Assessment

The Family Journal: Counseling and

Therapy for Couples and Families
The Communication Patterns 18(3) 275-287
ª 2010 SAGE Publications
DOI: 10.1177/1066480710370758
Questionnaire–Short Form: A Review http://tfj.sagepub.com

and Assessment

Ted G. Futris1, Kelly Campbell2, Robert B. Nielsen3, and

Stephanie R. Burwell1

Abstract
The Communication Patterns Questionnaire–Short Form (CPQ-SF) is an 11-item self-assessment of spouses’ perceptions of
marital interactions. A cited reference review of the CPQ-SF literature revealed no formal assessment of its psychometric
properties and that researchers are imprecise in their use, reporting, and referencing of the measure. Toward improving the use
of the CPQ-SF in research and practice, the factor structure and psychometric properties of this scale were examined with data
collected from a diverse sample of 477 married individuals. Three latent constructs were identified: criticize/defend, discuss/avoid,
and positive interaction patterns. Suggestions for a more precise use of the CPQ-SF in research and practice conclude the article.

Keywords
assessment, communication patterns questionnaire–short form, marital communication, conflict, interaction patterns,
demand–withdraw

In effort to understand various marital outcomes, researchers CPQ consists of 35 items in which couple members are
have conducted observational studies of married couples and asked to independently self-report on their typical interaction
their interaction patterns (Gottman & Notarius, 2000). patterns. The CPQ assesses interactions across three time peri-
Although observational methods provide exceptional assess- ods: when an issue or problem arises, during discussions of the
ment of marital interactions, they are costly and time consum- issue or problem, and after discussion of the issue or problem.
ing (Hahlweg, Kaiser, Christensen, Fehm-Wolfsdorf, & Groth, Respondents read descriptions of how conflict might typically
2000) and unnatural in that couples are prompted to engage in a be addressed in their relationship and use a 9-point Likert-type
conflict discussion while being videotaped in a laboratory set- scale to indicate the likelihood of that particular pattern
ting (Eldridge & Christensen 2002; Roberts, 2000). This limits occurring. Each pattern is reflective of either complementary
the ability to capture conflict occurring across different time behavior, where partners exhibit different behaviors (e.g.,
points (e.g., a problem occurring in the morning might be dis- ‘‘Man tries to start a discussion while woman tries to avoid a
cussed in the evening) and multiple settings (e.g., the car, the discussion’’), or symmetrical behavior, where partners
bedroom). As well, couples are not likely to engage in beha- exchange similar behavior (e.g., ‘‘Both members avoid discuss-
viors reflective of withdrawal and avoidance when instructed ing the problem’’).
to discuss a topic for a specified amount of time. Finally, obser- Researchers have conceptualized the CPQ subscales differ-
vational methods rely on researchers’ codings of the interac- ently over time. Initially, Christensen (1988) theoretically
tion, which may not reflect the couple members’ perceptions
of the conflict discussion.
Christensen (1987, 1988; Christensen & Sullaway, 1984) 1
developed the Communication Patterns Questionnaire (CPQ) Department of Child and Family Development, University of Georgia, Athens,
GA, USA
to address the aforementioned limitations of observational 2
Department of Psychology, California State University, San Bernardino, CA,
assessments of couple interactions. Items for the CPQ were USA
3
developed by Christensen and Sullaway (1984). Drawing from Department of Housing and Consumer Economics, University of Georgia,
their extensive clinical experience and research and the work of Athens, GA, USA
Fogarty (1976), Gottman (1979), and Peterson (1983), they
Corresponding Author:
composed an initial list of items representing couple interac- Ted G. Futris, Department of Child and Family Development, University of
tions and communication. These items were then refined Georgia, 225 Hoke Smith Annex. Athens, GA 30602, USA
through a series of empirical studies with couples. The final Email: tfutris@uga.edu

275
Downloaded from tfj.sagepub.com at UNIVERSITY OF SASKATCHEWAN LIBRARY on January 29, 2015
276 The Family Journal: Counseling and Therapy for Couples and Families 18(3)

Table 1. Components of the Communication Patterns Questionnaire–Short Form (CPQ-SF)

Complementary (C) versus

Items Symmetrical (S) Pattern

When issues or problems arise, how likely is it that . . . .

1. Both spouses avoid discussing the problem Mutual avoidance (S)
2. Both spouses try to discuss the problem Mutual discussion (S)
3. Female tries to start a discussion while male tries to avoid a discussion F-discusses/ M-avoids (C)
4. Male tries to start a discussion while female tries to avoid a discussion M-discusses/F-avoids (C)
During a discussion of issues or problems, how likely is it that . . .
5. Both spouses express feelings to each other Mutual expression (S)
6. Both spouses blame, accuse, or criticize each other Mutual blame (S)
7. Both spouses suggest possible solutions and compromises Mutual negotiation (S)
8. Female pressures, nags, or demands while male withdraws, becomes silent, or F-demands/M-withdraws (C)
refuses to discuss the matter further
9. Male pressures, nags, or demands while female withdraws, becomes silent, or M-demand/F-withdraws (C)
refuses to discuss the matter further
10. Female criticizes while male defends himself F-criticizes/M-defends (C)
11. Male criticizes while female defends herself M-criticizes/F-defends (C)

Note. Items 3, 8, and 10 are summed for the female demand/male withdraw subscale; Items 4, 9, and 11 are summed for the male demand/female withdraw sub-
scale; Items 3, 4, 8, 9, 10, and 11 are summed for the total demand/withdraw subscale; and Items 2, 5, and 7 are summed for the overall positive interaction
subscale. Christensen and Heavey (1990) did not indicate how to score Items 1 and 6.

organized 11 of the CPQ’s 35 items into three subscales: communication patterns (i.e., mutual blame, mutual threat, and
(a) demand/withdraw communication (6 items), in which one verbal aggression by the man and by the woman). Overall,
partner initiates discussion, demands, criticizes, or nags, the CPQ and its subscales are used widely in the study of couple
while the other partner avoids or withdraws from discussion; communication and therapy and demonstrate acceptable validity
(b) demand/withdraw roles, which consists of the same six and reliability (Christensen, Eldridge, Catta-Preta, Lim, &
interaction patterns but men’s scores are subtracted from Santagata, 2006; Doss, Thum, Sevier, Atkins, & Christensen,
women’s scores to identify sex-typed patterns of demand/ 2005; Roberts, 2000). Studies have also demonstrated consis-
withdraw; and (c) mutual constructive communication (5 tency between partner self-reports, self-reports, and observer rat-
items), in which both partners contribute to the discussion and ings (e.g., Bodenmann, Kaiser, Hahlweg, & Fehm-Wolfsdorf,
try to solve problems. Subsequently, Noller and White (1990) 1998; Christensen, 1988; Hahlweg et al., 2000).
identified four subscales using all 35 items: (a) destructive
process, which is similar to Christensen’s (1988) demand/
withdraw communication subscale; (b) coercion, which inclu- The CPQ-SF
ded threatening, aggressive, and pressuring/resist interaction Christensen and Heavey (1990) developed a condensed version
patterns; (c) mutuality, which consisted of symmetrical interac- of the CPQ, the CPQ–Short Form (CPQ-SF). The scale asks
tion patterns such as mutual avoidance, discussion, expression, spouses to identify their typical communication patterns for
negotiation, and withdrawal; and (d) post-conflict distress, two of the original three time periods: when an issue or prob-
which included post-conflict interaction patterns in which one lem arises and during discussions of the issue or problem. Items
partner feels guilty or tries to reconcile while the other feels from the CPQ that pertain to these two time periods and that
hurt or withdraws. assess demand/withdraw and positive interactions were
Next, Christensen and Shenk (1991), like Christensen included in the CPQ-SF. As shown in Table 1, the short form
(1988), divided the CPQ into a mutual constructive communi- consists of 11 items, 6 to assess complementary interaction pat-
cation subscale and a demand/withdraw communication sub- terns between spouses and 5 to assess symmetrical interaction
scale, but separated the latter into two scores reflective of patterns. The complementary patterns include one partner dis-
sex-specific patterns (female demand/male withdraw and male cusses while the other avoids, one partner demands while the
demand/female withdraw). They also added the mutual avoid- other withdraws, and one partner criticizes while the other
ance subscale, which consisted of three symmetrical patterns in defends. Each of these patterns is assessed with two items that
which both partners avoid, withdraw, or withhold from discus- present the female and male in alternating roles. The symmetri-
sion. Finally, Heavey, Larson, Zumtobel, and Christensen cal patterns reflect mutual discussion, expression of feelings,
(1996) reconceptualized and examined the psychometric prop- negotiation, avoidance, and blame.
erties of the constructive communication subscale, which con- Christensen and Heavey (1990; Heavey, Layne, &
sisted of the sum of 3 items reflecting positive communication Christensen, 1993) conceptually organized the CPQ-SF into 4
patterns (i.e., mutual discussion, expression, and negotiation) subscales: (a) female demand/male withdraw (sum of Items
being subtracted from the sum of 4 items assessing destructive 3, 8, and 10); (b) male demand/female withdraw (sum of Items

