The Initiator Style Questionnaire: A scale to assess initiator tendency in couples

24
The Initiator Style Questionnaire: A scale to assess initiator tendency in couples WAYNE H. DENTON AND BRANT R. BURLESON University of Texas Southwestern Medical Center; Purdue University Abstract The tendency of partners in a dyad to initiate or avoid relationship problem discussions has proven to have both theoretical and clinical significance. This tendency is conceptualized here as initiator tendency, which is defined as the propensity to initiate relationship-focused discussions with one’s partner or avoid such discussions. The current paper reports 3 studies summarizing the development of a measure to assess initiator tendency: the Initiator Style Questionnaire (ISQ). The ISQ provides self-report assessments of self and partner’s perceived initiator tendencies. The studies indicate that the 2 parallel 10-item measures exhibit strong unidimensionality, internal consistency, and test–retest reliability, as well as appropriate discriminant validity and good convergent and construct validity. Limitations and applications are discussed. When any two people are married or in an intimate relationship, it is only natural that there will be times when they find themselves having a difference of opinion or being unhappy with their partner’s behavior (e.g., Cummings & Davies, 2002). One important difference that distinguishes couples is the extent to which each of the partners expresses their discontent or keeps these feelings private. We term this difference ‘‘initiator tendency,’’ which we define as an individual’s tendency in a specific intimate relationship to either initi- ate discussions of relationship problems or to avoid discussion of such problems. We assume that initiator tendency becomes a relatively stable individual difference within a particular intimate relationship as that relationship acquires an increasingly predictable form over time; indeed, patterns of initiation and avoid- ance may contribute to the stabilization of the relationship, as manifested in both actual pat- terns of interaction and partners’ interaction expectations for self and other (see Eldridge & Christensen, 2002; Noller, Feeney, Bonnell, & Callan, 1994). Thus, we conceptualize initiator tendency as a relatively stable proclivity to initiate (or avoid initiating) discussions of relationship problems with one’s partner in a specific intimate relationship (such as mar- riage or cohabitation). Persons relatively high in initiator tendency are referred to as ‘‘initia- tors,’’ whereas those relatively low are ‘‘avoiders’’ (Denton, Burleson, Hobbs, Von Stein, & Rodriguez, 2001). Growing research suggests both the theo- retical and clinical significance of initiator tendency as aspects of what we term initiator tendency have been found to be associated with relationship distress (Christensen, 1987; Noller et al., 1994; Sullaway & Christensen, 1983; Walczynski, 1997), domestic violence (Berns, Jacobson, & Gottman, 1999; Holtzworth- Munroe, Smutzler, & Stuart, 1998), substance abuse (Caughlin & Malis, 2005), and depression Wayne H. Denton, Department of Psychiatry, University of Texas Southwestern Medical Center; Brant R. Burleson, Department of Communication, Purdue University. This research was supported in part by R24-MH51552 from the National Institute of Mental Health (principal investigator, Burton V. Reifler). We thank the following people for assistance in data collection: Barbara Hobbs, Christopher Rodriguez, Reem Utterback, and Margaret Von Stein. Correspondence should be addressed to Wayne H. Denton, University of Texas Southwestern Medical Center, Department of Psychiatry, Dallas, TX 75390-9121, e-mail: [email protected]. Personal Relationships , 14 (2007), 245–268. Printed in the United States of America. Copyright Ó 2007 IARR. 1350-4126=07 245

Transcript of The Initiator Style Questionnaire: A scale to assess initiator tendency in couples

The Initiator Style Questionnaire: A scale to assess

initiator tendency in couples

WAYNE H. DENTON AND BRANT R. BURLESON

University of Texas Southwestern Medical Center; Purdue University

AbstractThe tendency of partners in a dyad to initiate or avoid relationship problem discussions has proven to have both

theoretical and clinical significance. This tendency is conceptualized here as initiator tendency, which is defined as

the propensity to initiate relationship-focused discussions with one’s partner or avoid such discussions. The current

paper reports 3 studies summarizing the development of a measure to assess initiator tendency: the Initiator Style

Questionnaire (ISQ). The ISQ provides self-report assessments of self and partner’s perceived initiator tendencies.

The studies indicate that the 2 parallel 10-item measures exhibit strong unidimensionality, internal consistency,

and test–retest reliability, as well as appropriate discriminant validity and good convergent and construct validity.

Limitations and applications are discussed.

When any two people are married or in an

intimate relationship, it is only natural that

there will be times when they find themselves

having a difference of opinion or being

unhappy with their partner’s behavior (e.g.,

Cummings & Davies, 2002). One important

difference that distinguishes couples is the

extent to which each of the partners expresses

their discontent or keeps these feelings private.

We term this difference ‘‘initiator tendency,’’

which we define as an individual’s tendency in

a specific intimate relationship to either initi-

ate discussions of relationship problems or to

avoid discussion of such problems.We assume

that initiator tendency becomes a relatively

stable individual difference within a particular

intimate relationship as that relationship

acquires an increasingly predictable form over

time; indeed, patterns of initiation and avoid-

ance may contribute to the stabilization of the

relationship, as manifested in both actual pat-

terns of interaction and partners’ interaction

expectations for self and other (see Eldridge &

Christensen, 2002; Noller, Feeney, Bonnell, &

Callan, 1994). Thus, we conceptualize initiator

tendency as a relatively stable proclivity to

initiate (or avoid initiating) discussions of

relationship problems with one’s partner in

a specific intimate relationship (such as mar-

riage or cohabitation). Persons relatively high

in initiator tendency are referred to as ‘‘initia-

tors,’’ whereas those relatively low are

‘‘avoiders’’ (Denton, Burleson, Hobbs, Von

Stein, & Rodriguez, 2001).

Growing research suggests both the theo-

retical and clinical significance of initiator

tendency as aspects of what we term initiator

tendency have been found to be associated

with relationship distress (Christensen, 1987;

Noller et al., 1994; Sullaway & Christensen,

1983; Walczynski, 1997), domestic violence

(Berns, Jacobson, & Gottman, 1999; Holtzworth-

Munroe, Smutzler, & Stuart, 1998), substance

abuse (Caughlin &Malis, 2005), and depression

Wayne H. Denton, Department of Psychiatry, University ofTexas Southwestern Medical Center; Brant R. Burleson,Department of Communication, Purdue University.

This research was supported in part by R24-MH51552from the National Institute of Mental Health (principalinvestigator, Burton V. Reifler). We thank the followingpeople for assistance in data collection: Barbara Hobbs,Christopher Rodriguez, Reem Utterback, and MargaretVon Stein.

Correspondence should be addressed to Wayne H.Denton, University of Texas Southwestern Medical Center,Department of Psychiatry, Dallas, TX 75390-9121, e-mail:[email protected].

Personal Relationships, 14 (2007), 245–268. Printed in the United States of America.Copyright � 2007 IARR. 1350-4126=07

245

(Uebelacker, Courtnage, & Whisman, 2003)

among other significant phenomena (e.g.,

Ellis, 1997; Pasch, 1994; Shoham, Rohrbaugh,

Stickle, & Jacob, 1998). There is evidence

that these patterns are manifested in other

cultures as well (e.g., Rehman & Holtzworth-

Munroe, 2006). As we conceptualize it, initia-

tor tendency is related to constructs such as

demand-withdraw communication (Eldridge

& Christensen, 2002) and conflict engagement

and avoidance (Roloff & Ifert, 2000). Initiator

tendency, however, is also distinguishable in

important ways from these related constructs.

The initial section of the present paper de-

lineates the initiator tendency construct and

details its similarities to and differences from

related notions. The remainder of the paper then

reports three studies undertaken to develop and

evaluate a simple, self-report assessment of ini-

tiator tendency.

The Construct of Initiator Tendency

As developed here, initiator tendency focuses

rather narrowly on the perceived likelihood of

undertaking (or avoiding) discussion with

one’s intimate partner about felt problems in

that specific relationship. Thus, initiator ten-

dency references the inclination to initiate (or

avoid) a specific type of interaction (discus-

sion of relationship problems) with a particular

person (one’s relationship partner). As such,

initiator tendency is narrower in scope than

constructs such as Burgoon’s (1976) ‘‘unwill-

ingness to communicate’’ and Swann’s ‘‘blir-

tatiousness’’ (Swann & Rentfrow, 2001). Both

of these measures index more general tenden-

cies to enjoy interpersonal communication,

express one’s viewpoints readily, and inaugu-

rate interactions with a variety of others on

diverse topics. Initiator tendency is also dis-

tinct from the construct of conflict style, most

assessments of which tap an individual’s gen-

eral approach to waging conflict across people,

topics, and settings (Putnam, 2006). As we

conceptualize it, initiator tendency is closely

related to the propensity of an individual to

give voice to or withhold relational complaints

in a specific intimate relationship (Roloff &

Ifert, 2000; Roloff & Solomon, 2002). As

developed here, the concept of initiator ten-

dency applies to serious, committed relation-

ships where patterns of interaction have had

time to develop; we have doubts that this con-

struct would be applicable to casual dating

relationships.

Initiator tendency is conceptualized as a sta-

ble individual difference within the context of

a particular intimate relationship (i.e., it is

a relationship-specific trait or disposition).

For example, an individual may be an avoider

in one relationship and an initiator in the next

one. The extent to which individuals avoid or

initiate in a specific relationship is deter-

mined partially as a result of the commu-

nication style of their partner. A person may

avoid problem discussions with a partner who

is critical and insensitive yet initiate discus-

sions with a partner who is more patient and

understanding. Within a specific relationship,

however, the initiator tendency will tend to

be relatively stable and consistent.

The construct of initiator tendency bears

some relation to the well-known construct of

demand-withdraw communication (Christensen

& Heavey, 1990) but differs from this latter

construct in important respects. Most funda-

mentally, demand-withdraw communication

refers to a behavioral pattern characteristic of

couples, whereas initiator tendency refers to an

internal process in individuals which (like vir-

tually all other traits or dispositions) may, or

may not, be manifested in observable behavior

in any given context. More specifically, most

of the assessments of demand-withdraw com-

munication treat it as a dyadic variable (i.e., as

a property of a couple), whereas we conceptu-

alize initiator tendency as a relationship-

specific individual difference. For example,

the most popular measure of demand-withdraw

communication, the Communication Patterns

Questionnaire (CPQ; Christensen & Heavey;

Christensen & Sullaway, 1984), asks research

participants to indicate how likely it is that they

and their partners exhibit particular patterns of

interaction when discussing problems (e.g.,

‘‘Woman pressures, nags, or demands while

Man withdraws, becomes silent, or refuses to

discuss the matter further’’).

The CPQ and related assessments of

couple interaction patterns have proven utility

(e.g., Christensen & Heavey, 1990; Heavey,

246 W. H. Denton and B. R. Burleson

Layne, & Christensen, 1993; Holtzworth-

Munroe et al., 1998; Klinetob & Smith,

1996) and established reliability and validity

(e.g., Hahlweg, Kaiser, Christensen, Fehm-

Wolfsdorf, & Groth, 2000; Heavey, Larson,

Zumtobel, & Christensen, 1996; Noller &

White, 1990). We believe, however, that there

are research and clinical contexts in which

individual differences in interaction tenden-

cies are of interest. For example, it might be

of importance to determine whether individ-

uals with a particular clinical condition tend

to be more initiating or avoiding in their

relationships than those without the condition.

The CPQ does not allow for such measure-

ments. Eldridge and Christensen (2002) have

noted the potential utility of being able to

‘‘. separate the components of demand-

withdraw’’ (p. 296). To address concerns such

as these requires an instrument that specifi-

cally assesses individual differences in initia-

tor tendency.

