Você está na página 1de 23

Personal Relationships, 1 (1994), 309-331. Printed in the United States of America.

Copyright 0 1994 Cambridge University Press. 1350-4126/94 $5.00 + .OO

Prototype-based assessment of
laypeople’s views of love

BEVERLEY FEHR
University of Winnipeg

Abstract
Empirical work on love has focused mainly on romantic/passionate love. Recent research suggests that other
kinds of love, such as friendship love and familial love, may be more salient to laypeople (Fehr & Russell,
1991). One purpose of this research was to offer a more complete picture of how laypeople conceptualize love
by exploring a broad range of types of love. The other major purpose was to develop a methodology for
studying laypeople’s conceptions of different kinds of love. Four studies were conducted. In Study 1,dating
couples were presented with prototypes of 15 different types of love. They were asked to rate how similar the
conception of love depicted in each prototype was to their own view of love. Study 2 was a replication of Study
1with subjects who were not dating one another. In Study 3, the validity of the prototype measures was
explored by asking subjects to determine the kind of love depicted in each prototype. Study 4 investigated the
relation between the prototypes of love and existing love scales and therefore addressed issues of convergent
and discriminant validity. These prototype-based measures showed considerable promise as valid, reliable
instruments for assessing people’s views of love. They revealed that laypeople regard friendship love and
familial kinds of love as closest to their own view. Passionate kinds of love (e.g., passionate, infatuation, puppy
love) received the lowest ratings. The measures also provcd useful in elucidating the relation between
laypeople’s and experts’ conceptions of romantic and passionate love.

Attempts to assess people’s views of love sess the experience of passionate love.
date back to at least 1944 when Gross de- Scales to distinguish love from liking were
veloped a scale to measure romantic love published by Rubin (1970). More recently,
beliefs. Since then a number of scales to Hendrick and Hendrick (1986) created
measure romantic beliefs or attitudes have scales for each of six love styles as concep-
been constructed (e.g., Dion & Dion, 1985; tualized by Lee (1973):Eros (romantic, pas-
Knox & Sporakowski, 1968; Munro & sionate love), Storge (friendship love),
Adams, 1978; Sprecher & Metts, 1989).This Ludus (game-playing love), Pragma (logi-
work also spawned the construction of cal, practical love), Mania (possessive, de-
measures of related kinds of love and love pendent love), and Agape (altruistic love).
experiences. For example, Kanin, Davidson, Most of the scale construction and vali-
and Scheck (1970) and Hatfield and Spre- dation research continues to focus on ro-
cher (1986) developed instruments to as- mantic/passionate love, and the result has
been a marked increase in our knowledge
of romantic love and its relation to impor-
This rcsearch was supported, in part, by a University of
Winnipeg-Social Sciences and Humanities Research tant relationship processes (see, e.g., Stern-
Council grant. I am indebted to Mark Baldwin, Keith berg & Barnes, 1988). However, one won-
Davis, and three anonymous reviewers for their very ders whether romantic love is as salient to
helpful comments on an carlicr draft of this article. the participants in these studies as it is to
Correspondence concerning this article should be love researchers. Does romantic love come
sent to Bcvcrley Fehr, Department of Psychology, Uni-
versity of Winnipeg, Winnipcg,Manitoba,Canada,R3B to mind when research subjects are asked
2E9. to think about love? Some hints are pro-

309
310 B. Fehr

vided in a series of studies conducted by Overview


Fehr and Russell (1991). We asked laypeo- Four studies were conducted. In Study 1,
ple to list types of love (Fehr & Russell, subjects rated how similar the conception of
1991, Study 1). Romantic love was, in fact,
love depicted in 15 prototypes was to their
one of the most frequently listed kinds of own view of love. The stability of these rat-
love-it ranked eighth in a list of over 200 ings over time was assessed. Study 2 served
types. However, it was preceded by friend- as a replication of Study 1, with a different
ship love (ranked first) and various kinds of sample and a longer time interval between
familial love (parental, maternal, broth- ratings. In Study 3, the validity of this proto-
erly). In a subsequent study (Fehr & type measure was explored by asking sub-
Russell, 1991, Study 2), subjects rated 20 of jects to determine the kind of love depicted
these types according to how repre- in each prototype. Study 4 investigated the
sentative they were of the concept ‘‘love.’’ relation between the prototypes of love and
Romantic love was seen as a prototypical existing love scales and therefore addressed
kind of love-it received the fifth highest issues of convergent and discriminant valid-
rating. However, once again, maternal love, ity of the prototype measures.
parental love, friendship love, and sisterly
love were seen as more representative, or
closer to the core meaning, of the concept. Study 1. Assessing Views of Love:
These findings suggest a need for research A Prototype-Based Strategy
on the kinds of love that laypeople deem
most central to the concept (e.g., maternal A number of prototype strategies have been
love, friendship love), in addition to extant used to refine predictor and criterion meas-
investigations of romantic love. ures in personality assessment research (e.g.,
One purpose of this research, therefore, Broughton, 1984;Buss & Craik, 1980,1983).
was to explore a broader range of types of A recent development has been the presen-
love than is usually considered in social sci- tation of scenarios depicting characters who
ence research. I wanted to include types of perform various prototypical, trait-based
love that are regarded by laypeople as acts (Broughton, 1986, 1990). A subject’s
highly representative of the concept (e.g., standing on a given trait is assessed by hav-
maternal and other kinds of familial love), ing the subject rate how similar he or she is
as well as types that are seen as less central, to the hypothetical character exemplifying
but nevertheless considered part of the that trait. A similar approach to assessment
layperson’s conception. For example, in- of views of love was taken here, given that
fatuation, puppy love, platonic love, and scales based explicitly on laypeople’s views
sexual love have received little, if any, em- are not currently available. Subjects were
pirical attention. One of the reasons for so presented with prototypes of different kinds
little research on these types of love is the of love and were asked to rate how similar
absence of scales with which to measure the view of love depicted in each prototype
them. Thus, a second, related purpose of this was to their own view. Two questions were of
research was to develop a methodology particular interest. First, would subjects be
based on lay conceptions for assessing peo- able to make meaningful judgments about
ple’s views of various kinds of love. Such an their views of love using this similarity-to-
assessment device would then enable ex- prototype measure? Second, would such
ploration of other important issues pertain- ratings be reliable?
ing to people’s conceptions of love, includ- In Phase 1 of this study, subjects listed
ing gender and personality differences (see types of love. In Phase 2, another group of
Fehr & Broughton, 1994) and the relation subjects listed features for 15 of these types.
between experts’ and laypeople’s conceptu- In Phase 3 , subjects were dating couples
alizations (see Study 4). who rated how similar each of the 15 fea-
Assessment of views of love 311

ture lists (derived in Phase 2) was to their then coded by three judges, and a final list of
own view of love. features for each type of love was derived.
(As in similar studies, any differences in
opinion among the judges were discussed
Method until unanimous agreement was reached;
The first two phases of this study will be see, for example,Fehr, 1988;Fehr & Russell,
described only briefly because the data 1984; Holliday & Chandler, 1986.) For a
were collected in the context of a larger more detailed description of this phase, see
research project described elsewhere (Fehr Fehr and Russell (1991, Study 6). Examples
& Russell, 1991). of feature lists appear in Table 1.1