276
Downloaded from tfj.sagepub.com at UNIVERSITY OF SASKATCHEWAN LIBRARY on January 29, 2015
Futris et al. 277

4, 9, and 11); (c) total demand/withdraw (sum of Items 3, 4, and a thorough reading of the articles provides clues to the
8–11); and (d) overall positive interaction (sum of Items 2, 5, methodological path taken by the researchers. Like most studies
and 7). The demand/withdraw subscales of the CPQ-SF consist using the 35-item CPQ measure, all 21 studies examined
of the same items used in the demand/withdraw communi- demand/withdraw communication patterns. As shown in Table
cation subscale of the CPQ. The overall positive interaction 2, 5 studies computed only a total demand/withdraw score
subscale represents three of the five symmetrical interaction (Studies 5, 6, 7, 8, and 11), 13 studies computed only female
patterns from the original mutual constructive communication demand/male withdraw and male demand/female withdraw
subscale (Christensen, 1988). Christensen and Heavey do not scores (Studies 2, 3, 4, 9, 10, 13, 14, 15, 17, 19, and 20), and
indicate how the remaining two symmetrical interaction 3 computed all three scores (Studies 1, 12, and 21). Most of these
patterns in the CPQ-SF (i.e., mutual avoidance and mutual studies adhered to the conceptualized scoring of these subscales;
blame, Items 1 and 6) fit into the subscales or explain why they however, five studies that only examined sex-specific demand
are excluded. To date, no formal assessment of the CPQ-SF’s and withdrawal communication patterns (Studies 3, 9, 10, 13,
factor structure or psychometric properties has been published. and 14) intentionally excluded Items 3 and 4 (see Table 1), albeit
We conducted a cited reference search using the Web of Sci- for different reasons: Heffner et al. (2006) explained that
ence to identify studies citing Christensen and Heavey (1990, they were primarily interested in communication patterns
1993) and Heavey et al. (1993) where the CPQ-SF was origi- during a discussion, whereas Caughlin and colleagues noted that
nally presented. A total of 231 authors cited the article by the term ‘‘discussion’’ in these items ‘‘does not reflect the nega-
Christensen and Heavey (1990), 155 authors cited Heavey tive affect implied by the demanding behaviors inherent in
et al.’s (1993) article, and another 39 cited the book chapter demand/withdrawal’’ (Caughlin & Huston, 2002, p. 100). Over-
by Christensen and Heavey (1993). From these publications, all, across the 21 studies, the demand/withdraw subscales exhib-
authors of 21 studies clearly indicated they had administered ited moderate to high internal consistency (Cronbach’s a ranging
the CPQ-SF to participants who reported on their couple rela- from .50 to .85), with the exception of three of the five studies
tionships (see Table 2). The other articles referenced the that computed 2-item subscale scores (Cronbach’s a ¼ .33 in
CPQ-SF publications but did not administer the scale to parti- Studies 9, 13, and 14).
cipants, administered the full CPQ, or were not clear on Still, our review revealed a fundamental challenge for
whether they used the short or full version of the CPQ. Numer- researchers administering or interpreting the CPQ-SF scoring.
ous researchers who administered the full CPQ to participants, Specifically, of the remaining 5 items, it is unclear why Chris-
or variants of it, reported using only the 6 demand/withdraw tensen and Heavey (1990) included Items 1 (mutual avoidance)
items in their analyses; these 6 demand–withdraw items are and 6 (mutual blame) in the scale but excluded them from scor-
identical in both versions of the scale. All 21 studies used con- ing (see Table 1). As summarized in Table 2, 13 of the 21 stud-
venience samples and, with the exception of five studies (Stud- ies computed a positive interaction score using the three
ies 1, 6, 7, 12, and 16), most used the CPQ-SF with nonclinical mutually constructive communication items and reported rela-
populations. We used these 21 empirical studies to identify tively high internal consistency (Cronbach’s a ranging from .68
what is currently known about the CPQ-SF, including its factor to .91; Studies 1, 5, 6, 7, 8, 11 12, 15, 16, 17, and 18). In addi-
structure and psychometric properties. tion, although subsequent researchers have included the two
In our review, we noticed a number of inconsistencies in the destructive communication items in their administration of the
use and referencing of the CPQ-SF. For example, the CPQ-SF CPQ-SF, only three studies included them in their analyses,
was referred to by different acronyms (Studies 7 and 11), albeit as separate single-item scores (Studies 16, 17, and 18).
described as consisting of 8 (Studies 5 and 7) or 7 (Study 12) As a result, the literature provides little, and often inconsistent,
items rather than 11, and scored with differing Likert-type scale guidance for CPQ-SF users.
anchor points (i.e., 7 points rather than 9; Study 18). The overall
positive interaction subscale was referred to by different names
including mutual constructive communication (Studies 1, 6, 12, Alternate Model
and 16), constructive communication (Studies 7 and 17), posi- We hope to clarify issues about the use and scoring of the
tive communication (Studies 11 and 15), symmetrical positive CPQ-SF by proposing a scale structure derived from Gottman’s
communication (Studies 8 and 20), and mutually integrative extensive research on couple interactions. Whereas CPQ-SF
interaction (Study 18). At times, researchers did not cite users have traditionally conceptualized the scale in terms
the appropriate sources when describing the CPQ-SF and/or of two factors, demand/withdraw and positive interactions,
cited studies to support the CPQ-SF’s validity and reliability Gottman suggests that criticize/defend (i.e., conflict engaging)
that pertained to the CPQ, not the CPQ-SF (Studies 2, 5, 7, 15, patterns should not be conceptualized as a type of demand/
18, and 19). Only five studies (Studies, 1, 6, 8, 11, and 16) withdraw behavior (i.e., conflict avoiding). Gottman (1994)
used the full 11-item CPQ-SF and cited the appropriate sources identified four dysfunctional behavioral processes (i.e., ‘‘The
(i.e., Christensen & Heavey, 1990, 1993; Heavey et al., 1993). Four Horsemen of the Apocalypse,’’ p. 110) that contribute
These inconsistencies in the use and referencing of the to the demise of intimate relationships (in order of least to most
CPQ-SF present challenges for researchers who desire to either detrimental): criticism, contempt, defensiveness, and stonewal-
replicate or place prior research into context, but in most cases, ling. Defensiveness typically occurs in response to criticism

277
Downloaded from tfj.sagepub.com at UNIVERSITY OF SASKATCHEWAN LIBRARY on January 29, 2015
 278
Table 2. Results From Empirical Studies Using the Communication Patterns Questionnaire–Short Form (CPQ-SF)