Initiator tendency also differs from the

demand-withdraw construct in that it is not

assumed that either initiating or avoiding prob-

lem discussions is intrinsically negative

behavior. Christensen and his colleagues

(Heavey, Christensen, & Malamuth, 1995)

have stated explicitly that the concept of

demanding involves negative engagement,

and the CPQ items tapping ‘‘demand’’ clearly

reflect this (e.g., demand is represented as

pressuring, nagging, and criticizing). Some-

what similarly, withdraw behavior appears to

be conceptualized as a negative behavior in

that it involves silence, a refusal to discuss

a partner’s concern, or an active defense of

oneself.

Although the initiator tendency construct

encompasses the idea that problem discussions

can be initiated or avoided through negative

behaviors, these discussions can also be initi-

ated (or avoided) through more benign—even

prosocial—behaviors. For example, problem

discussions can be initiated through polite

requests, nonabrasive expressions of concern,

or even jokes and gentle teases; likewise, these

discussions can be avoided through a polite

decline to discuss the issue at the present time,

a skillful topic change, or a preemptive apol-

ogy. Moreover, ‘‘withdrawal’’ is a type of

avoiding in response to a partner’s attempt to

initiate discussion but ‘‘avoid’’ might also

apply when individuals have concerns that

they do not voice. In sum, demand-withdraw

communication represents one particular (and

negative) behavioral mode through which

couples conduct problem discussions. In con-

trast, initiator tendency represents a broader

dispositional orientation of individuals with

respect to initiating and avoiding problem

discussions—an orientation that can be exe-

cuted through a variety of behavioral strate-

gies, including those that may regularly have

positive, neutral, or negative outcomes.

Assessment Options for

Initiator Tendency

Information about the tendency to initiate or

avoid problem discussions with one’s partner

can be extracted from several forms of mea-

surement (for a review, see Noller & Feeney,

2004) including observational coding systems

for couple interaction such as theMarital Inter-

action Coding System (Weiss & Summers,

1983), the Rapid Couples Interaction Scoring

System (Krokoff, Gottman, & Hass, 1989),

and the Couples Interaction Rating System

(Heavey, Gill, & Christensen, 1996). The

research paradigm used with these instruments

involves having couples select a problem and

then discuss it for a specified amount of time in

the effort to resolve the issue or concern. Rat-

ings (by expert raters) of initiating and avoid-

ing tendencies in couples’ interactions

generated with these coding systems have

been associated with relationship distress

(for the Couples Interaction Rating System,

Christensen & Heavey, 1990; Heavey et al.,

1993; Noller et al., 1994) and with consider-

ation of or actual divorce (for theMarital Inter-

action Coding System and Rapid Couples

Interaction Scoring System, Gottman, 1993).

A difficulty with inferring ‘‘avoiding’’ from

laboratory discussions is that researchers in-

struct participants to discuss problems and this

may significantly influence the amount or type

of withdraw behavior that typically occurs

(Roberts, 2000). Eldridge and Christensen

(2002) noted that ‘‘. observation of video-

taped interactions may not provide the best

Initiator Style Questionnaire 247

indication of withdrawal or avoidance, as par-

ticipants are not likely to engage in many of

those behaviors, when they are instructed by

an experimenter to discuss a topic for a speci-

fied amount of time. Conflict avoidance may

occur in different ways naturalistically .’’

(p. 312). Additionally, the use of observational

scales requires the investment of considerable

resources into training raters and maintaining

their reliability.

A different approach to assessing initiator

tendency is through interviews. Eldridge and

Christensen (2002) have pointed out that

‘‘Interview-based assessment of marital inter-

action. is underused, and provides an oppor-

tunity for enhanced assessment’’ (p. 312).

Denton et al. (2001) used such an approach

in a study of initiate-avoid communication

and cardiovascular reactivity. Study staff

interviewed couples conjointly using the Com-

munication Patterns Interview (CPI; described

below). Raters coded the videotaped inter-

views according to a scoring manual that

guided the labeling of each individual as either

an initiator or an avoider. This method yielded

complete interrater agreement for a subsample

of 12 couples. These results suggest that a prac-

tical, reliable assessment of initiator tendency

can be obtained through semistructured inter-

views. As with the observer coding methods,

however, the interview method requires a rela-

tively large investment of time and resources,

especially in comparison to a self-report

questionnaire.

There has been some use of self-report

questionnaires that has relevance to the as-

sessment of initiator tendency. Roloff and

Solomon (2002) asked participants to respond

to two items asking ‘‘When your partner does

something that irritates you, how willing are

you to tell him or her?’’ and ‘‘How assertive

are you in arguments with your partner?’’

Responses to these two items were internally

consistent (a ¼ .74) and were positively cor-

related with a report of the number of com-

plaints expressed and negatively correlated

with a report of the number of complaints

unexpressed. Roberts (2000) developed the

Interaction Response Questionnaire to study

withdrawal and avoidance behaviors that are

difficult to study observationally; the fre-

quency with which couples used three differ-

ent types of withdrawing behavior predicted

marital distress. These successful uses of

self-report questionnaires suggest that this

method holds promise as a means of assessing

initiator tendency.

In sum, it appears there is a need for a brief

self-report questionnaire to assess initiator ten-

dency in individuals who are in a relationship.

The remainder of this paper describes the

development and validation of a new instru-

ment, the Initiator Style Questionnaire (ISQ),

which aims to provide convenient, reliable,

and valid assessments of perceived self and

partner initiator tendency. We report three

studies that provide initial evidence of the reli-

ability and validity of the ISQ. Study 1 focused

on item development, reliability assessment,

and initial validation. Study 2 supplied addi-

tional assessments of the reliability and valid-

ity of the ISQ, and Study 3 further validated

the ISQ by comparing its self-report assess-

ments with coder ratings based on a semistruc-

tured interview.

Study 1

The focus of Study 1 was on developing a set

of internally consistent, face-valid items with

which to obtain self-report assessments of self

and partner initiator tendencies. Participants

responded to a preliminary version of the

ISQ that consisted of items describing the ten-

dencies of self and partner with respect to ini-

tiating or avoiding discussions of relationship

concerns. Analyses focused on assessing inter-

item coherence and dimensionality.

Study 1 also provided an initial assessment

of the construct validity of the ISQ through an

evaluation of gender differences in reports of

self and partner initiate-avoid tendencies.

Although more similarities than differences

characterize the problem discussions and

conflict behaviors of men and women in

their intimate relationships (see Canary &

Emmers-Sommer, 1997), considerable research

indicates that the woman-demand, man-

withdraw pattern is more common than

the man-demand, woman-withdraw pattern

(Christensen & Heavey, 1990; Christensen &

Shenk, 1991; see review by Sagrestano,

248 W. H. Denton and B. R. Burleson

Heavey, & Christensen, 2006). More generally,

extensive research suggests that women are

more likely than men to initiate relationship

discussions with their partners and to use more

forceful strategies, whereas men are more likely

than women to avoid or withdraw from con-

flicts with their relationship partners (see

reviews by Cupach & Canary, 1995; Gottman

& Levenson, 1988; Kelly, Fincham, & Beach,

2003). A measure of initiate-avoid tendencies

should reflect this gender difference. Thus, we

hypothesized that both female and male partic-

ipants would see women as more likely than

men to initiate relationship discussions.

Method

Participants

Given the focus of the present study on provid-

ing an initial assessment of the ISQ’s reliability

and validity, we deemed the use of a conve-

nience sample appropriate. Thus, for Study 1

we recruited 151 married adults (73 men and

78 women) who resided in the state of North

Carolina in the United States. We recruited

these participants as individuals from university

employees, church groups, and a retirement

community. We did not collect any demo-

graphic information other than gender.

Procedure

We provided participants the initial version of

the ISQ with an addressed and stamped enve-

lope to return bymail if they chose to participate.

We did not collect data on the response rate.

Instrumentation

The initial version of the ISQ consisted of 28

items, 14 describing the participant’s own ten-

dencies in the context of relationship-focused

discussions and 14 items describing the ten-

dencies of the participant’s partner. We devel-

oped these items by drawing from the CPQ and

the experience of the investigators. To mini-

mize social desirability effects, we did not use

words that might seem pejorative (e.g.,

demand, withdraw, nag, or avoid). Instructions

for the questionnaires indicated that ‘‘we are

interested in how you (your partner) typically

respond to problems in your relationship (i.e.,

problems between you and your partner).’’

Within each set, we phrased an equal num-

ber of items to describe both initiating behav-

iors (e.g., ‘‘I generally feel comfortable

discussing relationship problems’’; ‘‘My part-

ner usually expresses any feelings about our

relationship to me’’) and avoiding behaviors

(e.g., ‘‘I usually keep my feelings about our

relationship private’’; ‘‘My partner usually

keeps feelings about our relationship private’’)

in relationship discussions. We presented each

item to participants on a 9-point Likert-type

scale with the scale poles anchored by strongly

disagree and strongly agree. Prior to analysis,

we reverse scored the avoid behaviors, so that

high scores for each item reflected a greater

tendency to initiate relationship discussions.

Results

We initially subjected the two sets of items

(self-rating and partner rating) to a set of inter-

nal consistency analyses. These analyses

examined, within each item set, interitem cor-

relations, corrected item-to-total correlations,

the squared multiple correlation between each

item and all other items in the set, and the

Cronbach’s alpha for the item set (a measure

of the overall internal consistency of a set of

items) when each item was deleted. These

analyses indicated that four items should be

deleted from each set. The final set of items

is presented in Appendix A.

Next, we subjected each set of 10 items to

an exploratory factor analysis (principal axis

extraction with a varimax or orthogonal rota-

tion) to assess the dimensionality of the items.

The factor analysis on the self item set

extracted two factors with eigenvalues greater

than 1.0. The second eigenvalue, however,

was barely greater than 1.0 (1.04) and was

considerably smaller than the first eigenvalue

(5.16). Moreover, inspection of a scree plot of

eigenvalues clearly suggested that the 10 items

were best described by a single factor. Thus,

we conducted a second factor analysis, forcing

Initiator Style Questionnaire 249

a one-factor solution. All 10 items correlated

acceptably with this factor, with loadings rang-

ing from .59 to .84 (item loadings ranged from

.52 to .83 for men and from .61 to .86 for

women). The internal consistency (Cronbach’s

alpha) for these 10 items was excellent, a¼ .89

(.86 for men and .89 for women). Thus, we

accepted a unidimensional solution for the 10

self items. We found the identity coefficient

(Van de Vijver & Leung, 1997) for the pattern

of item loadings for men and women to be

quite high (e ¼ .984), indicating a high degree

of similarity in the factor matrices for the two

sexes.

The factor analysis on the partner item set

extracted only one factor, which had a large

eigenvalue (6.66) and explained 66.6% of the

variance in the correlations among the 10

items. All 10 items correlated strongly with

this factor, with loadings ranging from .61 to .90

(item loadings ranged from .58 to .87 for

men and from .65 to .92 for women). We found

the internal consistency (Cronbach’s alpha) for

these 10 items to be excellent, a ¼ .94 (.93 for

men and .95 for women). Thus, we accepted

a unidimensional solution for the 10 partner

items.Avery high identity coefficient (e ¼ .995)

indicated great similarity in the factor matrices

for men and women.

We found the participants’ scores on the

two subscales to be positively, but modestly,

correlated (r ¼ .21, p , .05). This suggests

that participant judgments of self and partner

initiator tendencies are largely independent of

one another.

To assess mean differences in ISQ

responses as a function of participant sex

(male vs. female) and role (self vs. partner),

we performed a 2 � 2 mixed-model analysis

of variance (ANOVA) on initiator tendency.

This ANOVA detected the predicted interac-

tion between participant gender and role,

F(1, 149) ¼ 23.87, p , .001, g2 ¼ .14. Means

for this interaction are shown in Figure 1. Con-

sistent with our hypothesis, a planned com-

parison indicated that men rated themselves

as less likely to initiate relationship discus-

sions (M ¼ 6.06, SD ¼ 1.39) than did women

(M ¼ 6.84, SD 1.67), t(149) ¼ 3.11, p , .002,

d ¼ 0.51. Women viewed their partners (part-

ner rating by women) (M ¼ 5.42, SD ¼ 2.28)

as less likely to initiate discussions than men

viewed their partners (partner rating by men)

(M ¼ 6.37, SD ¼ 1.72), t(149) ¼ 2.89, p ,

.004, d ¼ 0.48. Additionally, women viewed

themselves as more likely to initiate relation-

ship discussions than they viewed their part-

ners, t(77) ¼ 5.12, p , .001, d ¼ 0.71.