Phase 1: Generating types of love Phase 3: Sinzilarity ratings


Subjects ( N = 84) simply were asked to list Subjects. Eighty dating couples (160 sub-
types of love. In all, 216 items were gener- jects) participated in this phase. At least one
ated, 93 of which were listed by more than member of each couple was a student in an
one subject. Each subject generated an av- Introductory Psychology class at the Uni-
erage of 8.69 types. (See Study 1 in Fehr & versity of Winnipeg. Students were told that
Russell, 1991, for a more complete descrip- they were eligible for this study if they were
tion of this phase.) currently dating (i.e., “going out” with some-
one, but not engaged, married, or cohabit-
Phase 2: Listing features of 15 types of love ing). Subjects who were students received
Subject Pool credit for their participation.
Fifteen target types of love were chosen. As
is usual in prototype research (e.g., Fehr &
Russell, 1984, 1991; Russell & Fehr, 1994), Materials. Subjects received the Views of
these included the 10 types that were gener- Love Questionnaire (along with additional
ated with the highest frequency in Phase 1 measures not relevant here) with the fol-
(friendship love, sexual love, parental love, lowing instructions:
brotherly love, sibling love, maternal love,
passionate love, romantic love, familial love, In a previous study, we asked people to tell us
and puppy love). Also in line with past prac- their views of love. Specifically,they were told to,
tice, an additional 5 types were chosen from “List the features or attributes of love.” Below
the low-frequency items to ensure that a are the responses of some of the people in our
broad range of responses was sampled. earlier study. The features or characteristics of
These additional types were selected on the love are listed roughly in order of importance
basis of two criteria: (1) they were interper- (i.e., people tended to list what they thought was
most important,or central to love, first). What we
sonal in nature (i.e., they represented a way
do not know from that study is how accurately
two people could love one another, thus ex- their responses reflect your view of love. Please
cluding such types as love for one’s country, read each of the descriptions of love below. After
love for one’s pet, love of work,etc.), and (2) you have read each one, rate how similar that
they were listed by more than one subject in person’s view of love is to your own view. In other
Phase 1.The following 5 types were chosen: words, if you had been asked to list the features of
affection, committed love, sisterly love, in- love, how similar is this list to the one you would
fatuation, and platonic love. Subjects ( N = have given? Keep in mind that we are interested
40) were asked to list the features for 10 in this study in what you think love is, rather than
types of love.Which 10 types were presented what you think love should be.
varied from one subject to the next, such
that each of the 15 types of love was de- 1. Complete feature lists for all 15 prototypes are
scribed by 20 subjects. The features were available from the author.
312 B. Fehr

Table 1. Sample ,feature lists f o r types of love

Friendship Love Romantic Love

Feel free to talk about anything Candlelight dinners


Caring Taking walks
Helping Happiness
Honesty Caring
Doing things together Idealistic
Trust Want to be with the person all the time
Sharing Giving gifts
Understanding Dream about the person
Sharing emotions Like a fairy tale
Being there for the person Feel relaxed with the person
Good times Good times
Happiness Physical attraction
Supportive Sitting in front of a fireplace
Longlasting Sharing
Loyalty Courtship
Openness Doing things together
Sharing thoughts Don’t know the person very well
A bond Kissing
Closeness Your world revolves around the
Common interests person
Concern Honesty
Feel relaxed with the person Glowing
Listening to each other Giving
Respect Friendship
Sadness Feel free to talk about anything
Sense of belonging Problems seem to vanish
Sharing experiences Sexual
Sweaty palms
Soft music
Think about the person all the time

Note: The features are listed in order of frequency; those that were listcd with the greatest fre.
quency appear first.

The feature lists were presented simply the statement that the attributes were listed
as descriptions of love, rather than as the roughly in order of centrality or importance
specific types of love they depicted, for the was included because a positive correlation
following reasons. First, subjects might be is found routinely between prototypicality
reluctant to endorse the description of and frequency of free listing (e.g., Fehr,
some types, such as infatuation or puppy 1988;Fehr & Russell, 1984).*
love, if they were labeled as such, even Each of the 15 descriptions of love ap-
though the feature list might be quite con- peared on a separate sheet, followed by the
sistent with their own view. (They might
feel compelled to give a description labeled 2. It could be argued that the statement that features
“committed love,” for example, a higher werc listcd roughly in order of importance could
rating.) Second, some of the types, if la- scrvc as a demand characteristic, prompting sub-
beled, might seem more appropriate for jects to respond in linc with the previous subjects’
one gender. Men, for example, might resist ratings. Thus, this statemcnt was removed from thc
instructions in anothcr study (see Fehr &
assigning high ratings to a description la- Broughton, 1994,Study 4). Similarity ratings in that
beled “maternal love” even if they agreed study correlated .96 and highcr with the scts of rat-
with the content of that feature list. Finally, ings gathered in the present studies.
Assessment of views oflove 313

