Download as pdf or txt
Download as pdf or txt
You are on page 1of 31

Human Communication Research

Vol. 11, No. 2,Winter 1984, 771-201

“SECRET TESTS”
Social Strategies for Acquiring Information
About the State of the Relationship

LESLIE A. BAXTER
Lewis and Clark College
WILLIAM W. WILMOT
University of Montana

The current study was undertaken to develop a typology of “secrettests, ”--that is,
social strategies that people use to acquire knowledge about the state of their
opposite-sex relationships. Furthermore, the influences of relationship type and
respondent sex on strategy use were assessed. Both qualitative and quantitative
methods were employed with data obtained from a total of 181 respondents.
Findings suggest 14 basic categories orsecret tests ”thatcomprise 7 cluster types in
a two-dimensional spatial representation. Passive, active, and interactive strategy
types were evident. Females reported more secret test strategies than did males, and
people in opposite-sex relationships that were in transition from platonic to
romantic reported more strategies than people in either platonic cross-sex or
romantic cross-sex relationships. Differences were found as well in the type of
secret test most likely to be employed as a function of respondent sex and
relationship type.

Substantial research in the area of social cognition addresses the ways in


which social knowledge is organized by people (Forgas, 1981; Harvey,
Ickes, & Kidd, 1976, 1978, 1981; Heider, 1958; Higgins, Herman, &
Zanna, 1981; Jones & Davis, 1965; Kelley, 1967; Kelly, 1955). How-
ever, examination of how social knowledge is organized pre-
supposes the logically prior research issue of how social information is
acquired to begin with. Basically, two research traditions can be

Leslie A. Baxter (Ph.D., University o f Oregon, 1975) is an associate professor o f


communications at Lewis and Clark College, Portland, OR. William W. Wilmot (Ph.D.,
University of Washington, 1970) is a professor of interpersonal communication at the
University of Montana, Missoula, MT.

0 1984 International Communication Assn

171
172 HUMAN COMMUNICATION RESEARCH / Winter 1984

identified that address the question of social information acquisition.


The first and more prevalent research tradition focuses on cognitive
strategies of information acquisition, such as perceptual selectivity
(Nisbett & Ross, 1980). The second research tradition focuses on
behavioral strategies that people employ in acquiring their information
(Berger, 1979, 1982; Berger & Bradac, 1982; Berger & Douglas, 1981;
Berger & Kellermann, 1983; Berger & Perkins, 1978, 1979; Kellermann
& Berger, 1984; Snyder, 1981; Snyder & Campbell, 1980; Snyder &
Cantor, 1979; Snyder & Swann, 1978a, 1978b; Trope & Bassok, 1982).
The current study is consistent with the second tradition’s focus on
behavioral strategies of information acquisition in its examination of
the social strategies that people employ to acquire information about
the state of their interpersonal relationships, or what we have come to
call “secret tests.”
Extant work in the behavioral strategies of social information
acquisition has concentrated on how people gain insight into a
stranger’s personality. Snyder and his colleagues examined one informa-
tion acquisition strategy in particular-the use of questions directed
toward the target-stranger-and found that people ask questions
seeking to confirm their operating hypotheses about the other’s
character as opposed to asking neutral questions that might disconfirm
their hypotheses (see, for example, Snyder, 1981). However, more
recently Trope and Bassok (1982) failed to find evidence supportive of
Snyder’s posited confirmatory bias.
Berger and his colleagues examined a broader domain of social
strategies than the single strategy of direct questions asked of a stranger-
acquaintance (for example, see Berger & Bradac, 1982). Collectively,
this research program posits three basic types of strategies by which
people acquire information about a stranger-target: (1) passive stra-
tegies, in which the person unobtrusively observes the target; (2) active
strategies, characterized by intervention into the target-stranger’s
environment in some way, including asking third parties about the
target person or structuring the target’s environment in order to conduct
an “experiment” in which to observe the target’s behavior; and (3)
interactive strategies, in which the person directly interacts with the
target-stranger, including interrogation of the target through use of
questions, intentional deviation from social rules of interaction in order
to observe the target’s reactions, and self-disclosure intended to evoke
reciprocal disclosure from the target. To date, research has focused in
Baxter, Wilmot / SECRET TESTS 173

particular on passive strategies (Berger & Douglas, 1981; Berger &


Perkins, 1978, 1979) and on direct interrogation strategies (Berger,
Gardner, Clatterbuck, & Schulman, 1976; Berger & Kellermann, 1983;
Kellermann & Berger, 1984).
Although both the Snyder and Berger research programs are
insightful, they are limited to the acquisition of individual-level
information and are confined to the relationship context of initial
acquaintances. Berger (1982) has posited that the information acquisi-
tion strategies comprise basic elements of well-learned “information
acquisition scripts” that may generalize to the acquisition of infor-
mation about nonstrangers. However, extant work has not examined
the validity of this claim. Furthermore, the parties in nonstranger
relationships acquire information about the state of their relationship in
addition to acquiring individual-level information about one another.
Berger and Bradac (1982, p. 99) noted the difference between these two
types of information that can be acquired:

Whereas previously John attempted to reduce his uncertainty about


Mary, now he attempts to reduce his uncertainty about Mary and John as
a unit. This is an important shift, because relational uncertainty is of a
different “logical type” than is uncertainty about individuals. . . . A
relationship is a higher-order abstraction than is one’s conception of an
individual. Accordingly, relational uncertainty reduction should be
comparatively difficult.

The extent to which these two “logical types”of information rely on the
same basic repertoire of information acquisition strategies has not been
determined to date.
Despite the absence of direct empirical work that speaks to the
generalizability of tne information acquisition strategies repertoire to
nonstranger relationships, substantial work indirectly supports this
expectation. In general, regardless of the “logical type” of the infor-
mation people should be motivated to increase their information-
seeking behavior as a function of information uncertainty (Berger &
Calabrese, 1975). Although the initial statement of uncertainty theory
focused on initial interactions, it suggested the presence of uncertainty
about the other in later stages of relationship development as well
(Berger & Calabrese, 1975). Others have recognized the pervasiveness of
relational uncertainty in relationships. As Parks and Adelman (1983)
recently observed, the notion of relational uncertainty is implicit, if not
174 HUMAN COMMUNICATION RESEARCH / Winter 1984

explicit, in all of the major theories of relationship development.


Whether one is considering the experimenting stage that follows initial
acquaintanceship, the transformations of relationships from acquaint-
ances to friends or lovers, or the reverse transformations that charac-
terize relationship disengagement, the significant feature to note about
relationships is that they experience change and its shadow companion,
uncertainty (Knapp, 1984; Wilmot, 1979). Even so-called stable rela-
tionships may be subject to relationship-level uncertainty. At a minimum,
Kurth (1970, p. 164) noted, relationship parties periodically check to see
if the other party is “holding up his [or her] end.”
Furthermore, existing relationships research suggests that people are
likely to have a multifaceted repertoire of information strategies by
which to cope with relationship uncertainty,just as they have developed
a complex repertoire of passive, active, and interactive strategies in
order to reduce uncertainty about strangers. Extant work suggests that
relationship parties are hesitant to rely on direct talk about the state of
their relationship. In a recent study, Baxter and Wilmot (1984) observed
that the “state of the relationship” was the “taboo topic” mentioned
most frequently among relationships at various stages of development,
emerging as a taboo for about two-thirds of the study’s respondents.
Overwhelmingly, these respondents provided negative conceptions of
direct relationship talk as warrants for avoiding discussion of the state
of the relationship. This absence of direct relationship talk is consistent
with others’ work in metacommunication (Cline, 1979; Watzlawick,
Beavin, & Jackson, 1967; Wilmot, 1980). In the absence of exclusive
reliance on direct talk about the relationship’s state parties must, by
default, develop alternative information acquisition strategies if they are
to cope successfully with relationship uncertainty.
Interestingly, the situational conditions that must be present in order
to implement passive, active, and interactive strategies of information
acquisition are even more pronounced in ongoing relationships than
they are in initial interactions between strangers. The observation of the
other party in the presence of others (a passive strategy) assumes that
one has such social opportunities available. Similarly, asking third
parties about the other party (an active strategy) assumes that one has
familiarity with the other’s social network. In fact, Parks and Adelman
(1983) recently posited that it is such strategic use of the other’s social
network that may have produced their finding that contact with the
social network of one’s relationship partner is the single best predictor of
uncertainty reduction about the other person. Because relationship
parties use more self-disclosure after the initial encounter (Knapp,
Baxter, Wilrnot / SECRET TESTS 175

