining, stubbornness, and withdrawal, particularly withdrawal by husbands, the conflict
was more likely to contribute to long-term relationship deterioration.
Gottman's Taxonomy of Dysfunctional Romantic Relationship Patterns
Gottman found abundant support for the greater potency of negative than positive
behaviors (Gottman, 1979; Gottman, 1994a; Gottman, 1999). In one study, Gottman
(1994a) videotaped married couples in their homes and in a laboratory setting as the
couples discussed several topics, including the nutritional value of food, how their days
had gone, general marital issues, and a specific issue or situation within their marriage
about which they had continuing disagreement. Fifteen minute segments of the tapes
were viewed, rated, and categorized by researchers, who identified different patterns that
the couples used while dealing with conflict. The categories included positive and
negative as well as verbal and nonverbal dimensions. Judges watched and rated the
degree to which partners engaged with each other during conflict, the ratio of the positive
to negative interactions exchanged between them, and partners' attempts at persuading
each other during their discussions. Gottman identified several categories of couples
based on their conflict interactions. Four of these categories were labeled: volatiless,
avoiderss," and hostiless.
Gottman combined the volatiles with the validators because the way that these
volatile couples expressed a high degree of both positive and negative emotions. The
husbands in this category were extremely involved and although they expressed a great
deal of negativity while engaging in discussion, they also expressed humor and affection
toward their partners.
With the validators, even though there was some belligerence
expressed by them during arguments, there were also attempts made by partners to
persuade each other and conflict between partners tended to be characterized by ease and
calm. In addition, interactions of the validators included validation of partners'
experiences, attempts to understand each other, and expressions of empathy for the
other's point of view.
In his work with avoiders, Gottman had a difficult time setting up conflict
situations. This was because avoiders tended to be unengaged with each other and
expressed very little emotion-regardless of whether the expected emotion was positive
or negative. These couples tended to "solve" problems on their own or they relied upon
the passage of time for resolving issues instead of having strategies to work through and
resolve conflict together.
When avoiders did come up with solutions, their solutions
typically were not specific. Gottman found that these individuals were not very
introspective or psychologically minded. In addition, when there were discussions
between them, after one or both finished stating their cases, the discussion was
considered over and the issue closed. Thus, exchanges between these couples involved
little give and take and, in general, no attempts were made to persuade each other.
When Gottman compared those categories (validators, volatiles, and avoiders)
hostile couples engaged actively with each other, others were more disengaged, detached,
and uninvolved. The interactions of the engaged hostile couples consisted of more
frequent direct conflict and more defensiveness. Interactions of the detached or
uninvolved hostile couples tended to include briefer episodes of reciprocated attacking
and defensiveness often over trivial issues. Hostile couples were also more likely to
experience "flooding," which has been defined as an unexpected eruption of negative
emotions that overwhelms and disorganizes partners (Holman & Jarvis, 2000). In
addition, hostile couples displayed more criticism, defensiveness, withdrawal, and
contempt in their interactions with each other.
Gottman's Four Horsemen of the Apocalypse
According to Gottman (1994a), the four processes of criticism, defensiveness,
withdrawal, and contempt greatly undermine the stability of relationships. Because
Gottman considered these four processes so destructive, he labeled them "The Four
Horsemen of the Apocalypse."
Gottman characterized criticism as a focused attention on
and attacking of the character or the personality of a partner rather than focusing on the
issues at hand. In his research, Gottman found that criticism was often followed by
contempt. Contempt included hostile humor, insults, and mockery. Gottman (1994a)
found that spouses tend to respond to perceived attacks, including criticism and contempt,
by either becoming defensive or by stonewalling. A defensive response to a perceived
attack may be for a partner to cross-complain, make a counter-attack, or to make excuses
for one's own behavior. Some partners, particularly men, make ill-fated attempts to
worse. This type of stonewalling behavior claimingg up or withdrawing) communicates to
partners a lack of concern, unresponsiveness, smugness, and disapproval and it can be
infuriating (Gottman, 1994a; Zadro & Williams, 2000).
In addition to the four horsemen (criticism, contempt, defensiveness, and
stonewalling) just described, Gottman later added belligerence to the list of processes that
undermine the stability of relationships (Gottman, Coan, Carrere, & Swanson, 1998).
According to Gottman (Brehm et al., 2002; Gottman et al., 1998), belligerence is defined
as an angry expression of the problem, stated in its most extreme form. Belligerence is
expressed as an attempt to provoke, intimidate, or aggressively reject one'
such, belligerence communicates total rejection of the other partner and challenges the
's power and authority. Gottman's
subsequent research supported the
addition of belligerence because belligerence tends to ultimately follow the other four
horsemen (Gottman et al., 1998; Gottman, 1999).
Another Look at Behavior Exchange: What Predicts Relationship Demise?
Gottman and his colleagues (1994a; 1994b; 1999; Gottman et al., 1998; Gottman
& Levenson, 1992) observed couples' interactions, paying specific attention to the five
destructive processes just described and found that they could reliably distinguish
between couples headed for divorce and couples in stable relationships. These researchers
found that the volatile, validator, and avoider couples all tended to share a ratio of
positive to negative interactions (including criticism, contempt, defensiveness,
stonewalling, and belligerence) during conflict situations that was approximately five to
negative interactions was approximately one to one. That is, when hostile couples
interacted with each other there was one positive interaction for every negative
interaction that was expressed between them. In addition, when couples rated relationship
satisfaction Gottman found that volatile, validator, and avoider couples had much higher
ratings of relationship satisfaction than hostile couples.
Other studies (Barnett & Nietzel, 1979; Gottman, 1993; Gottman & Levenson,
1992; Wills et al., 1974) have investigated romantic relationships and looked at
behavioral balance theories. These theories posit that marriages function with a set point
balancing positivity and negativity. For example, in one study (Wills et al., 1974) 73
engaged couples had three 15-minute conversations with each other and discussed a daily
event, a conflict situation, and a pleasant topic. Four years later the participants were re-
contacted and asked to complete questionnaires assessing relationship satisfaction and
marital status. Results showed that while the initial positive behaviors that were displayed
between the couples co-varied with ratings of subsequent relationship satisfaction
particularly for couples who were happy (Wills et al., 1974), the exchange of negative
behaviors between partners was a more potent predictor of subsequent decreases in
relationship satisfaction ratings, especially for the couples who were distressed (Barnett
& Nietzel, 1979). These findings lend support for the ability of negative behaviors to
predict later relationship satisfaction.
In another study, Gottman (1999) found that the quality of a couple's relationship
was better predicted by the presence or absence of negative behaviors than it was by the
decreasing one (either increasing or decreasing positive behaviors or increasing or
decreasing negative behaviors) does not necessarily increase or decrease the other. An
important implication for Gottman
finding was that although increasing positive
behaviors does impact relationship quality, the impact of increasing positive behaviors is
not nearly as significant as would be the impact of decreasing negative behaviors. This is
because the assessed negative impact of negative behaviors on relationship quality was
greater than the positive impact of positive behaviors.
In a study looking at videotaped segments of the interactions of 124 newlywed
couples who were engaged in a discussion about a conflict, Carrere and Gottman (1999)
were able to make predictions with 83% accuracy about which couples would divorce six
years later after only watching three minutes of the videotaped segments. Their prediction
accuracy increased for the husbands when 12 more minutes of videotape was watched. In
the segments, couples who were headed for divorce exhibited noticeably more
defensiveness, contempt, and belligerence toward each other than did the couples who
In another study, Gottman et al. (1998) looked at the interactions of 130
newlywed couples to determine which processes were the most accurate long-term
predictors of relationship stability and satisfaction for happy and unhappy married
couples. The couples were followed over a 6-year period. During the course of the study,
their interactions were videotaped and their affect during interactions was rated as being
either positive (e.g., affection, humor, interest, joy, and validation), negative (e.g., anger,
couples, including physiological data and participants' ratings of their own interactions.
Consistent with prior results from similar studies, results of this study showed that anger
was not predictive of divorce nor did it discriminate happy from unhappy couples.
However, for both husbands and wives, belligerence, defensiveness, and contempt were
the most destructive patterns utilized during conflict resolution. These patterns were most
predictive of divorce, but they did not discriminate between the happily and unhappily
married stable couples. Further, Gottman et al.'s (1998) findings suggest that the
reciprocity of negative affect does not generally predict negative marital outcome except
for during the first few months of marriage. That is, early in a marriage wives in
marriages that are headed toward instability were more likely to reciprocate negative
affect (e.g., anger, domination, disgust, fear, sadness, stonewalling, and whining)
compared to the wives who were in marriages that were stable. In addition, husbands'
refusal to accept influence from their wives, and husbands'
inability to self-soothe and
de-escalate from low intensity negativity (e.g., anger, disgust, domination, fear, sadness,
stonewalling, and whining) were predictive of negative relationship outcomes. Positive
affect (e.g., affection, humor, interest, joy, and validation) and the ability to de-escalate
predicted positive relationship outcomes. Another study looking at 19 couples who were
part of a previous study, found evidence suggesting that husbands, who were in
dissatisfying marriages, were more likely to withdraw emotionally much earlier than
wives were (Levenson & Gottman, 1985).
Further Support for the Detrimental Impact of Bad Behaviors on Relationships
al., (2001, p. 325) defined bad as a concept representing states or consequences that are
"undesirable, harmful, or unpleasant." Bad is contrasted with good, which is defined in
this context as a desirable state or consequence or as experiences that are beneficial or
pleasant. According to Baumeister et al., (2001) the impact of experiencing something
bad is greater than the impact of experiencing something good because bad is more
consistent, more multifaceted, and longer lasting.
In an extensive review, which referenced over 200 articles from the psychological
literature and covered multiple domains and disciplines, Baumeister et al., (2001)
provided impressive evidence supporting the premise that bad is stronger than good.
Because there is so much support for the greater power and impact of bad compared to
the impact of good, Baumeister referred to bad is stronger than good as a psychological
principle. In addition, these researchers found that this principle is applicable to the area
of close relationships.
Rozin and Royzman (2001), also discussed the greater power of negative events
and described another principle known as the negativity bias, which holds that in most
situations, negative events are more salient, potent, and dominant and more efficacious
than positive events (p. 297). These authors put forth a taxonomy for understanding this
phenomenon suggesting that the negativity bias is manifested by the following principles:
(a) negative potency, which posits that negative entities are stronger than positive
entities; (b) steeper negative gradients, which posits that the negativity of negative events
grows more quickly with regard to time or space than the positivity of positive events; (c)
valences would predict; and (d) negative differentiation, which posits that negative
entities are more complex, varied, and elicit more response options than positive entities.
Consistent with the principles of Rozin and Royzman (2001) and the findings of
Gottman (1999), Baumeister et al. (2001) concluded that when equal amounts of good
and bad behavior are compared, the impact of bad is more potent because bad is more
damaging, harmful, and hastens the demise of close relationships more powerfully than
good behavior can offset.
Every Day Aversive Interpersonal Behaviors
In her book, Behaving badly, Kowalski (2001) looked at bad behaviors that
occurred within the context of close, interpersonal relationships. Kowalski (2001)
conceptualized aversive or bad behaviors as behaviors that are enacted by one individual,
received by another individual, and perceived by the recipient as being stressful.