Similarly, men viewed themselves as less likely

to initiate relationship discussions than they did

their partners, although this difference (d ¼0.20) was not significant, t(72)¼ 1.45, p . .10.

Discussion

In Study 1, we largely achieved the goal of

developing face-valid, internally consistent,

and dimensionally simple sets of items for

assessing self and partner’s initiator tenden-

cies. We developed separate 10-item scales

4

4.5

5

5.5

6

6.5

7

7.5

8

Self Partner

Rated Person

Ra

te

d T

en

de

nc

y to

In

itia

te

D

is

cu

ss

io

ns

MenWomen

Figure 1. Sex differences in ratings of self and partner initiator tendency in Study 1.

250 W. H. Denton and B. R. Burleson

to measure each of these tendencies. The con-

tent of the items composing these two scales is

obviously relevant to the constructs being

assessed; thus, the two scales possess reason-

able face validity. Factor analyses indicate that

the items composing each scale are unidimen-

sional, and strong internal consistency coeffi-

cients are observed for each scale.

Study 1 also sought to provide an initial

assessment of the construct validity of the

ISQ by assessing whether it captured patterns

of gender differences documented in much of

the literature (e.g., Cupach & Canary, 1995;

Kelly et al., 2003; Sagrestano et al., 2006).

As predicted, the ISQ is sensitive to antici-

pated gender differences; women see them-

selves as more likely to initiate problem

discussions than men see themselves, and

women see themselves as more likely to initi-

ate such discussions than they see their part-

ners. Women also view their partners as less

likely to initiate than men view their partners.

Men, however, see themselves and their part-

ners as equally likely to initiate problem dis-

cussions. Consistent with other research on sex

differences in communication and social

behavior (see reviews by Aries, 2006; Canary

& Hause, 1993; Dindia, 2006), the magnitude

of these sex differences is modest, explaining

between 1% and 11% of the variability in ini-

tiator ratings (and averaging 6% of the vari-

ability). Overall, the findings obtained with the

ISQ in the current study are largely consistent

with the results of previous research that indi-

cate women are more likely than men to initi-

ate problem discussions with their intimate

partners, and these findings thus provide some

support for the validity of the ISQ.

Study 2

The purpose of Study 2 was to provide addi-

tional assessments of the reliability and valid-

ity of the ISQ. In Study 2, we aimed to

replicate Study 1 by assessing item dimension-

ality and internal consistency with a new sam-

ple. In addition, in Study 2 we sought to extend

the results of Study 1 by providing an assess-

ment of test–retest reliability.

Study 2 evaluated the convergent, discrim-

inant, and construct validity of the ISQ by

assessing associations of its two subscales with

several demographic, personality, interaction,

and relationship variables. The demographic

variables examined in Study 2 included gen-

der, age, ethnicity, family income, years mar-

ried, number of marriages, and number of

divorces. Although we expected to replicate

the pattern of gender differences observed in

Study 1, we did not expect any of the other

demographic variables to influence ISQ scores.

The personality measures obtained in Study

2 included the Marlowe-Crowne Social Desir-

ability Scale (Crowne & Marlowe, 1964) and

measures of trait argumentativeness (Infante &

Rancer, 1982) and verbal aggressiveness

(Infante & Wigley, 1986). Infante and Wigley

defined verbal aggressiveness as the trait of

attacking the self-concepts of people instead

of, or in addition to, their positions on issues.

This trait would seem inconsistent with the

attribute the ISQ seeks to tap, which is the

tendency to initiate or pursue discussions

about one’s relationship with one’s intimate

partner, presumably for the purpose of improv-

ing that relationship. The tendency, however,

to initiate relationship discussions might well

be associated with trait argumentativeness,

which is conceptualized as the tendency to

present and defend positions on controversial

issues while attempting to refute the contrary

positions taken by others (Infante & Rancer).

The discriminant validity of the ISQ subscales

would be supported by their independence

from the need for social approval and trait ver-

bal aggressiveness and by their positive asso-

ciation with trait argumentativeness. It is

desirable that scores on the ISQ not correlate

too highly with the need for social approval as

an indication of the independence of the two

constructs.

The CPQ (Christensen & Heavey, 1990)

taps perceptions of couple interaction patterns,

whereas the ISQ taps perceptions of individu-

als’ initiator tendencies. We wished, however,

to explore whether scores from the ISQ’s two

subscales can be combined to generate mea-

sures of dyadic patterns conceptually similar

to those obtained by the CPQ. For example,

the absolute value of the difference between

the ISQ subscales provides an index of the

total amount of demand-withdraw (or, more

Initiator Style Questionnaire 251

precisely, initiate-avoid) communication per-

ceived in a relationship. As discussed previ-

ously, we view our notion of initiating (or

avoiding) problem discussions with one’s

intimate partner as a broader construct that

subsumes the narrower construct of demand-

withdraw. If this assumption has merit, then

measures of dyadic patterns generated by com-

bining scores from the ISQ’s subscales should

moderately correlate with indexes obtained

from the CPQ. Thus, to assess the convergent

validity of the ISQ, we examined correlations

between dyadic scores derived from the ISQ

and comparable scores from the CPQ.

The ability to discuss relationship concerns

with ones partner is one sign of a healthy mar-

riage (see Kelly et al., 2003), so participants

who believe that they and their partners can

freely initiate relationship discussions are

likely to be more satisfied with their marriages

than those who feel less able to initiate discus-

sions. Support for this notion would contribute

to the construct validity of the ISQ. In addi-

tion, extensive research has found that marital

satisfaction declines as the perceived amount

of demand-withdraw communication increases

(Bodenmann, Kaiser, Hahlweg, & Fehm-

Wolfsdorf, 1998; Caughlin & Huston, 2002;

Christensen & Heavey, 1990; Noller & White,

1990). Further, some research has found posi-

tive associations between marital satisfaction

and an index of mutual constructive commu-

nication derived from the CPQ (e.g., Heavey,

Larson, et al., 1996). Thus, to further explore

the convergent validity of the ISQ, we exam-

ined whether parallel indexes derived from it

and the CPQ would correlate similarly with

marital satisfaction.

Method

Participants

Given the focus of Study 2 on further assess-

ments of the ISQ’s reliability and validity, we

deemed the use of a convenience sample

appropriate. We thus recruited 120 married

adults (48 men, 69 women, and 3 participants

who did not indicate their gender) as individ-

uals from patients attending a general psychi-

atry clinic and a marriage and family therapy

clinic to be participants in Study 2. Partici-

pants ranged in age from 23 to 70 years

(M ¼ 37.62) and were largely homogeneous

with respect to ethnicity (114 European

Americans, 6 other ethnicities). Median annual

family income was $50,000. Most (n ¼ 104)

had been married only once and had never

been divorced, 14 had been married twice

and divorced once, 1 reported being married

three times and divorced twice, and 1 did not

answer questions about marital history. Years

married for these participants ranged from less

than 1 year to 47 years (M ¼ 11.24 years).

Procedure

We told potential participants that we were

conducting the study to evaluate a new ques-

tionnaire; if interested in participating, we

provided them a questionnaire booklet, infor-

mation sheet, prepaid return mailer, and a post-

card by which they could indicate their

willingness to complete the ISQ a second time

approximately 4 weeks after its initial admin-

istration. Four weeks after receiving the post-

card, we sent participants another copy of the

ISQ with a prepaid return mailer and we asked

them to write the same number they had ini-

tially used. In this way, the initial and follow-

up ISQs could be matched without knowing

who had completed them. We did not collect

data on the return rate.

Instrumentation

Demographics. Participants provided in-

formation on gender, age, ethnicity, family

income, years married, number of marriages,

and number of divorces.

Personality assessments. Participants com-

pleted three individual-difference (personal-

ity) measures: the Marlowe-Crowne Social

Desirability Scale (Crowne & Marlowe, 1964),

the Trait Argumentativeness Scale (Infante &

Rancer, 1982), and the Trait Verbal Aggres-

siveness Scale (Infante & Wigley, 1986). We

utilized a short-form (13-item) version of the

Social Desirability Scale and obtained a reason-

able internal consistency coefficient for this

version of the instrument (a ¼ .74). We used

252 W. H. Denton and B. R. Burleson

two 20-item scales with items presented in 5-

point Likert format to assess trait argumenta-

tiveness (a ¼ .91) and verbal aggressiveness

(a ¼ .89).

Interaction assessments. Participants re-

ported perceptions of their own and their

partners’ interactive behavior using the short

form of the CPQ (Christensen & Heavey,

1990) and the two 10-item ISQ subscales

developed in Study 1 (reliabilities for the latter

are reported below in the Results). We

produced an index from the ISQ for men’s

perceptions of the man-demand, woman-

withdraw pattern by subtracting men’s score

for partner initiator tendency from their own

initiator tendency score (ISQself 2 ISQpartner);

this taps the extent to which men see them-

selves as initiating more than their partners.

We produced an index for women’s percep-

tions of the man-demand, woman-withdraw

by subtracting women’s score for their own

tendency from their partner tendency score

(ISQpartner 2 ISQself); this provides an index

of the extent to which women see their male

partners initiating more than they do. Simi-

larly, we produced an index from the ISQ for

women’s perceptions of the woman-demand,

man-withdraw pattern by subtracting women’s

score for partner tendency from their own ten-

dency (ISQself 2 ISQpartner); we produced an

index for men’s perceptions of the woman-

demand, man-withdraw by subtracting men’s

score for their own tendency from their partner

tendency score (ISQpartner 2 ISQself). We pro-

duced an index for total demand-withdraw

communication in a relationship by calculat-

ing the absolute value of the difference

between the two ISQ subscales (|ISQself 2

ISQpartner|) for all participants. Finally, we

formed an index of mutual constructive com-

munication in a relationship by summing the

two ISQ subscales (ISQself 1 ISQpartner) for all

participants; this index reflects the extent to

which both partners are perceived as initiating

discussions about problems in the relationship.

The short form of the CPQ has been used

extensively in marital interaction research and

is probably the most popular self-report

method for assessing the demand-withdraw

communication pattern. Summed responses

to three questions provide an assessment of

the man-demand, woman-withdraw pattern;

summed responses to parallel versions of these

three questions provide an assessment of the

woman-demand, man-withdraw interaction

pattern. Summed responses to all six of these

questions provide an assessment of total

demand-withdraw communication. Finally,

summed responses to three additional ques-

tions provide an assessment of mutual con-

structive communication (both members try

to discuss the problem, both members express

their feelings to each other, both members sug-

gest possible solutions and compromises). In

the current study, reliabilities (Cronbach’s

alpha) for the CPQ assessments of these

four interaction patterns were man-demand,

woman-withdraw¼ .56, woman-demand, man-

withdraw ¼ .74, total demand-withdraw ¼ .69,

and mutual constructive communication ¼ .83.

Relationship assessment. Norton’s (1983)

Quality of Marriage Index is a six-item scale

that generates a comparatively pure measure

of marital satisfaction in that all of its items

pertain to the participant’s overall evaluation

of the marital relationship (e.g., ‘‘We have

a good relationship’’; ‘‘Our relationship is

strong’’). In the current study, participants

responded to items one through five on a

7-point Likert scale and to item six on a 10-

point Likert scale. Extensive research supports

the validity and reliability of the Quality of

Marriage Index (Norton; Schumm et al.,

1986), and scholars (Bradbury, Fincham, &

Beach, 2000) specifically recommend this

measure as a global assessment of marital

satisfaction. The internal consistency of the

6-item Quality for Marriage Index was excel-

lent, a ¼ .94.