1984),they should be afforded greater opportunity to use self-disclosure


strategically in order to solicit a reciprocal disclosure from the other
party (an interactive strategy). Similarly, the greater one’s familiarity
with the other’s idiosyncratic rule system, the greater one’s opportunity
for deviation testing (an interactive strategy). In short, nonstranger
relationships provide even greater opportunities for imple-
mentation of passive, active, and interactive information acquisition
strategies than do initial encounters between strangers.
Although relationship uncertainty and the conditions conducive to
the use of information acquisition strategies potentially are present in all
relationship types and stages, this study focuses in particular on platonic
and romantic cross-sex relationships and relationships that are in
transition from platonic t o romantic (“romantic potential”
relationships).’ Relationship uncertainty seems especially likely among
these relationship partners. As Rawlins (1982) noted, the opposite-sex
relationship, in all of its forms, is especially problematic for people
socialized in the American culture. The opposite-sex feature of the
relationship predisposes people to categorize the relationship as roman-
tic and sexual. Thus, a platonic cross-sex relationship requires careful
monitoring to ensure that it does not become romantic, just as a
romantic cross-sex relationship necessitates monitoring to assure that it
does not become platonic. The definitional tension characteristic of
cross-sex relationships is exacerbated by the fact that platonic and
romantic relationship types overlap with many common attributes. As
Wilmot and Baxter (1984) recently observed, platonic friendships and
romantic relationships are not mutually exclusive; rather, they share
certain qualities such as trust, openness, caring, and comfort. Such
shared or conjunctive attributes often make for fuzzy distinctions
between the categories of platonic and romantic relationships that
produce relational uncertainty for the two parties. In addition, opposite
sex pairs are often plagued by the presence of gender differences in the
interpretive rules for social behaviors displayed in the relationship
(Maltz & Borker, 1982). If the man interprets the woman’s touching as a
signal of transformation of the relationship to a romantic basis and she
only intended a friendly display of nonromantic affection, relationship
uncertainty is likely to increase for the pair.
The discussion thus far can be summarized in the first research
hypothesis undertaken in the study:

HI: Passive, active, and interactive information acquisition strategy types


will generalize to the acquisition of information about the state of the
relationship between opposite-sex relationship parties.
176 HUMAN COMMUNICATION RESEARCH / Winter 1984

In addition to aglobal examination of the repertoire of strategies for


acquiring information about the status of opposite-sex relationships,
this study also considers possible differences in reported strategies as a
function of respondent gender and relationship type. Specifically, the
following hypothesis was advanced in terms of gender:

Hz: Females will report more information acquisition strategies by which to


discern the state of the relationship, both with regard to strategies
attributed to self as well as strategies attributed to the other, than wiI1
males.

First, females may be motivated more than males to seek information


about the status of their relationships because of their greater social-
ization to value interpersonal relationships (Gilligan, 1982). Second,
females may be more aware of relationship problems and sources of
relational uncertainty than are their male counterparts (Hill, Rubin, 8c
Peplau, 1976; Peplau, 1983). Thus, given the possibility of greater
awareness of relationship uncertainty coupled with a greater motivation
to reduce that uncertainty, females may employ more information
acquisition strategies than do males. Furthermore, because females may
monitor relationship dynamics more closely than males (Fishman, 1978;
Peplau, 1983),they may also be more aware than males of the strategies
used by their relationship partners.
In addition to a hypothesized difference in the number of reported
strategies, we explore possible differences in the types of strategies that
are employed by males and females. The compliance-gaining literature
appears to support a stylistic gender difference, with females using less
directness than males (see Baxter, 1984). However, the question of
whether such a stylistic difference will generalize to the nonpersuasive
goal of acquiring information about the state of the relationship remains
unanswered. On one hand, females bring a greater expectation of
openness to their relationships than do males (see Peplau, 1983), and
such an expectation may lead to more directness in information
acquisition on the part of females. On the other hand, however, Baxter
and Wilmot (1984) recently observed a tendency for females more than
males to report direct talk about the relationship as a taboo topic in their
developing opposite-sex relationships. Although their female respon-
dents displayed a tendency for fewer reservations in disclosing personal
information, direct relationship talk was perceived as risky if not
damaging. Because of the difficulty in predicting a gender difference in
the directness of information acquisition strategy use, and because of
Baxter, Wilmot / SECRET TESTS 177

the desire to explore other possible strategy use differences as well, the
following research question was advanced:

RQ1: Do males and females differ in the types of information acquisition


strategies that they attribute to themselves?

Although cross-sex relationships are characterized in general by


relational uncertainty, we argue that such uncertainty will be greatest
for relationships that are in between platonic and romantic types (i.e.,
the romantic potential relationship). Unlike the more stable platonic
and romantic types, in which both parties have agreed on the definition
of the relationship, the romantic potential relationship is in transition.
One or both of the relationship parties desires a romantic relationship,
but such a transformation has not been explicitly negotiated by both
parties. Thus, relational uncertainty is high because of the absence of
recognized consensus on the relationship definition. Because infor-
mation-seeking increases with uncertainty (Berger & Calabrese, 1975).
we would expect a greater number of reported information acquisition
strategies reported among romantic potential relationship partners as
opposed to those in the other two relationship types. This argument is
formalized as follows:

Hj: The transitional relationships (i.e., the romantic potential relation-


ships) will have more reported information acquisition strategies than
either the platonic or romantic relationship types.

The final goal of this study is to explore possible differences in the


types of reported information acquisition strategies among platonic,
romantic potential, and romantic relationships. Information acqui-
sition is framed in underlying schemata that assist the actor in deciding
which information is relevant and how it should be interpreted (Wilmot
& Baxter, 1984). In acquiring information about the state of the
relationship, opposite-sex relationship parties have implicit belief
systems about what the relationship is and should be. To the extent that
people have a confirmatory bias in their information acquisition
behavior, and to the extent that platonic and romantic relationship
types are characterized by different attributes, one might reasonably
expect that different strategies would emerge reflecting the underlying
schemata differences. However, as discussed above, platonic and
romantic relationships are characterized by both conjunctive and
disjunctive attributes, contributing to overlap in the two schemata.
Furthermore, the confirmatory bias research has been limited to only
178 HUMAN COMMUNICATION RESEARCH / Winter 1984

the strategy ofdirect interrogation (see Snyder, 1981) and results are not
completely consistent across studies (Trope & Bassok, 1982). Hence, the
issue of differences in strategy type as a function of relationship type is
cast in the form of a research question rather than a research hypothesis:

RQzDo platonic, romantic potential, and romantic relationshipsdiffer in the


types of reported information acquisitionstrategies by which the state of
the relationship is discerned?

METHODS

The study is composed of three distinct stages in which qualitative


and quantitative methodologies were triangulated against one another
in order to afford richer insight into the information acquisition
strategies by which persons discern the state of their relationships (Jick,
1983). The three research stages were as follows:

(I) Qualitative analysis of 90 ethnographic interviews in which respondents


discussed a current or recent opposite-sex relationship, resulting in a
grounded typology of basic types of information acquisition strategies.
(2) Quantitative analysis through cluster analysis and multidimensional
scaling (MDS) of prototypical strategies from Stage 1 that were sorted
on the basis of similarity by an additional 91 research subjects.
(3) A reanalysis of the 90 ethnographic interviewsfrom Stage 1 in light of the
typology of strategies that emerged in Stage 2, examining respondent
gender and relationship type against types of information acquisition
strategies.