Cronbach’s a Coefficients

Study Sample and Data Collection Methods Partner 1 Partner 2/Combined

1. Denton & Burleson, 2007 Convenience samples of 120 married individuals (69 .56 P1D/P2W
women) recruited from university, community, and .74 P2D/P1W
psychiatric/therapy clinics; mail-in survey .69 TDW
.83 Pi
2. Vogel, Murphy, Werner-Wilson, Cutrona, & Convenience sample of 72 married couples recruited .75 P1D/P2W .66 P1D/P2W
Seeman, 2007 through a university e-newsletter; in-person written .60 P2D/P1W .70 P2D/P1W
survey
3. Heffner et al., 2006 Convenience sample of 31 married couples recruited .76 P1D/P2Wa
through newspaper ads, senior centers, and partici-
pant referrals; in-person written survey
4. Malis & Roloff, 2006 Convenience sample of 219 individuals (137 women) .82 P1D/P2W
recruited from university classes; in-person written .82 P2D/P1W
survey
5. Troy, Lewis-Smith, & Laurenceau, 2006b Convenience samples of 118 (Study 1) and 109 (Study .58/.68 TDW .58/.68 TDW
2) dating couples at a university; in-person written .72/.80 Pi .72/.68 Pi
survey.
6. Byrne, Carr, & Clark, 2004 Convenience sample of 60 couples recruited from NRc
mental health clinics and the community; mail-in survey
7. Cook, Riggs, Thompson, Coyne, & Sheikh, 2004b Convenience sample of 331 ex-POWs and their .85 TDW
partners recruited through American Ex-Prisoner of .85 Pi
War Association; mail-in survey
8. Kurdek, 2004 Convenience sample of 101 married heterosexual .70 TDW
couples recruited from published marriage licenses .91 Pi
and 111 cohabiting gay/lesbian couples recruited
through advertisements; mail-in survey
9. Caughlin, 2002 Snowball/network sample of 46 married couples .65 P1D/P2Wa .69 P1D/P2Wa
recruited through participant and network referrals; .57 P2D/P1Wa .33 P2D/P1Wa
telephone interview
10. Caughlin & Huston, 2002 Convenience sample of 90 married couples recruited .74 P1D/P2Wa .69 P1D/P2Wa
through public marriage license records; telephone .58 P2D/P1Wa .64 P2D/P1Wa

Downloaded from tfj.sagepub.com at UNIVERSITY OF SASKATCHEWAN LIBRARY on January 29, 2015

interview
11. Kurdek, 2001 Convenience sample of 101 married heterosexual .70 TDW
couples recruited from published marriage licenses .91 Pi
and 222 cohabiting gay/lesbian couples recruited
through advertisements; mail-in survey
12. Byrne & Carr, 2000b Convenience sample of 28 couples recruited from .69 P1D/P2W
general practitioners; mail-in survey .74 P2D/P1W
.70 TDW
.71 Pi
13. Caughlin & Vangelisti, 2000 Snowball/network sample of 57 married couples .67 P1D/P2Wa .68 P1D/P2Wa
recruited through participant and network referrals; .63 P2D/P1Wa .33 P2D/P1Wa
in-person survey

(continued)
 Table 2 (continued)

Cronbach’s a Coefficients

Study Sample and Data Collection Methods Partner 1 Partner 2/Combined

14. Caughlin & Vangelisti, 1999 Snowball/network sample of 57 married couples .67 P1D/P2Wa .68 P1D/P2Wa
recruited through participant and network referrals; .63 P2D/P1Wa .33 P2D/P1Wa
in-person survey.
15. Vogel, Wester, & Heesacker, 1999 Convenience sample of 118 individuals (60 women) .71 P1D/P2W
recruited from university classes; in-person survey .66 P2D/P1W
.87 Pi
16. Holtzworth-Munroe, Smutzler, & Stuart, 1998 Convenience sample of 119 married men recruited .63 P1D/P2W .
through marital violence treatment programs, and .67 P2D/P1W
community and newspaper advertisements; in- .75 Pi
person survey
17. Kiecolt-Glaser, et al., 1997 Convenience sample of 31 married couples recruited NRd
through newspaper ads, senior centers, and partici-
pant referrals; telephone interview.
18. Kluwer, Heesink, & Van de Vliert, 1997 Convenience sample of 494 couples (75% married; .65 P1D/P2W
25% cohabiting) recruited through midwives (female .58 P2D/P1W
participants were all pregnant); mail-in survey .74 Pi
19. Klinetob & Smith, 1996 Convenience sample of 50 married couples recruited .79 P1D/P2We(i) .71 P1D/P2We(i)
from university classes; in-person survey .77 P1D/P2We(ii) .72 P1D/P2W e(ii)
.57 P2D/P1We(i) .65 P2D/P1We(i)
.74 P2D/P1We(ii) .72 P2D/P1We(ii)
20. Heavey, Layne, & Christensen, 1993 Convenience sample of 29 married couples recruited .71 P1D/P2W
from fliers sent to preschools in the Los Angeles area; .66 P1D/P2W
in-person survey .87 Pi
21. Christensen & Heavey, 1990 Convenience sample of 31 married couples recruited .85 P1D/P2W .71 P1D/P2W

Downloaded from tfj.sagepub.com at UNIVERSITY OF SASKATCHEWAN LIBRARY on January 29, 2015

through medical professionals, newspaper advertise- .50 P2D/P1W .72 P2D/P1W
ments, and local schools; in-person survey. .73 TDW .74 TDW
.78 Pi .73 Pi
Note. Partner 1 ¼ female or self (if couple data not collected); Partner 2/Combined ¼ male or averaged couple report; P1D/P2W ¼ Partner 1 demand and Partner 2 withdraw; P2D/P1W ¼ Partner 2 demand and
Partner 1withdraw; Pi ¼ overall positive interaction or constructive communication; TDW ¼ total demand-withdraw.
a
Based on 2 items from demand/withdraw subscale; excluded Items 1 and 2 (see Table 1). b Shortened version of CPQ-SF with 8 items (Studies 5 and 7) and 7 items (Study 12). c Computed TDW and PI; Cronbach’s a
coefficients were reported as being above .70. d Computed P1D/P2W, P2D/P1W, PI, and negative behaviors (Items 10 and 11). e Each partner completed CPQ-SF twice with respect to (i) one issue identified by the
wife and (ii) one issue identified by the husband.

279
280 The Family Journal: Counseling and Therapy for Couples and Families 18(3)