question “How similar is this person’s view love prototype to a high of 7.89 for friend-
of love to your own view?” Broughton ship love. Thus, these prototypes of love
(1986), in his research on personality as- elicited a relatively wide range of similarity
sessment, found that although subjects are ratings (see Table 2 for the means and
reluctant to ascribe negative traits to them- standard deviations for the 15 prototypes at
selves, they are quite willing to admit that each testing session). The size of the stand-
they are similar to a hypothetical person ard deviations indicates that the variability
demonstrating these traits. Given that it in similarity ratings for any given prototype
might be seen as more socially desirable to was acceptable as well. With regard to reli-
endorse some of the descriptions than oth- ability, there was evidence that subjects
ers (e.g., committed love vs. puppy love), it agreed with one another in their ratings. An
was hoped that subjects would respond intraclass correlation coefficient (ICC)
more honestly if the descriptions of the 15 computed for the Time 1 ratings was .86
types of love were attributed to another (using the formula presented in Winer,
person. Similarity ratings were made on a 1 9 6 2 , ~128).
. For the Time 2 ratings, the ICC
scale ranging from (1) not at all similar to was .89. Moreover, the mean ratings as-
(9) extremely similar to my view of love.3 signed to the 15 prototypes at Time 1 and
Time 2 were highly correlated, r = .97. To
Procedure. Subjects participated in the ex-
determine the stability of individual sub-
periment with their dating partner. They
ject’s ratings across the two testing sessions,
completed the questionnaires individually
correlations between Time 1 and Time 2
in separate rooms. Subjects were assured
ratings were computed for a subsample of
that under no circumstances would their re-
50 subjects. Obviously, the ratings for indi-
sponses be shown to their partners, nor
vidual subjects are more unstable than are
would their partners’ responses be shown
those based on the entire sample. However,
to them. The 15 descriptions were adminis- even these ratings showed remarkable sta-
tered to subjects again 3 to 4 months later.
bility; the average across the 50 test-retest
(Those whose dating relationship ‘termi-
correlations was .73 ( r to 2 to Y trans-
nated after the first experimental session
formed).
were scheduled to return to the second ses-
The high degree of agreement between
sion alone.) These ratings were obtained
raters at each testing session was especially
from 76 men and 77 women.
heartening, given that many of these dating
Results and Discussion relationships were rather unstable. On the
other hand, it could be argued that the de-
Mean similarity ratings ranged from a low gree of consensus in the prototype ratings
of 2.41 on the 9-point scale for the sexual was inflated because the subjects were in-
volved in a close relationship with one an-
3. Ratings were made in terms of one’s vicw of love other. Given long-standing evidence that
more generally, rather than in terms of the specific
type of love depicted in a given prototype. Other-
partners choose one another on the basis of
wise, this procedure would have undermined the similarity (e.g., Brehm, 1985), a sample of
earlier decision to prcscnt the feature lists without dating couples might show greater agree-
labels (to minimize socially desirable icsponding). ment in their views of love than would sub-
Rcscarch by Sternberg and Grajck (1984) has jects who are not one another’s relationship
showcd that, dcspitc the fact that love for one’s
mother may fcel different from love for a friend or
partners. To examine this possibility, for
lover, the structure of thcse different kinds of love is each of the 80 couples, the male and female
rciiiarkably similar. (They described this general partner’s ratings of the 15 prototypes of
factor or common core of l o w as “one of interper- love were correlated. These correlations
sonal communication, sharing, and support” [p. ranged from .13 to .90, with an average of
3271.) Thus, it seemed defensible to ask subjccts to
rate dcscriptions of specific types of love in terms of .63 ( r to 2 to r transformed). (The average
similarity to their gcneral view of love. correlation at Time 2 was very similar, aver-
314 B. Fehr

Table 2. Mean ratings f o r 15 prototypes of love

Study 1 Study 1 Study 2 Study 2 Study 2


Time 1 Time 2 Time 1 Time 1" Time 2

TypeofLove Mean SD Mean SD Mean SD Mean SD Mean SD


~

Friendship 7.89 1.20 7.53 1.38 7.31 1.37 7.49 1.17 7.04 1.37
Maternal 7.24 1.59 6.71 1.83 6.89 1.75 6.90 1.69 6.55 1.74
Sisterly 6.77 1.57 6.84 1.61 6.48 1.75 6.31 1.87 6.73 1.76
Parental 6.33 1.92 6.37 1.79 6.31 1.88 6.43 1.72 6.31 1.87
Familial 6.37 1.78 6.80 1.55 6.28 1.73 6.06 1.85 6.51 1.57
Brotherly 6.63 1.64 5.98 1.70 6.06 1.82 5.96 1.87 5.59 1.69
Committed 6.07 2.18 6.24 2.10 5.69 2.22 5.71 2.22 5.78 1.77
Affection 5.91 1.76 5.90 1.67 5.20 1.93 5.43 1.71 5.37 1.70
Platonic 5.22 2.00 5.51 1.95 5.15 2.16 5.61 2.14 5.29 1.83
Romantic 5.64 2.08 5.25 2.21 4.90 2.06 5.27 1.86 5.20 1.95
Sibling 4.88 1.85 5.74 1.77 4.74 1.87 4.92 1.78 5.55 1.36
Passionate 4.28 2.04 4.10 2.05 3.72 2.09 3.75 1.87 3.78 1.69
Infatuation 2.67 1.85 2.43 1.53 2.52 1.83 2.22 1.40 2.65 1.52
PUPPY 2.37 1.59 2.52 1.56 2.51 1.70 2.39 1.58 2.63 1.55
Sexual 2.41 1.83 2.56 1.52 2.43 1.83 2.24 1.63 2.75 1.47
N= 160 N = 153 N = 227 N=51 N=51

"Based on only subjects who also complctcd the ratings at Time 2 ( N = 51).

age r = h5.) Thus, overall, there was a mod- prototype-based strategy could be used for
erate level of intracouple agreement. assessing views of a wide range of love con-
It is not possible to establish the internal cepts were encouraging.
consistency of the ratings for individual
prototypes through standard procedures Study 2. Ratings of Love Prototypes:
(e.g., alpha coefficients), given that they are A Non-Couple Sample
single-item measures. However, in recent
research (Fehr, 1994), subjects have been In the previous study, subjects were dating
asked to rate the features of each individual couples who displayed substantial agree-
prototype, rather than provide a single rat- ment in their views of love. However, it was
ing for the entire feature list. Alpha coeffi- questionable whether adequate interrater
cients based on these ratings range from .70 agreement would obtain if subjects were
to .89 (alpha = .89 for 5 prototypes; alphas not relationship partners. Thus, in this study,
exceeded .80 for 11 of the 15 prototypes), the Views of Love questionnaire was ad-
thereby demonstrating satisfactory internal ministered to a new group of respondents
consistency. The single-item format seemed who were not dating one another. In addi-
preferable for early investigations, how- tion, a longer time interval between the first
ever, because it then was possible to gather and second testing sessions was introduced
ratings of a large number of types of love as an even more stringent test of the reli-
from the same subjects on more than one ability of the prototype ratings.
occasion.
To summarize, the 15 prototypes of love Method
elicited different ratings. Subjects agreed
with one another, suggesting that they
Subjects
found this a meaningful task. Moreover, Subjects were students ( N = 227: 111 men
ratings of the 15 prototypes were stable and 114 women; 2 subjects did not indicate
over time. All in all, indications that this gender) enrolled in various psychology
Assessment of views of love 315

classes at the University of Winnipeg. Al- obtained in Study 1generalized to a sample