STAGE 1

The first stage of the research project involved the collection and
analysis of interview data. A research team of 6 students was trained in
ethnographic interviewing procedures by one of the authors. Three
versions of an open-ended interview schedule were pilot tested in the
field by this research team before the final interview schedule was
developed to collect the actual data of the study. The schedule consisted
of background information about a current (or recently concluded)
opposite-sex relationship in which the respondent was involved, and
open-ended questions that probed any information acquisition stra-
tegies that the respondent could recall using o r that were used by the
other party in the opposite-sex relationship. Additional questions in the
Baxter, Wilrnot / SECRET TESTS 179

interview addressed facets of the relationship beyond the focus of this


particular study. The interviews were approximately 60 to 90 minutes
long. Although interviewers approached each interview with a schedule
of suggested questions and the order in which issues could be
approached, the nature of ethnographic interviewing necessitated
variation from interview to interview as adaptations were made to
respondent remarks (Spradley, 1979). Nonetheless, all interviewers
employed a common introduction of the portion of the interview
relevant to this study:

When people want to assess their relationship, that is, where the
relationship stands and how the parties feel about it, we suspect that they
say or do things designed to shed insight on the matter. For ease of
reference, let’s call these actions that people say or do tests. I’d like to ask
you some questions about your experiences with testing in the opposite-
sex relationship we’ve been talking about. Do you have any questions
before we begin? Do you understand what I mean by a test?

Each interviewer composed detailed accounts based on notes taken


during the interviews. All respondents were assured anonymity in their
interview reponses.
A probability-based sample of 40 males and 50 females was drawn
from the undergraduate population at the institution of one of the
authors. Each interviewer completed 15 interviews from this sample
during September and October of 1983, producing a data corpus of 90
total interviews. During the background portion of the interview
respondents were given descriptions of platonic, romantic potential,
and romantic relationships as the terms were used in the study (see
below) and were asked to categorize appropriately the relationship they
had selected for purposes of the interview. Based on the responses the
sampled relationships were distributed as follows:

( 1 ) Platonic: Relationships in which the respondent indicated that both


relationship parties viewed their relationship as platonic with no desire
by either party to become romantic partners. (n = 12)
( 2 ) Romantic potential: Relationships that were described by the respon-
dent as “more than friends” but that were not yet mutually recognized
and defined as romantic. Typically, these relationships were in transi-
tion, with the respondent reporting that at least one of the parties wanted
to become romantically involved but mutual consensus on redefinition
had not yet been achieved. (n 25)
180 HUMAN COMMUNICATION RESEARCH / Winter 1984

(3) Romantic. Relationships in which the respondent indicated that both


parties explicitly agreed in defining their relationship as romantic. (n =
53)

Although it would have been more desirable to have the relationship


perceptions of both parties from a given relationship, such dual
perception data were not available. However, interviewers probed
respondents for “evidence” of their perceptions to encourage carefully
considered categorizing decisions. The frequency distribution of relation-
ship types did not differ for males and females (x2 2.29; 2 df; p > .30).
The mean duration of the relationships was 20.8 months, with no
significant differences in duration as a function of respondent gender (F
= .0004; 1,84 df; p > .90), relationship type (F = 2.03; 2,84 df; .10 < p <
.20) or gender X type (F = 1.64; 2, 84 df; .20 < p < .30).
A total of 158 information acquisition strategies (or tests, as they
were called in the interviews) were identified by the interview team
members in reading the written records of the 90 interviews. Using
Bulmer’s ( 1 979) method of analytic induction, we independently derived
categories of tests and then sorted the tests into their respective category
sets. Basically, the method of analytic induction involves the devel-
opment of categories from a subset of the data that are then applied to
additional data and modified as necessary until a category set is
developed that captures the entire data set. A total of 143 of the 158
strategies were successfully categorized into 15 basic category types; the
remaining 15 represented miscellaneous categories with small n’s. The
overall agreement in the independent sortings of the tests was .67
(computed as the number of agrees divided by the total number of sorted
items). That is, approximately two-thirds of the tests were placed in
equivalent categories from our independently derived category sets. We
jointly discussed the sorting discrepancies and decided on logical
groupings of the category types into 7 broader types of strategies.

STAGE 2

Stage 2 of the study was undertaken in order to determine the


representational validity (Poole & Folger, 1979) of the typology of tests
developed in Stage 1 of the study. That is, we sought to determine
whether our perceptual organization of the strategies corresponded to the
structure perceived by members of the same population from which the
ethnographic interviews were collected. Both methodological and
theoretical reasons warrant the decision to engage in Stage 2 of the
Baxter, Wilrnot / SECRET TESTS 181

study. Methodologically, the .67 sorting agreement is sufficiently low to


make reliability suspect without additional evidence. Theoretically, the
study is grounded in the ethnographic perspective (Spradley, 1979) and
the social cognition perspective (Roloff & Berger, 1982), both of which
argue the necessity of understanding a phenomenon from the viewpoint
of the insider.
Two pure prototypical instances of each of the 15categories from our
Stage 1 typology were selected for presentation to an additional 91
undergraduates for sorting on the basis of perceived similarity. The
rationale for selecting two prototypical instances per category is
threefold: (1) The original set of 158 tests was regarded as too large for
sorting by the second sample; (2) The researchers sought t o reduce
“noise” in the data by using category prototypes in order to maximize
the opportunity for reliable category recognition; (3) More than one
instance per category was often necessary to capture variations in the
particulars by which a given strategy type was implemented.
The sorters, drawn from introductory communication classes, were
each presented with a card deck composed of 30 tests (2 per type X 15
category types). They were asked to place similar strategies into the
same pile and to use as many or as few piles of cards as they found
necessary to capture their perceptions. After completing the sort
participants were asked to provide labels and short descriptions/ expla-
nations of each pile of tests/ cards.
A data matrix was derived from the sorts that represented the
frequencies with which all pairs of strategies were sorted into the same
category by the 91 sorters. This matrix provides a form of proximity
d a t a appropriate for submission t o cluster analysis and multi-
dimensional scaling (Shepard, 1972;Rapoport & Fillenbaum, 1972). An
agglomerative hierarchical cluster analysis (SPSS, 1983) was employed
on the data matrix to determine the basic groupings of strategies. The
average linkage method was used in the cluster analysis as the
compromise between the extremes afforded by single-linkage and
complete-linkage methods (Bailey, 1974; Krippendorff, 1980). The data
matrix was also submitted to MDSCAL-SM, a nonmetric multi-
dimensional scaling algorithm (Kruskal & Carmone, 1971), in order to
provide a visual representation of the groupings or clusters.

STAGE 3

Of our 15 category types, 14 were validated in Stage 2 of the study.


Because the researchers had already made test-by-test categorization
182 HUMAN COMMUNICATION RESEARCH / Winter 1984

decisions in Stage I, those data were returned to in this final stage of the
study; only the category that was not validated required reanalysis, but
the instances of that category type in Stage 1 were relatively few in
number. The tests from the original 90 interviews were thus organized
into the 7 conceptual types that emerged from Stage 2 of the study and
were assessed for possible differences by respondent gender and
relationship type using the appropriate statistical procedures.

RESULTS

STAGE 1

As described above, we independently sorted the 158 tests generated


in the 90 interviews into inductively derived categories; 15 categories
emerged and were subsequently grouped into 7 basic conceptual types.
Discussion of the 15 categories is presented below within the organi-
zational framework of the 7 broader conceptual types. Each category is
illustrated with the two prototype instances that were selected for
presentation to research participants in Stage 2 of the study. Table 1
summarizes the typology.
The first of 7 conceptual types was labeled Direct Questioning. The
category involves direct relationship talk initiated by one of the parties.
This interview excerpt from a female respondent in a romantic potential
relationship illustrates the category: [Interviewer: Have you ever used
any tests in this relationship?] “Never. Whenever I wanted to know how
he felt, I just asked him” (#103). A male in a platonic relationship
provided the second prototypical response: “Usually, just point-blank
asking questions-‘How do you feel about me?’ ‘How do you see our
relationship?’ (#I 12).