Results

Dimensionality and reliability assessments

Dimensionality and internal consistency.

We assessed the dimensionality of the ISQ

items with Time 1 data. We subjected the

items for the two subscales to a principal axis

factor analysis. The factor analysis on the self

item set extracted only one factor, which had

Initiator Style Questionnaire 253

a large eigenvalue (5.99) and explained 59.9%

of the variance in the correlations among the

10 items. All 10 items correlated acceptably

with this factor, with loadings ranging from

.51 to .86 (.55–.91 for men and .46–.83 for

women). A high identity coefficient (e ¼ .986)

indicated substantial similarity in the pattern

of factor loadings for men and women. The

internal consistency for these 10 items was

excellent, a ¼ .92 (.94 for men and .89 for

women). The factor analysis on the partner

item set also extracted only one factor, which

had a large eigenvalue (7.41) and explained

74.10% of the variance in the correlations

among the 10 items. All 10 items correlated

strongly with this factor, with loadings ranging

from .69 to .92 (.68–.92 for men and .74 to .93

for women). Once more, a very high identity

coefficient (e ¼ .996) indicated substantial

similarity in the pattern of factor loadings for

men and women. The internal consistency for

these 10 items was excellent (a ¼ .96; .95 for

men and .97 for women). Thus, replicating the

results of Study 1, we observed unidimen-

sional and internally consistent item structures

for both self and partner ratings. Participants’

ratings on the two subscales were not signifi-

cantly correlated (r ¼ .15, p , .12).

Test-retest reliability. Of the 119 partici-

pants who completed the ISQ at Time 1, 69

(58%) completed the ISQ at Time 2. We cal-

culated internal consistency coefficients

(Cronbach’s alpha) for each of the two 10-item

subscales at Time 2: for the self item set,a¼ .90

and for the partner item set, a ¼ .91. The test–

retest reliability for the self subscale was r ¼.81, and the test–retest reliability for the partner

subscale was r ¼ .77. These represent good

test–retest reliabilities (Carmines & Zeller,

1979; DeVellis, 2003).

Validity assessments

Gender differences. To evaluate the gender

hypothesis, we conducted a 2 (men vs.

women) � 2 (self vs. partner) mixed-model

ANOVA with repeated measures on the sec-

ond factor. There was a significant inter-

action between participant gender and role,

F(1, 115) ¼ 23.87, p , .001, g2 ¼ .14. Means

for this interaction are shown in Figure 2.

Planned comparisons indicated that men rated

themselves as less likely to initiate relationship

discussions (M ¼ 6.53, SD ¼ 1.74) than did

women (M ¼ 7.28, SD ¼ 1.27), t(115)¼ 2.55,

p , .02, d ¼ 0.50. Women viewed their part-

ners (partner rating by women, M ¼ 5.76,

SD ¼ 2.24) as less likely to initiate discussions

than men viewed their partners (partner rating

by men,M ¼ 6.50, SD ¼ 1.87), t(115) ¼ 1.95,

p ¼ .05, d ¼ 0.36. Further, women viewed

themselves as more likely to initiate relation-

ship discussions than they did their partners,

t(68) ¼ 5.49, p , .001, d ¼ 0.87. Men, how-

ever, did not view themselves as less likely to

initiate relationship discussions than they did

their partners, t(47)¼ 0.09, p . .90, d ¼ 0.00.

These results are largely consistent with the

hypothesized gender differences and directly

replicate the results of Study 1.

Other demographic differences. Other than

gender, none of the demographic variables had

any significant association with ratings for self

initiator tendency (we did not include ethnicity

in these analyses due to the absence of varia-

tion in this variable). Ratings of partner behav-

ior, however, were significantly and negatively

correlated with both age (r ¼ 2.27, p , .01)

and years married (r ¼ 2.21, p , .05). Age

and years married were, unsurprisingly,

strongly correlated (r ¼ .82, p , .001); thus,

controlling for age had the effect of eliminating

the association between ratings of partner

behavior and years married (rpartial ¼ .00).

4

4.5

5

5.5

6

6.5

7

7.5

8

Rated

T

en

den

cy to

In

itiate D

iscu

ssio

ns

Self Partner

Rated Person

MenWomen

Figure 2. Sex differences in ratings of self

and partner initiator tendency in Study 2.

254 W. H. Denton and B. R. Burleson

Personality assessments. Correlational anal-

ysis indicated that the ISQ assessment of self

initiator tendency was negatively associated

with both need for approval (r ¼ 2.20, p ,

.05) and verbal aggressiveness (r ¼ 2.30,

p , .001). Ratings of partner initiator tendency

were not associated with either need for ap-

proval or verbal aggressiveness. Neither ISQ

scale was associated with argumentativeness.

Interaction assessments. We calculated

correlations between the four assessments of

interaction patterns obtained from the CPQ

and parallel indexes generated from the ISQ.

For men, the correlation between the ISQ and

CPQ assessments of the man-demand, woman-

withdraw pattern was r ¼ .48 (p , .001),

whereas for women the correlation between

the two assessments of the man-demand,

woman-withdraw pattern was r ¼ .12 (p .

.30). For women, the correlation between the

ISQ and CPQ assessments of the woman-

demand, man-withdraw pattern was r ¼ .58

(p , .001), whereas for men the correlation

between the two assessments of the woman-

demand, man-withdraw pattern was r ¼ .35

(p , .02). For the entire sample, the correla-

tion between the ISQ and CPQ assessments

of total demand-withdraw communication

was r ¼ .39 (p , .001). Finally, for the entire

sample, the correlation between the ISQ and

CPQ assessments of mutual constructive com-

munication was r ¼ .79 (p , .001). Overall,

these mostly moderate to strong correlations

provide good support for the convergent val-

idity of the ISQ.

Relationship assessment. As predicted,

correlations between the Quality of Marriage

Index and ISQ were significant and positive:

for the Quality of Marriage Index and ISQ self

ratings (r ¼ .35, p , .001), and for the Quality

of Marriage Index and ISQ partner ratings

(r ¼ .36, p , .001). We also computed corre-

lations between the Quality of Marriage Index

and the parallel indexes of total demand-

withdraw and mutual constructive communi-

cation generated by the ISQ and CPQ. As

expected, marital quality was negatively cor-

related with the total demand-withdraw index

generated by each instrument at an approxi-

mately equivalent level; the ISQ total

demand-withdraw correlation with the Quality

of Marriage Index was r ¼ 2.44 (p , .001),

whereas the CPQ total demand-withdraw cor-

relation with the Quality of Marriage Index

was r ¼ 2.47 (p , .001). The correlation

between the Quality of Marriage Index and

the ISQ total demand-withdraw index re-

mained significant even after controlling

for the CPQ total demand-withdraw index

(rp ¼ 2.31, p , .001). Also as expected,

marital quality was positively correlated with

the mutual constructive communication index

generated by each of the two interaction

assessments; the ISQ mutual constructive

communication correlation with the Quality

of Marriage Index was r ¼ .46, p , .001,

whereas the CPQ mutual constructive com-

munication correlation with the Quality of

Marriage Index was r ¼ .57, p , .001. Con-

trary to expectations, when controlling for the

effect of the CPQ mutual constructive commu-

nication index, the correlation between the

Quality of Marriage Index and ISQ mutual

constructive communication index was re-

duced to near zero (rp ¼ .02, p . .50).

Discussion

The demographic, personality, interaction, and

relationship assessments obtained in Study 2

are largely associated in expected ways with

assessments of initiator tendency generated by

the ISQ and thus provide supportive evidence

for the convergent, discriminant, and construct

validity of this instrument. In addition, the ISQ

subscales exhibit strong unidimensionality and

excellent reliability.

Replicating the results of Study 1, the items

for both subscales of the ISQ are found to be

unidimensional and internally consistent, with

alpha coefficients for both subscales exceed-

ing .90. Further replicating the results of Study

1, very similar patterns of factor loadings are

found for both men and women. Both sub-

scales exhibit excellent test–retest reliability,

with these correlations approaching .80 for

partner tendency and exceeding .80 for self-

tendency.

Construct validity of the ISQ is supported

by the expected gender differences and by

Initiator Style Questionnaire 255

associations with marital quality. Replicating

the results of Study 1, and consistent with

research examining sex differences in conflict

behavior, social influence, and demand-

withdraw communication (see Cupach &

Canary, 1995; Kelly et al., 2003; Sagrestano

et al., 2006), both men and women generally

rated women in this sample as being more

likely to initiate relationship discussion than

are men. The construct validity of the ISQ is

further supported by the positive associations

of both of its subscales with marital quality.

These positive associations confirm the ex-

pectation that the perceived ability to discuss

relationship issues contributes to marital satis-

faction (see Fincham, 2004; Kelly et al.).

We expected that assessments of initiator

tendency would be positively associated with

trait argumentativeness, but this is not the case.

The items for the argumentativeness scale are

context free (e.g., ‘‘I enjoy a good argument

over a controversial issue’’; ‘‘I consider an

argument an exciting intellectual challenge’’),

whereas the items for the ISQ all refer, natu-

rally, to interactions with one’s intimate part-

ner. The context specificity of the attributes

assessed by the ISQ may help explain the lack

of association with the argumentativeness

scale; the general trait of argumentativeness

does not, apparently, influence the perceived

likelihood of self or partner initiating a discus-

sion about relationship issues.

The ISQ demonstrates good discriminant

validity by its independence from most demo-

graphic variables (other than gender). There

is a small, negative association between par-

ticipant age and ratings of partner initiator

tendency, which indicates that older partici-

pants see their partners as less likely to initi-

ate a relationship discussion than do younger

participants. This association could be the

product of either a developmental effect (as

people progress through a relationship, they

tend to see their partners as less likely to ini-

tiate discussions) or a cohort effect (people

born earlier in the 20th century see their part-

ners as less likely to initiate discussions than

people born later in the 20th Century). In any

event, the small magnitude of this association

does not appear to compromise the validity of

the ISQ.

The modest negative correlations between

ratings of one’s own initiator tendency and the

traits of need for approval and verbal aggres-

siveness, although unanticipated, are interest-

ing in that they suggest that those who see

themselves as initiating relationship discus-

sions are less likely to see themselves as acting

in socially desirable ways, or as trying to hurt

others by attacking their self-concept. People

with a high need for social approval may be

somewhat less inclined to report initiating

problem discussions with their partners than

those with a lower need for approval; however,

the modest amount of variance shared (4%)

suggests that the approval motive does not

substantially confound the ISQ’s assessment

of own initiator tendency. Moreover, the neg-

ative association of own initiator tendency

with verbal aggressiveness suggests that the

tendency to initiate problem discussions is

not inherently a form of aggression.

Finally, the ISQ exhibits encouraging con-

vergent validity. Several indices derived

from the ISQ subscales correlate at moderate

to strong levels with assessments of dyadic

interaction patterns obtained with the well-

established CPQ. A particularly strong corre-

lation (r ¼ .79) is observed between the ISQ

and CPQ assessments of mutual constructive

communication; a less strong but still respect-

able correlation (r ¼ .39) is observed between

the ISQ and CPQ assessments of total demand-

withdraw. These correlations, and especially

that for the total demand-withdraw index, are

not so strong as to suggest that the ISQ and

CPQ are parallel measures. Indices from the

ISQ and CPQ, however, correlate in a very

similar manner with the criterion of marital

quality: Consistent with previous theorizing

about communication patterns and marital

quality (see reviews by Eldridge & Christensen,

2002; Fincham, 2004), indices of demand-with-

draw (or initiate-avoid) communication from

both instruments are negatively associated with

marital quality, while indices of mutual initia-

tion (mutual constructive communication) from

both instruments are positively associated with

marital quality. These results provide additional

convergent validation for the ISQ. The index of

mutual constructive communication derived

from the ISQ is not associated with marital

256 W. H. Denton and B. R. Burleson

quality when controlling for the parallel index

of mutual constructive communication derived

from the CPQ; the nonsignificant partial corre-

lation here stems from the large covariance

between the two assessments of mutual con-

structive communication. The index of total

demand-withdraw derived from the ISQ, how-

ever, remains moderately associated with mar-

ital quality when controlling for the parallel

index of total demand-withdraw derived from

the CPQ, a finding that underscores that assess-

ments of marital communication derived from

these two instruments, although related, are not

duplicative. Overall, then, the results of Study 2

suggest that the ISQ provides reliable, valid

measures of the constructs it intends to tap.