Although many aversive behaviors would be considered socially inappropriate, many are
not clearly socially inappropriate (Kowalski, 1997). In addition to being stressful,
aversive behaviors interfere with people's
basic psychological need to belong, and
negatively impact people's sense of control and their sense of self. Aversive behaviors
contribute to disconnections, disruptions, and to the demise of relationships (Kowalski,
According to Kowalski (2001), when most people think about aversive behaviors
they call to mind extreme, egregious acts of aggression or violence such as murder, rape,
or physical abuse. However, these behaviors are quite rare. Instead, most people are
relationship satisfaction and to the stability of relationships, especially when the
behaviors are exchanged between close, intimate partners (Kowalski, 1997
A list of mundane aversive interpersonal behaviors compiled by Kowalski (2001)
includes betrayal, chronic complaining, conflict, contempt, criticism, deception,
defensiveness, disappointment, excessive reassurance-seeking, forgetting commitments,
gossiping, hurting other's feelings, incompetence, infidelity, intentional embarrassment,
jealously, lying, moodiness, neglect, not listening, ostracism, passive-aggressiveness,
rudeness, spreading rumors, swearing, teasing, using the silent treatment, and violating
confidentiality. According to Kowalski (1997; 2001), these behaviors do occur frequently
in relationships. In addition, they are often done inadvertently and much of the time the
hurt that results is unintentional. Kowalski'
focus has been on understanding the
conditions under which these behaviors occur as well as the consequences of these
behaviors for the individuals and for the relationship.
In addition to the term,
aversivee interpersonal behaviors"
2001), researchers have used terms such as,
"relational transgressions" (Metts, 1994) and
"social undermining" (Vinokur & van Ryn, 1993; Rook, 1998) to describe bad behaviors
that occur between partners in interpersonal relationships. Relational transgressions have
been defined as violations of rules for appropriate relational conduct (Metts, 1994).
Social undermining refers to interactions characterized by aversive behaviors (Vinokur &
van Ryn, 1993).
satisfaction, is an example of one method used by couples researchers. Cross-sectional
data from studies utilizing daily records have been analyzed to evaluate the relationship
between positive and negative interactions, and relationship satisfaction (Barnett &
Nietzel, 1979; Broderick & O'Leary, 1986; Wills,
Weiss, & Patterson, 1974).
from these studies have lent support to the greater power of negatively valenced
behaviors over positively valenced behaviors for predicting relationship satisfaction,
particularly for distressed couples. Although positive behaviors do co-vary with
relationship satisfaction, these studies have also shown that the strength of the
relationship is weaker for distressed than for happy couples (Wills et al.,
Another method that has been utilized in couples'
research is to compare the
behaviors of couples in distressed vs. happy relationships (Birchler, Weiss, & Vincent,
Gottman, Markman, & Notarius, 1977).
When Gottman et al., (1977) compared
distressed with nondistressed couples they found a sex difference for the nondistressed
Wives in nondistressed relationships were less likely to respond negatively to
negative messages from husbands than were husbands in nondistressed relationships who
were responding to negative messages from their wives.
Several studies investigating romantic relationships have used longitudinal
research methods to look at the association between interactional behaviors and
relationship satisfaction. These longitudinal studies took information about couples
gathered early in their relationships to make predictions about later relationship
satisfaction (Kelly, Huston, & Cate,
1985; Markman, 1979,
1981). Generally these
early in marriage are powerful predictors of future relationship dissatisfaction and
Karney and Bradbury have utilized longitudinal research methods in their work
on why some marriages succeed while others fail, and to look at the origins of distress
and divorce in relationships. In one study, Karney & Bradbury (1997) recorded the
interactions of 60 newlywed couples and followed them over a span of four years,
contacting and assessing them eight times. At the end of the fourth year, 38 of the
original relationships were intact, 18 had dissolved, and four had withdrawn from the
study. Analyses of the data showed that for husbands and to a greater degree for wives,
behavior directed toward a partner during the initial interaction predicted rates of change
in marital satisfaction. Additionally, a husband'
negative behavior during problem
solving discussions was more predictive of rapid declines in a wife's marital satisfaction
over the first four years of marriage. This was true even when the wife'
initial levels of
satisfaction were controlled. A wife's negative behaviors were also predictive of declines
in a husband'
and a wife's marital satisfaction over the first four years of marriage, but
to a lesser degree. This study demonstrated that problem-solving behavior was strongly
associated with rates of change in marital satisfaction, independent of initial levels of
Given the strong inverse association between negative behaviors and relationship
satisfaction, Huston and Vangelista (1991) became interested in causality. They wanted
to know if ratings of relationship satisfaction were responsible for negative behaviors
examined the relationship between the expression of positive and negative behaviors and
relationship satisfaction. Participants were interviewed approximately two months after
their marriages and then semi-annually over the next two years. The researchers
conducted phone interviews to gather information about the couples' behavioral
interactions and face-to-face interviews to collect information about relationship
satisfaction. The data were analyzed both cross-sectionally and longitudinally. Cross-
sectional data analyses assessed the relationship between satisfaction and the expression
and receipt of positively- and negatively-valenced behaviors. Longitudinal analyses
explored whether changes in relationship satisfaction could better predict behavior or
whether behavior better predicted relationship satisfaction.
Vangelista and Huston found
that for both husbands and wives, the expression and receipt of negativity was
consistently and inversely associated with marital satisfaction, even for couples married
for as little as two months. The behavior of wives, whose husbands expressed relatively
more negative behaviors at the beginning of marriage, became more negative. Further,
the behavior of husbands whose wives were less satisfied at the beginning of marriages
became more negative over time. They also found that whenever negative behaviors were
expressed in marriages regardless of which spouse was expressing the negativity, wives
were less satisfied with the relationship (Huston & Vangelista, 1991).
Support for Intervention
Huston and colleagues (2001) also found support for what has been called the
enduring dynamics model, which posits that certain interpersonal patterns and problems
predictive of how happily married couples would be 13 years later if the couple stayed
together. In addition, when relationships terminated, researchers were able to predict
reliably how long the marriages would last prior to separation or termination based on
' initial interactions. Several other studies (Huston & Chorost, 1994; Huston &
Houts, 1998; Karney & Bradbury, 1997
Markman, Renick, Floyd, Stanley, & Clements,
1993; Noller & Feeney, 1998) have shown similar results, demonstrating that the initial
differences between couples did persist over time and that these differences were
predictive of later ratings of relationship satisfaction and of relationship stability.
In a study comparing early interventions, Markman and his colleagues (1993)
looked at the causes of marital distress, focusing on behavioral factors as opposed to
personality factors. They found that when there were deficits in the manner in which
couples communicated early in the relationship, there would be subsequent declines in
relationship satisfaction. Other researchers have found similar results showing that the
more positively couples initially interacted with each other, the more satisfied they would
be in the future (Markman, 1979; 1992; Markman, Floyd, Stanley, & Storaasli, 1988;
Markman, Renick, Floyd, Stanley, & Clements, 1993; Storaasli & Markman, 1990).
In a longitudinal study Markman (1979) initially assessed 14 newlywed couples
and gathered information on their interactional patterns, the intensity of their problems,
and their ratings of relationship satisfaction. Markman (1979) used this information to
make predictions about their ratings of relationship satisfaction
2 2 years later. He found
that the more positively one partner's initial interactions were, the more satisfied that
development of future relationship distress. In another study Markman (1981) assessed
26 couples who were engaged to be married. The couples were then re-contacted and re-
interviewed three times over a 5 2 year period. At the end of the study, 21 out of the
original 26 couples' relationships were intact. Nine of those couples completed all of the
research data. Results from this study showed that there was a significant correlation
between early communication problems and distress later in the marriage.
Several other studies have investigated partners' communication patterns and
styles (Gottman, 1979; Riskin & Faunce, 1970; Rusbult, Johnson, & Morrow, 1986;
Stafford & Canary, 1991; Ting-Toomey, 1983; Wills,
Weiss, & Patterson, 1974). Results
of these studies have provided convincing support for the generally held assumption that
positive communication is associated with higher levels of relationship satisfaction.
These studies have also demonstrated that there is a correlation between people's
reported satisfaction and how positively or negatively they communicate with each other.
Markman and his colleagues (Markman et al., 1993) were interested in whether a
pre-marital intervention would have any impact on subsequent relationship quality. They
followed 114 engaged couples who were part of a larger study matching them on several
variables and assigning them to a treatment group, a control group, or a group of couples
who declined treatment. The treatment group underwent a five session intervention
designed to teach couples effective conflict management and communication skills. All
of the couples taking part in the study received follow-up contacts 1 V, 3,
4, and 5 years
after the study began. These contacts included completing questionnaires and
husbands at the 5-year follow-up. Those who had not received the intervention had lower
levels of relationship satisfaction then the participants who received the intervention. In
addition, couples who received the intervention were less likely than control or decline
group couples to terminate their relationships. Markman and colleagues (1993) also
found that participation in the intervention gave couples a significant future advantage in
communicating and conflict management skills even four years after receiving the
intervention. Married partners who received the intervention displayed less withdrawal,
denial, dominance, negative affect, conflict, and negative communication toward each
other than did the partners in the control group. Markman et al. (1993) attributed these
differences between the groups to the intervention, because the intervention specifically
and directly addressed these aforementioned behaviors. Another finding from this study
was that the tendency for partners to engage in physical relationship violence, which
Markman and Kraft (1989) defined in part as the result of partners' inability to handle
conflict constructively, was reduced for the participants who received the intervention
versus the control and decline treatment groups. Markman and his colleagues (1993)
again attributed this difference to the intervention, as these couples exhibited more
positive and fewer negative conflict management skills and communication behaviors.
The Current Study
Receipt of negative or aversive behaviors has an impact on people. It affects their
sense of self, sense of control, and sense of belonging. Receipt of aversive behaviors is
stressful. Negative behaviors decrease people's satisfaction with their relationships and
their relationship to feeling loved and relationship satisfaction. The following chapter
details the methods used to accomplish this goal.
This chapter is divided into three sections and details the methods used in scale
development. Each of the sections discusses one of three phases. The first section
describes the methods utilized in Phase 1 for item generation. The second section
describes the methods used in Phase 2 for scale item identification and assessment of
factor structure. The third section describes the methods utilized in Phase 3 for
assessment of psychometric properties of the resulting scale. Each section discusses
sample characteristics, measures, procedures, and data analyses as applicable.
Response data from 102 (69.61% female, 30.39% male) students enrolled in one
of two sections of an undergraduate psychology course in Abnormal Psychology or
Personality Psychology at the University of Florida were used in the first phase of the
study. The mean age of participants responding to the questionnaire item asking for age
was 20.03 (SD
1.51). Participants' ages ranged from 17 to 28 years of age. A majority
of the participants identified as White (52.94%), with 12.75% reporting Hispanic/Latino,
11.76% African Descent/Black, 9.80% Asian/Pacific Islander, and 12.75% identifying as
bi-racial, multicultural, or other. A majority (98.04%) of the sample identified as
participants reported being freshmen, 31 (30.39%) reported sophomore, 48 (47.06%)
reported junior, 16 (15.69%) reported senior, and one (0.98%) indicated post-
baccalaureate student status.
Fifty seven (55.88%) of the participants reported currently being in a romantic
relationship. Fifty six, or 98.25% of those in current relationships indicated that their
current relationship was heterosexual; one participant (1.75%) reported that the current
relationship was with a same sex partner. The mean length for current relationships was
months (SD = 20.2 months). The range was from less than one month to 10 years.
Forty six (80.70%) of those in current relationships reported exclusivity with their current
partners. Six (10.53%) of those in current relationships reported non-exclusivity in their
relationship with their current partners. Five (8.77%) of those in current relationships did
not indicate whether their relationships were exclusive or not. Seven (12.28%) of those in
current relationships reported living with their current partners, two (3.51%) reported
being engaged to their current partners, and one (1.75%) reported being married to the
Regardless of whether participants were in a current romantic relationship, 70
participants (68.6%) indicated they have had at least one prior romantic partner. The
average length of prior romantic relationships was 12.74 months (SD = 12.37 months)
with a range from one month to 60 months. All participants who reported having had a
prior romantic relationship indicated that their relationship was heterosexual. Four
(3.92%) participants reported that they were neither currently in nor have they previously
Eighty (78.43%) of the 102 participants agreed to being re-contacted by the
researcher for a follow-up phone interview. The study's author randomly selected three
females and three males from those who agreed.