most half of the subjects (n = 104) were of individuals who were not in close rela-
solicited from summer session courses to tionships with one another. Intraclass cor-
ensure that a wider range of age and expe- relation coefficients were computed to as-
rience would be represented. These stu- sess the extent to which subjects who were
dents volunteered their participation. The not dating one another would agree in their
remaining students ( n = 123) completed ratings. These values remained high: .83 for
the questionnaires in the fall in their Intro- the first testing session and .83 for the sec-
ductory Psychology class and received Sub- ond session ( N = 51). A further comparison
ject Pool credit for their participation. The with the Study 1 dating sample was made
mean age for the entire sample was 24 by creating a set of 80 couples (i.e., the same
years, with a range of 17 to 70 years. number as in Study l),randomly pairing 80
female with 80 male subjects. As in Study 1,
Procedure for each impromptu pair, a correlation was
computed between the male and the female
The summer students were invited to par-
partner’s ratings. The average of these 80
ticipate in what was described as a series of
correlations was .57 ( r to 2 to r trans-
ongoing experiments in the University’s
formed), which is lower than that found in
Psychology Department. They were given,
Study 1 ( r = .63), although still moderate.
in random order, the Revised Interpersonal
Thus, the level of agreement found in the
Adjective Scales (IAS-R) (Wiggins, Trap-
dating sample may mostly reflect cultural
nell, & Phillips, 1988;see Fehr & Broughton,
(or at least university student) consensus
1994, Study 2) and the Views of Love ques-
about the meaning of love, rather than
tionnaire from Study 1 (along with an unre-
agreement unique to relationship partners.
lated measure, rating photographed facial
Even though a half year had elapsed
expressions, to give the impression that all of
since the first testing session, the mean rat-
the tasks administered in the experimental
ings of the 15 prototypes at each session
session were independent). The regular ses-
were highly correlated, r = .98. When test-
sion Introductory Psychology students com-
retest correlations were computed indi-
pleted the measures in class. They were
vidually for each of the 51 subjects who had
given the scales on separate days,early in the
completed ratings at both testing sessions,
term when they were completing a variety of
standard scales (for inclusion in their Sub-
ject Pool file). The Views of Love question-
1
the average of th se correlations was -70,
which is comparab e to the level of stability
that was observed in Study 1.
naire was completed again by 51 Introduc-
tory Psychology students approximately 6
months later. Study 3. Face Validity of the
Prototypes of Love
Results and Discussion
The ratings of the 15 prototypes of love in
The mean ratings (and standard deviations) Studies 1 and 2 demonstrated satisfactory
for the 15 prototypes at each testing session psychometric properties; they elicited an ap-
appear in Table 2. The range of ratings and propriate range of similarity ratings, and
their variability werc comparable to those these ratings were stable over time. How-
found in Study 1.The sexual love prototype ever, the validity of this measure remains to
again received the lowest similarity ratings, be demonstrated. Study 3 focused on face
while the friendship love prototype re- validity. Given that this measure is based on
ceived the highest ratings. Intermediate rat- laypeople’s views, it was particularly impor-
ings were assigned to types of love such as tant that the feature lists did, in fact, portray
affection, romantic, and committed love. to the respondents the view of love that was
Thus, the pattern of the prototype ratings purportedly being measured. Thus, subjects
316 B.Fehr

in this study were asked to match the feature words, the pattern of incorrect responses
lists with their labels. One would not expect should not be random. To test this empiri-
perfect agreement, given increasing evi- cally, first a matrix containing the propor-
dence that love is not a classically defined tion of features each prototype shared with
concept, but rather possesses the properties all of the other prototypes was constructed.
of prototype concepts (e.g., Aron & West- The diagonal of this matrix consisted of 1s
bay, 1991; Buss, 1988; Fehr, 1988; Fehr & (i.e., each prototype shares all of its features
Russell, 1984,1991;Fitness & Fletcher, 1993; with itself). Next, a matrix of subjects’ ac-
Shaver, Schwartz, Kirson, & O’Connor, tual choices was constructed (Table 3). The
1987;see Fehr, 1993, for a review). In other proportion of subjects who chose the cor-
words, the concept of love has blurry rect target prototype appeared along the
boundaries, and the features of one type of diagonal. The remaining entries were the
love overlap with the features of other types. confusions (i.e., the proportion of subjects
However, subjects should still be able to who chose each of the 14 non-target labels
identify the prototype from its features. Fur- for each prototype). If subjects’ choices,
ther, any errors in identification should be were, in fact, driven by the “correct” labels,
predictable and orderly (e.g., mistaking one one would expect a high degree of corre-
kind of familial love for another because of spondence between these two matrices.
their overlapping features), not random. To This result was obtained, r = .79,p < .001. A
test this idea, a confusion matrix (e.g.,Camp- subsequent analysis focused only on the
bell & Clark, 1988) was constructed in which subjects’ incorrect responses in order to de-
subjects’ pattern of choices was compared to termine the extent to which confusions
a matrix of the proportion of overlapping were attributable to feature overlap. One
features among the prototypes. would expect that this correlation would be
substantial, if predictable confusions are
Method
made, but lower than in the previous case
Subjects (because correct responses, i.e., the diago-
nals, are not included). The resultant r of .52
Thirty-nine students (21 women and 18
indicates that, as predicted, false positives
men) in a second-year psychology class par-
were likely to occur to the extent that the
ticipated in this study. Their mean age was
target prototype shared features with the
21.68 years, with a range of 19 to 30 years.
(incorrectly) chosen prototype.
In conclusion, there is evidence that these
Procedure prototypes of love possess face validity. Sub-
Subjects received the 15 prototypes of love jects were able to determine which kind of
and a list of 15 labels corresponding to the love was being depicted from the feature
types of love. They were asked to think lists; in each case, the modal response was
about what kind of love was depicted in the correct label. Moreover, any errors that
each feature list and then to choose the did occur were predictable from a prototype
appropriate label. perspective. Thus, despite the fact that many
of the prototypes shared features with one
Results and Discussion another, subjects still were able to identify
The correct label was the modal choice for with considerable success the kind of Iove
all 15 prototypes (see Table 3). As men- depicted in each prototype.
tioned earlier, the target type of love should
Study 4. Convergent and Discriminant
be the modal response, and any confusions
Validity of the Prototype Measures
between the target and other kinds of love
should result from shared features between This study examined the extent to which the
the target and other prototypes. In other prototypes of love map onto existing meas-
Table 3. Subjects’ choice of labels for 15prototypes of love