Study 3

We conducted Study 3 to obtain further eval-

uations of the convergent, construct, and dis-

criminant validity of the ISQ. To accomplish

this goal, we included several additional

assessments of demographic, personality,

interaction, and relationship variables in this

study, as well as an assessment of mood. We

particularly focused in Study 3 on convergent

validity of the ISQ; this study examined the

extent to which ISQ assessments of self and

partner initiator tendencies converged with

assessments of initiator tendencies provided

by trained judges who coded videotaped con-

joint interviews conducted with the spouses.

Study 3 also included additional assessments

of the dimensionality and internal consistency

of the ISQ subscales.

We examined the demographic variables

of gender, age, family income, level of edu-

cation, religious denomination or affiliation,

number of years married, number of mar-

riages, number of divorces, and number of

children in the home. We expected to repli-

cate the pattern of gender differences

observed for the ISQ in Studies 1 and 2, with

both male and female participants perceiving

women as more likely than men to initiate

relationship discussions. We did not expect

any of the other demographic variables to

influence the ISQ assessments of own or part-

ner initiator tendency.

The personality measures obtained in Study

3 included an assessment of the value placed

on conflict management skills derived from

Burleson and Samter’s (1990) Communication

Functions Questionnaire, and an assessment of

religiosity. The tendency to initiate discus-

sions with one’s partner about relationship

issues indicates a positive, healthy orientation

toward the management of interpersonal con-

flict (Fincham, 2004; Kelly et al., 2003). Thus,

we expected ISQ assessments of initiator ten-

dencies, especially self initiator tendency, to

be positively associated with the value placed

on conflict management skill in close relation-

ships. There is no basis for expecting ISQ

assessments of initiator tendencies to be asso-

ciated with religiosity, the extent to which

individuals identify themselves as being reli-

gious. We assessed mood by the Beck Depres-

sion Inventory (Beck, Ward, Mendelson,

Mock, & Erbaugh, 1961). Research has indi-

cated that depressed individuals are less likely

to initiate or actively participate in com-

munication with intimates (Segrin, 1998;

Uebelacker et al., 2003) and we expected to

find depression negatively associated with

ISQ assessments of initiator tendency, espe-

cially self initiator tendency.

We derived the assessment of interaction

included in Study 3 (other than the ISQ) from

the CPI (Denton et al., 2001). Clinical observers

initially described patterns of initiation and

avoidance in marital interaction, and Denton

et al. designed the CPI to simulate a clinical

interview with some modifications so that it

could be utilized with nondistressed couples.

Assessments from the CPI thus provide a basis

for examining the extent to which ISQ assess-

ments of self and partner initiator tendency

converge with assessments of initiator ten-

dency based on a clinical-type interview.

The relationship quality assessments ob-

tained in Study 3 included Spanier’s (1976)

Dyadic Adjustment Scale and O’Leary’s

(O’Leary, Fincham, & Turkewitz, 1983) Pos-

itive Feelings Questionnaire. The Dyadic

Adjustment Scale provides an overall assess-

ment of marital satisfaction, whereas the Pos-

itive Feelings Questionnaire taps the degree of

liking individuals have for their relational

partners. The ability to discuss relationship

Initiator Style Questionnaire 257

concerns with one’s partner is one sign of a

healthy marriage (e.g., Kelly et al., 2003), so

participants who believe that they and their

partners can freely initiate relationship discus-

sions are likely to be more satisfied with their

marriages and think more positively about

their partners than those who feel less able to

initiate discussions.

Method

Participants

We recruited 60 married couples (N ¼ 120) as

participants from newspaper advertisements

and referrals from clinicians. Clinicians were

residents in psychiatry and graduate students

in marriage and family therapy as well as fac-

ulty psychiatrists and psychologists. All clini-

cians worked in an academic department of

psychiatry. Technical difficulties led to the

loss of data from two couples; in addition,

seven men and nine women did not complete

the ISQ. Thus, the analyzed sample consisted

of 49 couples for dyadic analyses or 100 per-

sons (51 men and 49 women) for individual

analyses. We paid participants $40 (USD)

per couple for their participation. Age of par-

ticipants ranged from 23 to 71, with a mean

age of 39. Participants had a mean of 1.3 mar-

riages with a range of 1–6; the average couple

had been married 12 years with a range of less

than 1 year to 47 years. Couples averaged 1.2

children living in the home with a range of

none to four. All but three of the participants

had at least a high school education and 97

had some education beyond high school. The

average participant reported a yearly family

income between $40,000 and $49,999 (USD).

The participants included two couples in which

both members were African American, one

couple where one member was a native Asian,

and the remainder of the participants were

Caucasian.

Procedure

Participant couples came jointly to a one-time

assessment session. After obtaining informed

consent, we separated them and they

responded to several questionnaires assessing

demographics, personality, interaction style,

and marital quality. Each participant next

completed several tasks that generated physi-

ological measures (blood pressure and heart

rate); these measurements are not relevant to

the present study and are not discussed further.

We then reunited the participant couples in

an interview room for the CPI (to determine

the pattern of interaction between spouses in

the dyad). After having completed training,

one of two therapists conducted the interviews.

We videotaped the CPI and the interview

lasted 10–20 min. The protocol for the CPI

has the couple seated together and the inter-

viewer stating, ‘‘Every couple occasionally

has things come up that they don’t exactly

agree on or see eye to eye on or maybe one

person does something that the other person

doesn’t like—maybe its a big thing or it could

be a small thing—and every couple has a dif-

ferent way of handling these disagreements—-

maybe they try to resolve them or maybe they

just ignore them and decide they will live with

it or maybe they try to resolve them and they

find that they can’t. What I would like to get

a picture of during this discussion is how the

two of you handle these situations. I know that

you might handle different situations differ-

ently but most couples have a ‘usual’ or ‘most

common’ way of handling differences. What I

am most interested in is the usual way you

would handle one of these situations.’’ Denton

et al. (2001) provided additional details

regarding the CPI protocol.

Instrumentation

Demographic assessments. Participants

initially completed a background information

questionnaire. Included were questions about

gender, age, level of education, family income,

religious denomination or affiliation, number

of years married, number of marriages, num-

ber of divorces, and number of children.

Personality and mood assessments. Per-

sonality and mood assessments obtained in

Study 3 included the Communication Func-

tions Questionnaire, a simple measure of reli-

giosity, and the Beck Depression Inventory.

258 W. H. Denton and B. R. Burleson

The Communication Functions Questionnaire

provides an assessment of communication val-

ues or the importance participants assign to a

variety of communication skills (Burleson &

Samter, 1990); in the present study, we exam-

ined the value participants placed on skill for

managing interpersonal conflict. Participants

read five items describing the exercise of con-

flict management skill (e.g., ‘‘Makes me feel

that I can be really honest about the things in

our relationship that produce conflict’’) and

rated each item on 5-point scales anchored

by somewhat important and extremely impor-

tant. Reliability of this five-item measure of

value for conflict management was a ¼ 63.

The validity of the CFQ has been demon-

strated in numerous studies in which it has

been found to be appropriately associated with

demographic (Burleson, Kunkel, Samter, &

Werking, 1996), personality, and relationship

assessments (Burleson & Samter).

To assess religiosity (i.e., the extent to

which individuals identify themselves as being

religious) participants responded to a single,

5-point item to which they indicated that they

were antireligious, not religious, slightly reli-

gious, moderately religious, or deeply reli-

gious. This item closely resembles the key

item in Pfeifer and Waelty’s (1995) Religious

Commitment Scale and exhibits similarities to

the core items in other measures of religious

commitment and involvement (see Hill &

Hood, 1999). The Beck Depression Inventory

is a commonly used self-report measure of

depression (Beck et al., 1961) containing

21 items assessing different aspects of depres-

sion. Extensive evidence supports the validity

of the Beck Depression Inventory (see Richter,

Werner, Heerlien, Kraus, & Sauer, 1998); in

the current study, the reliability for this mea-

sure was a ¼ .88.

Interaction assessments. We obtained two

assessments of initiate and avoid interaction

patterns in the present study: the ISQ ques-

tionnaire assessment of self and partner initi-

ator tendencies and the CPI assessment of

initiator tendency. Participants reported per-

ceptions of their own and their partners’ ini-

tiator tendencies using the two 10-item ISQ

subscales developed in Study 1 (reliability

assessments for these subscales are presented

below).

To generate an assessment of participants’

initiator tendencies from the CPI interview,

two graduate students in marriage and family

therapy coded the videotaped interviews. The

coders had no information about the partici-

pants other than what could be observed on

the CPI videotapes. The first author developed

a manual for coding the videotapes and

instructed the raters that they would be asked

to label each individual participant as either an

‘‘initiator’’ or an ‘‘avoider.’’ The first author

provided didactic instruction to the coders in

the concepts to be assessed and showed and

discussed videotapes of sample CPIs for fur-

ther instruction in the assignment of initiator

tendency. Raters assigned labels based on

what the participants reported during the

course of the interview about their handling

of differences. At the conclusion of the train-

ing process, the coders each coded 20% of the

sample (videotapes for 12 couples) and had

perfect agreement on labeling each participant

(n ¼ 24) as an avoider or an initiator. We then

divided the remaining 46 videotapes between

them for coding. There were no additional

evaluations of the raters’ ratings made after

they jointly rated the first 20% of recordings

(e.g., a second person confirming their rating)

so it is not possible to rule out subsequent drift

of the ratings after the initial training period.

Marital quality assessments. Participants

completed two measures of marital quality,

Spanier’s (1976) Dyadic Adjustment Scale

and O’Leary’s (O’Leary et al., 1983) Positive

Feelings Questionnaire. The Dyadic Adjust-

ment Scale is a 32-item self-report inventory

and is one of the most widely used instruments

for the assessment of marital adjustment. In

the current study, reliability for the entire scale

was a ¼ .94. Validity of the Dyadic Adjust-

ment Scale has been demonstrated by its abil-

ity to discriminate married from divorced

couples and its high correlation with other

measures of marital adjustment (Bradbury

et al., 2000; Spanier & Filsinger, 1983). The

Positive Feelings Questionnaire is a 17-item

self-response inventory designed to assess

the amount of positive affect toward a spouse

Initiator Style Questionnaire 259

(O’Leary et al.). All 17 items discriminate

between clinic and nonclinic couples (O’Leary

et al.). Reliability for the Positive Feelings

Questionnaire was a ¼ .95.

Results

Dimensionality and reliability assessments

Dimensionality and internal consistency.

We assessed the dimensionality of the ISQ

items by subjecting the items for each of the

two subscales to a principal axis factor analy-

sis. The factor analysis on the self item set

extracted only one factor, which had a large

eigenvalue (5.66) and explained 56.6% of the

variance. All 10 items correlated acceptably

with this factor, with loadings ranging from

.50 to .82 (from .54 to .88 for men and from

.51 to .84 for women). A high identity coeffi-

cient (e ¼ .990) indicated strong similarity in

the pattern of factor loadings for men and

women. The internal consistency (Cronbach’s

alpha) for these 10 items was excellent, a¼ .91

(.90 for men and .91 for women). The factor

analysis on the partner item set also extracted

only one factor, which had a large eigenvalue

(6.52) and explained 65.2% of the variance in

the correlations among the 10 items. All 10

items correlated strongly with this factor, with

loadings ranging from .50 to .89 (from .52 to

.91 for men and from .53 to .89 for women).