When the attempts were made to re-
contact those who were selected, all six were reached and all consented to participate in
the follow-up phone interviews.
Prior to meeting with the participants, a research questionnaire (Appendix A) was
developed. The questionnaire included an instruction intended to elicit response items.
The instruction asked participants to "Take a moment to think about your current
romantic relationship or a former romantic relationship. Using the space provided below,
record your responses to:
What could your partner DO that would make you feel
unloved? What could your partner SAY that would make you feel unloved?" Space was
provided for the participants to record their responses on the questionnaire following the
instruction and prompt items.
After recording the elicited response items, participants were instructed to
respond to a scale that assessed background, demographic, and relationship status
(Appendix B). Items on the scale included questions about participants' sex, age,
academic status, sexual orientation, racial and ethnic background, relationship status,
relationship satisfaction, and relationship length.
The six participants who participated in the follow-up phone interviews were
randomly selected and contacted by phone. These participants were asked to respond
Prior to data collection, the study's author contacted course instructors to discuss
details about the study and arrange times to meet with students. At the scheduled times,
the study's author met with students and provided them with written informed consent
(Appendix C) and information about the study. Students were informed that the study
was about romantic relationships, that participation was voluntary and involved recording
their responses anonymously to questionnaire items, and that they would be given extra
credit in their class by their instructor if they agreed to participate. Students were also
informed that if they were agreeable, they could participate further in the research study.
Further participation would involve being contacted by phone for a follow-up interview
related to the study questionnaire that would last for approximately five minutes. Those
who decided that they wanted to participate in this way provided their consent by
recording their names, phone numbers, and the best times for them to be reached, on a
form provided to them separately from the questionnaire.
All students present consented to participate in the study. Each was given a
questionnaire (Appendix A).
Questionnaires provided instructions to participants for
recording their responses to questionnaire items. After recording responses, participants
placed questionnaires into one box and consent forms agreeing to be contacted for a
follow-up phone interview into a separate box. Each participant then signed a form that
was later given to course instructors for the provision of extra credit. Prior to leaving,
participants were given a list of couples' counseling resources (Appendix D), thanked,
14 days after questionnaire administration. Randomly selected phone interviewees were
each contacted separately by the study's author, who reintroduced herself, informed each
participant that participation was still voluntary, and that responses would be recorded
but identity and contact information would not be associated with responses. Each of the
six-randomly selected participants agreed to be interviewed by phone. Each was asked,
"What could your partner do or say that would make you feel unloved?" and responses
were recorded verbatim by hand.
The questionnaire generated 542 items. The interviews generated 19 items.
Redundant items were eliminated from the 561 resulting items. Items were corrected for
spelling and grammar, and made non-gender specific. The remaining list of 135 items
was labeled the Preliminary Undesirable Behavior Scale (PUBS; Appendix E).
Data from 220 students (30.84% male, 69.16% female) were used in Phase 2 of
the study. The mean age of participants was 19.76 (SD
1.96) with a range of 16 to 30
years of age. A majority of participants identified as White (62.62%). The remaining
participants identifying as African Descent/Black (12.62%), Asian/Pacific Islander
(10.75%), Hispanic/Latino (10.75%), and "other" (3.27%). Ten (4.69%) participants
reported that they were international students. A majority of the participants responding
to the sexual orientation item identified as heterosexual (90.70%), with the remaining
identifying as bisexual (5.12%) and gay/lesbian (4.19%).
With regard to academic year in
One-hundred twelve (53.33%) participants reported that they were currently in a
romantic relationship and 98 (46.67%) reported they were not. Twenty-four (19.20%) of
the participants reported current relationship length of less than three months; 19
(15.20%) reported current relationship length of three to six months; 21 (16.80%)
reported current relationship length of six to 12 months; 29 (23.20%) reported current
relationship length of one to two years; and
(25.60%) reported current relationship
length of more than two years.
Eighty-seven (73.73%) respondents indicated that they were dating their current
relationship partners exclusively. Seventeen (14.41%) reported that they were in non-
exclusive relationships. Seven (5.93%) reported that they lived with their current
partners; two (1.69%) reported being engaged to their current partner, and five (4.24%)
reported being married to their current partner.
Regardless of whether participants were in a current romantic relationship, 174
(83.25%) indicated that they have had at least one prior romantic relationship. Thirty-
eight (18.36%) participants reported that the length of their past relationship was less than
three months; 42 (20.29%) reported that the length of their past relationship was three to
six months; 56 (27.05%) reported that the length of their past relationship was six to 12
months; 37 (17.87%) reported that the length of their past relationship was more than one
When asked when their most recent past relationships ended, 30 (14.35%)
participants reported that their most recent past relationships ended less than three months
between six and 12 months ago, and 81 (38.76%) reported that their most recent past
relationships ended over one year ago.
A questionnaire (Appendix F) was developed and utilized in the second phase of
the study to collect response data. The Phase 2 questionnaire included instructions, the
Preliminary Undesirable Behavior Scale (PUBS; Appendix E), which was the list of
items elicited from Phase 1 participants, the Desired Loving Behavior Scale (DLBS;
Heesacker, Smith, & Lawrence, 1998; Appendix G), which is a scale comprised of 39
desired loving behaviors, the Inventory of Desirable Responding in Relationships (IDRR;
Loving & Agnew, 2001; Appendix H), which is a scale comprised of 10 items assessing
people's tendencies to deceive themselves and others about negative aspects of their close
relationships, and a scale that assessed demographic, background, and relationship status
Preliminary Undesirable Behaviors Scale
The PUBS (Appendix E) is a list 135 items generated by Phase 1 participants that
would elicit the feeling that one is not loved by one's partner. Phase 2 participants were
asked to read each of the PUBS items and rate how unloved they would feel if they
received these items from a romantic partner. Items were rated on a 5-point Likert scale
with values ranging from 0 to 4 with lower scores indicating that respondents would feel
more unloved and higher scores indicating respondents would feel more loved.
Desired Loving Behavior Scale
process of constructing the DLBS was similar to the process utilized in scale
development in the present study. Specifically, a group of participants generated items
that reportedly elicited the feeling of being loved by their romantic partners when the
behaviors were received from their partners. After items were generated a list of 158 non-
redundant items was compiled. A second group of participants rated the items indicating
how unloved they would feel if the items were received from a partner. Factor analysis
and an oblique rotation of the 158 items resulted in 39 items achieving simple structure
and loading on to one of four factors. These four factors were labeled: Caring Actions,
Relationship Support, Scripting, and Sex.
Since the DLBS was developed, the psychometric properties of this instrument
have been assessed in a series of studies and the DLBS has demonstrated adequate
reliability and validity. In addition, the DLBS has been utilized as an assessment
instrument in research and in clinical settings to identify specific behaviors that are
desired from relationship partners. Higher DLBS scores indicate that items are more
desired and lower scores indicate items are less desired. In clinical and research settings
the DLBS may be administered to one or both relationship partners. In addition, the
DLBS may be administered more than once to participants varying the instructions and
having respondents not only rate how much of each behavior is desired by them but also,
how often the behaviors are received from relationship partners.
When the scale is
administered in this way, a want-get discrepancy score can be calculated for either the
total scale, each of the subscales, or for individual items. Discrepancy scores are
When discrepancy scores are calculated, therapeutic interventions may be
targeted in areas where the discrepancy between what is desired and what is received is
greatest, thus maximizing the potential for significantly impacting and improving the
As previously stated, prior research investigating the DLBS has demonstrated that
this instrument possesses ample validity and reliability. For example, the magnitude of
the DLBS discrepancy is significantly related in an inverse direction with relationship
satisfaction. This means that as the want-get discrepancy decreases, relationship
satisfaction increases and as the want-get discrepancy increases, relationship satisfaction
decreases. In one study, 42.27% of the variance measuring how satisfied participants
were with their relationships could be accounted for by the total want-get discrepancy
(Smith, 2000). Test-retest reliability for the DLBS as a whole measured .65 across a 26-
day interval for a sample of 158 college students. Internal consistency reliability for the
overall scale and for the subscales were also high (overall scale a = .91, (Caring Actions
a-= .78, Relationship Support a
a = .90, and Sex
The 39 DLBS (Appendix G) items were included and placed randomly among the
PUBS items (Appendix E) on the Phase 2 questionnaire (Appendix F) to assess the
construct validity of the Love Crushing Behavior Scale (LCBS; Appendix J), which was
being constructed as a result of Phase 2 data analyses. All DLBS (Appendix G) and
PUBS (Appendix E) items were rated using a 5-item Likert scale in order to be
consistent. Participants were asked to read each of the items and rate how loved (or
feel less loved. It was anticipated that if there were significant correlations between the
PUBS and the DLBS items, those correlations would be inverse.
Inventory of Desirable Responding in Relationships
The Inventory of Desirable Responding in Relationships (IDRR, Loving &
Agnew, 2001, Appendix H) is a ten-item scale that was developed for use in dyadic
research. The scale measures socially desirable responding patterns, which are people'
tendencies to deceive themselves and deceive others about negative aspects of, in this
case, their close relationships. Prior to the development of the IDRR, most instruments
measuring socially desirable responding patterns were developed to assess individual'
general predispositions to misrepresent information about themselves. The IDRR was
developed with a more specific aim in mind, which was to measure individual'
tendencies to misrepresent aspects of their romantic relationships.
The IDRR was developed and validated over a series of 5 studies. The 10 items
making up the IDRR can be factored into two subscales. These two factors are
Impression Management (IM) and Self-Deception (SD). Impression Management, also
known as misrepresentation, is the tendency for respondents to directly make an effort to
enhance their image in the eyes of others. A 7-point Likert scale is used to rate the five
Impression Management items. Likert scores range from 1 to 7
, with higher scores
indicating more effort is being put forth to present the self in the best possible light by
exaggerating performance on desirable behaviors and discounting performance on
undesirable behaviors. It is believed that when people respond in this manner it is the
settings that are anonymous to settings that are more public in nature (Loving & Agnew,
The second factor, Self-Deception, is also made up of five items. These items
capture the extent to which individuals hold positive illusions about themselves.
Impression Management measures conscious misrepresentations that are made by
respondents, Self-Deception measures unconscious misrepresentations made by
respondents about their relationships. This is because respondents often actually believe
the enhancing reports that are being presented to others about themselves.
respondents rate the items on the self-deception subscale, a Likert scale is used with
values ranging from 1 to 7. Higher scores indicate that respondent hold more positive
illusions about themselves.
Initial research by Loving and Agnew (2001) on the IDRR showed that there is a
significant intercorrelation among the five Self-Deception items (a =
.86) and among the
Impression Management items (a = .81), and that there are nonsignificant correlations
between the Self-Deception subscale and the Impression Management subscale (r = -.18,
ns). In research with both dating partners and married couples, the Self-Deception
subscale was found to correlate positively with measures of commitment, satisfaction,
investment, and dyadic adjustment, and this scale correlated negatively with alternatives.