The second conceptual type-Asking Third Parties-involves interro-


gation of social network members about the perceptions of the
relationship by one’s partner. A male respondent in a romantic
relationship provided one prototypical instance of this test type: “I
asked my girlfriend’s sister to tell me anything she could about how my
girlfriend felt about me and our future” (#70). A female in a romantic
potential relationship provided the second instance of this category: “I
went to some of his friends and asked them what he thought of me after
we had dated a few times” (#I43).
The third broad conceptual type of strategies is composed of 3 related
categories of tests and is labeled Trial Intimacy Moves. The first
category of tests contained in this conceptual type is Escalated Touch. A
prototypical instance of this category was provided by a male
Baxter. Wilmot / SECRET TESTS 183

TABLE 1
Summary of Stage 1 Typology of Tests

Category Labels Test #‘s

I. Direct Questioning 103 112


II. Asking Third Parties 70 143
Ill. Trial Intimacy Moves
A. EscalatingTouch 21 31
B. Self-Disclosures 39 114
C. Public Presentation 110 115
I V . Taken for Granted Tests
A. Joking 10 47
B. Initiation Induction 105 128
C. Self-Putdown 27 88
D. Hinting 5 156
V. Endurance Tests
A. Forced Choice 69 151
B. Physical Separation 91 94
C. Testing Limits 17 155
VI. Jealousy Tests
A. Describing Alternatives 6 148
B. Beginning Alternatives 60 72
VII. Fidelity Checks 73 86

respondent in a romantic potential relationship: “By touching her, I


found out whether she was ready for our relationship to go beyond
friends” (#21). A female in a romantic potential relationship described
the second test categorized as Escalated Touch.

When we were just becoming romantic, he kept getting more and more
bold with his touching-first his arm around my shoulder, then moving in
real close, etc. He was waiting to see if I would tell him when to stop as a
sign of how much I liked him. (#3 1)

The second category in the Trial Intimacy Moves group was labeled
Self-Disclosures and involved personal disclosure with the hope of
pressuring the partner into a reciprocal disclosure. A female in a
romantic potential relationship provided one typical instance of this test
type:

I was having trouble getting him to open up to me about his family and his
background. So I started talking about my folks and my sisters, hoping
184 HUMAN COMMUNICATION RESEARCH / Winter 1984

that he would reciprocate and tell me about himself as a sign that he


trusted me and wanted me to know more about him. (#39)

A female in a romantic potential relationship described the Self-


Disclosure test employed by her partner in the following way:

Yeah, he really spilled his guts, telling me how he felt about me. It was to
back me into a corner, to get me to react by saying directly how I felt about
him. (#114)

The final test category grouped in the cluster of Trial Intimacy Moves
was Public Presentation. This test involves a public presentation of the
relationship type by one party to see how the partner will react.
Reflecting on his romantic relationship, a male respondent provided
this illustration of the test type:

At the time, I didn’t know it was a test, but it was. She invited me to her
family’s home for a week. They started to talk to me in terms of beingjust
like a son, how much they hoped we wanted kids someday, etc. She knew
that was coming from them and wanted to see if I would get scared by
hearing her parents talk like that. (#I 10)

A female in a romantic potential relationship provided another instance


of the test type:

1 intentionally introduced him as my boyfriend to my best friend who


came to visit and then watched how he reacted at the time and then
later-you know, did he seem embarrassed? Did he act upset later that I
had made “us” public? (# 1 15)

The fourth conceptual grouping of tests was labeled Taken f o r


Granted and consists of 4 test categories that collectively represent 2
meanings of the phrase, “taken for granted.” Consistent with Hopper’s
(1981) use of the phrase, some of the categories in this grouping rely on
the target’s tacit social knowledge. In addition, however, some of the
categories test the target’s willingness to expend extra attention or effort
directed toward the actor (i.e., not taking the person for granted). The
first category in this grouping-the Joking Test-relies primarily on the
first sense of the phrase. The test type is illustrated in this account
excerpt from a female respondent describing her platonic relationship:
“Whenever I wanted to know how he felt about the relationship, I
wouldn’t ask him outright. Instead, I would use jokes, so that I could
Baxter, Wilmot / SECRET TESTS 185

back out of it if the topic made him uncomfortable” (#lo). Another


female described the Joking Test that her boyfriend employed in their
romantic relationship as follows:

He wants us to get married, but I don’t yet. So to see if I still feel the same
about it, he will joke around and say things like “Figured out yet what
we’ll do with the .5 child in our family of 2.5 children?” If I joke about it,
he knows that I’m not ready to talk seriously about marriage yet, but also
that I’m not pissed off that he keeps pushing me. (#47)

The second type of test grouped in the Taken for Granted category
was labeled Initiation Induction and illustrates the second meaning of
“taken for granted.” Basically, this test involves structuring the situation
so that the burden of meeting rests on the other party. A female
respondent in a romantic relationship described an instance of the test
employed by her boyfriend: “When I returned from Switzerland, he
didn’t call me even though he knew exactly when I was returning. He
wanted to see if I still felt the same way about him by calling him first”
(#105). A male described the same test that he used with his girlfriend: “I
would often wait to see how long it would take her to make contact with
me to see how serious she was about the relationship” (#128).
The third test type that was grouped with the Taken for Granted
cluster was labeled Self-putdown and also illustrates the second sense of
the phrase. In general, this test involves self-deprecation to elicit a
positive statement from the other party. A female in a romantic
potential relationship described the test type in her partner’s behavior:
“He was really getting down on himself, saying that he wasn’t a very
capable or nice person. He said it to get me to compliment him and
verbalize how much I thought of him” (#27). A male described his
actions in his romantic potential relationship as follows: “I sounded
really down to see if she would give me any feedback about how much
she cared for me” (#88).
The fourth category in the Taken for Granted type was labeled
Hinting and presumes that tacit social knowledge will be brought to
bear by the target in interpreting its use. A female in a romantic potential
relationship described the test in this way: “I flirt a lot with him in
letters-you get a lot of courage when you write instead of it being in
person. I wait and see what his reaction is t o the flirting-whether he’s
ready to extend the relationship” (#5). A male in a romantic potential
relationship described adifferent version of the same test: “I met her at a
party. Later in the evening, I went up to her and asked her if she was
186 HUMAN COMMUNICATION RESEARCH / Winter 1984

having a good time at the party and stuff-kind of hinting that I was
ready to leave and was inviting her” (#156).
The fifth grouping of tests was labeled Endurance Tests and consists
of 3 related categories of test types-Forced Choice, Physical Separa-
tion¶and Testing Limits. These 3 categories share the following logic: If
the relationship is made costly to the other party and he or she still stays
in the relationship, that demonstrates the other’s depth of commitment
and feeling. The Forced Choice test type involves forcing the other party
to make a costly choice on behalf of the relationship. A male in a
romantic potential relationship described his partner’s use of the test
type as follows:

During the break, she wrote me to see if she could come and visit me for a
few days. The thing was that I would have to drive 500 miles to pick her
up. It was a test to see how much inconvenience I would stand in the
relationship. (#69)

Another male in a romantic relationship described the same test type


from his girlfriend, but with different particulars:

She came to where I work out and said that she was really depressed but
didn’t want to bother me. She really wanted to see if I would stop what I
was doing and show my concern, indicating how much I care for her.
(#151)

The Physical Separation test type involves a separation from the


other party to determine whether the other’s feelings withstand the
absence. A male respondent employed the test in his romantic rela-
tionship: “Although I didn’t intend it as a ‘test,’ I used our separation
while away at school as a test of how much we cared about each other”
(#91). A female illustrated an intentional use of the same test type in her
romantic relationship: “To see if our relationship was really strong, I
tested it by intentionally separating us by going overseas for a few
months” (#94).
The third category of Endurance Tests-Testing Limits-involves
rule deviation to determine how much the other party will withstand. A
female in a romantic potential relationship described the test type in her
partner’s behavior: “To see if I Really liked him a lot, he kept pushing me
too far to see if I would still like him despite his actions”(#17). Another
female described the same test employed by her boyfriend: “He will test
Baxter, Wilmot / SECRET TESTS 187

how much I love him by showing his worst side to me. If I’m still around
after that, it means that my feelings are not just superficial” (#155).
Two test categories comprise the sixth grouping of Jealousy Tests.
The first category-Describing Alternatives-involves the verbal de-
scription of a potential competitor to the other party. A female
respondent in a romantic relationship provided a typical instance of the
test category: “I talked about my other boyfriend back home in his
presence to see how he’d react-whether he would get jealous or not”
(#6). Another female provided this description.