and involves a denial of responsibility. This pattern serves to Survey Research Center. Respondents were screened to assure
escalate conflict, but is not as destructive as stonewalling, they were 18 years of age or older, currently married, and shar-
which is an emotional, psychological, and/or physical with- ing a residence with their spouse. To balance the number of male
drawal from the interaction. Defensive behaviors therefore keep and female spouses participating in the study, a household-level
partners engaged in communication, whereas stonewalling/with- random selection procedure was used to determine whether the
drawal serves to minimize or terminate an interaction. A couple male or female spouse would complete the survey. Finally, to
who is seeking therapy or near breakup may have interactions ensure that rural respondents were well represented, rural tele-
characterized by more demand/withdraw behaviors, whereas a phone exchanges were oversampled, resulting in 53.5% of
couple who is less distressed may exhibit fewer demand/with- respondents who lived in nonmetropolitan areas.
draw and more criticize/defend behaviors. Based on Gottman’s
work, we believe it is appropriate to conceptualize the
CPQ-SF according to three, rather than two factors: criticize/ Sample
defend, demand/withdraw, and positive interactions. Of the 517 married respondents, 477 provided complete data
Discrepancies in the use and referencing of the CPQ-SF are on the CPQ-SF and were included in the analyses reported
likely due to the lack of formal, published research on the here. The sample ranged in age from 18 to 85 years (M ¼
assessment’s psychometric properties. Therefore, the goals of 50.5 years, SD ¼ 15.2 years) and 60% were female. Eighty
this study are threefold. Our first goal is to critically examine percent were Caucasian, 17% were African American, and
the factor structure of the CPQ-SF as originally conceptualized. 3% classified themselves as ‘‘Other.’’ Most of the sample
Although previous research on the factor structure of the CPQ (67%) had completed at least some college or had one or more
is limited (Noller & White, 1990), similar analyses with the college degrees, 28% had only completed high school or had a
CPQ-SF do not exist, and CPQ-SF (and CPQ) users have been GED, and 5% had less than a high school diploma. The major-
primarily dependent on the original conceptualization of the ity of respondents (62%) were in first-time marriages with the
6-item demand/withdraw subscales. Thus, the current study remainder (38%) in a marriage in which one or both partners
examines both the two-factor structure originally concept- had been previously married. The duration of participants’
ualized by Christensen and Heavey (1990) consisting of current marriage ranged from <1 year to 66 years (M ¼
demand/withdraw (complementary interaction patterns) and 22.9 years, SD ¼ 16.2 years).
overall positive interactions (symmetrical interaction patterns),
and the three factor structure later proposed by Christensen and
Shenk (1991) where demand/withdraw is separated into two Measures
factors based on sex-specific patterns. Our second goal is to test CPQ-SF. The CPQ-SF is a condensed version of the Com-
an alternate model in which the positive interaction subscale munication Patterns Questionnaire consisting of 11 items
remains consistent with the original conceptualization of (Christensen & Heavey, 1990, 1993). Individuals were read
Christensen and Heavey, but the demand/withdraw subscale descriptions of interaction patterns over the telephone and used
is divided in two factors to distinguish conflict engaging beha- a 9-point Likert-style scale (1 ¼ very unlikely; 9 ¼ very likely)
viors (i.e., criticize/defend) from conflict avoiding (i.e., to indicate the representativeness of that description for the
demand/withdraw) behaviors. This distinction may be useful conflict and communication patterns in their relationship. As
to researchers and clinicians interested in understanding the described more fully in the results section, items corresponding
level and severity of negativity present in a couple’s interaction to each of originally conceptualized and alternate subscales
patterns, given that withdrawal or stonewalling is the most were assessed: (a) male demand/female withdraw (Items 4, 9,
severe behavior in terms of detriment to the relationship and 10); (b) female demand/male withdraw (Items 3, 8, and
(Gottman, 1994). The final goal of this study is to formally 11); (c) original total demand/withdraw (Items 3, 4, 8–11);
examine the psychometric properties of the CPQ-SF using a (d) alternate demand/withdraw (Items 1, 3, 4, 8, and 9); (e) cri-
diverse sample of 477 individuals from married and remarried ticize/defend (Items 6, 10, and 11); and (f) positive interaction
households. Our study will extend the existing CPQ-SF (Items 2, 5, and 7). Higher scores on each subscale indicate a
research by recruiting participants through random sampling greater likelihood of using that communication pattern during
methods. Prior studies (see Table 2) have relied on convenience conflict interactions.
or snowball/network sampling methods, which are prone to
external validity challenges. Demographics. Participants provided information on sex,
age, race, ethnicity, education, years married, marital status
(i.e., first marriage for both spouses vs. repeat marriage for
Method respondent and/or spouse), and number of children.
The current data were drawn from a larger study that examined
married individuals’ perceptions of their relationship behaviors Revised Dyadic Adjustment Scale (RDAS). The RDAS (Busby,
and well-being. A total of 517 computer-assisted telephone Crane, Larson, & Christensen, 1995) is a condensed version
interviews were obtained from a random-digit dialed sample of the widely used Dyadic Adjustment Scale (DAS). It consists
of Georgia households conducted by the University of Georgia of 14 items that assess relationship adjustment across three

280
Downloaded from tfj.sagepub.com at UNIVERSITY OF SASKATCHEWAN LIBRARY on January 29, 2015
Futris et al. 281

dimensions: dyadic consensus, satisfaction, and dyadic Model 3. Specifically, Model 4 illustrates the individual path
cohesion. An overall sum score was computed (possible range: and overall model fit differences that occur when Items 8 and
0–69), with higher scores indicating greater relationship quality 9 simultaneously inform the two latent constructs criticize/
(Cronbach’s a coefficient ¼ .78). Consistent with the scoring defend and demand/withdraw, and Model 5 restricts these
of the RDAS (Crane, Middleton, & Bean, 2000), participants 2 items from informing any latent construct.
were divided into two groups: those who scored 48 and above As shown by the individual path coefficients in Table 3, the
were categorized as high in marital adjustment (n ¼ 386, M ¼ direction and magnitude of the specified paths were consistent
55.76, SD ¼ 4.32) and the remaining participants were categor- with our expectations. Path coefficients in each of the models
ized as low in marital adjustment (n ¼ 91; M ¼ 40.40, SD ¼ were positive and significant at a 95% confidence level. Each
7.06). There were no statistically significant differences individual path informed the subscales as operationalized by
between the two groups in terms of the respondents’ age, edu- either Christensen and Heavey (1990) or our hypothesized
cation, marital status, years married, and the presence and total alternatives based on Gottman (1994). However, a comparison
number of children. However, a higher proportion of female of the overall quality of the models offers important insights
respondents were categorized as low (69.2%) versus high into the relative quality of the models. As shown by the multi-
(57.5%) in marital adjustment compared to males (30.8% vs. ple indicators of overall model fit, the alternative 11-item mod-
42.5%, respectively), w2 ¼ 4.20, p ¼ .04, and a higher propor- els that include criticize/defend, demand/withdraw, and
tion of respondents who specified their race as African Amer- positive interaction (Models 3 and 4) best fit the data. Indeed,
ican or ‘‘Other’’ were categorized as low (33.6%) versus high on all indicators of fit, the alternative models provide a better
(16.5%) in marital adjustment compared to Caucasians overall fit of the data than the Christensen and Heavey
(66.3% vs. 83.5%, respectively), w2 ¼ 15.85, p ¼ .001. These demand/withdraw and positive interaction models (Models 1
results are consistent with previous research that found that and 2). For example, although the root mean square error of
relative to men and Caucasian individuals, women and ethnic approximation (RMSEA) values of .075 and .071 for Models
minority populations tend to report lower marital satisfaction 3 and 4, respectively, indicate a modest fit (Byrne, 1998), each
(Adelman, Chadwick, & Baerger, 1996; Whisman, Uebelacker, is better than the poorly fitting .156 and .154 for Models 1 and
& Weinstock, 2004). 2, respectively.
Similar differences in the quality of absolute model fit
indicators were found for the estimated goodness-of-fit index
Results (GFIM1 ¼ .83, GFIM2 ¼ .83, GFIM3 ¼ .95, GFIM4 ¼ .95, and
GFIM5 ¼ .82) and the adjusted goodness-of-fit index that
Confirmatory Factor Analyses accounts for model complexity (AGFIM1 ¼ .74, AGFIM2 ¼
To assess whether the 11 items on the CPQ-SF perform in .74, AGFIM3 ¼ .91, AGFIM4 ¼ .92, and AGFIM5 ¼ .73).
a manner consistent with the two originally conceptualized sub- Furthermore, the comparative fit index (CFI) offers evidence
scales (i.e., demand/withdraw and positive interaction) or a that the alternative Gottman-inspired models offer superior fit
Gottman-inspired three-factor solution (criticize/defend, compared to the original models (CFIM1 ¼ .80, CFIM2 ¼ .80,
demand/withdraw, and positive interaction), five alternative CFIM3 ¼ .96, CFIM4 ¼ .96, and CFIM5 ¼ .80). Finally, esti-
confirmatory factor analysis models were specified using mates of the adequacy of the sample size required to fit the
LISREL 8.80 (Jöreskog & Sörbom, 2006). See Appendix for the alternative models, as indicated by Hoelter’s Critical N, sug-
covariance matrix. As shown in Table 3, Models 1 and 2 reflect gest that the sample size of 477 is adequate given the specified
the demand/withdraw and positive interaction constructs as orig- models, with the alternative models nearly meeting (Model 3)
inally conceptualized by Christensen and Heavey (1990). Model or exceeding (Model 4) the conventional sample size ade-
1 includes the sex-specific demand/withdraw items whereas quacy indicator of a Critical N of 200 or higher (CNM1 ¼
Model 2 estimates a total demand/withdraw factor. Consistent 58.40, CNM2 ¼ 60.21, CNM3 ¼ 196.98, CNM4 ¼ 214.17, and
with Christensen and Heavey, Models 1 and 2 were both speci- CNM5 ¼ 58.65).
fied without Item 1 (mutual avoidance) or 6 (mutual blame). Together, the individual path and model fit indicators sug-
The remaining three models were specified as alternative gest that the individual subscales specified in Models 1 and
three-factor models that included Gottman-inspired (1994) cri- 2, the original Christensen and Heavey demand/withdraw and
ticize/defend (conflict engaging), demand/withdraw (conflict positive interaction models, offer a less compelling fit of the
avoiding), and positive interaction subscales. These three data than what is achieved with Models 3 and 4, our alternative
remaining models reflect our working hypotheses. Specifically, Gottman-inspired models with criticize/defend, demand/with-
Model 3 uses all 11 items in a model with three factors: criti- draw, and positive interaction constructs. In addition, based
cize/defend, demand/withdraw, and positive interaction. Each on these results, the inclusion of all 11 items is advised.
individual item informs only one hypothesized latent construct.
Because the literature reviewed earlier suggested that Items 8
and 9 (the sex-specific demand/withdraw items) may appropri-
Reliability
ately inform both criticize/defend and demand/withdrawal To assess the reliability of the original and alternative
behaviors, Models 4 and 5 provide alternatives to compare with CPQ-SF subscales, Cronbach’s a coefficients were computed.