Aff. Bro. Com. Fam. Frnd. Infat. Mat. Par. Pass. Plat. Pup. Rom. Sex. Sib. Sis.
Affection 56.41 0 0 2.56 0 0 10.26 10.26 0 15.38 0 0 0 0 5.13
Brotherly 5.13 43.59 2.56 10.26 2.56 0 5.13 2.56 0 17.95 0 0 0 7.69 2.56
Committed 0 0 87.18 0 2.56 0 7.69 0 2.56 0 0 0 0 0 0
Familial 2.63 13.16 5.26 36.84 5.26 0 0 2.63 5.26 5.26 0 0 0 23.68 0
Friendship 7.89 7.89 0 2.63 60.53 0 0 2.63 0 0 0 0 0 2.63 15.79
5 Infatuation 5.13 0 0 0 0 84.62 0 0 2.56 2.56 2.56 2.56 0 0 0
Maternal 5.26 5.26 2.63 7.89 0 0 39.47 10.53 0 5.26 0 2.63 2.63 10.53 7.89
Parental 2.70 2.70 8.11 0 0 0 27.03 56.76 0 0 0 0 0 2.70 0
Passionate 0 0 0 0 0 5.13 0 0 79.49 2.56 2.56 2.56 7.69 0 0
Platonic 5.13 10.26 2.56 5.13 23.08 0 2.56 2.56 0 38.46 0 0 0 7.69 2.56
PUPPY 0 0 0 0 2.63 5.26 0 0 0 0 92.11 0 0 0 0
Romantic 0 0 0 0 0 0 0 2.56 2.56 2.56 0 92.31 0 0 0
Sexual 0 0 2.56 0 0 2.56 0 0 2.56 0 2.56 0 89.74 0 0
Sibling 5.26 18.42 0 18.42 0 0 2.63 13.16 0 0 0 0 0 39.47 2.63
Sisterly 2.56 5.13 0 5.13 10.26 2.56 2.56 0 0 5.13 0 0 0 2.56 64.10
Note: Numbers represent 96 subjects choosing each target ( N = 39).
318 B. Fehr

ures of love. Would there be any relation with experts’ constructions is an empirical
between experts’ scales of romantic love, for question that can and should be answered.
example, and a scale based on the layper- To explicate the relations between the
son’s prototype of romantic love? Would prototype and extant measures as fully as
other prototypes of love relate to existing possible, a large number of love scales was
measures in a meaningful way? For exam- included. It was not expected that many of
ple, laypeople’s prototypes of infatuation or the correlations would exceed .3 in magni-
passionate love should be more closely re- tude, given that the prototype measures
lated to measures of romantic love than share little method variance with standard
their prototypes of platonic or brotherly love scales and that any correlations with
love. Stated differently, this study explored single-item measures would be attenuated.
where the prototype measures fit in the no- The important question was whether the
mological net of existing measures of 10ve.4 obtained pattern of relations would make
It was hoped that there would be sufficient conceptual sense. Thus, this study was de-
correspondence between the prototype signed to assess convergent and discrimi-
measures and experts’ scales so that the pat- nant validity of the prototype measures, as
tern of relations between them could be ex- well as to shed light on the issue of lay
amined to evaluate convergent and dis- versus experts’ conceptions of love.
criminant validity. Unfortunately, little, if
any, effort has been expended to determine Method
the extent to which the conceptions of love Subjects
underlying the experts’ scales map onto
their respondents’ conceptions.This is not to Subjects were 357 (98 men and 252 women;
say that they should or must (see Fehr & 7 subjects did not indicate their gender) In-
Russell, 1991, for a discussion of this issue). troductory Psychology students who par-
However, the assumption implicit in roman- ticipated in this study for Subject Pool
tic love scalcs, for example, is that what the credit. They ranged in age from 17 to 52
investigator is measuring corresponds with years, with a mean of 20.26 years.
the notion of romantic love in the mind of
the research participant.The extent to which Materials
laypeople’s conceptions of love correspond Subjects were given a variety of love scales.
These included:

4. Given that the purpose of this research was to de-


1. Hobart’s (1958) Attitudes Toward Ro-
velop a mcthodology for assessing views of love, I manticism scale. This is a 12-item meas-
dccidcd to incliidc only scales that specifically pur- ure of the extent to which one endorses
ported to incastire love, rather than the nature or romantic beliefs such as “To be truly in
quality of intimate rclatioiiships.Tlitis, scalcssuch as love is to be in love forever.”
the Relationship Rating Form (Davis & Todd,1985)
and attachmcnt style scales (c.g., Collins & Read, 2. Fengler’s (1974) Romantic Idealist Fac-
1990; Hazan & Shaver, 1987; Simpson, 1990),which tor. This 6-item scale contains items
arc corrclatcd with love scales, but were not explic-
itly dcvclopcd to nicasurc love, were not included in
such as “A deep love for one another
this initial invcstigation. It should also be pointed can compensate for differences in relig-
out that Sternbcrg’s love scales wercstill in the proc- ious and economic background.”
ess of development when these data wcrc gathered.
Howcvcr, thc large number of publishcd scales 3 . Spaulding’s (1970) Romantic Love
available at thc time of thc study seemed sufficient Complex. Romantic beliefs are meas-
to establish discriminant and convergent validity. ured in this scale (e.g., “True love is
The relation bctwccn the love prototypes and nicas- known at once by the people in-
tires such as those of Davis and of Stcrnbcrg and the
nicasurcs of attachmcnt style will be cxplorcd in fu-
volved”). Of the 11 items, 4 overlap
turc rcsearch. with Hobart’s scale.
Assessment of views of love 319

4. Munro and Adams’s (1978) Attitudes (romantic, passionate love), Agape (al-
Toward Love scale. This 26-item scale is truistic love), Ludus (game-playing
comprised of three subscales: Roman- love), Pragma (practical love), and Ma-
tic Power (9 items; e.g., “True love nia (dependent, obsessive love). Each
never dies; it surmounts all obstacles”), style is measured by 7 items.
Romantic Idealist (9 items; e.g., “Many
years of being in love deepens rather 9. Rubin’s (1970) Love and Liking scales.
than exhausts the sense of pleasure that The 9-item version of each scale was
the partners feel for each other”), and used.
Conjugal-Rational (8 items; e.g., 10. Hatfield and Sprecher’s (1986) Pas-
“Good companionship is more impor- sionate Love scale. This scale focuses
tant than romantic love in making a on the passionate experience of love
good marriage”). (e.g., “I sense my body responding
5. Knox and Sporakowski’s (1968) Atti- when touches me”). The
tudes Toward Love scale. This 29-item short version (15 items) was used.
scale assesses the extent to which one 11. Kanin et al.3 (1970) Love Reactions
endorses a romantic (vs. a conjugal) measure. This measure consists of 8 in-
view of love. Agreement with items tense, passionate love reactions or ex-
such as “Love at first sight is often the periences (e.g., “floating on a cloud”;
deepest and most enduring type of “want to run, jump, scream”).
love” indicates a romantic view of love;
disagreement suggests a conjugal view. 12. The Views of Love questionnaire, in
which subjects rate how similar 15 pro-
6. Sprecher and Metts’s (1989) Romantic totypes of love are to their own view of
Beliefs scale. This 15-item scale meas- love.
ures the extent to which one endorses
romantic beliefs such as “The relation-
ship I will have with my ‘true love’ will Thus, a total of 22 scales (or subscales)
be nearly perfect.” Some items for this was included along with the 15 prototypes.
scale were borrowed or adapted from For the standard love scales, subjects indi-
scales such as Hobart’s and Spaulding’s. cated their responses on a 7-point scale,
A factor analysis of the items revealed with 7 indicating agreement with or en-
four underlying factors. The total scale dorsement of the item. The 9-item scale for-
score was used here. mat (9 = extremely similar to m y view of
love) was retained for the prototype ratings.
7. Dion and Dion’s Romantic Love ques-
tionnaire (e.g.,Dion & Dion, 198S).The Results and Discussion
16-item Attitudes toward Romantic
A table of correlations between the 22 love
Love scale in this questionnaire was in-
scales and each of the 15 love prototypes
cluded here. Dion and Dion typically
would present a rather overwhelming inter-
factor-analyze these items and use the
pretive task. Thus, the love scales were fac-
resultant factors in their research (K.L.
tor-analyzed to elucidate the relations
Dion, personal communication, 1993).
among them and the prototype measures
Items include “True love is mysterious
more manageably. The results of the factor
and cannot be explained by reason. It
analysis will be presented first, followed by a
can be felt, not explained.”
discussion of the relation between the love
8. Hendrick and Hendrick’s (1986) Love scale factors and the prototype measures.
Attitudes scale. This 42-item scale The specific question raised earlier-
measures six different love styles: namely, which experts’ scales of roman-
Storge (friendship-based love), Eros tic/passionate love are closest to layperson’s
320 B. Fehr