When we had been dating a while and I had decided that I was serious
about him, I talked about my ex-boyfriend to see if I would make him
angry. That way, I’d know that he was serious about me, too (#148)

The second category of Jealousy Tests was labeled Beginning Alterna-


tives. In contrast to simply describing possible rivals, this test type
involves the physical presence of a potential competitor. A male
respondent described the test type as follows in reflecting on his
behavior in a romantic relationship: “I tested her limits by going out
with other women and making sure that she found out about it. I wanted
to know what kind of relationship commitment she wanted” (#60). A
female used the same test against her boyfriend-to-be: “Before we
became boyfriend and girlfriend, I intentionally went out with another
guy and made sure that he would find out about it. I wanted him to get
jealous” (#72).
The final conceptual cluster was labeled Fidelity Checks, and consists
of a single category of the same label. A female respondent provided a
detailed description of the test type in reflecting on her romantic
relationship:

I would intentionally leave him alone in the room with my roommate and
then ask him when I returned“What have the two of you been up to?”If he
acted uncomfortable, I’d know that he wasn’t faithful (or at least that he
was thinking of being unfaithful to me). (#73)

A male provided an equally detailed description of the same test used


against his girlfriend:

We went for a walk with two brothers who are friends of mine. I walked
ahead with the younger brother and left her to walk with the older brother
who had expressed a more than passing interest in her as a lover. I wanted
to see her reactions. (#86)
188 HUMAN COMMUNICATION RESEARCH / Winter 1984

The 15 categories of tests, and their logical grouping into 7 broader


conceptual types, reflected the analyses of the researchers. As discussed
above, the second stage of the study was conducted to determine the
representational validity of this inductively derived typology.

STAGE 2

The 30 tests presented above were given to 91 additional under-


graduates who were asked to sort the tests into piles on the basis of their
similarity. The composite frequencies of all pairwise test combinations
were submitted to hierarchical cluster analysis, Figure 1 presents the
dendrogram of successive clusters that emerged from this statistical
procedure. The figure displays the emergence of hierarchical clusters,
beginning with the left-most margin in which there are 30 “clusters,”
each of which has a within-group distance of zero. Obviously, the
distance for each of these 30 clusters is zero, because each of these initial
clusters consists of a single test item. As one scans Figure 1 from left to
right, it is possible to trace the successive emergence of larger clusters
through combinations of items.
Of the 15 basic categories identified in Stage 1, 14 were validated by
the research participants who sorted tests. The category of Forced
Choice (tests #69 and #151) did not emerge in the cluster analysis.
Instead, the category was split, with each item clustering with a different
grouping. Test #69, a request that the other party drive 500 miles to
provide a ride, displays a physical separation of the relationship parties
that perhaps accounts for its clustering with other tests involving
physical separation. Item #151, inconveniencing the other party by
suggesting that he sacrifice his workout, tests the other’s limits and
hence was associated with the other endurance-related tests.
The process of determining the number of clusters to interpret is
highly subjective (Bailey, 1974). In examining Figure 1, it is apparent
that anywhere from 5 to 14 clusters could be used to summarize the data.
However, the primary goal of cluster analysis is parsimony, suggesting
the interpretation of a smaller as opposed to a larger number of clusters.
Simultaneously, however, the number of clusters that is accepted should
be easily interpretable (i.e., the accepted clusters should each be
characterized by logical coherence). Given these two criteria, a 7-cluster
solution was accepted. These 7 clusters are indicated in Figure 1 and are
also listed in Table 2. As Figure 1 illustrates, an alternative analysis
would be the acceptance of 5 rather than 7 clusters. In the 5-cluster
Baxter. Wilmot / SECRET TESTS 189

SPSS-X Re-scaledDistance for Cluster ComMnatlons

Test tl

I 770

~ 112
1111 39

94
IV 1 105

v i' 155
27
88

Figure 1 Dendrogram of emergent hierarchical clusters with average linkage method.

solution, Separation Tests and Endurance Tests merge into a single


cluster and Public Presentation Tests merge with Indirect Suggestion
Tests. Although more parsimonious, this 5-cluster solution lacks the
theoretical coherence of the 7-cluster solution that was accepted. Both
of the mergers involved in the 5-cluster solution involve ignoring the
distinction between active and interactive strategies that is evident in
extant work in information acquisition strategies (see Berger & Bradac,
1982). Separation Tests and Public Presentation Tests are active
strategies, whereas the other two clusters represent interactive strate-
gies.
The 7 clusters both confirm and alter the typology developed in Stage
1 of the study. The first Stage 2 cluster-Asking Third Parties-is
identical to grouping I1 from the Stage 1 analysis. The second cluster-
labeled Triangle Tests because they involve 3-person triangles-is a
merging of groupings VI and VII (Jealousy Tests and Fidelity Checks,
respectively) from the Stage 1 analysis. The third cluster-labeled
190 H U M A N C O M M U N I C A T I O N RESEARCH / Winter 1984

TABLE 2
Summary of Stage 2 Typology of Tests
-~ - ~~.

Category Labels Test s'#


~ -

I. Asking Third Part Tests 70 143


It. Triangle Tests
A. Fidelity Checks 73 86
B. Jealousy Tests 6 60
72 148
Ill. Directness Tests
A. Direct Questioning 103 112
B . Self-Disclosures 39 114
I V . Separation Tests
A. Physical Separation 91 94
B. Initiation Induction 105 128
C. Forced Choice #1 69
V. Endurance Tests
A. Testing Limits 17 155
6 . Self-putdown 27 88
C. Forced Choice #2 151
VI: Public Presentation Tests 110 115
V I I . Indirect Suggestion Tests
A. Joking 10 47
8. EscalatingTouch 21 31
C. Hinting 5 156

Directness Tests-consists of Direct Questioning (group I from Stage 1


analysis) and the Self-Disclosures category. Rather than perceiving
Self-Disclosures as Trial Intimacy Moves, the sorters clearly focused on
the openness quality of the disclosures. The fourth cluster to emerge-
Separation Tests-consists of the categories of Physical Separation and
Initiation Induction, plus test #69 from the Forced Choice category.
All of these tests involve physical distance between the relationship
parties, and it is this feature that is captured in the cluster label. The fifth
cluster-Endurance Tests-consists of tests that dramatically reduce
the rewards to the other party in the relationship. The category of
Testing Limits is consistent with the Stage 1 analysis. However, it is
apparent that the sorters perceived an endurance test in the Self-
Putdown tests as well as in Forced Test #151. Public Presentation
emerged as a unique cluster in Stage 2, in contrast to its categorization as
Baxter, Wilrnot / SECRET TESTS 191

a Trial Intimacy Move in the Stage 1 analysis. The last cluster-Zndirect


Suggestion Tests-contains the Joking and Hinting categories that were
grouped as Taken for Granted in the Stage 1 analysis. However, the
category of Escalated Touch was perceived as indirect suggestion rather
than as a trial intimacy move. Furthermore, the category of Initiation
Induction was not perceived by sorters as a Taken for Granted test.
To provide an alternative representation of the clusters, the data
matrix submitted to hierarchical cluster analysis was also submitted to
nonmetric multidimensional scaling (MDS). According to Krippen-
dorff (1980, p. 276):

By far the most appealing form of representationdepicts the proximities


among objects in some space and indicates clusters by drawing their
boundaries. Since proximities are an essential ingredient of Gestalt
perception, groupings are much easier to visualize when similarities,
correlation, and the like are expressed as distances.