Impression Management was not significantly associated with any relationship quality
index. According to Loving and Agnew (2001), this suggests that when Impression
Management occurs, it may be measuring a natural desire to enhance the public image of
In addition to responding to the aforementioned scales, participants read and
responded to demographic and background items as well as to items assessing
relationship satisfaction and status (Appendix I). The demographic and background items
assessed participants' sex, age, academic status, sexual orientation, race, and cultural
identity. The scales used to rate relationship satisfaction ranged from 1 to
with 1 =
extremely unsatisfied to 5 = extremely satisfied. Rating scales used to assess love for
partner, sense of love from partner, sense of caring from partner, and sense of feeling
loved when thinking about partner ranged from 1 to 5, with 1 = to the most negative
response option (i.e., not at all or strongly disagree) and 5 = to the most positive response
option (i.e., very much or strongly agree). In analyzing these data, a mean score for love
was calculated by averaging the sum of participants' scores on these items: sense of love,
sense of caring, and feeling loved when thinking about partner. The averaged score for
love ranged from 1 to
with 1 equal to the most negative response and
equal to the
most positive response.
A procedure similar to the one utilized in the last phase was used to recruit the
majority of the Phase 2 participants whose response ratings on the PUBS (Appendix E)
would be used to identify which items would comprise the Love Crushing Behavior Scale
(LCBS; Appendix J). In addition to the participants enrolled in psychology courses, a few
participants were recruited through the Psychology Department's Introductory
(Appendix K) information. All students consenting to participate in Phase 2 were given
instructions, questionnaires (Appendix F), and scanable answer sheets for recording
responses. After responding to questionnaires, participants placed questionnaires and
answer sheets into a box and signed their names to a form indicating that they
participated in the study. Their signed forms were separated from their questionnaires and
answer sheets. The resulting list of names was used later to provide extra credit to
participants. Prior to leaving, all participants were handed a debriefing statement
(Appendix L) and a list of couples'
counseling resources (Appendix D), thanked, and
PUBS items (Appendix E) were subjected to exploratory factor analysis to
determine the factor structure and scale items. Scores on the IDRR (Appendix H) and the
DLBS (Appendix G) were correlated with the PUBS. Results of analyses are presented in
Data from 277 students (31.88% male, 68.12% female) were used in the third
phase of the study. The majority of the participants (80.87%) were enrolled in one of four
sections of undergraduate psychology courses; 19.13% were obtained through the
Psychology Department's Introductory Psychology Participant Pool.
With regard to age,
3 (1.09%) participants reported being less than 18 years of age, 52 (18.84%) reported
years of age, and
(0.72%) reported age as 40 years of age or more. The mean age for
participants was 19.68 (SD = 1.35).
With regard to academic status, 60 (21.74%)
participants reported being freshmen, 80 (28.99%) reported being sophomores, 92
(33.33%) reported junior, 41 (14.86%) reported senior, and three (1.09%) reported having
post-baccalaureate student status. The mean rating for academic status was 2.45 (SD =
1.02) with 1
= sophomore; 3
= junior; 4
= senior; and 5 = post-
baccalaureate student. Two-hundred fifty-seven. (93.12%) participants identified
themselves as heterosexual, 9 (3.26%) identified themselves as bisexual, and 10 (3.62%)
identified themselves as gay/lesbian. A majority of participants identified themselves as
White (67.87%). The remaining participants identified themselves as African
Descent/Black (6.59%), Asian/Pacific Islander (8.36%), Hispanic/Latino (10.55%), and
(6.63%). Five (1.81%) participants indicated that they were international
One-hundred fifty-one, or 54.51% of all respondents, indicated that they were
currently in romantic relationships; 115 (41.52%) reported that they were not currently in
romantic relationships. Eleven did not indicate whether or not they were in current
romantic relationships. The mean length for those in current relationships was 3.82
Fifteen (5.66%) participants reported that the length of their current
relationship was 1 month or less, 9 (3.40%) reported that the length of their current
2 months, 13 (4.91%) reported that the length of their current
relationship was 3 months, 13 (4.91%) reported that the length of their current
relationship was 9 months to 1 year, 46 (17.36%) reported that the length of their current
relationship was 1 to 2 years, 26 (9.81%) reported that the length of their current
years to 3 years, and 18 (6.79%) reported that the length of their current
relationship as 3 years or more.
Regardless of whether participants were in a current relationship,
indicated that they have had a prior romantic relationship and 49 (17.69%) reported that
they have not. Eleven (4.15%) participants reported that their most recent past
relationship terminated within the past month, 12 (4.53%) reported that their most recent
past relationship ended
relationship ended 3 months ago,
months ago, 11 (4.15%) reported that their most recent past
(9.43%) reported that their most recent past
relationship ended between 4 and 5 months ago, 27 (10.19%) reported that their most
recent past relationship ended between 6 and 8 months ago, 21 (7.92%) reported that their
most recent past relationship ended between 9 months and 1 year ago, 49 (18.49%)
reported that their most recent past relationship ended between 1 and 2 years ago, 32
(12.08%) reported that their most recent past relationship ended between
and 3 years
ago, and 40 (15.09%) reported that their most recent past relationship ended more than 3
Twenty-seven (10.23%) of the participants who reported having a past
relationship indicated that the most recent past relationship lasted for one month or less,
21(7.95%) reported that their most recent past relationship lasted two months, 23 (8.71%)
reported that their most recent past relationship lasted three months,
22 (8.33%) reported
that their most recent past relationship lasted nine months to one year, 39 (14.77%)
reported that their most recent past relationship lasted one to two years, 27 (10.23%)
reported that their most recent past relationship lasted two to three years, and 11 (4.17%)
reported that their most recent past relationship lasted three years or more.
The Phase 3 Questionnaire (Appendix M) consisted of the Love Crushing
Behavior Scale (LCBS; Appendix J) and several other measures mainly included to
LCBS construct validity. The additional measures were the Positive and Negative
Affect Scale (PANAS; Watson, Clark, & Tellegen, 1988; Appendix N), the Relationship
Assessment Scale (RAS; Hendrick, 1988; Hendrick, Dicke, & Hendrick, 1998; Appendix
0), the Global Measure of Relationship Satisfaction Scale (GMREL; Lawrance & Byers,
1995; Lawrance & Byers, 1998; Appendix P), the Relationship Happiness Scale, (RHS;
Fletcher, Fitness, & Blampied, 1990; Appendix Q), the Perceived Relationship Quality
Component Scale-short version (PRQC-short version; Fletcher, Simpson, & Thomas,
2000; Appendix R), Predictions about the Future Scale (Aarestad, 2000; Appendix S), the
Partner-rated Self-disclosure Index (Miller, Berg, & Archer, 1983; Appendix T), the Self-
rated Self-disclosure Index (Miller, Berg, & Archer, 1983; Appendix U), the Mutual
Psychological Development Questionnaire: Form A (MPDQ: Form A; Genero, Miller, &
Surrey, 1992; Appendix V), and a scale assessing demographic, background, and
relationship status (Appendix W).
The Love Crushing Behavior Scale
presented to Phase 3 participants twice. Participants rated the items using a 5-point Likert
scale both times the items were presented. The first time participants indicated how loved
(or unloved) they would feel if romantic partners did or said each of the items. Likert
scale values ranged from 0 to 4, with higher scores indicating that participants would feel
loved and lower scores indicating that they would feel unloved. The second time,
respondents indicated how often behaviors were received from romantic partners, using a
Likert scale with values ranging from 0 to 4. Higher scores indicated more frequent
receipt of LCBS behaviors, whereas lower scores indicated less frequent receipt of LCBS
The Positive and Negative Affect Scale
The Positive and Negative Affect Scale (PANAS; Watson, Clark, & Tellegen,
1988; Appendix N) is a measure made up of 20 words that describe 10 positive affect
items (e.g., inspired, interested, strong) and 10 negative affect items (e.g., distressed,
nervous, upset). The positive affect (PA) items refer to the extent which people feel
enthusiastic, active, and alert. High PA is characterized by experiencing high energy,
pleasurable engagement, and full concentration. Low PA indicates the presence of
sadness and lethargy. Negative affect (NA) assesses subjective distress and unpleasurable
contact with others. High NA includes feelings of contempt, disgust, guilt, fear, anger,
and nervousness. Low NA is characterized by a state of calmness and serenity.
In responding to the PANAS participants are asked to rate each of the emotion
words as they were presented and indicate to what extent they feel "this way.
and "in general" or "on the average." A 5-point scale with values ranging from "very
slightly" or "not at all" to "extremely" is used by respondents to rate the emotion words.
The PANAS is brief, easily administered, and has demonstrated reliability and
For example, in several different studies students were given different
temporal instructions when they rated the 20 PANAS emotion words and coefficient
alphas ranged between .86'and .90 for PA and .84 and .87 for NA. In these studies the
reliability of the scales was unaffected by the time instructions given to participants
(Watson, Clark, Tellegen, 1988). A non-student sample of participants was asked to rate
each of the affective items and to indicate how they had been feeling over the past few
weeks. Internal consistency reliabilities ranged between .86 and .87
reliabilities ranged from .47 to .71, with a 2-month time interval between administrations
for all time instructions. Retest stability tended to increase as the rated time frame
lengthened. These results indicate that the PANAS is a stable instrument at appropriate
levels over a 2-month time period. Convergent validity for the scales was demonstrated
by the significantly positive correlations (.76 to .92) between the PANAS and several
other positive and negative affect scales (including Diener & Emmons, 1985; McAdams
& Constantian, 1983; Stone, Hedges, Neale, & Satin, 1985). The PA and NA scales are
largely uncorrelated. PA-NA intercorrelations ranged from -0.12 to -0.23 with student
samples and -0.09 with a non-student sample. These discriminant values indicate that the
PA and NA are quasi-independent dimensions.
The PANAS was included on the Phase 3 questionnaire. However, the data
the data will be utilized in a subsequent study. No results for the PANAS are included in
The Relationship Assessment Scale
The Relationship Assessment Scale (RAS; Hendrick, 1988, Hendrick, Dicke, &
Hendrick, 1998; Appendix O) is a 7-item unifactorial measure of relationship satisfaction
that has been used in both clinical and research settings by Hendrick and others (e.g.,
Fletcher, Simpson, & Thomas, 2000; Flora & Segrin, 2000; Meeks, 1996; Osland, 2001;
Shi, 1999). Examples of RAS items are: "How well does your partner meet your needs?"
and "How often do you wish you hadn't gotten in this relationship?" The 7-item RAS
uses a 5-point Likert scale, with values ranging from 1 to
Higher RAS scores indicate
greater relationship satisfaction. The RAS is scored by reversing participants' scores on
items 4 and 7 and calculating an average score for all items.
The RAS has demonstrated both adequate reliability and validity. RAS scores
over 4.0 tend to indicate that partners are non-distressed, whereas scores closer to 3.5 for
males and between 3.0 and
for females tend to indicate greater relationship distress
and relationship dissatisfaction (Hendrick et al., 1998). The RAS also effectively
discriminates between dating couples and couples who have broken up (Hendrick, 1988).
RAS inter-item correlation is .49 and internal consistency reliability has been estimated
to be between .86 and .90 (Hendrick, 1988). To assess test-retest reliability, the RAS was
administered twice to a sample of 65 undergraduate students with a 6-7 week time
interval between administrations. Test-retest reliability for this sample was .85 (Hendrick,
(KMSS, Schumm et al., 1986). In a study of 84 married couples, correlations between the
RAS and the DAS ranged from
for Anglo-American couples, .72 for Bi-cultural
couples, and .64 for Hispanic-American couples (Hendrick, Dicke, & Hendrick, 1998).
Validity was also demonstrated when the RAS and the KMSS were correlated.
Correlations between these two instruments were .74 for females and .64 for males.