Again, a high identity coefficient (e ¼ .981)

indicated strong similarity in the pattern of

factor loadings for men and women. The inter-

nal consistency (Cronbach’s alpha) for these 10

items was excellent (a ¼ .94; .95 for men and

.89 for women). Thus, replicating the results of

Studies 1 and 2, we observed unidimensional

and internally consistent item structures for

both own and partner ratings of initiator

tendency.

Because the participants in Study 3 were

married couples, it is possible that their ratings

of initiator tendency were not independent of

one another. Nonindependence of partner

responses would require treating the couple

rather than the individual as the unit of analy-

sis (Kenny & Kashy, 1991). Thus, to assess

whether husbands’ and wives’ ratings of their

own and their partners’ initiator tendencies

were independent of one another, we corre-

lated husbands’ and wives ratings of self initi-

ator tendency (r ¼ .17, p . .20) and partner

initiator tendency (r ¼ .24, p ¼ .10). The non-

significant associations indicated that partici-

pant ratings of self and partner initiator status

were largely independent of the ratings their

spouses provided; this warrants treating indi-

viduals as the unit of analysis in subsequent

statistical analyses.

Validity assessments

Gender differences. To evaluate the gender

difference hypothesis, we conducted a 2 (men

vs. women)� 2 (self vs. partner) mixed-model

ANOVA on perceived initiator tendency with

4

4.5

5

5.5

6

6.5

7

7.5

8

Ra

te

d T

en

de

nc

y to

In

itia

te

D

is

cu

ss

io

ns

Self Partner

Rated Person

MenWomen

Figure 3. Sex differences in ratings of self and partner initiator tendency in Study 3.

260 W. H. Denton and B. R. Burleson

repeated measures on the second factor. Means

for this analysis are shown in Figure 3. The

main effect for gender was significant,

F(1, 98) ¼ 12.80, p , .001, g2 ¼ .12; women

(M ¼ 6.63, SD ¼ 2.00) viewed both them-

selves and their partners as more likely to ini-

tiate relationship discussions than did men

(M ¼ 5.62, SD ¼ 1.96). In addition, there

was a significant main effect for role,

F(1, 98) ¼ 8.52, p , .001, g2 ¼ .08. Overall,

participants viewed themselves as initiators

(M ¼ 6.39, SD ¼ 1.60) more strongly than

they did their partners (M ¼ 5.86, SD ¼1.87). Unlike Studies 1 and 2, the interaction

between participant gender and role was not

significant, F(1, 98) ¼ 0.03, p . .80. These

results provide partial support for the predicted

gender differences. Consistent with expecta-

tions, female participants viewed themselves

as more likely to initiate discussions than did

male participants; further, women viewed

themselves as more likely to initiate discus-

sions than their partners. Men, however, also

viewed themselves as more likely to initiate

discussions than their partners.

Other demographic differences. Other than

participant gender, none of the demographic

variables (age, family income, level of educa-

tion, religious denomination or affiliation,

years married, number of marriages, number

of divorces, number of children at home) was

associated with ratings for own or partner ini-

tiator tendency (we did not include ethnicity in

these analyses due to the absence of variation

in this variable). In particular, and distinct

from Study 2, ratings of partner initiator ten-

dency were not correlated with age (r ¼2.12,

p . .20).

Personality and mood assessments. We

computed correlations between the two ISQ

measures and the value for conflict manage-

ment skill, religiosity, and mood. As expected,

ratings for self initiator tendency were signif-

icantly and positively associated with the

value placed on conflict management skill

(r ¼ .26, p , .01); ratings of partner initiator

style were not associated with this variable

(r ¼ .10). Also, as expected, religiosity was

not associated with ratings of both self and

partner initiator tendencies. Ratings for self

initiator tendency were significantly, albeit

weakly, negatively associated with depression

(r ¼ 2.17, p , .05, one-tailed test). Depres-

sion was not associated with ratings of partner

initiator tendency (r ¼ 2.11).

Interaction assessments. To assess the

interaction hypothesis, a point-biserial correla-

tion was computed between each participant’s

initiator tendency, as coded by judges who

rated the CPI interviews, and the participant’s

self initiator tendency rating from the ISQ. As

expected, we observed a strong, positive asso-

ciation (r ¼ .57, p , .001). These results sup-

port the convergent validity of the ISQ.

Because the participants in Study 3 were

married couples, both of whom completed rat-

ings for own and partner initiator tendencies, it

was possible to examine the extent to which

the participant’s ratings for own style con-

verged with his or her spouse’s ratings of part-

ner style. Husbands’ ratings of their own

initiator tendency were not significantly corre-

lated with their wives’ ratings of partner initi-

ator tendency (r ¼ .13). Wives’ ratings of their

own initiator tendency, however, were signif-

icantly and positively correlated with their

husbands’ ratings of partner initiator tendency

(r ¼ .30, p , .05).

Relationship assessments. We calculated

correlations between the Dyadic Adjustment

Scale and Positive Feelings Questionnaire

assessments of marital quality and the meas-

ures of self and partner initiator tendency pro-

vided by the ISQ. As predicted, these

correlations were significant and positive: rat-

ings for self initiator tendency were positively

associated with both the Dyadic Adjustment

Scale (r ¼ .26, p , .01) and Positive Feelings

Questionnaire (r ¼ .22, p , .03); similarly,

ratings of partner initiator tendency were pos-

itively associated with both the Dyadic Adjust-

ment Scale (r ¼ .35, p , .001) and Positive

Feelings Questionnaire (r ¼ .22, p , .03).

Discussion

The results of Study 3 provide additional evi-

dence supporting the reliability and validity of

Initiator Style Questionnaire 261

the ISQ. Replicating the results of Studies

1 and 2, the items for both subscales of the

ISQ are found to be unidimensional and inter-

nally consistent, with alpha coefficients for

both subscales exceeding .90.

Construct validity of the ISQ is supported in

Study 3 by some (but not all) of the expected

gender differences. Associations between ISQ

assessments of initiator style and value for con-

flict management skill, depression, and marital

quality are modestly but statistically significant

and in the predicted directions. Replicating the

results of our first two studies, and consistent

with research using other measures of marital

interaction (e.g., Gottman & Levenson, 1988;

Heavey et al., 1993), women rate themselves as

more likely to initiate relationship discussion

than do the men; women also rate themselves

as more likely to initiate discussion than their

partners. Although these perceptions are shared

by men in Studies 1 and 2, in Study 3 men rate

themselves as more likely to initiate discussion

than they rate their partners. Still, these men

rate themselves as less likely to initiate discus-

sions than their wives rate them.

The construct validity of the ISQ is further

supported by the negative associations of self

initiator tendency with depression, and the

positive association of the self ratings with

the value placed on conflict management skill.

These associations, although of modest mag-

nitude, suggest that people who are less

depressed and who value constructive conflict

management also perceive themselves as more

likely to initiate relationship discussions. The

construct validity of the ISQ is also supported

by the positive correlations between both of its

subscales and the two assessments of relation-

ship quality (the Dyadic Adjustment Scale and

the Positive Feelings Questionnaire). These

correlations confirm the expectation that the

perceived ability to discuss relationship issues

is associated with marital satisfaction (e.g.,

Roberts, 2000). The ISQ demonstrates dis-

criminant validity in Study 3 by its indepen-

dence from a variety of demographic variables

and religiosity.

A major interest in Study 3 is the conver-

gent validity of the ISQ with an interview

assessment of initiator tendency. Self-reports

of participants’ own initiator tendencies

obtained by the ISQ are strongly associated

(r ¼ .57) with judges’ coding of initiator ten-

dency from videotaped couple interviews.

These findings provide additional support for

the convergent validity of the ISQ.

Because intact couples are employed in

Study 3, we are able to explore the extent to

which partners’ ratings of each other’s initiator

tendency converge with self ratings. Hus-

bands’ ratings of their wives’ initiator tenden-

cies are only moderately correlated (r ¼ .30)

with wives’ self ratings, and wives’ ratings of

their husbands’ initiator tendencies are uncor-

related with husbands’ self ratings. It is not

uncommon to find limited agreement among

spouses about their behavioral styles and ori-

entations (e.g., Ptacek, Pierce, Ptacek, &

Nogel, 1999). Future research should seek to

identify both determinants and consequences

of agreement by spouses about each other’s

tendencies to initiate problem discussions.

For example, the greater convergence between

husband partner ratings and wife self ratings of

initiator tendency may stem from the fact that

wives tend to be more demanding when dis-

cussing their concerns than do husbands when

discussing their concerns (see Sagrestano

et al., 2006), which may lead to initiations by

wives having greater salience for their hus-

bands than do initiations by husbands for their

wives.

General Discussion

The tendency to initiate (or avoid) discussions

about perceived problems in the relationship

with one’s partner is emerging as a variable

of theoretical and clinical relevance. The ISQ

offers a new option for the assessment of this

individual difference. The ISQ exhibits excel-

lent reliability with the two subscales display-

ing strong unidimensionality across all three

studies and robust internal consistency with

alpha coefficients approaching or exceeding

.90. In all three studies, the pattern of item

loadings is very similar for men and women

on both subscales and virtually equivalent

internal consistencies are observed for men

and women on both subscales. The results of

Study 2 find good test–retest reliability coef-

ficients for both the self scale (r ¼ .81) and the

262 W. H. Denton and B. R. Burleson

partner scale (r ¼ .77). Although reliability of

the ISQ should continue to be assessed, it

appears that the current version of both sub-

scales provides consistent measurement of the

constructs tapped.

The items of the ISQ have obvious rele-

vance to the theoretical constructs of interest,

evidencing the face validity of the instrument.

That is, they clearly appear to be measuring the

perceived tendencies by self and partner to ini-

tiate discussions about relationship issues. Face

validity is further supported by the unidimen-

sional character of the items for each scale.

The three studies reported here provide ini-

tial evidence of the construct validity of the

ISQ assessments of initiate-avoid tendency.

Research examining gender differences in

how intimate partners negotiate, influence,

and manage conflict (see Cupach & Canary,

1995; Kelly et al., 2003), as well as research

examining gender differences in demand-

withdraw communication (see Sagrestano

et al., 2006), led us to expect that women

would be rated as having a stronger tendency

to initiate relationship discussions than men.

We tested this hypothesis in different samples

and it is generally supported. Across all three

of our studies, women consistently see them-

selves as more likely to initiate problem dis-

cussions than men see themselves; further,

women consistently see themselves as more

likely to initiate than they see their partners

as likely to initiate. There is only one notable

difference in the sex differences observed over

the three studies: In Studies 1 and 2, men see

themselves and their partners as equally likely

to initiate discussions when experiencing

a problem, whereas in Study 3 men see them-

selves as more likely than their partners to

initiate a discussion when experiencing a prob-

lem. In Study 3, couples are aware that their

partners were completing the research ques-

tionnaires; perhaps, the men in this study want

to project an image as powerful and influential

as their wives (see Carli, 2004) and thus do not

want to depict themselves as less willing to

initiate discussions about problems than they

depict their wives.

Additional evidence supporting the con-

struct validity of the ISQ comes from its asso-

ciations with relationship quality. Previous

research has found that demand-withdraw

communication is associated with relationship

dissatisfaction (e.g., Bodenmann et al., 1998;

Caughlin & Huston, 2002; Christensen &

Heavey, 1990; Noller &White, 1990). Consis-

tent with these results, we found in Study 2 that

an index of demand-withdraw (i.e., initiate-

avoid) communication formed from the two

ISQ subscales is negatively associated with

marital quality.