MDS solutions were obtained for 1 to 5 dimensions, with stress


(formula 2) values of .453, .240, .188, .162, and .144, respectively. The
criteria for selecting the appropriate MDS solution are numerous and
could all be characterized as rules of thumb. All 5 of the solutions have
sufficient recovery capability based on the ratio of the number of
stimulus points to number of dimensions (Sherman, 1972). Shepard’s
(1972) visualizability criterion suggests no more than a three-dimen-
sional solution. Kruskal’s (1964) guideline of a .05 stress value for an
excellent solution and a .10 level for a good solution leads to the
rejection of all 5 solutions; however, as Young (1970) has observed, this
guideline is unnecessarily stringent. The “elbow of the stress curve”
criterion (Kruskal, 1964; Isaac & Poor, 1974) suggests acceptance of the
two-dimensional solution. The constraint index advanced by Isaac and
Poor (1974) also suggests adoption of the two-dimensional solution.
The two-dimensional solution was accepted and is reported in Figure
2. This figure provides avisual display of the Stage 2 results. Boundaries
are drawn around the 7 clusters based on the results of the hierarchical
cluster analysis presented above. Based on a subjective reading of the
labels and descriptions/ explanations that sorters provided with their
card sort decks, the X-axis in Figure 2 appears to capture the perceived
directness of the test, with Triangle Tests and Directness Tests
representing the two polar extremes. The Y-axis in Figure 2 appears to
reflect whether or not the parties were in one another’s presence for the
192 HUMAN COMMUNICATION RESEARCH / Winter 1984

200

100

-100

-200

Figure 2 2-Dimensional MDS solution for 30 test prototypes.

test, with Separation Tests and Indirect Suggestion Tests representing


the extremes along this dimension.

Stage 3

Because the 158 tests generated in the Stage 1 interviews had already
been categorized into the I5 basic test types, they were simply recoded
during this phase of the study into the appropriate cluster based on the
results of the Stage 2 hierarchical cluster analysis that indicated which
cluster a given test category was in. The single exception to this
procedure was the category that was not validated in Stage 2-Forced
Choice. The 8 tests in addition to #69 and #151 were categorized into
either Cluster IV or Cluster V.
Baxter, Wilmot / SECRET TESTS 193

A two-way ANOVA was used to assess possible gender and


relationship type differences in the number of reported strategies. The
sex X type interaction was not significant ( F = .19; 2, 84 df; p > .80).
However, both main effects were significant beyond the .05 level.
Females (mean = 1.98) reported significantly more tests than males
(mean = 1.48; F = 7.96; I , 84 df; p < .01; eta2 = .078). However, the
number of reported tests included tests attributed to self as well as tests
attributed to the other. A paired t-test indicated that respondents were
significantly more likely to report self-enacted tests (mean = 1.32) as
opposed to other-enacted tests (mean = .43; t = 7.23, 89df; p = .001).
Males (mean = .425) and females (mean = .440) did not differ on the
number of reported tests attributed to their relationship partners (t = .26,
88df; p > .70). However, females (mean = 1.54) reported significantly
more self-attributed tests than did males (mean = 1.05; t = 2.92, 88df;
p=.005).
The main effect for relationship type was significant as well (F = 6.41,
2, 84df; p < .01; eta2 = .13). Subsequent Schefft: tests indicated that
romantic potential relationships (mean = 2.32) reported significantly
more tests than either platonic relationships (mean = 1.17) or romantic
relationships (mean = 1.62). Self-attributed and other-attributed tests
were not distinguished from one another because the analysis of
relationship type is a relationship-level, not individual-level, variable.
Table 3 reports the proportion of respondents who reported at least
one test in each of the 7 test clusters. Overall, Endurance Tests were the
most widely reported, followed closely by Triangle Tests and Indirect
Suggestion Tests. Overwhelmingly, respondents employed secret tests,
given that only 20.9% of them reported a test that was perceived as
direct.
Table 4 provides a breakdown of the proportion of each gender
reporting at least one self-enacted test in each of the 7 clusters. Because
there were only 39 total tests attributed to the other party, a table
reporting other-enacted test types by respondent gender is not provided.
Because respondents could report multiple tests, including more than
one test in a given cluster, chi-square tests were performed on each
cluster type separately, comparing the proportions of males who did and
did not report at least one test in a given cluster against the proportions
of females who did and did not report at least one test in the cluster.
Females were significantly more likely to report self-enacted Triangle
Tests and Separation Tests than were males.
Table 5 provides a summary of the proportions of platonic, romantic
potential, and romantic relationship types reporting at least one test in
194 H U M A N C O M M U N I C A T I O N RESEARCH / Winter 1984

TABLE 3
Proportion of Respondents Reporting a t Least
One Test in Each Cluster

PToportion of
Cluster Respondents

Asking Third Party Tests 12.8


Triangle Tests 31.4
Directness Tests 20.9
Separation Tests 24.4
Endurance Tests 34.9
Public Presentation Tests 5.8
Indirect Suggestion Tests 27.9

NOTE: Proportions do not sum to 100.0 because a given respondent could report
multieie tests.

each of the 7 clusters. Because of the same nonindependence issue in the


data described above for gender comparisons, separate chi-square tests
were performed for each cluster type and are reported in Table 5 .
Romantic potential relationships were more likely to employ Separa-
tion Tests and Indirect Suggestion Tests than were the platonic and
romantic relationship types.

DISCUSSION

Through a combination of qualitative and quantitative methods, we


developed a set of 14types of information acquisition strategies that were
also perceived by members of the respondent population, by which
people determine the state of their opposite-sex relationships. These 14
strategy types were further reduced to 7 clusters of strategies, with some
discrepancy between our “outsider” organizing framework and the
“insider” organizing framework of the undergraduate participants.
Given the grounding of the current study in ethnographic and social
cognition perspectives, we opted for insider perceptions above our own
in assessing reported strategy use as a function of respondent gender and
relationship type. After all, it was the use of the strategies by our
research participants that was of interest, and their actions were guided
by their own perceptual frames, not ours.
The participants’ perceptions were also configured through MDS
analysis, producing two apparent underlying dimensions by which the
strategy types and clusters were perceptually organized: Direct/ Indirect
Baxter, Wilmot / S E C R E T T E S T S 195

TABLE 4
Summary of Proportion of Males and Females
Reporting Self-EnactedTest Types

Chi-square
Cluster Label Males Females Value‘

Asking Third Party 7.5 10.0 -**


Triangle Tests 7.5 28 .O 4.83***
Directness Tests 25 .O 12.0 1.76
Separation Tests 10.0 30.0 4.19***
Endurance Tests 15.0 20.0 .1 1
Public Presentation 2.5 2.0 -*I
Indirect Suggestion 20.0 24.0 .04

*All chi-square tests Include Vate’s correction with 1 df.


**Chl-square violatlon with more than 20% o f cells wlth expected frequency less
than 5 .
***p < .05.

and Interactive/Not Interactive. However, the relatively informal


procedure brought to bear in labeling these two underlying dimensions
should lead to caution in their acceptance. Subsequent work needs to
employ more structured methods, such as rating scales, to determine the
dimensions of strategy perceptions. Nonetheless, these two dimensions
have been noted by others. The Interactive/ Not Interactive dimension
appears to correspond to the distinction between active and interactive
strategies made by Berger and his colleagues in their work on
information acquisition strategies (see Berger & Bradac, 1982). And the
Direct/ Indirect dimension has been posited as a universal feature of
communicative behavior (Brown & Levinson, 1978).
The repertoire of information acquisition strategies that people
employ to discern the state of their opposite-sex relationships can aptly
be referred to as secret tests given the relatively low proportion of Stage
1 respondents who reported direct strategies. Perhaps the interview
procedure of Stage 1 biased respondents against reporting straight-
forward relationship talk in its use of the term test. However, the
apparent pervasiveness of secret tests is consistent with extant work in
metacommunication that suggests that direct relationship talk is an
infrequent phenomenon (Cline, 1979; Wilmot, 1980).
The widespread reporting of Endurance Tests and Triangle Tests by
Stage 1 respondents perhaps suggests that these are the most likely
“scripted”strategies that will be called up when a person is experiencing
relational uncertainty. On the assumption that secret tests are framed in
196 HUMAN COMMUNICATION RESEARCH / Winter 1984