281
Downloaded from tfj.sagepub.com at UNIVERSITY OF SASKATCHEWAN LIBRARY on January 29, 2015
 282
282
Table 3. Alternative Confirmatory Factor Analysis models (N ¼ 477)
Model 1a Model 2b Model 3c Model 4d Model 5e

Item MDFW FDMW PI TDW PI CD ADW PI CD ADW PI CD ADW PI

1. Mutual avoidance – – – – – 0.95 1.05 1.13

(0.14) (0.15) (0.15)
6.58 7.12 7.38
2. Mutual discussion 1.12 1.12 1.17 1.17 1.13
(0.12) (0.12) (0.12) (0.12) (0.12)
9.23 9.23 9.83 9.76 9.37
3. F-discusses/M-avoids 1.41 1.41 1.66 1.85 2.06
(0.13) (0.13) (0.13) (0.13) (0.15)
10.90 10.83 12.81 13.78 13.77
4. M-discusses/F-avoids 1.31 1.32 1.49 1.58 1.52
(0.11) (0.12) (0.12) (0.12) (0.13)
11.43 11.43 12.91 13.20 11.77
5. Mutual expression 1.23 1.22 1.23 1.24 1.26
(0.12) (0.12) (0.11) (0.11) (0.12)
10.42 10.35 10.71 10.77 10.74
6. Mutual blame – – – – – 1.67 1.68 1.64
(0.11) (0.11) (0.11)
15.21 15.37 14.80
7. Mutual negotiation 1.51 1.51 1.45 1.44 1.46
(0.13) (0.13) (0.13) (0.13) (0.13)
11.36 11.34 11.36 11.36 11.25
8. F-demands/M-withdraws 1.56 1.59 1.76 0.55 1.32 – – –
(0.12) (0.12) (0.12) (0.16) (0.17)
13.27 13.59 15.03 3.45 7.84
9. M-demands/F-withdraws 1.66 1.68 1.83 0.78 1.18 – – –
(0.11) (0.11) (0.11) (0.15) (0.15)
15.02 15.21 16.44 (5.26) 7.66
10. F-criticizes/M-defends 1.99 2.04 2.22 2.22 2.22
(0.11) (0.11) (0.11) (0.11) (0.11)
17.36 18.47 21.01 21.00 20.55
11. M-criticizes/F-defends 2.02 2.04 2.22 2.21 2.23
(0.11) (0.11) (0.10) (0.10) (0.10)
18.79 19.35 21.89 21.85 21.56
Cronbach’s a 0.709 0.658 0.605 0.814 0.605 0.826 0.714 0.605 0.832 0.714 0.605 0.826 0.571 0.605
Model fit indicators

Downloaded from tfj.sagepub.com at UNIVERSITY OF SASKATCHEWAN LIBRARY on January 29, 2015

w2 560.64y 554.39y 150.69y 133.39y 556.20y
df 43 45 41 39 43
w2/df 13.04 12.32 3.68 3.42 12.93
Hoelter’s critical N 58.40 60.21 196.98 214.17 58.65
GFI 0.83 0.83 0.95 0.95 0.82
AGFI 0.74 0.74 0.91 0.92 0.73
RMSEA 0.156 0.154 0.075 0.071 0.158

Note. AGFI ¼ adjusted goodness-of-fit index; GFI ¼ goodness-of-fit index; RMSEA ¼ root mean square error of approximation; F ¼ Female; M ¼ Male. First row is estimated coefficient, second row is standard
error of the estimate, third row is the t value. All path estimates significant at greater than 99% confidence level.
a
Model 1 ¼ 3 factors, 9 items per Christensen and Heavey (1990): (1) M demand/F withdraw, (2) F demand/M withdraw, and (3) positive interaction. b Model 2 ¼ 2 factors, 9 items per Christensen and Heavey
(1990): (1) total demand/withdraw and (2) positive interaction. c Model 3 ¼ 3 factors, all 11 items: (1) criticize/defend, (2) alternate demand/withdraw, and (3) positive interaction. d Model 4 ¼ 3 factors, all 11 items,
Items 8 and 9 cross-loading: (1) criticize/defend, (2) alternate demand/withdraw, and (3) positive interaction. e Model 5 ¼ 3 factors, Items 8 and 9 dropped: (1) criticize/defend, (2) alternate demand/withdraw, and
(3) positive interaction.
y
p < .001.
The Family Journal: Counseling and Therapy for Couples and Families 18(3)
Futris et al. 283

Table 4. Mean (SD) and Intercorrelations for CPQ-SF Subscales and RDAS

CD ADW FDMW MDFW TDW PI RDAS

Criticize/defend (CD) 1.000 0.552 0.736 0.753 0.809 0.266 0.397

Alternate demand/withdraw (ADW)) 1.000 0.839 0.796 0.888 0.422 0.436
F-demand/M-withdraw (FDMW) 1.000 0.697 0.924 0.361 0.412
M-demand/F-withdraw (MDFW) 1.000 0.918 0.318 0.419
Total demand/withdraw (TDW) 1.000 0.369 0.451
Positive interaction (PI) 1.000 0.431
Revised Dyadic Adjustment Scale (RDAS) 1.000
Overall sample mean (SD) 8.56 (6.67) 14.03 (9.14) 8.75 (6.22) 7.92 (6.03) 16.68 (11.28) 23.22 (4.91) 52.83 (7.81)
Low marital adjustment sample mean (SD) 13.11 (7.12) 21.22 (9.70) 13.18 (6.04) 12.24 (6.88) 25.42 (11.03) 19.55 (5.79) 40.40 (7.06)
High marital adjustment sample mean (SD) 7.49 (6.09) 12.33 (8.13) 7.71 (5.79) 6.91 (5.33) 14.62 (10.32) 24.08 (4.25) 55.76 (4.32)

Note. All correlations and F values significant at p < .001. Low marital adjustment (RDAS 47 or lower) sample, n ¼ 91. High marital adjustment (RDAS 48 or higher)
sample, n ¼ 386.