The Global Measure of Relationship Satisfaction
The Global Measure of Relationship Satisfaction (GMREL; Lawrance & Byers,
1998; Appendix P) is one of the subscales comprising the Interpersonal Exchange Model
of Sexual Satisfaction Scale (IEMSS; Lawrance & Byers, 1998). The IEMSS is a self-
report instrument made up of three subscales and a checklist of specific sexual rewards
and costs that might occur in a relationship. The GMREL subscale assesses overall
relationship satisfaction using the prompt,
"In general, how would you describe your
overall relationship with your partner?" Participants use a seven-point bipolar scale
anchored by the five dichotomous variables "good-bad," "pleasant-unpleasant,"
and "valuable-worthless." Higher response
scores indicate endorsement of more positive relationship qualities or values (e.g., good,
pleasant, positive, satisfying, and valuable) whereas lower response scores indicate
endorsement of negative relationship qualities.
Reliability and validity for the GMREL was established in a study using a sample
of university students who had been dating and cohabitating for more than one year
(Lawrance & Byers, 1992) and a study using a community sample of married and
(a = .91) and the community sample (a = .96).
Validity of the GMREL is supported by
its significant positive correlation with the Dyadic Adjustment Scale (DAS; Spanier,
1976), (r=.69, p
The Relationship Happiness Scale
The Relationship Happiness Scale (RHS; Fletcher, Fitness, & Blampied, 1990;
Sumer, 1996; Appendix Q) is comprised of six items and was developed specifically to
be used by dating partners. The six items represent perceptions of love, happiness,
general satisfaction, relationship stability, seriousness of problems, and level of
commitment. Examples of RHS items are "I have a good relationship with my partner"
and "My relationship with my partner makes me happy." Respondents use a 7-item Likert
scale to rate each of the six items. Scores range from 1 to 7, with higher scores indicating
greater relationship happiness. In order to score the RHS, ratings for the seven items are
summed. Respondent ratings tend to range from the low 20s to the high 30s (J. Fitness,
personal communication, July 23, 2003). In a sample of 568 students the reliability of the
RHS was .87 and test re-test reliability was .90 across two administrations a couple of
weeks apart (J. Fitness, personal communication, July 23, 2003). In another study, a
sample of 71 undergraduate students who were in long-term dating relationships rated
hypothetical interactive behaviors within relationships that were either positively or
negatively valenced. The partners who were happy in their relationships provided
attributions that were relationship enhancing. The unhappy partners' attributions
maintained their level of distress. In this study the relationship between relationships
convergent validity with the components of Sternberg's (1986) Triangular Love Scale:
intimacy (r =
.82), passion (r =
.72), and commitment (r =
The Perceived Relationship Quality Component Scale
The Perceived Relationship Quality Component Scale (PRQC, Fletcher, Simpson,
& Thomas, 2000; Appendix R) is a six component, 18-item instrument. The scale
measures perceptions of specific evaluative domains in close relationships. These
components are: satisfaction, commitment, intimacy, trust, passion, and love. Each of the
six components is comprised of three items. According to the scale developers, there are
several advantages to utilizing the PRQC, including that three-item subscales are brief,
reliable, and possess high face validity as the scale measures perceptions of specific
evaluative domains in close relationships (Fletcher et al., 2000). The six components are
domain-specific, quasi-independent constructs. Separately, the components form semi-
independent, lower-level factors. Together, the components load onto one second-order
factor that has been labeled global perceived relationship quality. Three examples of
PRQC items are: "How sexually intense is your relationship?" "How dependable is your
partner?" "How much do you cherish your partner?"
In responding to the PRQC, respondents rate each of the items using a 7-point
Likert scale with values ranging from 1 to 7
. Higher scores indicate greater satisfaction,
commitment, intimacy, trust, passion, and love. The model (six first-order factors loading
on one second-order factor) was tested to determine whether the factor structure was
stable and could be replicated with different samples and separate studies. Results of the
results showed that there was excellent replication of the results at every stage of the
confirmatory factor analytic procedure.
Although it is possible to administer the entire 18-item, 6-factor scale, the scale
developers have recommended in some situations using only the six best exemplars for
each of the relationship quality components, instead of the full scale. For example, the
scale developers stated that if researchers want to use self-report measures to assess
relationships evaluations, then the brief scales would be appropriate. However, if a
researcher wants to measure a single construct (e.g., trust) then a scale measuring that
particular construct would be preferable. For the purposes of this study, the short version
of the PRQC (Appendix R) composed of the six best exemplars of the PRQC was
utilized. The short form items are: Item 1 (measuring satisfaction): "How satisfied are
you with your relationship?" "Item 4 (measuring commitment): How committed are you
to your relationship?" "Item
(measuring intimacy): How intimate is your relationship?"
"Item 10 (measuring trust): How much do you trust your partner?" "Iteml3 (measuring
passion): How passionate is your relationship?" and "Item 16 (measuring love): How
much do you love your partner?" In a study using a sample of individuals in long-term
relationships, the PRQC short form demonstrated adequate aggregated internal reliability
.88, item-total rs ranged from .48 to .77). In another study using a sample of
individuals in short term (< 4 weeks) relationships internal reliability results were also
item-total rs ranged from .45 to .80).
Predictions about the Future Scale
their relationships. Participants' ratings on items 1, 2, and 3 reflect their prediction about
what the status of their relationships will be in the future. For example, the items ask
respondents to make predictions about (1) six months from now, (2) one year from now,
and (3) five years from now about whether they and their partner will be a couple or not
be a couple or whether they will be married. Items are rated using a 6-point Likert scale
with values ranging from 1 to 6. Lower values indicate a prediction that the relationship
will terminate, whereas higher scores predict stability and marriage. The fourth item asks
respondents to predict the likelihood that they will be together in one year. The fifth item
assesses the likelihood that the couple will marry in the future. Items 4 and
using 7-point Likert scales with values ranging from 1 to 7. Higher scores reflect more
positive predictions about future outcome. To score the scale all item ratings are summed.
Higher scores reflect predictions of more stable, long-lasting relationships. In one study,
internal consistency reliability for the scale was .90 for females and .91 for males
(Aarestad, 2000). In addition, the total model accounted for 18.3% (F (5,197)
.001) of the variance for females' and 21.8% (F (6, 193)
= 8.96, p < .001) of the variance
for males' predictions about the future of their relationship (Aarestad, 2000).
The Partner-rated Self-disclosure Index and the Self-rated Self-disclosure Index
The Partner-rated Self-disclosure Index (Miller, Berg, & Archer, 1983; Appendix
T) and The Self-rated Self-disclosure Index (Miller, Berg, & Archer, 1983; Appendix U)
are measures that were developed to assess individual's perceptions of how much
partners disclose to a specific target person (Miller et al., 1983; Meeks, Hendrick, &
one's partner to oneself. These scales assess whether respondents' partners talk to them
and whether respondents talk to their partners about ten different subjects that represent
self-disclosure. Examples of Partner-rated Self-disclosure items are: My partner has
talked to me about the following subjects: "Things my partner wouldn't do in public;"
Examples of Self-rated Self-disclosure items are: I have
talked to my partner about the following subjects: "What I like and dislike about myself;"
My worst fears."
Each item is rated on a 5-point Likert scale with values ranging from 1 to
score the scales the ratings for all scale items are summed. Higher scores on both the
scales represent greater self-disclosure (self to partner or partner to self). Coefficient
a for both the Self-rated and Partner-rated Self-disclosure indices have ranged between
.86 and .93 for both men and women.
Variations in coefficient a estimates were
dependent on the target (Miller et al., 1983). In addition, the scale has demonstrated
convergent validity with the classic Jourard Self-Disclosure Questionnaire (JSDQ;
Jourard & Resnick, 1970).
The Mutual Psychological Development Questionnaire: Form A
The Mutual Psychological Development Questionnaire: Form A (MPDQ-Form A;
Genero, Miller, & Surrey, 1992; Appendix V) is an instrument that assesses mutuality in
close relationships. There are two forms for the MPDQ: MPDQ-Form A and MPDQ-
Form B. Each form consists of twenty-two items representing the six key conceptual
elements of mutual interaction described by Miller (1986) and Surrey (1985). These key
because of the bi-directionality of mutuality (e.g., mutuality includes responding to
another person and being responded to by another person). One response set has self-
prompts. For example: When we talk about things that matter to my spouse/partner, I am
likely to ...
"be receptive;" "get impatient;" "try to understand;" "get bored"). The other
set has other-prompts. For example: When we talk about things that matter to me, my
spouse/partner is likely to ...
"pick up on my feelings;" "feel like we're not getting
anywhere;" "show an interest;" get frustrated".
In responding to and rating MPDQ items, a 6-point Likert scale with values
ranging from 1 to 6 is used. Higher scores represent more mutuality in the relationship.
To compute mutuality scores, negative items are reversed and then the average is
calculated by dividing the summed ratings by the total number of items rated.
The reliability and validity of the MPDQ is supported by studies that have looked
at the relationship between spouses, dating partners, and relationships between close
friends. Research suggests that the MPDQ has excellent inter-item reliability, with
a scores ranging from .89 to .92. Construct validity has been demonstrated through
significantly positive correlations between perceptions of mutuality and measures of
social support, relationship satisfaction, and cohesion. In addition, self-reported ratings of
depression have been negatively correlated with relationship mutuality. For the current
study, Form A of the MPDQ was used.
Phase 3 participants also responded to a scale assessing demographics and
Prior to Phase 3 data collection the study's author contacted course instructors and
arranged mutually suitable times to meet students. In addition, times were scheduled
through the Psychology Department's Introductory Psychology Participant Pool. At the
prearranged times, the study's author and a research assistant orally provided participants
with details about the study and garnered written informed consents (Appendix X).
Students were told that the study was about romantic relationships, that participation was
voluntary, that responses to questionnaire items would be anonymous, and that those
agreeing to participate would receive extra credit. Questionnaires (Appendix M) and
scanable answer sheets were provided to participants. Participants were instructed to read
and follow the directions on the questionnaire. Upon completion, participants placed
questionnaires and answer sheets into a box and signed a form, which was separated from
questionnaires and answer sheets, indicating that they had participated in the study so that
extra credit could be provided to them. Before leaving, participants were given a list of
couples' counseling resource information (Appendix D), a debriefing statement
(Appendix Y), thanked, and dismissed.
Means and standard deviations were calculated for the Love Crushing Behavior
Scale (LCBS; Appendix J), the Relationship Assessment Scale (RAS; Appendix 0), the
Global Measure of Relationship Satisfaction Scale (GMREL; Appendix P), the
Relationship Happiness Scale, (RHS; Appendix Q), the short version of the Perceived
Mutual Psychological Development Questionnaire (MPDQ: Form A; Appendix V) and
are reported in the next chapter.
Construct validity of the LCBS was assessed by examining the relationship
between the LCBS and relationship constructs. The relationship constructs were
measured using the scales just described. The constructs are: relationship satisfaction,
relationship happiness, relationship quality, relationship mutuality, predictions about the
future of the relationship, and self-disclosure between relationship partners. The value
utilized for the LCBS was initially computed by calculating the absolute value of the
difference between participants' ratings of how unloved they would feel from how often
they reported receiving the love crushing behaviors. Linear regression analyses were
conducted to assess the association between this LCBS discrepancy and the relationship
constructs. After the correlational analyses were conducted and the results showed little
association with the measures of related constructs, a second series of linear regression
analyses was performed to assess the association between the LCBS and the relationship
constructs using a recalculated LCBS score. The recalculated LCBS score was computed
by multiplying participants'
ratings of how unloved they would feel upon receiving a
particular behavior by how often they reported receiving that love crushing behavior and
summing the resulting products.