Table 4. Factor analysis of Dion and Dion’s Attitudes toward Romantic Love scale
Item Factor Loading
Factor 1:Idealistic view
14. True love lasts forever. -78
2. There is only one real love for a person. .64
4. True love leads to almost perfect happiness. .64
Factor 2: Cynical view
6. Romantic love is an o u t m o d e d a n d unrealistic concept. .71
3. There’s no room in m o d e r n marriages for t h e old idea of romance. .67
Factor 3: Romantic experience
10. True love is mysterious a n d cannot be explained by reason. It can b e felt, not .66
explained.
11. W h e n one is i n love, h e or s h e lives almost entirely for t h e o t h e r person. .63
Factor 4: Pragmatic view
7. People marry as often for things like money, status, a n d social considerations
as they d o for love. .75
8. Conflict can be a real part of love. .73

conceptions of these kinds of love-was ad- analysis with varimax rotation. The number
dressed by examining the simple correla- of factors retained was based on the results
tions between the individual love scales and of Longman, Cota, Holden, and Fekken’s
the prototype ratings. (1989) parallel analysis. Four factors, ac-
counting for 28.0,11.8,9.0, and 7.0% of the
variance, respectively, were retained. Total
Factor analysis of the love scales
variance accounted for was 55.8%. The .60-
First, a factor analysis of Dion and Dion’s .40 criterion for item (scale) retention was
Attitudes toward Romantic Love scale was applied (i.e., an item must have a primary
conducted (see Table 4).5 The resultant four loading of at least .60 and not load above
factors were then included with the remain- .40 on any other factors) (McCroskey &
ing 21 love scales in a principal components Young, 1979). This rule resulted in a clean

5. First it was necessary to pcrform a factor analysis of to have high loadings on other factors. Based on this
Dion and Dion’s Attitudcs toward Romantic LOVC criterion, two items were discarded from Factor 1-
scale (K. L. Dion, pcrsonal communication). The 16 itcm 13 (prcvious love affair may have been real love;
items were subjected to a principal components - .47 loading on own factor and .40 on Factor 3) and
analysis with varimax rotation. The number of fac- item 5 (doubt may enter real love; -.43 on own factor
tors chosen was bascd on Longman et al.’s (1989) and .33 on Factor 4). Item 1was dropped from Factor
proceduic. Esscntially, this procedurc dcterinines 2 (true love is known at once; .49 loading on own
the eigcnvalties that would obtain if thc data wcrc factor and .35 on Factor 3). Itcm 12 was dropped from
random. One then retains those factors for which Factor 3 (Iovc constrains one’s independence; .53 on
thc cigenvalues excccd thosc dcrived from random own factor and .44 on Factor 2). No items wcrc
data. Bascd on this criterion, four of thc fivc factors dropped from Factor 4. Results wcrc quite similar to
wcrc rctained (see Tablc 4). The percent variancc the Idealistic (Factor l), Cynical (Factor 2), and Prag-
adcountcd for by each of the factors was 18.9,11.3, matic (Factor 4) factors found by Dion and Dion
9.2, and 7.8, respectivcly. The total variance ac- (1973). The major differencc was that in this analysis,
counted for was 47.3%. In determining which items two of the Idcalistic items formed a separate factor;
to rctain for cach factor, the .60-.40 criterion was Item 10 refers to the affective experience of romantic
uscd: Thc item must load at least .60 on its own love (mysterious and inexplicable), whereas item 11
factor and not above .40 on the othcr factors speaks of living almost entirely for thc other person
(McCroskey & Young, 1979). This procedure re- when in lovc. This factor was labeled Romantic Expc-
sultcd in the cleanest solution, given that thosc rience because it sccmed to tap thc experiential as-
itcms which loaded less than .60 on a factor tended pect of romantic love.
Assessment of views of love 321

Table 5. Factor analysis of love scales

Love Scale Factor Loading


Factor 1:Romantic beliefs
Romantic Love Complex (Spaulding, 1970) .81
Romantic Beliefs scale (Sprecher & Metts, 1989) .80
Attitudes Toward Love scale (Knox & Sporakowski, 1968) .78
Idealistic View factor (Dion & Dion; see fn. 5) .76
Romantic Power subscale (Munro & Adams, 1978) .70
Romantic Experience factor (Dion & Dion; see fn. 5) .63
Factor 2: Companionate love
Liking scale (Rubin, 1970) .7 1
Love scale (Rubin, 1970) .67
Ludus love style (Hendrick & Hendrick, 1986) - .66
Factor 3: Experience of passionate love
Mania love style (Hendrick & Hendrick, 1986) .73
Love Reactions (Kanin, Davidson, & Scheck, 1970) .68
Passionate Love scale (Hatfield & Sprecher, 1986) .66
Factor 4:Pragmatic love
Pragma love style (Hendrick & Hendrick, 1986) .81
Romantic Idealist Factor (Fengler, 1974) - .78