TABLE 5
Summary Proportions of Relationship Types Reporting Test Types

Romantic Chi-square
Cluster Label Platonic Potential Romantic Value

Asking Third Parties a .3 a .o 15.1 -


Triangle Tests 16.7 28 .o 34.0 1.46
Directness Tests 16.7 16.0 22.6 .56
Separation Tests 8.3 44.0 17.0 8.67**
Endurance Tests 41.7 32.0 32.1 .43
Public Presentation 8.3 4.0 5.7 -*
Indirect Suggestion 16.7 52.0 17.0 12.85""'
-
*Chi-square violation with more than 20% of the cells havlng expected frequenctes
less than 5 .
**p < .05.
***p < .01.
underlying relationship schemata, these secret test types perhaps suggest
that fidelity and loyalty at all costs are salient attributes in these
respondents' opposite-sex relationship schemata. However, one cannot
dismiss the possibility that these test types were reported by many
simply because of a method artifact. By asking Stage 1 respondents to
report test instances retrospectively, interviewers were likely to solicit
only the most salient and memorable examples (Nisbett & Ross, 1980).
Certainly, jealousy and escalated relational costs are likely to be
memorable.
In turning to the hypotheses and research questions, it is apparent
that the first hypothesis was supported. Both Stage 1 and Stage 2
typologies contain features similar to those discussed in existing work
(Snyder, 1981; Berger & Bradac, 1982). Several of the specific secret test
categories replicate strategies noted in extant research; specifically,
Asking Third Parties, Direct Questioning, Testing Limits, and Self-
Disclosures. At a more general level passive, active, and interactive
features are also evident in the secret test types. Fidelity Checks and
Physical Separation Tests may be either passive or active in nature,
depending on whether the actor takes advantage of conditions already
present or proactively creates appropriate circumstances for the tests.
Asking Third Parties, Triangle Tests, Separation Tests, and Public
Presentation Tests are likely to involve proactive environmental
structuring of one kind or another. Last, Directness Tests, Endurance
Tests, and Indirect Suggestion Tests are characterized by their inter-
active quality. 'Thus, all of the features posited by Berger and his
colleagues have emerged in the secret test strategies, and several of the
specific strategies have emerged as well.
Baxter, Wilmot / SECRET TESTS 197

On the face of it, then, it would appear that the information


acquisition strategies do generalize across different logical types of
information-knowledge about an individual person in extant work
and knowledge of the state of the relationship in the current study.
However, a closer examination of the secret tests indicates that most
were strategies directed at understanding an individual-level phenom-
enon-another person’s attitudes and feelings toward the relationship.
With the possible exception of the Separation Tests, actors did not test
the relationship as a unit; rather, the other party was tested. To test the
relationship as a unit, a person would have to consider jointly enacted
actions by the two parties.
The second hypothesis was partially supported. Females reported
significantly more self-attributed secret tests than did males. However,
females and males did not differ in the number of secret tests attributed
to their relationship partner. The use of tests by females more than by
males may reflect their greater relationship monitoring that was
referenced in the first part of the article. It is through secret tests that
females monitor a relationship’s pulse. Furthermore, one might expect
the monitoring female to report more tests by her partner, whereas there
were no differences between males and females in attributing tests to the
other. This may be accounted for in two ways. First, females are more
relationally attuned, but males use fewer tests. The net result is no
difference because the less attuned males do not perceive the presence of
more tests by the females, and the more sensitive females have fewer
male tests to identify. Or, alternatively, it may simply be that secret tests
are so secret that the test recipient does not categorize the behavior as a
test, thereby obviating any male-female differences in detection of tests.
The third hypothesis was supported. Relationships in transition
between platonic and romantic types had the largest number of secret
tests. This finding may reflect the high uncertainty that characterizes a
relationship undergoing change. However, the uncertainty may be
intensified for opposite-sex relationships that are shrouded in a “sexual
agenda” (Rawlins, 1982) that complicates the platonic-romantic
transformation.
In response to the first research question, gender differences were
noted on the type of secret test reported. Females more often than males
reported use of Triangle Tests and Separation Tests. The difference on
Separation Tests may be a function of the lingering belief that males, not
females, should be relationship initiators. And the difference on
198 HUMAN COMMUNICATION RESEARCH / Winter 1984

Triangle Tests may be a by-product of gender socialization difference in


which boys function easily in social groups of three or more, whereas
girls organize their social network in dyads and experience greater
difficulty with triads or larger social units (Maltz & Borker, 1982).
However, no significant difference emerged in the use of Directness
Tests. When the topic is the state of the relationship, females do not
appear to display the openness identified with them in other communica-
tion situations.
The second research question received an affirmative response as
well. Romantic potential relationships were more likely than platonic or
romantic relationships to display two test types, Separation Tests and
Indirect Suggestion Tests. The more widespread use of separation-
related tests is understandable given that the parties are probably still in
the process of negotiating time together versus time apart. Unlike in
established relationships, the failure to initiate contact may still be a
relatively common behavior that allows the test to be secret and, thus,
safe. Indirect tests, such as joking and hinting, also appear to be
relatively low-risk secret tests. Relationships undergoing change are
likely to be fragile, with the parties unwilling to engage in risky behavior.
Directions for subsequent work are several. First, it would be useful
to determine the extent to which people display Snyder’s (1981)
confirmatory bias in their testing behavior. If a person wants to see
evidence of transformation to a romantic relationship, will biased tests
be employed that make it more difficult for the other party to “fail’?
Second, as noted above, it would be useful to determine whether and
how people test their relationship as a unit in contrast to acquiring
information about the other party’s feelings. Third, research should
consider possible links between the underlying relationship schemata
that people hold and their secret testing behavior; secret tests may
provide rich meta-information about the beliefs that people have about
relationships.

NOTE

1. We recognize that relationships can also experience a transition from romantic to


platonic status. However, as Duck (1982) has observed, former romantic partners may
describe their status as one of friendship for purposes of public “gravedressing” only.
Because of the difficulty in distinguishing actual friendship from grave-dressing friend-
ship, we avoided this form of relationship transition.
Baxter, Wilrnot / SECRET TESTS 199

REFERENCES

BAILEY, K. D. (1974). Cluster analysis. In D. R. Heise (Ed.), Sociologicalmethodology


1975. San Francisco: Jossey-Bass.
BAXTER, L. A. (1984). An investigation of compliance-gaining as politeness. Human
Communication Research, 10,427-456.
BAXTER, L. A., & WILMOT, W. W. (in press). “‘Taboo topics’in close relationships:
An exploratory investigation.” Journal of Social and Personal Relationships.
BERGER, C. R. (1979). Beyond initial interaction: Uncertainty, understanding, and the
development of interpersonal relationships. In H. Giles & R. St. Clair (Eds.), Language
and social psychology. Oxford: Blackwell.
BERGER, C. R. (1982). Social cognition and the development of interpersonal
relationships: The quest for social knowledge. Paper presented at the First
International Conference on Personal Relationships, Madison, WI.
BERGER, C. R. & BRADAC, J. J. (1982). Language and social knowledge: Uncertainty
in interpersonal relations. London: Arnold.
BERGER, C. R., & CALABRESE, R. J. (1975). Some explorations in initial interaction
and beyond: Toward a developmental theory of interpersonal communication.
Human Communication Research, 1. 99-1 12.
BERGER, C. R., & DOUGLAS, W. (1981). Studies in interpersonal epistemology, 111:
Anticipated interaction, self-monitoring, and observational context selection. Commu-
nication Monographs, 48, 183-196.
BERGER, C. R., GARDNER, R. R., CLATTERBUCK, G. W., & SCHULMAN, L. S.
(1976). Perceptions of information sequencing in relationship development. Human
Communication Research, 3, 29-46.
BERGER, C. R., & KELLERMANN, K. A. (1983). To ask or not to ask Is that a
question? In R. N. Bostrom (Ed.), Communication yearbook 7. Beverly Hills, CA:
Sage.
BEKGER, C. R., & PERKINS, J. W. (1978). Studies in interpersonal epistemology, I:
Situational attributes in observational context selection. In B. D. Ruben (Ed.),
Communication yearbook 2. New Brunswick, NJ: Transaction Books.
BERGER, C. R., & PERKINS, J. W. (1979). Studies in interpersonal epistemology, II:
Selfmonitoring, involvement, facial affect, similarity. and observational context
selection. Paper presented at the annual convention of the Speech Communication
Association, San Antonio, TX.
BROWN, P., & LEVINSON, S. (1978). Universals in language usage: Politeness
phenomena. In E. Goody (Ed.), Questions and politeness: Strategies in social
interaction. New York: Cambridge University Press.
BULMER, M. (1979). Concepts in the analysis of qualitative data. Sociological Review,
27, 651-677.
CLINE, T. (1979). A Markhov analysis of strangers’, room-mates’,and married couples’
conversational focus on their relationships. Southern Speech Communication Jour-
nal, 45, 55-68.
DUCK, S. (1982). A typology of relationship disengagement and dissolution. In S. Duck
(Ed.), Personal relationship 4: Dissolving personal relationships. New York:
Academic.
200 HUMAN COMMUNICATION RESEARCH / Winter 1984