What follows is a description of how each hypothesis
was tested statistically using both the LCBS discrepancy score and the LCBS multiplier
Hypotheses la and lb predicted that there would be a statistically significant
analysis was performed to evaluate the relationship between the LCBS discrepancy and
relationship satisfaction as measured by the averaged scores on the Relationship
Assessment Scale (RAS; Appendix O). To test Hypothesis lb a Pearson product-moment
correlations coefficient analysis was performed to evaluate the relationship between the
LCBS discrepancy and relationship satisfaction as measured by the summed scores on the
Global Measure of Relationship Satisfaction (GMREL; Appendix P). Hypotheses la and
lb were reevaluated. To retest Hypotheses la and lb Pearson product-moment
correlations coefficient analyses were performed to evaluate the relationships between the
LCBS multiplier score and relationship satisfaction as measured by the summed scores
on both the RAS and the GMREL.
predicted that there would be a statistically significant inverse
relationship between the LCBS and ratings of relationship happiness. To test Hypothesis
a Pearson product-moment correlations coefficient analysis was performed to evaluate
the relationship between the LCBS discrepancy score and relationship happiness as
measured by summed scores on the Relationship Happiness Scale (RHS; Appendix Q).
Hypothesis 2 was retested using a Pearson product-moment correlations coefficient
analysis to evaluate the relationship between the LCBS multiplier score and RHS.
Hypothesis 3 predicted that there would be a statistically significant inverse
relationship between the LCBS discrepancy score and perceived relationship quality. To
test Hypothesis 3 a Pearson product-moment correlations coefficient analysis was
performed to evaluate the relationship between the LCBS discrepancy and the short form
analysis to evaluate the relationship between the LCBS multiplier score and the PRQC
Hypothesis 4 predicted that there would be a statistically significant inverse
relationship between the LCBS discrepancy score and predictions about the future of the
relationship. To test Hypothesis 4 a Pearson product-moment correlations coefficient
analysis was performed to evaluate the relationship between LCBS discrepancy and the
summed scores on the Predictions about the Future Scale (Appendix S). Hypothesis 4
was retested using a Pearson product-moment correlations coefficient analysis to evaluate
the relationship between the LCBS multiplier score and predictions about the future.
Hypotheses 5a and 5b predicted that there would be a statistically significant
inverse relationship between the LCBS discrepancy and self-disclosure. Hypothesis 5a
was tested by performing a Pearson-product moment correlations coefficient analysis to
assess the relationship between the LCBS discrepancy with the summed scores on the
Partner-rated Self-disclosure Index (Appendix T). Hypothesis 5b was tested by
correlating the LCBS discrepancy with the summed scores on the Self-rated Self-
disclosure Index (Appendix U). Hypotheses 5a and 5b were retested using a Pearson
product-moment correlations coefficient analyses to assess the relationship between the
LCBS multiplier with partners'
self-disclosure and self-disclosure.
Hypothesis 6 predicted that there would be a statistically significant inverse
relationship between the LCBS and relationship mutuality. Hypothesis 6 was tested using
a Pearson product moment correlations coefficient analysis to assess the relationship
using a Pearson product moment correlations coefficient analysis to assess the
relationship between the LCBS multiplier score and relationship mutuality.
The correlational analyses described above were all conducted to explore the
association between the LCBS and the six criterion variables. The magnitudes of
associations and statistical significance of each association is described in the next
Phase 3 analyses of the data also included conducting a confirmatory factor
analysis via latent variable modeling to assess the stability of the uni-factorial structure of
the LCBS established during Phase 2 of the study. Goodness of fit was assessed by
examining the significance level of the chi-square and the scores on the comparative fit
index (CFI) and the non-normative fit index (NNFI). The CFI and NNFI are usually
considered to show a good fit when they are .90 or higher (Fletcher, Simpson, & Thomas,
2000; Bentler, 1995; Bentler & Bonnet, 1980). Coefficient a was calculated to assess the
internal consistency of the total scale.
The Love Crushing Behavior Scale (LCBS; Appendix J) was constructed to
identify behaviors that would elicit the feeling that one was unloved by romantic partners
when the identified behaviors were received from romantic partners. This chapter
describes the results of the analyses used to construct the scale. This chapter is divided
into three sections. Each of the three sections describes one of the three phases of the
study. The three phases of the study were: Phase 1, which included item generation;
which included scale item identification and assessment of factor structure; and
Phase 3, which included the assessment of the psychometric properties of the scale.
Results for each phase are presented below.
Phase 1: Item Generation
The primary purpose of the first phase of the study was to generate a list of
negative behavioral items that when provided by one romantic partner and received by
another partner resulted in the receiving partner feeling unloved by the providing partner.
For the first phase, data were collected from 102 participants who responded to the Phase
1 questionnaire (Appendix A) and six (of the 102) participants who were contacted by
phone for a follow up interview. All participants were asked to generate behaviors that
from the list of 561 items and the 135 remaining items were corrected for spelling and
grammar and made gender non-specific. This list was labeled the Preliminary
Undesirable Behavior Scale (PUBS; Appendix E).
Phase 1 participants' also responded to items asking them to rate how satisfied
they were with their relationships. Thirty three percent of the respondents who responded
reported being extremely satisfied with their current relationships, 44% reported being
very satisfied, 18% reported being satisfied, and 5% reported being neither satisfied nor
unsatisfied with their relationships. None of the respondents rated themselves as being
unsatisfied, very unsatisfied, or extremely unsatisfied with their current relationship.
Participant responses indicated that 59% of respondents strongly agreed with the
statement, "I am very loved by my romantic partner," 20% agreed, 14% neither agreed
nor disagreed, 0% disagreed, and 7% strongly disagreed. Participants' responses
indicated that 66% strongly agreed with the statement, "My romantic partner cares a great
deal about me,"
23% agreed with this statement, 4% neither disagreed nor agreed, 0%
disagreed, and 7% strongly disagreed. Participants' responses indicated that 57% strongly
agreed with the statement, "When I think about my romantic partner, I feel loved,
agreed, 5% neither disagreed nor agreed,
4% disagreed, and 7% strongly disagreed.
When participants were asked how much they loved their partners, participant
responses indicated that 64% reported loving their partners very much, 18% reported
loving their partners quite a bit, 9% reported a moderate amount of love toward their
partners, and 9% reported a little bit of love for their partners. None of the respondents
Phase 2: Scale Refinement
There were 231 participants who took part in Phase 2 of the study. These
participants completed a 202-item questionnaire (Appendix F) that included the 135-item
Preliminary Undesirable Behavior Scale (PUBS; Appendix E), which is a list of negative
behaviors generated by Phase 1 participants. In addition, the 39 item Desired Loving
Behavior Scale and additional items assessing such information as relationship
satisfaction and participant demographics were included. The primary purposes of the
second phase of the study were to identify which of the 135 items making up the
Preliminary Undesirable Behavior Scale (PUBS; Appendix E) would be retained to make
up the Love Crushing Behavior Scale (LCBS; Appendix J) and to evaluate the factor
structure of the LCBS. Following is a description of the data analyses.
Exploratory Factor Analysis and Item Selection
Exploratory factor analyses were conducted with the 135-items making up the
PUBS (Appendix E). Exploratory factor analysis has two general functions, which are to
explain and to reduce data (Floyd & Widaman, 1995). One of the ways that data are
explained is to determine whether there are underlying factors, separable dimensions, or
latent variables that exist within a main scale (Floyd & Widaman, 1995). In the case of
this study, exploratory factor analysis was used to discover whether the negative
behaviors that made up the PUBS (Appendix E) could be separated into underlying
subscales or factors. The second use of exploratory factor analysis is to reduce data
(Floyd & Widaman, 1995). Because of the large number of initial items, a first step in
given factor only if that item's loading on the factor had an absolute value of .40 or
greater and the loading was less than an absolute value of .30 on any of the other factors.
The 135 PUBS items (Appendix E) were first factor analyzed using common
factor analysis. A scree test suggested a four-factor solution, with a secondary scree
plateau at eight factors. Unfortunately, neither oblique nor orthogonal four-factor
rotations yielded interpretable factors. Instead, too few items achieved simple structure
across the four factors. The first factor yielded a sufficient number of conceptually related
items, but later factors tended not to. Oblique and orthogonal rotations of three, five, six,
seven, and eight factors showed similar simple structure problems. Because of these
failures to achieve simple structure with a sufficient number of items, the data were
reanalyzed with an unrotated common factor analysis. This approach yielded a first factor
with maximized variance. Items were retained if they achieved simple structure with
respect to that first, unrotated factor and all other unrotated factors with an eigenvalues of
1.0 or greater. In all, 22 items were retained as a result of this procedure, which yielded a
single factor. This single factor structure, though not suggested and indeed not
suggestable by a scree test, is consistent in a larger sense, with the eigenvalues of the
unrotated factor analysis. The first factor produced eigenvalues seven times larger than
the next largest factor, 35.85 vs. 4.92, suggesting a very large difference between the first
and all other factor eigenvalues. The magnitude of this difference may simply reflect that
despite many efforts at rotation, the items really taped a single construct.
The resulting 22-item scale was labeled the Love Crushing Behavior Scale
Descriptive Statistics for the 22-Item Love Crushing Behavior Scale (LCBS)
Item n M SD FL r
Verbally abuse me
Spend little or no time with me
say things that are hurtful to me
Tell me that they do not love me
Talk badly about me to other people
Want to hang out or be with other people
more than they want to be with me
Refuse to spend anytime with my family
Never do anything special for my special
occasions or holidays
Cheat on me
Do not express enough affection toward me
Look at other people longingly
Ignore me in front of others
Not respect my family
Was not honest with me
Bring up ex-love interests
Tell me that they hate me
was on other things when I am around
Not call me often
Tell me that they want to discontinue our
Share information about themselves with others
but not tell me
Not tell me that he/she loves me
Watrh TV while tallkina tin m
u r~ rll nl~ lu
Additional Phase 2 Analyses
Whereas the primary purpose of Phase 2 was to identify items and explore factor
structure of the Love Crushing Behavior Scale (LCBS, Appendix J), statistical analyses
were also performed on data provided by Phase 2 participants, specifically participants'
responses to the Desired Loving Behavior Scale (DLBS, Heesacker & Lawrence, 1994;
Appendix G), a social desirability scale (IDRR; Loving & Agnew, 2001; Appendix H),
and items assessing various relationship variables.
Overall, participants' relationship satisfaction response ratings indicated that 40%
of respondents reported being extremely satisfied with their relationships, 34% reported
being satisfied, 15% reported being neither satisfied nor unsatisfied with their current
relationship, 9% reported being unsatisfied, and 2% reported being extremely unsatisfied
with their current relationship.
Regarding the amount of love felt for their romantic partners, 60% of participants
reported they love their partners very much, 17% reported loving partners "quite a bit,"
12% reported a "moderate amount," 9% reported a "little bit," and 3% indicated they did
not love their partners at all. Means and standard deviations for these items are detailed in
Table 4.2, which follows.
With regard to how loved respondents felt, 53% indicated that they strongly
agreed with the statement,
"I am very loved by my romantic partner," 29% agreed with
the statement, 12% neither agreed nor disagreed, 3% disagreed, and 3% strongly
disagreed. Participants' responses indicated that 62% strongly agreed with the statement,
Means and Standard Deviations for Phase
2 Questionnaire Items
Variable n M SD
Love Crushing Behavior Scale (LCBS)
Desired Loving Behavior Scale (DLBS)
Love for Partner
Sense of Love from Partner
Sense of Caring from Partner
Feel loved when thinking of Partner
Love Mean (mean of previous 3 items)
Social Desirability (IDRR)
Total responding to item. M
= Mean item rating. SD
= Standard deviation.
T 11. l.~1. _t 1 1 rrll rn r II r 1 ,r 1 rn, I 1 '11
were averaged to calculate an overall level of love and caring participants felt from their
partners. This "love mean" was comprised of participants' sense of love from their
partners, sense of caring from their partners, and feeling loved when thinking of their
partners. Means and standard deviations are detailed in Table 4.2.