Although the internal consistency of the 3 items making up subscale, F(1,475) ¼ 81.448, p < .001; original demand/with-
the criticize/defend subscale was strong (a ¼ .83), the reliabil- draw subscale, F(1,475) ¼ 78.563, p < .001; female demand/
ity of the alternate demand/withdraw subscale was moderate male withdraw, F(1,475) ¼ 64.563, p < .001; and male
(a ¼ .71). The internal consistency for the 3 items in the pos- demand/female withdraw, F(1,475) ¼ 65.484, p < .001. In con-
itive interaction subscale was .61. Finally, a coefficients for trast, those in the high marital adjustment group reported more
the originally conceptualized male demand/female withdraw positive interactions than those in the low marital adjustment
(Items 4, 9, and 11), female demand/male withdraw (Items group, F(1,475) ¼ 72.106, p < .001.
3, 8, and 10), and total demand/withdraw (Items 3, 4, and
8–11) subscales were a ¼ .71, a ¼ .66, and a ¼ .81, respec-
tively. These coefficients are largely consistent with those Discussion
reported in previous studies (see Table 2). The study of couple interaction patterns remains an important
focus for researchers and clinicians because it predicts marital
outcomes such as intimacy (Cook, Riggs, Thompson, Coyne,
Validity & Sheikh, 2004), satisfaction (Caughlin 2002; Caughlin &
An assessment of the convergent validity of the CPQ-SF was Huston, 2002; Bodenmann et al., 1998; Heavey et al., 1993;
undertaken with the RDAS because the scales measure related Heavey et al., 1996), and dissolution (Gottman & Notarius,
constructs (i.e., marital interaction and marital adjustment). 2000; Gottman, 1994). The 35-item CPQ, the 11-item
First, a Pearson’s correlation coefficient was computed for each CPQ-SF, and other abbreviated versions of the CPQ are widely
of the CPQ-SF subscales and the RDAS. Although the mea- used to assess couple communication and interactions; yet, the
sures assess similar constructs, the instruments do not contain use and referencing of the abbreviated versions and specifi-
overlapping content. Prior research has shown that the full ver- cally the CPQ-SF have been imprecise. Inconsistencies may
sions of the CPQ and DAS are significantly correlated, with have resulted from the absence of empirical testing on the
coefficients ranging from .54 to .78 (Heavey et al., 1996). As scale’s factor structure and psychometric properties. We
shown in the Table 4, the CPQ-SF subscales used here, includ- sought to fill this gap by examining the factor structure, relia-
ing the ‘‘new’’ 3-item criticize/defend and demand/withdraw bility, and validity of the CPQ-SF with data from a large,
subscales (Model 3), are significantly correlated with RDAS diverse, and representative sample of married and remarried
in the expected direction, though the magnitude may be slightly individuals.
lower, with the absolute value of the Pearson coefficients rang- The results of our analyses provide several advancements
ing from .397 to .451. that offer guidance to researchers and clinicians who rely on the
Finally, to confirm that the subscales clearly discriminate CPQ-SF. When the CPQ and subsequent CPQ-SF were origi-
respondents with high (RDAS ¼ 48 or higher; n ¼ 386; range ¼ nally developed, the items were organized into subscales repre-
48–67) and low (RDAS ¼ 47 or lower; n ¼ 91; range ¼ 18–47) senting two underlying factors, demand/withdraw patterns and
marital adjustment, analyses of variance (ANOVAs) were positive interaction patterns (Christensen & Heavey, 1990).
computed to compare group means across the six CPQ-SF sub- Although all researchers have used the same 6 items to assess
scales. Consistent with similar assessments conducted with the the demand/withdraw pattern, whether using the CPQ or
full CPQ (Noller & White, 1990), results in Table 4 indicate CPQ-SF, our analyses revealed that neither the two-factor nor
that the groups significantly discriminated on all 6 subscales. the three-factor structure consisting of these items is supported.
Compared to respondents in the high marital adjustment group, The results support an alternative three-factor solution that
those in the low marital adjustment group reported more use of has not been conceptualized in previous studies using the
criticize/defend, F(1,475) ¼ 58.739, p < .001, and demand/ CPQ or CPQ-SF. Specifically, a distinct criticize/defend
withdraw communication patterns: alternate demand/withdraw factor was confirmed and was comprised of 3 items (2

283
Downloaded from tfj.sagepub.com at UNIVERSITY OF SASKATCHEWAN LIBRARY on January 29, 2015
284 The Family Journal: Counseling and Therapy for Couples and Families 18(3)

complementary and 1 symmetrical interaction pattern). In proposed. Still, congruent with prior research using the
other studies, the criticize/defend pattern is normally sub- CPQ-SF (e.g., Vogel, Murphy, Werner-Wilson, Cutrona, &
sumed within the demand/withdraw subscales (Christensen Seeman, 2007; Troy, Lewis-Smith, & Laurenceau, 2006),
& Heavey, 1990; Holtzworth-Munroe, Smutzler, & Stuart, both the male demand/female withdraw and female
1998). We based our hypothesis on the work of Gottman demand/male withdraw subscales (Model 1) showed accep-
(1994), who categorized couple interaction patterns according table levels of internal consistency (a ¼ .71 and .66, respec-
to varying degrees of dysfunction. Gottman’s research clearly tively). A number of studies have shown that the wife
distinguished patterns of couple engagement (i.e., criticizing demand/husband withdraw pattern is significantly more
and defending) as less severe than patterns of avoidance or common than the husband demand/wife withdraw pattern,
withdrawal (i.e., stonewalling). particularly when couples discuss wives’ issues (Christensen
In future applications, researchers and clinicians may wish & Heavey, 1990; Heavey et al., 1993; Vogel et al., 2007).
to score the CPQ-SF differently, depending on their goals and Future research should continue to explore the capacity of the
contexts. For example, clinicians treating couples who seek CPQ-SF (and CPQ) for effectively measuring the distinctive-
therapy as a final relationship-saving effort (Wolcott, 1986) ness of these interaction patterns.
may find little value in obtaining separate scores for criticize/ The current study also provides continued evidence of the
defend versus demand/withdraw patterns, because couples are construct validity of the original CPQ-SF subscales as well as
likely to score high on all negative communications. Clinicians evidence for the two newly specified subscales, criticize/
working with couples in this context might wish to use the orig- defend and demand/withdraw, based on their associations
inal two-factor structure to assess a couple’s total amount of with relationship quality. Overall, positive interaction was
negative communication (i.e., demand/withdraw behaviors) positively associated with relationship quality, whereas criti-
and then examine sex differences in these patterns. However, cize/defend, and all four versions of demand/withdraw were
using the alternative three-factor structure with proactive cou- negatively associated with relationship quality. The findings
ples in clinical settings may elucidate specific patterns of com- presented here demonstrate that all the CPQ-SF subscales
munication and interaction, which can inform and modify discriminate respondents with low and high marital adjust-
existing models of couple therapy to target the sequences of cri- ment, a pattern consistent with prior research (e.g., Caughlin,
ticize/defend and demand/withdraw patterns. Such a discrimi- 2002, Christensen et al., 2006; Kluwer, Heesink, & Van de
nating use of the CPQ-SF subscale structures can inform Vliert, 1997).
clinicians who work with couples in very different stages of
distress. Finally, depending on their goals, researchers may find
it useful to use the three-factor structure, which distinguishes
Limitations
criticize/defend from demand/withdraw patterns, or to subsume Although these findings provide meaningful direction to the
negative patterns together and score the scale according to pos- future use and adaptation of the CPQ-SF, this study is not with-
itive versus negative interaction patterns. out limitations. Participants in this study were asked to report
Regarding positive interaction, the results of the current on their dyadic relationships, and hence, the data are represen-
study provide evidence supporting Christensen and Heavey’s tative of only one spouse’s perception of the relationship. Pre-
(1990) conceptually constructed 3-item subscale. Across all vious research has found significant correlations between
models, the individual item coefficients for the three symmetri- husbands’ and wives’ reports on the CPQ-SF (Heavey et al.,
cal positive communication patterns (mutual discussion, 1996). Although a complete assessment of the CPQ-SF would
expression and negotiation) were consistent. The moderate include similar assessments of husbands’ and wives’ reports,
association between positive interaction and the other the data in this study preclude an assessment of inter-spouse
CPQ-SF subscales (r ¼ .27 to .42, see Table 4) reinforced agreement. Second, these data were collected as part of a larger
the unique nature of positive interaction patterns relative to effort that specifically sought to understand marital relation-
negative interaction patterns. These findings are supported by ships and thus excluded nonmarried couples (e.g., cohabiting,
prior studies that distinguish positive from negative affective same sex, and dating). Because relationships among nonmar-
behaviors and identify differing roles for each within intimate ried couples are increasingly varied, future research that
relationships (Bradbury & Karney, 2004; Christensen, 1988; explores the factor structure and psychometric features of the
Gottman & Notarius, 2000). The positive interaction subscale CPQ-SF with unmarried couples would contribute an important
was found to be internally consistent, although its a coefficient dimension to this literature. Another limitation of these results
of .61 was weak relative to previous studies. is reflected in the percentage of African Americans (n ¼ 80, or
A final important finding is that our analyses, like that of 17%) and participants of other racial backgrounds (n ¼ 19, or
others (Noller & White, 1990), yielded mixed support for 3%) relative to Caucasians (n ¼ 378, or 80%). Although the
separate constructs for the sex-specific interaction patterns. number and percentage of African Americans was large rela-
Model 1, which grouped all of the male demand/female with- tive to other research using the CPQ-SF, we did not assess the
draw complementary communication pattern items together psychometric properties by race. The CPQ-SF and demand/
and distinctly from the female demand/male withdraw items, withdraw subscales have been validated with cross-cultural
showed weaker fit relative to the alternative models samples (Christensen et al., 2006; Kluwer et al., 1997), but the