set of factors, given that scales which loaded Factor 1 was labeled Romantic Beliefs
below .60 on a factor tended to load sub- because measures of romantic beliefs pre-
stantially on other factors. Most of the items dominantly loaded on this factor. Items
(scales) met the inclusion criteria-only from these scales include: “True love is
one item was dropped from Factors 1and 4, known at once by the people involved,”
two from Factor 2, and none from Factor 3.6 “Love at first sight is often the deepest and
The scales that loaded on each factor ap- most enduring type of love,” “True love
pear in Table 5. never dies; it surmounts all obstacles,” “True
love leads to almost perfect happiness,” and
“There is only one real love for a person.”
6. The rule adopted for item (scale) inclusion was a pri- Factor 2-which consists of Rubin’s
mary loading of .60 or higher on its own factor and
.40 or lower on another factor. The item that was
Love and Liking scales, along with the
dropped from Factor 1(Romantic Bclicfs) was Ho- Ludus love style (loaded negatively)-was
bart’s (1958) Romanticism scale. It loaded .60 on its labeled Companionate Love. Hatfield and
own factor, but -.46 on Factor 4 (Pragmatic Love). Walster (1978) defined companionate love
The Eros lovc style subscale was droppcd from Fac- as “a lower-key emotion. It’s friendly affec-
tor 2 (Companionate Love) bccausc it loaded .53 on
that factor and .44 on Factor 3 (Passionate Experi- tion and deep attachment to someone”
ence). However, given the iniportance of this scale (p. 2). Similarly, Brehm (1985) listed re-
in the love literature, simplc correlations between spect, admiration, and trust as the major
Eros and the related love prototypes are reported in components of companionate love and de-
the General Discussion section. The Agapc love scribed it as a stable, secure kind of love in
style subscale was excluded as well because its load-
ing of .56 did not meet the .60 criterion. Simple cor- which the accompanying emotions are calm
relations rcvcalcd that this lovc style was most and steady. These conceptualizations seem
highly correlated with the committed love proto- consistent with Rubin’s Love scale, which
type ( r = .27),followed by the maternal ( r = .25),pa- measures the extent of one’s feelings of car-
rental ( r = .21), and brotherly love ( r = .20) proto- ing, intimacy, and needlattachment toward
types. No items (scales) were dropped from Factor 3.
Onc item was dropped from Factor 4: Dion and
one’s romantic partner (Rubin, 1973). The
Dion’s Cynicism factor-it loaded .40 on its own Liking scale focuses on feelings of admira-
factor and -.30 on Factor 2 (Companionate Love). tion and respect for one’s partner, which
322 B. Fehr

are also a part of companionate love. In fatuation). In fact, the highest correlation
fact, the Liking scale has been described as was found with the romantic prototype,
a measure of companionate love (e.g., followed by weaker, positive correlations
Brehm, 1985; Hatfield, 1982). The Ludus with the passionate, infatuation, and puppy
love style subscale, which loaded negatively, love prototypes. Also as expected, scores
contrasts sharply with the depiction of com- on this factor were not correlated with the
panionate love. Items on this scale refer to various kinds of family love. Scores on the
juggling more than one lover simultane- Romantic Beliefs factor were, however,
ously, resisting a partner’s attempts to se- correlated with ratings of the committed
cure a commitment, and so on. love prototype. This result was not unex-
Factor 3 captures the experience of pas- pected, given that high correlations (in the
sionate love. Passionate love is usually con- .7Os-.8Os) typically are found between
ceptualized as intense, romantic, highly measures of commitment to relationships
sexualized, emotionally volatile, and vul- and love scales (e.g., Hatfield & Sprecher,
nerable to disruption (e.g., Brehm, 1985). 1986; Lund, 1985).
This experience is depicted in the Mania
love style subscale (e.g., “Sometimes I get Companionate love. Scales on the Compan-
so excited about being in love that I can’t ionate Love factor present a view of love as
sleep”). Kanin e t a1.k (1970) and Hatfield steady, comfortable, pleasant, and not too
and Sprecher’s (1986) scales were designed intense. Thus, positive correlations were ex-
specifically to measure passionate love ex- pected with the familial kinds of love and
periences. Both loaded on this factor. friendship love prototype. In fact, familial,
Factor 4 included the Pragma love style maternal, sisterly, brotherly, and parental
subscale and the Fengler Romantic Idealist love were correlated with this factor, as was
Factor. The Pragma love style subscale pre- the friendship love prototype (allps < .OOl).
sents a view of love in which pragmatic, A significant correlation was also predicted
practical concerns are paramount, leaving (and found) with the committed love proto-
little room for passionate, “head in the type, given evidence that the concept of
clouds” emotions (e.g., “A main considera- commitment overlaps considerably with
tion in choosing a lover is how he/she re- companionate love (Fehr, 1988; Kelley,
flects on my family”). The Romantic Ideal- 1983). Further, Lund (1985) obtained high
ist factor, which advocates marrying only on correlations between her measure of com-
the basis of love and not being concerned mitment and Rubin’s Love scale, which was
with the person’s social or economic back- included in this factor. It was also expected
ground, loaded negatively on this factor. that prototypes depicting passionate love
Hence, this factor reflects a pragmatic view (passionate love, infatuation, puppy love)
of lovc. would be unrelated or even negatively cor-
related with scores on the Companionate
Love factor. The results were consistent with
Relation between love scale factors and the
these predictions. For example, the infatu-
p rototype M ensures
ation and puppy love prototypes, for which
Correlations between love scale factor negative correlations were obtained, were
scores, and ratings of the prototypes of love characterized by features, such as “not
appear in Table 6. Findings for each factor longlasting,” “initially strong but then
will be discussed in turn. fades,” and “based on first impressions,”
which are antithetical to the enduring,stead-
Romantic beliefs. It was anticipated that the fast nature of companionate love.
Romantic Beliefs factor would be most
closely related to the romantic love and re- Passionate experience. It was predicted that
lated prototypes (e.g., passionate love, in- the passionate love and related prototypes
Table 6. Correlations between 15prototypes of love and love scale factors
Factor 1 Factor 2 Factor 3 Factor 4
Romantic Beliefs Companionate Love Passionate Experience Pragmatic Love

Type of Love All Men Women All Men Women All Men Women All Men Women

Friendship .04 .10 .01 .21*** .24** .19** .lo* .15f .04 - .03 .06 -.07
Maternal .09f .21* .03 .26*** .41*** .23*** .15** .24* .11* - .03 -.16f .02
Sisterly - .06 - .04 - .05 .18*** .39*** .ost .04 .06 .02 - .07f -.19* -.01
Parental .07 .03 .08t .20*** .26*** .18** .09f .10 .08 .lo* .05 .14*
Familial .02 03 .02 .18*** .32** .14* .ost .25** .01 -.lo* -.12 -.09t
Brotherly .01 -.02 .01 .19*** .17* .20** .03 .08 - .01 .02 -.06 .05
Committed .21*** .23* .19** .29*** .32** .28*** .27*** .29** .26*** .oo -.02 .02
Affection .04 .14f - .oo .03 .09 .01 - .01 .17* -.11* .lo* .07 .11*
Platonic .06 - .05 .11* .07t .11 .05 - .07 .02 -.13* .16*** .13 .18**
Romantic .34*** .31*** .34*** .lo* .15t .lot .29*** .41*** .28*** - .03 -.02 -.02
Sibling .03 - .01 .03 .07 - .03 .11* .06 .14f .04 .13*** .22* .11*
Passionate .15** .11 .15** - .01 - .oo - .oo .12* .21* .12* - .07t .21* -.17**
Infatuation .18*** .20* .17** -.17** -.12 -.19** .14** .28** .lot .07 .16t .03
PUPPY .19*** .20* .19** -.20*** -.18* -.19** .07f .25** .03 .09* .19* .05
Sexual .09f .07 .09t - .07 - .08 -.06 .06 21 .04 .09* .19* .05
N=345 N=97 N=248
t p < .lo. *p < .05. **p < .01. ***p < ,001.
324 B. Fehr