Pearson product moment correlations were used to explore the relationships
between the 22 Love Crushing Behavior Scale (LCBS; Appendix J) items, social
desirability (IDRR; Loving & Agnew, 2001; Appendix H), the DLBS subscales, and
items assessing various relationship variables (e.g., relationship satisfaction, sense of love
from partners, sense of caring for partners, feeling loved when thinking of partners, and
amount of love for partners). Correlational analyses indicated no significant relationship
between the LCBS and the IDRR Impression Management subscale (r = -0.03, p < .72).
In other words, there was no indication that participants felt a need to enhance their
representations of themselves. This result is not surprising as impression management
scores have been found to decrease when going from public to anonymous research
settings (Loving & Agnew, 2001). The participants in this research study completed
questionnaires individually and anonymously. However, there was a statistically
significant inverse relationship between the IDRR self-deception subscale and the LCBS
items (r = -
p < .001), suggesting that increases in the magnitude of responses on
the LCBS were accompanied by decreases in the level of self-deception demonstrated by
the participants. This finding suggests that participants who held fewer positive illusions
(e.g., unconscious misrepresentations) about themselves were also more likely to rate the
the possible operation of depressive realism. Table 4.3 presents the means and standard
deviations for the IDRR subscales and items.
Statistics of the IDRR Items and Subscales
Note. Social Desirability items were scored on a 7-point scale with higher scores
reflecting higher levels of Relationship Self-Deception (REL-SD) and Relationship
Impression Management (REL-IM).
Total responding to item. M
= Mean item rating. SD
= Standard deviation.
0 an .-n f f~~l fl CAr~ A S' fr.lnC u ar n 4~ ar a~u a. a~ 1r n 4- a~ n -r 1~-. -t ~-4 in a1,
Variable n M SD
Social Desirability 204 4.63 0.87
Relationship Self Deception subscale 202 5.07 1.23
REL-SD Item 1 194 5.78 1.39
REL-SD Item 2 196 3.26 1.92
REL-SD Item 3 199 4.89 1.81
REL-SD Item 4 192 4.51 1.73
REL-SD Item 5 195 5.44 1.53
Relationship Impression Management 202 4.19 1.33
REL-IM Item 1 194 4.25 1.84
REL-IM Item 2 194 3.63 1.81
REL-IM Item 3 193 4.03 1.79
REL-IM Item 4 196 3.27 1.62
REL-IM Item 5 193 3.84 1.71
scores on the LCBS and their reported satisfaction with their relationships (r = 0.20, p
.033), love for their partners (r = -0.30, p <
.0009), sense of love from their partners
(r = -0.27, p < .0028), sense of caring from their partners (r = -0.36, p
< .0001), feelings
of love when thinking of their partners (r = -0.39., p < .0001), and their overall love
mean score, which is the mean of the three latter items, (r = -0.37, p < .0001). In other
words, participants who rated LCBS items as making them feel more unloved also
reported feeling less satisfaction with their current relationship, less love towards their
partners, and less love from their current partners.
The items from the Desired Loving Behavior Scale (DLBS; Heesacker &
Lawrence, 1994; Appendix G) were randomly embedded among the 135 PUBS items
(Appendix E) on the questionnaire. As anticipated, none of the DLBS items were retained
in the final factor structure of the LCBS, though the inclusion of the DLBS on the
questionnaire allowed for additional analyses. Specifically, Pearson product moment
correlation analyses were performed to explore the relationship between the LCBS and
the DLBS. Results of these analyses indicated several statistically significant inverse
correlations between the LCBS score and the DLBS scores. Specifically, the relationships
between the LCBS and the Relationship Support subscale (r = -0.60, p < .001), the DLBS
Scripting subscale (r = -0.62, p
< .001), the DLBS Caring Actions subscale (r = -0.58, p
< .001), and the DLBS total score (r = -0.62, p < .001) were all statistically significant
and inversely related. The relationship between the LCBS and the DLBS Sex subscale
analytic techniques as described in detail above to determine a factor structure and
identify scale items. Twenty two items yielding a single factor were retained. These
items were labeled the Love Crushing Behavior Scale.
Phase 3: Assessment of Psychometric Properties of Scale
Normative and Reliability Data
Responses to a 151-item questionnaire (Appendix M) were used for Phase 3
analyses. The questionnaire included the LCBS and the eight measures previously
described. Means, standard deviations, and ranges are detailed in Table 4.4.
Descriptive Statistics of the Phase 3 Scales
Scales n M SD Min Max
LC 277 1.89 .35 0 4
LCR 274 1.78 .48 1 5
LCDisc 274 0.43 .34 0
LCMult 274 3.72 1.11 1.61 9.12
RAS 245 3.71 .88 1 5
GMREL 244 25.74 9.34 1 36
RHS 244 25.62 8.98 1 37
PRQC 242 5.55 1.24 1 7
PAF 238 13.38 9.61 1 28
PRSD 243 2.39 1.14 1 5
SRSD '242 2.37 1.21 1 5
MPDQ 242 3.50 .68 1 5
Total responding to items. M
= Mean item ratings. SD
= Standard deviation.
Min = Minimal score. Max = Maximum score. LC = Love crush item ratings. LCR =
The purpose of this phase was to assess the psychometric properties of the LCBS.
With regard to internal consistency reliability, the Cronbach's
a coefficient for the LCBS
was .88 suggesting that for this administration with this sample the LCBS possessed
adequate measurement precision. An alpha with a maximum value of .90 is
recommended to avoid redundancy among items (Steiner, 2003). Inter-item correlations
for the scale items are provided in Table 4.5 which follows. Mean inter-item correlations
Item to Total Correlations for the LCBS in Phase 3
Does not tell me that he/she loves me
Refuses to spend anytime with my family .56
Does not call me often .54
Wants to hang out or be with other people more than they want to be with me .53
Never does anything special for my special occasions or holidays .51
Looks at other people longingly .51
Talks badly about me to other people .49
Knowingly says things that are hurtful to me .49
Verbally abuses me .49
Spends little or not time with me .48
Does not respect my family .48
Ignores me in front of others .47
Does not express enough affection toward me .46
Cheats on me .45
Shares information about themselves with others but does not tell me .44
Brings up ex-love interests .43
Tells me that they do not love me .42
Is not honest with me .38
Tells me that they want to discontinue our relationship .37
should ideally be between .15 to .20 for scales measuring broad characteristics and
between .40 to .50 for scales tapping narrower ones (Clark & Watson, 1995).
Construct validity of the Love Crushing Behavior Scale (LCBS) was assessed by
examining the relationship between LCBS and these constructs: relationship satisfaction,
relationship happiness, perceived relationship quality, predictions about the future of the
relationship, self-disclosure, and mutuality. Initially, to evaluate the relationships
between the aforementioned constructs and the LCBS, an LCBS discrepancy score was
calculated and correlated with the scores of each of the instruments utilized to measure
constructs. The LCBS discrepancy score was obtained by calculating the absolute value
of the difference between how loved/unloved respondents rated they reportedly would
feel if LCBS behaviors were received from their partners from the frequency of receipt of
the behaviors. For the most part, the results obtained when the data were analyzed in this
way were unexpectedly non-significant. Therefore, additional evaluations of the data
were conducted to examine further the relationship between LCBS and these related
constructs. Specifically, each of the previously listed variables was re-correlated with an
LCBS multiplier score. The LCBS multiplier score was calculated by multiplying
participants' ratings of how unloved they would feel upon receiving a particular behavior
by how often they reported receiving that love crushing behavior and summing the
Hypothesis 1 predicted that there would be a statistically significant inverse
relationship between the Love Crushing Behavior Scale (LCBS) and relationship
satisfaction. For Hypothesis la relationship satisfaction was measured using the
Relationship Assessment Scale (RAS; Hendrick, 1988, Hendrick, Dicke, & Hendrick,
1998; Appendix 0) and for Hypothesis lb relationship satisfaction was measured using
the Global Measure of Relationship Satisfaction Scale (GMREL; Lawrance & Byers,
1995; Lawrance & Byers, 1998; Appendix P).
To evaluate Hypothesis la a Pearson product-moment correlation coefficient
analysis was performed to evaluate the relationship between the LCBS using the LCBS
discrepancy score and the averaged scores on the Relationship Assessment Scale (RAS;
Hendrick, 1988, Hendrick, Dicke, & Hendrick, 1998; Appendix O). Results demonstrated
a low but statistically significant inverse relationship between the LCBS and relationship
satisfaction (r = -0.13, p
< .0495) when relationship satisfaction was measured using the
Hypothesis lb was initially evaluated using a Pearson product moment correlation
coefficient analysis to correlate the LCBS discrepancy score with relationship satisfaction
as measured by scores on the Global Measure of Relationship Satisfaction (GMREL;
Lawrance & Byers, 1995; Lawrance & Byers, 1998; Appendix P). The relationship
between the LCBS and relationship satisfaction as measured using the GMREL did not
reach significance (r = -0.12, p < .057), although results indicated a relationship between
the variables that fell just short of significance in the direction predicted by the
Hypothesis la was reevaluated using a Pearson product-moment correlation
coefficient to assess the relationship between LCBS using the LCBS multiplier score and
the scores on the RAS. Results indicated that as predicted, there was a statistically
significant inverse relationship between the LCBS and relationship satisfaction (r = -0.42,
p < .0001) as measured by the RAS. Hypothesis lb was reevaluated using a Pearson
product moment correlation coefficient analysis to assess the relationship between the
LCBS using the LCBS multiplier score and relationship satisfaction using the GMREL
scores. As predicted, results demonstrated a statistically significant inverse relationship
between the LCBS and relationship satisfaction as measured by the GMREL (r = -0.41, p
< .0001). Thus, Hypothesis la and lb were supported by the data when the LCBS
multiplier score was used. In other words, increases in how unloved one would feel
and/or increases in the receipt of love crushing behaviors as measured using the LCBS
were associated with decreases in relationship satisfaction as measured by both the RAS
and the GMREL.
Hypothesis 2 predicted that there would be a statistically significant inverse
relationship between the LCBS and ratings of relationship happiness, as measured by the
Relationship Happiness Scale, (RHS; Fletcher, Fitness, & Blampied, 1990; Appendix Q).
Initially, Hypothesis 2 was assessed using a Pearson product moment correlation
coefficient analysis to correlate the LCBS as measured by the LCBS discrepancy score
with relationship happiness as measured by the sum of the RHS scores. Results indicated
that although the LCBS discrepancy score and relationship happiness appeared to be
Hypothesis 2 was reevaluated using a Pearson product moment correlation
coefficient to correlate the LCBS using the LCBS multiplier score with relationship
happiness using the RHS scores. Results indicated that as predicted, there was a
statistically significant inverse relationship between the LCBS as measured using the
LCBS multiplier score and relationship happiness as measured using the scores on the
RHS (r = -0.46, p < .0001). Thus, Hypothesis 2 was supported by the LCBS multiplier
data. In other words, increases in how unloved one feels and/or increases in the receipt of
love crushing behaviors as measured by the LCBS were associated with decreases in
relationship happiness as measured by the RHS.
Hypothesis 3 predicted that there would be a statistically significant inverse
relationship between the LCBS and perceived relationship quality as measured by the
Perceived Relationship Quality Component Scale (PRQC; Fletcher, Simpson, & Thomas,
2000; Appendix R). Hypothesis 3 was assessed using a Pearson product moment
correlation coefficient to correlate the LCBS discrepancy score with the scores on the
short form of the PRQC. Results indicated that the relationship between the LCBS as
measured using the LCBS discrepancy score and perceived relationship quality as
measured using the short form of the PRQC was in the predicted direction although the
relationship did not achieve statistical significance (r = -.12, p
Hypothesis 3 was reassessed using a Pearson product moment correlation
coefficient analysis to assess the relationship between the LCBS using the LCBS
multiplier score and the short form of the PRQC. Results indicated that as predicted, the
Hypothesis 3 was supported by the LCBS multiplier data. In other words, increases in
how unloved one would feel and/or increases in the receipt of love crushing behaviors as
measured using the LCBS were associated with decreases in relationship quality as
measured by the PRQC short form.