284
Downloaded from tfj.sagepub.com at UNIVERSITY OF SASKATCHEWAN LIBRARY on January 29, 2015
Futris et al. 285

Appendix. Communication Patterns Questionnaire–Short Form (CPQ-SF) Item Means, Standard Deviations, and Covariances

M SD 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11.

1. Mutual avoidance 3.048 2.907 8.45

2. Mutual discussion 7.788 2.187 1.77 4.78
3. F-discusses/M-avoids 3.094 2.813 2.49 1.28 7.91
4. M-discusses/F-avoids 2.631 2.500 1.24 1.09 3.21 6.25
5. Mutual expression 7.889 2.089 1.31 1.24 1.67 0.64 4.36
6. Mutual blame 2.862 2.542 1.38 1.03 1.96 1.56 0.84 6.46
7. Mutual negotiation 7.541 2.293 1.13 1.69 1.21 0.93 1.93 1.15 5.26
8. F-demands/M-withdraws 2.738 2.609 1.45 1.28 3.06 2.30 1.38 2.59 1.38 6.81
9. M-demands/F-withdraws 2.516 2.526 1.40 1.62 2.47 2.94 1.05 3.04 1.19 3.28 6.38
10. F-criticizes/M-defends 2.920 2.641 1.32 1.04 2.26 2.02 0.75 3.64 1.02 3.15 2.87 6.97
11. M-criticizes/F-defends 2.778 2.557 0.84 0.91 2.40 2.35 0.95 3.61 1.03 2.56 3.30 4.98 6.54

literature could still benefit from an assessment of the CPQ-SF Declaration of Conflicting Interests
that is specifically designed to explore the applicability of its The authors declared no potential conflicts of interests with respect to
use with more diverse populations. the authorship and/or publication of this article.
Despite these limitations, our sample exhibits numerous
strengths. Notably, our relatively large sample was randomly
Funding
selected and represented residents of a large Southeastern state.
Funding was provided by the University of Georgia Office of the
As a result, many of the problems inherent with small, self-
Vice President for Research and the University of Georgia College
selecting, or purposively selected samples, which often include
of Family and Consumer Sciences. Generous in-kind support was also
monetary compensation to participants, were minimized. provided by the University of Georgia Survey Research Center.
Although the sample was limited to married couples, there was
notable diversity in other relevant areas, including the respon-
dent’s sex, age, education, prior marital history, number of References
years married, and residence in metro/nonmetro communities. Adelman, P. K., Chadwick, K., & Baerger, D. R. (1996). Marital qual-
In these ways, our research benefits from a much more diverse ity of Black and White adults over the life course. Journal of Social
sample of respondents compared to prior work. and Personal Relationships, 13, 361-384.
Bodenmann, G., Kaiser, A., Hahlweg, K., & Fehm-Wolfsdorf, G.
(1998). Communication patterns during marital conflict: A cross-
cultural representation. Personal Relationships, 5, 343-356.
Conclusion Bradbury, T. N., & Karney, B. R. (2004). Understanding and altering
The CPQ-SF has been shown to be a valid measure of the inter- the longitudinal course of marriage. Journal of Marriage and Fam-
personal communication and interactions between married ily, 66, 862-879.
couples, but further refinement of its use will assist researchers Busby, D. M., Crane, D. R., Larson, J. H., & Christensen, A. (1995).
and clinicians. Although we were able to identify a number of A revision of the Dyadic Adjustment Scale for use with dis-
studies that had used the CPQ-SF, the process often required tressed and nondistressed couples: Construct hierarchy and mul-
detective work due to inconsistent scale labeling and incorrect tidimensional scales. Journal of Marital and Family Therapy,
works cited. Still, our review of the literature confirmed exten- 21, 289-308.
sive use of the CPQ-SF and variant short versions of the CPQ Byrne, B. M. (1998). Structural equation modeling with LISREL,
and our formal assessment of the factor structure and psycho- PRELIS, and SIMPLIS: Basic concepts, applications, and
metric properties of the CPQ-SF provide guidance for those programming. Mahwah, NJ: Lawrence Erlbaum Associates.
interested in using this instrument. The brevity of the scale Byrne, M., & Carr, A. (2000). Depression and power in marriage.
makes it easy for clinicians to administer and interpret in their Journal of Family Therapy, 22, 408-427.
practice, and our results suggest that depending on one’s Byrne, M., Carr, A., & Clark, M. (2004). Power in relationships of
research goals the scale may be further condensed. In addition, women with depression. Journal of Family Therapy, 26, 407-429.
depending on the researchers’ interest and theoretical orienta- Caughlin, J. P. (2002). The demand/withdraw pattern of communica-
tion, our findings suggest that the CPQ-SF can provide a means tion as a predictor of marital satisfaction over time: Unresolved
to assess negative and positive affect in couple communication issues and future directions. Human Communication Research,
and interaction in multiple ways. Regardless of how future 28, 49-85.
researchers choose to structure the CPQ-SF and its subscales, Caughlin, J. P., & Huston, T. L. (2002). A contextual analysis of the
greater care must be taken when reporting the composition of association between demand/withdraw and marital satisfaction.
the scales. Personal Relationships, 9, 95-119.