Although demand-withdraw communica-

tion is associated with relationship dissatisfac-

tion, we reasoned that the ability to discuss

relationship concerns with one’s partner is

a sign of a healthy marriage (see Kelly et al.,

2003) and thus predicted that participants who

perceived themselves and their partners as

more likely to initiate relationship discussions

would evaluate their relationships more posi-

tively. Consistent with this expectation, we

find ISQ assessments of initiator tendency pos-

itively associated with three different mea-

sures of relationship quality in two different

studies. This particular result helps distinguish

our conceptual and methodological approach

to initiator style from work focused more spe-

cifically on demand-withdraw communication

because there is no evidence that ‘‘demand-

ing’’ is positively associated with relationship

quality.

We predicted that ISQ measures would be

associated with argumentativeness but not ver-

bal aggressiveness; however, we do not obtain

these findings. In retrospect, a case can be

made that the desire to discuss relationship

issues does not indicate an enjoyment of argu-

ing; indeed, people who initiate relationship

discussions almost universally indicate that

their intention is not to start an argument. It

might be guessed that there would be a positive

association between initiator tendency and

verbal aggressiveness, but we detect a negative

association between these variables: those who

avoid relationship discussions report higher

levels of verbal aggression. This finding is

consistent with a clinical observation that

one reason avoiders give for avoiding relation-

ship discussions is their concern that the dis-

cussion will be destructive. In fact, when

avoiders are finally forced into engagement

they may do so with shouting, cursing, etc.

Initiator Style Questionnaire 263

Thus, theoretically, avoiding may be a way for

avoiders to control the tendency to be more

aggressive than they would like; avoidance

may also be a means of maintaining control

of the relationship through a refusal to even

discuss a possibility of change (see Roberts,

2000; Roloff & Ifert, 2000).

The standard of discriminant validity main-

tains that the variables with which a measure

does not correlate can be just as important as

those with which it does correlate. Consistent

with our expectations, the ISQ assessments of

initiator tendency are not associated with demo-

graphic variables including family income,

years married, number of marriages, and num-

ber of divorces. There is a small correlation

with age in Study 2 but this is not replicated

in Study 3 and does not suggest a serious con-

found. The ISQ is also found not associated or

only weakly associated with several personality

variables from which it should be independent,

including need for approval and religiosity.

Finally, the ISQ demonstrates good conver-

gent validity with both an established self-

report instrument (the CPQ) and an interview

assessment (the CPI). Five of six associations

between measures of dyadic patterns derived

from the ISQ and CPQ are significant and in

the predicted direction. Additionally, the

dyadic indexes derived from the ISQ have

the same pattern of associations as the CPQ

indexes with marital quality. Clinical observ-

ers first described the tendencies to initiate and

avoid problem discussions with a relationship

partner and it was our intention to create an

instrument to capture what has been observed

clinically. Matched with a clinical-type inter-

view, the ISQ performs well, further support-

ing its validity as a measure of the desired

characteristic.

In conclusion, the ISQ appears to be a reli-

able and valid assessment of perceived tenden-

cies by self and partner to initiate relationship

discussions. Unlike observer coding or inter-

view assessments, the ISQ is brief and easily

administered and scored. It allows for the

assignment of scores to individuals and can

also generate assessments of perceived dyadic

communication patterns (demand-withdraw or

initiate-avoid communication; mutually con-

structive communication).

Limitations of the ISQ include that it has,

thus far, been used only with convenience

samples of married participants in the United

States. Because the participants in our three

studies were not randomly selected from the

population of married people in the United

States (or, for that matter, around the world),

important questions can be raised about the

generalizability of our results to this popula-

tion. For example, the three studies reported

here utilize samples composed almost exclu-

sively of Caucasians and the applicability of

the instrument to other ethnic, racial, cultural

groups needs future investigation. Although

the participants in our studies were not sam-

pled randomly from the population of married

persons, they did appear to be typical of mar-

ried, Caucasian persons residing in the south-

eastern United States, especially in terms of

demographics such as age, income, education,

time married. The consistency of results for

gender differences in patterns of ISQ ratings

observed across our three studies provides

some basis for confidence in the reliability

(i.e., generalizability) of these results, as does

the consistency of the absence of effects for

other demographic variables (e.g., education,

income, number of children). Of course, the

results of our three studies require replication

to increase confidence in the presently

reported findings; ideally, future studies will

be conducted with systematically selected

samples rather than convenience samples. Fur-

ther, future research should explore the psy-

chometric properties of the instrument with

other types of couples, including dating, cohab-

iting, same-sex, etc.

As noted above, interest in initiating and

avoiding behavioral patterns in married couples

was initially stimulated by couple therapists

describing what they observed in couple psy-

chotherapy sessions. The ISQ provides another

option for quantifying such observations, and

this has potential significance for clinical work,

especially clinical research. Some models of

couple therapy specifically target the alteration

of the demand-withdraw (initiate-avoid) pattern

such as emotion-focused therapy for couples

(Johnson & Denton, 2002). Whether the ISQ

would be sensitive to changes in therapy is

a future research question.

264 W. H. Denton and B. R. Burleson

Additionally, we are particularly interested

in the effects of initiator tendency on health

and illness. Thus, we are currently utilizing

the ISQ studies in exploring the role of initiator

tendency among patients and partners with

somatoform disorders, cardiac illness, and

depression. We hope that our presentation of

this new instrument will provide an additional

tool for other investigators interested in basic

and clinical research examining central aspects

of communication in close relationships.

References

Aries, E. J. (2006). Sex differences in interaction: A reex-amination. In K. Dindia & D. J. Canary (Eds.), Sexdifferences and similarities in communication (2nded., pp. 21–36). Mahwah, NJ: Erlbaum.

Beck, A. T., Ward, C. H., Mendelson, M., Mock, J., &Erbaugh, J. (1961). An inventory for measuringdepression. Archives of General Psychiatry, 4, 571.

Berns, S. B., Jacobson, N. S., & Gottman, J. M. (1999).Demand withdraw interaction patterns between differ-ent types of batterers and their spouses. Journal ofMarital and Family Therapy, 25, 337–347.

Bodenmann, G., Kaiser, A., Hahlweg, K., & Fehm-Wolfsdorf, G. (1998). Communication patterns dur-ing marital conflict: A cross-cultural representation.Personal Relationships, 5, 343–356.

Bradbury, T. N., Fincham, F. D., & Beach, S. R. H. (2000).Research on the nature and determinants of maritalsatisfaction: A decade in review. Journal of Marriageand the Family, 62, 964–980.

Burgoon, J. K. (1976). The unwillingness to communicatescale: Development and validation. CommunicationMonographs, 43, 60–69.

Burleson,B.R.,Kunkel,A.W., Samter,W.,&Werking,K. J.(1996). Men’s and women’s evaluations of communica-tion skills in personal relationships: When sex differ-encesmake a difference—Andwhen they don’t. Journalof Social and Personal Relationships, 13, 201–224.

Burleson, B. R., & Samter, W. (1990). Effects of cognitivecomplexity on the perceived importance of communi-cation skills in friends. Communication Research, 17,165–182.

Canary, D. J., & Emmers-Sommer, T. M. (1997). Sex andgender differences in personal relationships. NewYork: Guilford.

Canary, D. J., & Hause, K. S. (1993). Is there any reason toresearch sex differences in communication? Commu-nication Quarterly, 41, 129–144.

Carli, L. L. (2004). Gender effects on social influence. InJ. S. Seiter & R. H. Gass (Eds.), Perspectives onpersuasion, social influence, and compliance gaining(pp. 133–148). Boston: Allyn & Bacon.

Carmines, E. G., & Zeller, R. A. (1979). Reliability andvalidity assessment. Beverly Hills, CA: Sage.

Caughlin, J. P., & Huston, T. L. (2002). A contextualanalysis of the association between demand/withdrawand marital satisfaction. Personal Relationships, 9,95–119.

Caughlin, J. P., & Malis, R. S. (2005). Demand/withdrawcommunication between parents and adolescents: Con-

nections with self-esteem and substance use. Journalof Social and Personal Relationships, 21, 124–148.

Christensen, A. (1987). Detection of conflict patterns incouples. In K. Hahlweg & M. J. Goldstein (Eds.),Understanding major mental disorder: The contribu-tion of family interaction research. The Family Pro-cess Press monograph series (pp. 250–265). NewYork: Family Process Press.

Christensen, A., & Heavey, C. L. (1990). Gender andsocial structure in the demand/withdraw pattern ofmarital conflict. Journal of Personality and SocialPsychology, 59, 73–81.

Christensen, A., & Shenk, J. L. (1991). Communication,conflict, and psychological distance in nondistressed,clinic, and divorcing couples. Journal of Consultingand Clinical Psychology, 59, 458–463.

Christensen, A., & Sullaway, M. (1984). CommunicationPatterns Questionnaire. Unpublished manuscript,University of California, Los Angeles.

Crowne, D. P., & Marlowe, D. (1964). The approvalmotive: Studies in evaluative dependence. New York:Wiley.

Cummings, E. M., & Davies, P. T. (2002). Effects of mar-ital conflict on children: Recent advances and emerg-ing themes in process-oriented research. Journal ofChild Psychology and Psychiatry, 43, 31–63.

Cupach, W. R., & Canary, D. J. (1995). Managing conflictand anger: Investigating the sex stereotype hypothesis.In P. J. Kalbfleisch & M. J. Cody (Eds.), Gender,power, and communication in human relationships(pp. 233–252). Hillsdale, NJ: Erlbaum.

Denton, W. H., Burleson, B. R., Hobbs, B. V., VonStein, M., & Rodriguez, C. P. (2001). Cardiovascularreactivity and initiate/avoid patterns of marital com-munication: A test of Gottman’s psychophysiologicmodel of marital interaction. Journal of BehavioralMedicine, 24, 401–421.

DeVellis, R. F. (2003). Scale development: Theory andapplications (2nd ed.). Thousand Oaks, CA: Sage.

Dindia, K. (2006). Men are fromNorth Dakota, women arefrom South Dakota. In K. Dindia & D. J. Canary(Eds.), Sex differences and similarities in communica-tion (2nd ed., pp. 3–20). Mahwah, NJ: Erlbaum.

Eldridge, K. A., & Christensen, A. (2002). Demand-withdraw communication during couple conflict: Areview and analysis. In P. Noller & J. A. Feeney(Eds.), Understanding marriage: Developments inthe study of couple interaction (pp. 289–322). NewYork: Cambridge University Press.

Ellis, A. P. (1997). The effects of coping strategies andinteractional patterns on individual and dyadic adjust-ment to prostate cancer. Unpublished doctoral disser-tation, University of Kansas.

Fincham, F. D. (2004). Communication in marriage. InA. L. Vangelisti (Ed.), Handbook of family communi-cation (pp. 83–104). Mahwah, NJ: Erlbaum.

Gottman, J. M. (1993). The roles of conflict engagement,escalation, and avoidance in marital interaction: A lon-gitudinal view of 5 types of couples. Journal of Con-sulting and Clinical Psychology, 61, 6–15.

Gottman, J. M., & Levenson, R. W. (1988). The socialpsychophysiology of marriage. In P. Noller & M. A.Fitzpatrick (Eds.), Perspectives on marital interaction(pp. 182–200). Philadelphia: Multilingual Matters.

Hahlweg,K., Kaiser, A., Christensen,A., Fehm-Wolfsdorf,G., & Groth, T. (2000). Self-report and observationalassessment of couples’ conflict: The concordance

Initiator Style Questionnaire 265

between the communication patterns questionnaire andthe KPI observation system. Journal of Marriage andthe Family, 62, 61–67.

Heavey, C. L., Christensen, A., & Malamuth, N. M.(1995). The longitudinal impact of demand and with-drawal during marital conflict. Journal of Consultingand Clinical Psychology, 63, 797–801.

Heavey, C. L., Gill, D. S., and Christensen, A. (1996).The couples interaction rating system. Unpublishedmanuscript.

Heavey, C. L., Larson, B. M., Zumtobel, D. C., &Christensen, A. (1996). The Communication Pat-terns Questionnaire: The reliability and validity ofa constructive communication subscale. Journal ofMarriage and the Family, 58, 796–800.