Hypothesis 4 predicted that there would be a statistically significant inverse
relationship between the LCBS and predictions about the future of the relationship, as
measured by the Predictions about the Future Scale (Aarestad, 2000; Appendix S).
Hypothesis 4 was assessed using a Pearson product moment correlation coefficient
correlating the LCBS discrepancy score with the sum of the scores on the Predictions
about the Future Scale. Results indicated that the relationship between the two variables
(LCBS discrepancy and the Predictions about the Future Scale) tended in the predicted
direction but the relationship was not statistically significant (r = -.06, p < .369).
Hypothesis 4 was reassessed using a Pearson product moment correlation
coefficient analysis to evaluate the relationship between the LCBS using the LCBS
multiplier score and predictions about the future of the relationship using the sum of
scores on the Predictions about the Future Scale. Results indicated that as predicted, the
relationship between the variables, LCBS and predictions about the future, was
statistically significant in the direction predicted by the hypothesis (r = -.31, p < .0001).
Therefore, Hypothesis 4 was supported by the LCBS multiplier data. In other words,
increases in how unloved one would feel and/or increases in the receipt of love crushing
behaviors as measured using the LCBS were associated with decreases in favorable
Hypothesis 5a and 5b predicted that there would be a statistically significant
inverse relationship between the LCBS and self disclosure. Hypothesis 5a was assessed
by evaluating the relationship between the LCBS and ratings on the Partner-rated Self-
disclosure Index (Miller, Berg, & Archer, 1983; Appendix T), which measures disclosure
from one's partner. Hypothesis 5b was assessed by evaluating the relationship between
the LCBS and ratings on the Self-rated Self-disclosure Index (Miller, Berg, & Archer,
1983; Appendix U), which measures disclosures made to one's partner. To assess
Hypothesis 5a a Pearson product moment correlation coefficient analysis was used to
correlate the LCBS discrepancy score with the sum of the scores provided on the Partner-
rated Self-disclosure Index. Results indicated that the relationship between the variables
was not statistically significant and that the variables were not correlated in the direction
predicted by the hypothesis (r =
.10, p <.112). Hypothesis 5b was assessed using a
Pearson product moment correlation coefficient to look at the relationship between the
LCBS discrepancy score and the sum of the scores on the Self-rated Self-disclosure
Index. Results indicated that the relationship between these two variables was not
statistically significant and that the variables were not related in the direction predicted
by the hypothesis (r = .02, p <
Hypothesis 5a was reassessed using a Pearson product moment correlation
coefficient analysis to correlate the LCBS multiplier score with the summed scores on the
Partner-rated Self-disclosure Index. The results indicated that the relationship between
the two variables was statistically significant; however, these two variables were not
partners would feel and/or in how much they received love crushing behaviors as
measured using the LCBS. Hypothesis 5b was reassessed using a Pearson product
moment correlation to evaluate the relationship between the LCBS using the LCBS
multiplier score and the scores on the Self-rated Self-disclosure Index. Again, results
indicated that the relationship between these two variables was not in the direction
predicted by the hypothesis. In addition, the relationship between LCBS and self-rated
self-disclosure was statistically insignificant (r =
.11, p <.1035). Thus, regardless of the
manner in which the LCBS variable was calculated, Hypotheses 5a and 5b were not
supported by the data.
Hypothesis 6 predicted that there would be a statistically significant inverse
relationship between the LCBS and relationship mutuality. Hypothesis 6 was assessed
using a Pearson product moment correlation coefficient to correlate the LCBS
discrepancy score with scores on the Mutual Psychological Development Questionnaire
(MPDQ: Form A; Appendix V). Results indicated that the relationship between LCBS
discrepancy and relationship mutuality tended in the direction predicted, though the
relationship did not achieve statistical significance (r = -.09, p < .156).
Hypothesis 6 was reassessed using a Pearson product moment correlation
coefficient to evaluate the relationship between the LCBS using the LCBS multiplier
score and scores on the MPDQ. Results indicated that as predicted, the relationship
between these two variables, LCBS and relationship mutuality, was statistically
significant and, as predicted by the hypotheses, the variables were inversely related (r = -
crushing behaviors as measured with the LCBS were associated with decreases in
relationship mutuality as measured with the MDPQ.
Table 4.6 provides inter-item correlations for the Love Crushing Behavior Scale
(LCBS), the Relationship Assessment Scale (RAS), the Global Measure of Relationship
Satisfaction Scale (GMREL), the Relationship Happiness Scale (RHS), the short version
Perceived Relationship Quality Component Scale (PRQC), the Predictions about the
Future Scale, the Partner-rated Self-disclosure Index, the Self-rated Self-disclosure Index,
and the Mutual Psychological Development Questionnaire (MPDQ).
Confirmatory Factor Analysis
In order to assess the stability of the single factor structure across samples, for the
Love Crushing Behavior Scale (LCBS; Appendix J) a confirmatory factor analysis was
conducted. The results of the confirmatory factor analysis failed to confirm the single
factor structure. First, the chi square fit index was not only statistically significant,
< .0001, but it was also larger than double the n (562
> 209 X 2).
The two indicators, significant p value and Xvalue double the degrees of freedom,
indicates that the model does not fit the data. Likewise, Bentler's Comparative Fit Index
and Bentler and Bonett's Normative Fit Index were each smaller than .90 (.77 and .68,
respectively) again suggesting that this one-factor model fails to account for these data.
One important limitation on the use of confirmatory factor analysis and on the use
of these indicators of fit involves the distributions of the scores on items. Nonnormally
distributed items do not perform appropriately in CFAs. Unfortunately, most of the items
o N cag
ae a e o
o 0 c i 3 &
C) 0 004 O )
A 9 1| 4to
II oo \o -: RS
9 0 0 0 r
t 3 S *
o o 0 0 6 o o ,"
Co 0 0O N- 'C *x
2 d 0
0 00 0 i0
lI 0O m o- V N C N
-c 0 N e e n
Q) 0 0 0 0 0 0 0 oc
ra r + 9 00 .e
03 O O CTn \O O O~ M
cr .2 fro
ag + u +
av (> Ou (N ^f CT <-"r
S C9 o -^ ^-'^ c\3 m d '- enC *
.1 a S<
the Coefficient a for this sample was .88, which suggests a high degree of inter-item
correlation, consistent with a single factor structure.
The Love Crushing Behavior Scale
The LCBS items are presented below in Table 4.7.
The columns of the table
LCBS Item Rank Comparisons
Cheats on me
Tells me that they do not love me
Talks badly about me with other people
Verbally abuses me
Tells me that they hate me
Tells me that they want to discontinue our relationship
Knowingly says things that are hurtful to me
Ignores me in front of others
Wants to hang out or be with other people more than they want to be with me
Looks at other people longingly
Spends little or no time with me
Never does anything special for my special occasions or holidays
Is not honest with me
Does not respect my family
Does not express enough affection toward me
Shares information about themselves with others but does not tell me
Brings up ex-love interests
Does not call me often
Does not tell me that he/she love me
Refuses to spend anytime with my family
include rankings for each item. Column 1 rankings are based upon the Phase 2 mean
ratings. Column 2 rankings are based upon the Phase 3 mean ratings. Column C rankings
are based upon the Phase 2 correlations with total scores. Column D rankings are based
upon the Phase 3 correlations with total scores. The items' ranks (i.e.,
connote how unloving the behaviors were rated by participants with 1 = most unloving
behavior, 2 = next most unloving, and so on.
Key Findings/Integration with Existing Research
The main purpose of this study was to develop a love crushing behavior scale. Prior
to LCBS development, it was presumed that a scale of love crushing behaviors would
likely be factorable. One of the reasons for this presumption was that the DLBS is a
multi-factorial instrument. Another reason is that Gottman (1994a; 1994b; 1999), who
has done extensive work with couples on relationships found that the most destructive
relationship behaviors fit into categories. As a result, it was anticipated that the LCBS
would be a multi-factorial instrument and that the emergent factors might complement
the DLBS factors (e.g., caring actions, relationship support, scripting, and sex); or that
the LCBS factors might be similar to Gottman's categories (criticism, defensiveness,
stonewalling, and contempt). However, a very interesting finding from this study was that
the LCBS items did not fall into any factors that made sense and instead, results from the
analyses support a uni-factorial scale.
Despite the fact that the LCBS is made up of only one factor, when the individual
DLBS and LCBS items were reviewed and compared, many items on these two scales
seemed to complement each other. That is, the DLBS caring actions item "Take walks
with me during the day,
" and the DLBS relationship support item "Spend time talking to
The DLBS relationship support item,
"Remember my birthday," seems to complement
"Never does anything special for my special occasions or holidays.
the LCBS item,
DLBS scripting item,
the inverse of the LCBS items,
"Say, "I love you with, my whole heart and soul"," seems to reflect
"Tells me that they hate me," "Tells me that they do not
love me," and "Does not tell me that he/she loves me." The DLBS scripting items,
"I want to be with you forever"," and "Talk about our future together," seem to
reflect the opposite sentiment expressed by the LCBS item,
"Tells me that they want to
discontinue the relationships.
" The DLBS scripting item,
"Say to me,
"I enjoy spending
time with you more than any other person"," seems to be the inverse of the LCBS item,
"Wants to hang out or be with other people more than they want to be with me." Despite
examples of complementary items and the fact that there were statistically significant
inverse correlations between the LCBS and DLBS Caring Actions, Relationship Support,
and Scripting subscales there is no support for a reciprocal LCBS factor structure. In
addition, there did not appear to be any examples of DLBS Sex subscale and LCBS item
complements. Based on the study's results, this was not surprising as there was not a
statistically significant relationship between LCBS and the DLBS Sex subscale.
Gottman (1994a; 1999) identified categories of behaviors in his work with couples.
The types of behaviors that made up Gottman's categories were considered by him to be
the most destructive to relationship quality and to relationship longevity. Gottman's
categories, labeled the "Four Horseman of the Apocalypse," are criticism, defensiveness,
stonewalling, and contempt.
While reviewing the data and comparing Gottman's
when I am around" and "Wants to hang out with other people more than they want to be
might be interpreted as stonewalling behaviors.
"Is not honest with me," and
"Brings up ex-love interests," might be interpreted as behaviors displayed by one who is
behaving defensively. "Talks badly about me to other people"
might be considered a
criticism. In spite of the fact that these inferences could be made, these comparisons may
not be appropriate. However, comparisons between the LCBS items and the contempt
category appear to be more defensible.
Contempt has been described by Gottman (1999, pp. 128-129) as sulfuric acid for
love. Although some may think behaviors from other categories such as criticism are just
as bad as contempt, when Gottman observed partners showing contempt to each other he
considered it to be the single most important sign that the marriage was in trouble
(Gladwell, 2005). According to Gottman, contempt kills love and because he considers it
so detrimental to relationships, when Gottman interacts therapeutically with couples who
are behaving in contemptuous ways, he stops their behavior, identifies the contempt for
the couples, and labels it as "abusive"
(Gottman, 1999). Gottman describes contempt as
qualitatively different from criticism and the other categories because when partners are
contemptuous they are being hierarchical, speaking from a superior plane and
communicating that they are more important than their partners (Gladwell, 2005;
Any behavior with the power to kill love might also be a behavior that evokes the
feeling that one is unloved by one's partner. Indeed, when the contempt category was