285
Downloaded from tfj.sagepub.com at UNIVERSITY OF SASKATCHEWAN LIBRARY on January 29, 2015
286 The Family Journal: Counseling and Therapy for Couples and Families 18(3)

Caughlin, J. P., & Vangelisti, A. L. (1999). Desire for change in one’s Gottman, J. M. & Notarius, C. I. (2000). Decade review: Observing
partner as a predictor of the demand/withdraw pattern of marital marital interaction. Journal of Marriage and the Family, 62,
communication. Communication Monographs, 66, 66-89. 927-947.
Caughlin, J. P., & Vangelisti, A. L. (2000). An individual difference Hahlweg, K., Kaiser, A., Christensen, A., Fehm-Wolfsdorf, G., &
explanation of why married couples engage in the demand/with- Groth, T. (2000). Self-report and observational assessment of cou-
draw pattern of conflict. Journal of Social Personal Relationships, ples’ conflict: The concordance between the Communication Pat-
17, 523-551. terns Questionnaire and the KPI Observation System. Journal of
Christensen, A. (1987). Detection of conflict patterns in couples. In Marriage and the Family, 62, 61-67.
K. Hahlweg & M. J. Goldstein (Eds.), Understanding major mental Heavey, C. L., Larson, B. M., Zumtobel, D. C., & Christensen, A.
disorders: The contribution of family interaction research (1996). The Communication Patterns Questionnaire: The reliabil-
(pp. 250-265). New York, NY: Family Process Press. ity and validity of a constructive communication subscale. Journal
Christensen, A. (1988). Dysfunctional interaction patterns in couples. of Marriage and the Family, 58, 796-800.
In P. Noller & M. A. Fitzpatrick (Eds.), Perspectives on marital Heavey, C. L., Layne, C., & Christensen, A. (1993). Gender and con-
interaction (pp. 31-52). Philadelphia, PA: Multilingual Matters. flict structure in marital interaction—A replication and extension.
Christensen, A., Eldridge, K., Catta-Preta, A. B., Lim, V. R., & Journal of Consulting and Clinical Psychology, 61, 16-27.
Santagata, R. (2006). Cross-cultural consistency of the demand/ Heffner, K. L., Loving, T. J., Kiecolt-Glaser, J. K., Himawan, L. K.,
withdraw interaction pattern in couples. Journal of Marriage and Glaser, R., & Malarkey, W. B. (2006). Older spouses’ cortisol
Family, 68, 1029-1044. responses to marital conflict: Associations with demand/withdraw
Christensen, A., & Heavey, C. L. (1990). Gender and social structure communication patterns. Journal of Behavioral Medicine, 29,
in the demand-withdraw pattern of marital conflict. Journal of Per- 317-325.
sonality and Social Psychology, 59, 73-81. Holtzworth-Munroe, A., Smutzler, N., & Stuart, G. L. (1998). Demand
Christensen, A., & Heavey, C. L. (1993). Gender differences in mar- and withdraw communication among couples experiencing hus-
ital conflict: The demand/withdraw interaction pattern. In band violence. Journal of Consulting and Clinical Psychology,
S. Oskamp & M. Costanzo (Eds.), Gender issues in contemporary 66, 731-743.
society (pp. 113-141). Newbury Park, CA: SAGE. Jöreskog, K. G., & Sörbom, D. (2006). LISREL 8.8 for Windows
Christensen, A., & Shenk, J. L. (1991). Communication, conflict, and [Computer software]. Lincolnwood, IL: Scientific Software
psychological distance in nondistressed, clinic, and divorcing cou- International.
ples. Journal of Consulting and Clinical Psychology, 59, 458-463. Kiecolt-Glaser, J. K., Glaser, R., Cacioppo, J. T., MacCallum, R. C.,
Christensen, A., & Sullaway, M. (1984). Communications patterns Snydersmith, M., Kim, C., & Malarkey, W. B. (1997). Marital con-
questionnaire. Unpublished questionnaire. Los Angeles: Univer- flict in older adults: Endocrinological and immunological corre-
sity of California. lates. Psychosomatic Medicine, 59, 339-349.
Cook, J. M., Riggs, D. S., Thompson, R., Coyne, J. C., & Sheikh, J. I. Klinetob, N. A., & Smith, D. A. (1996). Demand-withdraw communi-
(2004). Posttraumatic stress disorder and current relationship func- cation in marital interaction: Tests of interspousal contingency and
tioning among World War II ex-prisoners of war. Journal of Fam- gender role hypotheses. Journal of Marriage and the Family, 58,
ily Psychology, 18, 36-45. 945-957.
Crane, D. R., Middletone, K. C., & Bean, R. A. (2002). Establishing Kluwer, E. S., Heesink, J. A. M., & Van De Vliert, E. (1997). The mar-
criterion scores for the Kansas Marital Satisfaction Scale and the ital dynamics of conflict over the division of labor. Journal of Mar-
revised Dyadic Adjustment Scale. American Journal of Family riage and the Family, 59, 635-653.
Therapy, 28, 53-60. Kurdek, L. A. (2001). Differences between heterosexual-nonparent
Denton, W. H., & Burleson, B. R. (2007). The Initiator Style Ques- couples and gay, lesbian, and heterosexual-parent couples. Journal
tionnaire: A scale to assess initiator tendency in couples. Personal of Family Issues, 22, 727-754.
Relationships, 14, 245-268. Kurdek, L. A. (2004). Are gay and lesbian cohabiting couples really
Doss, B. D., Thum, Y. M., Sevier, M., Atkins, D. C., & Christensen, A. different from heterosexual married couples? Journal of Marriage
(2005). Improving relationships: Mechanisms of change in couple and Family, 66, 880-900.
therapy. Journal of Consulting and Clinical Psychology, 73, 624-633. Malis, R. S., & Roloff, M. E. (2006). Demand/withdraw patterns in
Eldridge, K. A., & Christensen, A. (2002). Demand–withdraw com- serial arguments: Implications for well-being. Human Communi-
munication during couple conflict: A review and analysis. In cation Research, 32, 198-216.
P. Noller & J. A. Feeney (Eds.), Understanding marriage: Devel- Noller, P., & White, A. (1990). The validity of the Communication
opments in the study of couple interaction (pp. 289-322). Cam- Patterns Questionnaire. Psychological Assessment, 2, 478-482.
bridge, England: Cambridge University Press. Peterson, D. R. (1983). Conflict. In H. H. Kelley, E. Bersheid, A.
Fogarty, T. F. (1976). Marital crisis. In P. G. Guerin (Ed.), Family therapy: Christensen, J. H. Harvey, T. L. Huston, G. Levinger, . . . D. R.
Theory and practice (pp. 325-334). New York, NY: Gardner Press. Peterson (Eds.), Close relationships (pp. 360–396). New York,
Gottman, J. M. (1979). Empirical investigations of marriage. New NY: W. H. Freeman.
York, NY: Academic Press. Roberts, L. J. (2000). Fire and ice in marital communication: Hostile
Gottman, J. M. (1994). What predicts divorce? Hillsdale, NJ: and distancing behaviors as predictors of marital distress. Journal
Erlbaum. of Marriage and Family, 62, 693-707.

286
Downloaded from tfj.sagepub.com at UNIVERSITY OF SASKATCHEWAN LIBRARY on January 29, 2015
Futris et al. 287

Troy, A. B., Lewis-Smith, J., & Laurenceau, J. P. (2006). Interracial Vogel, D. L., Wester, S. R., & Heesacker, M. (1999). Dating relationships
and intraracial romantic relationships: The search for differences and the demand/withdraw pattern of communication. Sex Roles, 41,
in satisfaction, conflict, and attachment style. Journal of Social and 297-314.
Personal Relationships, 23, 65-80. Whisman, M. A., Uebelacker, L. A., & Weinstock, L. M. (2004).
Vogel, D. L., Murphy, M. J., Werner-Wilson, R. J., Cutrona, C. Psychopathology and marital satisfaction: The importance of
E., & Seeman, J. (2007). Sex differences in the use of demand evaluating both partners. Journal of Consulting and Clinical
and withdraw behavior in marriage: Examining the social Psychology, 72, 830-838.
structure hypothesis. Journal of Counseling Psychology, 54, Wolcott, I. H. (1986). Seeking help for marital problems before separa-
165-177. tion. Australian Journal of Sex, Marriage and Family, 7, 154-164.

287
Downloaded from tfj.sagepub.com at UNIVERSITY OF SASKATCHEWAN LIBRARY on January 29, 2015