Heavey, C. L., Layne, C., & Christensen, A. (1993). Gen-der and conflict structure in marital interaction: A rep-lication and extension. Journal of Consulting andClinical Psychology, 61, 16–27.

Hill, P. C., & Hood, R. W., Jr. (Eds.). (1999). Measures ofreligiosity. Birmingham, AL: Religious Education Press.

Holtzworth-Munroe, A., Smutzler, N., & Stuart, G. L.(1998). Demand and withdraw communication amongcouples experiencing husband violence. Journal ofConsulting and Clinical Psychology, 66, 731–743.

Infante, D. A., & Rancer, A. S. (1982). A conceptualiza-tion and measure of argumentativeness. Journal ofPersonality Assessment, 46, 72–80.

Infante, D. A., & Wigley, C. J. (1986). Verbal aggressive-ness: An interpersonal model and measure. Communi-cation Monographs, 53, 61–69.

Johnson, S. M., & Denton, W. (2002). Emotionallyfocused couple therapy: Creating secure connections.In A. S. Gurman & N. S. Jacobson (Eds.), Clinicalhandbook of couple therapy (3rd ed., pp. 221–250).New York: Guilford.

Kelly, A. B., Fincham, F. D., & Beach, S. R. H. (2003).Communication skills in couples: A review and dis-cussion of emerging perspectives. In J. O. Greene &B. R. Burleson (Eds.), Handbook of communicationand social interaction skills (pp. 723–752). Mahwah,NJ: Erlbaum.

Kenny, D. A., & Kashy, D. A. (1991). Analyzing interde-pendence in dyads. In B. M. Montgomery & S. Duck(Eds.), Studying interpersonal interaction (pp. 275–285).New York: Guilford.

Klinetob, N. A., & Smith, D. A. (1996). Demand-withdrawcommunication in marital interaction: Tests of inter-personal contingency and gender role hypotheses.Journal of Marriage and the Family, 58, 945–957.

Krokoff, L. J., Gottman, J. M., & Hass, S. D. (1989).Validation of a global rapid couples interaction scoringsystem. Behavioral Assessment, 11, 65–79.

Noller, P., & Feeney, J. A. (2004). Studying family com-munication:Multiple methods and multiple sources. InA. L. Vangelisti (Ed.), Handbook of family communi-cation (pp. 31–50). Mahwah, NJ: Erlbaum.

Noller, P., Feeney, J. A., Bonnell, D., & Callan, V. J.(1994). A longitudinal study of conflict in early mar-riage. Journal of Social and Personal Relationships,11, 233–252.

Noller, P., & White, A. (1990). The validity of the Com-munication Patterns Questionnaire. PsychologicalAssessment, 2, 478–482.

Norton, R. (1983). Measuring marital quality: A criticallook at the dependent variable. Journal of Marriageand the Family, 45, 141–151.

O’Leary, K. D., Fincham, F., & Turkewitz, H. (1983).Assessment of positive feelings toward spouse. Journalof Consulting and Clinical Psychology, 51, 949–951.

Pasch, L. A. (1994).Fertility problems andmarital relation-ships: The effects of appraisal and coping differenceson communication. Unpublished doctoral dissertation,University of California, Los Angeles.

Pfeifer, S., & Waelty, U. (1995). Psychopathology andreligious commitment: A controlled study. Psychopa-thology, 28, 70–77.

Ptacek, J. T., Pierce, G. R., Ptacek, J. J., & Nogel, C.(1999). Stress and coping processes inmenwith prostatecancer: The divergent views of husbands and wives.Journal of Social & Clinical Psychology, 18, 299–324.

Putnam, L. (2006). Definitions and approaches to conflictand communication. In J. G. Oetzel & S. Ting-Toomey(Eds.),Handbook of conflict communication (pp. 3–21).Thousand Oaks, CA: Sage.

Rehman, U. S., & Holtzworth-Munroe, A. (2006). A cross-cultural analysis of the demand-withdraw maritalinteraction: Observing couples from a developingcountry. Journal of Consulting and Clinical Psychology,74, 755–766.

Richter, P.,Werner, J., Heerlien, A., Kraus, A., & Sauer, H.(1998). On the validity of the Beck Depression Inven-tory: A review. Psychopathology, 31, 160–168.

Roberts, L. J. (2000). Fire and ice in marital communica-tion: Hostile and distancing behaviors as predictors ofmarital distress. Journal of Marriage and the Family,62, 693–707.

Roloff, M. E., & Ifert, D. E. (2000). Conflict managementthrough avoidance: Withholding complaints, suppress-ing arguments, and declaring topics taboo. In S. Petronio(Ed.), Balancing the secrets of private disclosures (pp.151–164). Mahwah, NJ: Erlbaum.

Roloff, M. E., & Solomon, D. H. (2002). Conditions underwhich relational commitment leads to expressing orwithholding relational complaints. The InternationalJournal of Conflict Management, 13, 276–391.

Sagrestano, L. M., Heavey, C. L., & Christensen, A.(2006). Individual differences versus social structuralapproaches to explaining demand-withdraw and socialinfluence behaviors. In K. Dindia & D. J. Canary(Eds.), Sex differences and similarities in communica-tion (2nd ed., pp. 379–396). Mahwah, NJ: Erlbaum.

Schumm, W. R., Paffbergen, L. A., Hatch, R. C., Obiorah,F. C., Copeland, J. M., Meens, L. D., et al. (1986).Concurrent and discriminant validity of the KansasMarital Satisfaction Scale. Journal of Marriage andthe Family, 48, 381–387.

Segrin, C. (1998). Interpersonal communication problemsassociated with depression and loneliness. In P. A.Andersen & L. A. Guerrero (Eds.), Handbook of com-munication and emotion: Research, theory, applica-tions, and contexts (pp. 215–242). San Diego, CA:Academic Press.

Shoham, V., Rohrbaugh, M. J., Stickle, T. R., & Jacob, T.(1998). Demand-withdraw couple interaction moder-ates retention in cognitive-behavioral versus family-systems treatments for alcoholism. Journal of FamilyPsychology, 12, 557–577.

Spanier, G. (1976). Measuring dyadic adjustment: Newscales for assessing the quality of marriage and similardyads. Journal of Marriage and the Family, 38, 15–28.

Spanier, G., & Filsinger, E. (1983). The Dyadic Adjust-ment Scale. In E. Filsinger (Ed.),Marriage and familyassessment. Beverly Hills, CA: Sage.

266 W. H. Denton and B. R. Burleson

Sullaway, M., & Christensen, A. (1983). Assessment ofdysfunctional interaction patterns in couples. Journalof Marriage and the Family, 45, 653–660.

Swann, W. B., Jr., & Rentfrow, P. J. (2001). Blirtatious-ness: Cognitive, behavioral, and physiological conse-quences of rapid responding. Journal of Personalityand Social Psychology, 81, 1160–1175.

Uebelacker, L. A., Courtnage, E. S., & Whisman, M. A.(2003). Correlates of depression and marital dissatis-faction: Perceptions of marital communication style.Journal of Social and Personal Relationships, 20,757–769.

Van de Vijver, F., & Leung, K. (1997). Methods anddata analysis of comparative research. In J. W. Berry,Y. H. Poortinga, & J. Pandey (Eds.), Handbook ofcross-cultural psychology, Vol. 1: Theory and method(2nd ed., pp. 257–300). Boston: Allyn & Bacon.

Walczynski, P. T. (1997). Power, personality, and conflictualinteraction: An exploration of demand/withdraw interac-tion in same-sex and cross-sex couples. Unpublished doc-toral dissertation, University of California, Los Angeles.

Weiss, R. L., & Summers, K. J. (1983). Marital InteractionCoding System-III. In E. Filsinger (Ed.),Marriage andfamily assessment (pp. 85–116). Newbury Park, CA: Sage.

Appendix A

Initiator Style Questionnaire

In this part of the questionnaire we are interested in how you typically responds to problems in

your relationship (i.e., problems that are between you and your partner). Please rate each item on

a scale of 1 (strongly disagree) to 9 (strongly agree).

1) When discussing a relationship problem,

I usually try to keep the discussion

going until we settle the issue

Strongly disagree Strongly agree

1 2 3 4 5 6 7 8 9

2) I usually express my feelings about

our relationship to my partner.

Strongly disagree Strongly agree

1 2 3 4 5 6 7 8 9

3) I usually keep my feelings about our

relationship private and do not share them

with my partner.

Strongly disagree Strongly agree

1 2 3 4 5 6 7 8 9

4) When I become aware of a problem in our

relationship, I usually do not say anything

about it.

Strongly disagree Strongly agree

1 2 3 4 5 6 7 8 9

5) I am the kind of person who generally feels

comfortable discussing relationship problems.

Strongly disagree Strongly agree

1 2 3 4 5 6 7 8 9

6) When my partner wants to talk about a

relationship problem, I am usually ready

to do so as well.

Strongly disagree Strongly agree

1 2 3 4 5 6 7 8 9

7) I usually become silent or refuse to

discuss a relationship problem further if my

partner pressures or demands that I do so.

Strongly disagree Strongly agree

1 2 3 4 5 6 7 8 9

8) When my partner wants to talk about a

relationship problem, I usually try to get

out of the discussion.

Strongly disagree Strongly agree

1 2 3 4 5 6 7 8 9

9) When I become aware of a problem in

our relationship, I usually try to start a

discussion of that problem.

Strongly disagree Strongly agree

1 2 3 4 5 6 7 8 9

10) I am the kind of person who generally

does not feel comfortable discussing

relationship problems.

Strongly disagree Strongly agree

1 2 3 4 5 6 7 8 9

(continued)

Initiator Style Questionnaire 267

Appendix A. (coninued)

Initiator Style Questionnaire

In this part of the questionnaire we are interested in how your partner typically responds to

problems in your relationship (i.e., problems that are between you and your partner). Please rate

each item on a scale of 1 (strongly disagree) to 9 (strongly agree).

11) When I want to talk about a relationship

problem, my partner usually tries to

get out of the discussion.

Strongly disagree Strongly agree

1 2 3 4 5 6 7 8 9

12) My partner usually expresses any feelings

about our relationship to me.

Strongly disagree Strongly agree

1 2 3 4 5 6 7 8 9

13) My partner is the kind of person who

generally feels comfortable discussing

relationship problems.

Strongly disagree Strongly agree

1 2 3 4 5 6 7 8 9

14) When my partner becomes aware of a

problem in our relationship, my partner usually

tries to start a discussion of that problem.

Strongly disagree Strongly agree

1 2 3 4 5 6 7 8 9

15) When discussing a relationship problem,

my partner usually tries to keep the discussion

going until we settle the issue.

Strongly disagree Strongly agree

1 2 3 4 5 6 7 8 9

16) If my partner and I are discussing an

important relationship issue, my partner usually

tries to keep discussing it even if it seems we are

beginning to become emotional.

Strongly disagree Strongly agree

1 2 3 4 5 6 7 8 9

17) My partner usually keeps feelings about

our relationship private and does not share

them with me.

Strongly disagree Strongly agree

1 2 3 4 5 6 7 8 9

18) My partner is the kind of person who

generally does not feel comfortable

discussing relationship problems.

Strongly disagree Strongly agree

1 2 3 4 5 6 7 8 9

19) When my partner becomes aware of a

problem in our relationship, my partner

usually does not say anything about it.

Strongly disagree Strongly agree

1 2 3 4 5 6 7 8 9

20) When I want to talk about a relationship

problem, my partner is usually ready

to do so as well.

Strongly disagree Strongly agree

1 2 3 4 5 6 7 8 9

Scoring: Items 3, 4, 7, 8, 10, 11, 17, 18, and 19 are reverse scored. Items 1–10 are then summed to produce the ISQ self

score and items 11–20 are summed to produce the ISQ partner score.

268 W. H. Denton and B. R. Burleson