FISHMAN, P. (1978). Interaction: The work women do. Social Problems, 25, 397-406.
FORGAS, J. P. (1981). Social cognition: Perspectives in everyday understanding. New
York: Academic.
GILLIGAN, C. (1982). In a different voice. Cambridge, MA: Harvard University Press.
HARVEY, J. H., ICKES, W., & KIDD, R. F. (1976). New directions in affriburion
research (Vol. 1). Hillsdale, NJ: Erlbaum.
HARVEY, J. H., ICKES, W., & KIDD, R. F. (1978). New directions in attribution
research (Vol. 11). Hillsdale, NJ: Erlbaum.
HARVEY, J. H., ICKES, W., & KIDD, R. F. (1981). New directions in attribution
research (Vol. 111). Hillsdale, NJ: Erlbaum.
HEIDER, F. (1958). Thepsychology of interpersonal relations. New York: John Wiley.
HIGGINS, E. T., HERMAN, C. P., & ZANNA, M. P. (1981). Social cognition: The
Ontario symposium (Vol. 1). Hillsdale, NJ: Erlbaum.
HILL, C. T., RUBIN, Z., & PEPLAU, L. A. (1976). Breakups before marriage: The end of
103 affairs. Journal of Social Issues. 32, 147-167.
HOPPER, R. (1981). The taken-for-granted. Human Communication Research, 7,
195-211.
ISAAC, R., & POOR, D. (1974). On the determination of appropriate dimensionality in
data with error. Psychometrika, 39, 91-109.
JICK, T. D. (1983). Mixing qualitative and quantitative methods: Triangulation in action.
In J. Van Maanen (Ed.), Qualirative methodology. Beverly Hills, CA: Sage.
JONES, E. E., & DAVIS, K. E. (1965). From acts to dispositions: The attribution process
in person perception. In L. Berkowitz (Ed.), Advances in experimental social
psychotogy. New York: Academic.
KELLERMANN, K. A., & BERGER, C. R. (1984). Affecr and the acquisition of social
information: Sit back, relax. and tell me about yourself: Paper presented at the
International Communication Association Convention, San Francisco.
KELLEY, H. H.(1967). Attribution theory in social psychology. In D. Levine (Ed.),
Nebraska symposium on motivation. Lincoln: University of Nebraska Press.
KELLY, G. H. (1955). The psychology ofpersonal constructs. New York: Academic.
KNAPP, M. L. (1984). Interpersonal communication and human relationships. Boston:
Allyn & Bacon.
KRIPPENDORFF, K. (1980). Clustering. In P. R. Monge & J. N. Cappella (Eds.),
Multivariate techniques in human communication research. New York: Academic.
KRUSKAL, J . B. (1964). Multidimensional scaling by optimizing goodness of fit to a
nonmetric hypothesis. Psychometrika, 29, 3-10.
KRUSKAL, J . B., & CARMONE, F. (1971). How to use M-D-SCAL (version SM)and
other useful information. Murray Hill, NJ: Bell Telephone Laboratories.
KURTH, S. (1970). Friendships and friendly relations. In G. McCall et al., Social
relationships. Chicago: Aldine.
MALTZ, D. N., & BORKER, R. A. (1982). A cultural approach to male-female
miscommunication. In J. J. Gumperz (Ed.), Language and social identity. New York:
Cambridge University Press.
NISBETT, R. E., & ROSS, L. (1980). Human inference: Strategies and shortcomings of
socialjudgment. Englewood Cliffs, NJ: Prentice-Hall.
PARKS, M. R., & ADELMAN, M. B. (1983). Communication networks and the
development of romantic relationships: An expansion of uncertainty reduction theory.
Human Communication Research, 10, 55-79.
Baxter, Wilmot / SECRET TESTS 201

PEPLAU, L. A. (1983). Roles and gender. In H. H. Kelleyet al., Closerelationships. San


Francisco: W. H. Freeman.
POOLE, M. S., & FOLGER, J. P. (1979). A model for establishing the representational
validity of interaction coding schemes, opus I: Do we see what they see? Paper
presented at the annual convention of the Speech Communication Association, San
Antonio, TX.
RAPOPORT, A., & FILLENBAUM, S. (1972). An experimental study of semantic
structures. In A. K. Romney, R. N. Shepard, & S. B. Nerlove (Eds.), Multidimensional
scaling: Theory and applications in the behavioral sciences (Vol. 2). New York:
Seminar Press.
RAWLINS, W. (1982). Cross-sex friendship and the communicative management of
sex-role expectations. Communication Quarterly, 30, 343-352.
ROLOFF, M. E., & BERGER, C. R. (1982). Social cognition and communication.
Beverly Hills, CA: Sage.
SHEPARD, R. N. (1972). A taxonomy of some principal types of data and of
multidimensional methods for their analysis. In R . N. Shepard, A. K. Romney, & S. B.
Nerlove (Eds.), Multidimensional Scaling: Theory and applications in the behavioral
science (Vol. 1). New York: Seminar Press.
SHERMAN, C. R. (1972). Nonmetric MDS: A Monte Carlo study of the basic
parameters. Psychometrika, 37,323-355.
SNYDER, M. (1981). Seek and ye shall find: Testing hypotheses about other people. In E.
T. Higgins, C. P. Herman, & M. P. Zanna (Eds.), Social cognition: The Ontario
symposium. Hillsdale, NJ: Erlbaum.
SNYDER, M., & CAMPBELL, B. (1980). Testing hypotheses about other people: The
role of the hypothesis. Personality and Social Psychology Bulletin, 6,421-426.
SNYDER, M., & CANTOR, N. (1979). Testing hypotheses about other people: The use of
historical knowledge. Journal of Experimental Social Psychology, 15, 330-342.
SNYDER, M., & SWANN, W. B., Jr. (1978a). Behavioral confirmation in social
interaction: From social perception to social reality. Journal of Experimental Social
Psychology, 14, 148- 162.
SNYDER, M., & SWANN, W. B., Jr. (1978b). Hypothesis-testing processes in social
interaction. Journal of Personality and Social Psychology, 36, 1202-1212.
SPRADLEY, J. (1979). The ethnographic interview. New Y o r k Holt, Rinehart &
Winston.
SPSS, Inc. (1983). SPSSX: User’s guide. New York: McGraw-Hill.
TROPE, Y., & BASSOK, M. (1982). Confirmatory and diagnosing strategies in social
information gathering. Journal of Personality and Sociaf Psychology, 43, 22-34.
WATZLAWICK, P., BEAVIN, J., & JACKSON, D. (1967). Pragmatics of human
communication. New York: Norton.
WILMOT, W. (1979). Dyadic communication (2nd ed.). Reading, MA: Addison-Wesley.
WILMOT, W. (1980). Metacommunication: A re-examination and extension. In D.
Nmmo (Ed.), Communication yearbook 4. New Brunswick, NJ: Transaction Books.
WILMOT, W., & BAXTER, L. (1984). Definingrelationships: The interplay of cognitive
schemata and communication. Paper presented at the annual convention of the
Western Speech Communication Association, Seattle, WA.
YOUNG, F. W. (1970). Nonmetric MDS: Recovery of metric information. Psycho-
metrika, 35,455-473.

You might also like