(e.g., romantic love, infatuation) would cor- sistent with the down-to-earth,nonromantic
relate most highly with the Passionate Ex- orientation to love portrayed by this factor.
perience factor. Scores on the passionate Thus, a meaningful pattern of correla-
love prototype were positively correlated tions emerged between the prototype
with this factor, although surpassed by the measures and the love scale factors, sug-
correlation with the romantic love proto- gesting that the prototype measures possess
type. Hatfield and Sprecher’s (1986) high both convergent and discriminant validity.
correlations between a measure of commit- The magnitude of these correlations is, of
ment and their Passionate Love scale ( r = course, attenuated because the prototype
.87 for men; .73 for women) suggested that scales were single-item measures. As one
a significant correlation might be found be- would expect, the size increases when the
tween Passionate Experience (which in- correlations are corrected for attenuation.
cluded their Passionate Love scale) and the To give some examples, for the Romantic
committed love prototype. This hypothesis Beliefs factor, the correlations of .34, .15,
was supported. The more companionate- and .18 with the romantic, passionate, and
like familial and friendship prototypes were infatuation prototypes respectively, in-
not expected to correlate with Passionate crease to .48, .23, and .33, respectively, when
Experience, The results were largely consis- corrected for attenuation. For the Compan-
tent with predictions, except for an unan- ionate Love factor, the correlations of .26,
ticipated positive correlation between Pas- .29, and .21 with the maternal, committed,
sionate Experience and the maternal love and friendship love factors increase to .52,
prototype, particularly for men. There is no .51, and .45, respectively.
obvious explanation for this finding (al-
though one could imagine a Freudian inter-
Relation between experts’ and lay
pretation). Features in the maternal love
conceptions of romanticlpassionate love
prototype such as “hugging, kissing, touch-
ing” may have triggered this association. As mentioned earlier, most love scales fo-
cus on romantic love, including the six Hen-
drick and Hendrick (1986,1989) love styles.
Pragnzatic love. Finally, it was expected that Do laypeople’s views of romantic/passion-
correlations with the Pragmatic Love factor ate love correspond with those presented
would be weak, but positive, for family love by the experts in these scales? The answer
prototypes, which included features that is a qualified yes. All of the scales purport-
downplay the expression of emotion and ing to measure romantic love (or some as-
take a rather practical approach to relation- pect of it, e.g., beliefs) were positively cor-
ships. The parental love prototype, with fea- related with the romantic love prototype,
tures such as “obligation, discipline, and although to varying degrees. The highest
teaching,” as well as the sibling love proto- correlation was obtained with Knox and
type where a dominant theme was not show- Sporakowski’s (1968) Attitudes Toward
ing one’s feelings, would be included here. Love scale, r = .36,p < .001(.33 for men; .36
Results supported these predictions. The for women). This scale contains statements
highest correlation was with the platonic such as, “When you fall head-over-heels-in-
love prototype, which depicted a very low love, it’s sure to be the real thing”; “Love at
key, nonintense kind of love with features first sight is often the deepest and most en-
such as “feel relaxed with the person” and during kind of love”; and “Love is best de-
“normal behavior.” Similarly, a weak, posi- scribed as an exciting thing rather than a
tive relation was found with the affection calm thing.” The next highest correlations
prototype, which captured feelings of fond- were with two Hendrick and Hendrick love
ness that were neither intense nor sexual. styles: Eros, I = .26,p < ,001 (.34 for men;
Both platonic love and affection seem con- .24 for women) and Mania, r = .28,p < .001
Assessment of views of love 325

(-28 for men; .27 for women). Sprecher and These correlations highlight which of the
Metts’s (1989) Romantic Beliefs scale cor- experts’ constructions of love show the
related similarly, r = .23,p < .001 (.18 for greatest amount of overlap with lay con-
men; .24 for women), as did Dion and ceptions. As acknowledged earlier, these
Dion’s Romantic Experience factor, r = .21, correlations were attenuated because of
p < .001 (.22 for men; .20 for women). Fi- factors such as item length. However, some
nally, the Passionate Love scale (Hatfield & interesting findings emerged. For example,
Sprecher, 1986) was related to the romantic in terms of the layperson’s conception, the
love prototype, particularly for men, r = .21, Eros love style subscale is correctly de-
p < .OO1 (.38 for men; .18 for women). scribed as a measure of romanridpassionate
Some measures that might be expected love. Some of the highest correlations were
to correlate significantly with the romantic found between this love style scale and
love prototype did not. For example, both the romantic and passionate love pro-
Fengler’s (1974) Romantic Idealist Factor totypes. The Mania love style, which depicts
was virtually unrelated to the lay concep- “symptoms” of intense love experiences,
tion of romantic love ( r = .05). When one also overlaps with these two prototypes. In
examines the content of this scale, four of terms of the romantic love prototype per se,
the items focus on one’s willingness to Knox and Sporakowski’s (1968) Attitudes
marry someone of a different race, back- Toward Love scale is most closely related
ground, social class, etc. Another item refers to the layperson’s conception. With regard
to not caring whether one gets married. The to the passionate love prototype, Munro
remaining item (reverse-scored) states that and Adams (1978) come closest by captur-
the main function of the family is to edu- ing the opposite of what laypeople mean by
cate and protect its children. These state- passionate love in their Conjugal-Rational
ments seem quite far removed from the scale.
layperson’s view of romantic love.
For the passionate love prototype, the
strongest associations were with the Eros,
General Discussion
r = .23, p < .001 (.30 for men; -22 for
women), and Mania, r = .14,p < .001 (.20
Prototype measures
for men; .ll for women) love styles, espe-
cially for men. The highest correlation was Prototype-based measures of love hold con-
a negative one with the Conjugal-Rational siderable promise for assessing people’s
subscale of Munro and Adams’s (1978) At- views of love. As reported in Studies 1and 2,
titudes Toward Love scale, r = -.24, p < subjects exhibited high levels of agreement
.001 (--15 for men; -.27 for women). This in their ratings of the 15 prototypes a t each
subscale contains items that are antithetical testing session and across sessions. This
to the passionate love prototype (e.g., level of agreement was also evident in Study
“Love is feeling warm, close and involved, 4 (ICC = .78) and reached .84 in another set
but not necessarily sexually excited”; of prototype ratings collected by Fehr and
“Erotic and romantic feelings toward an- Broughton (1994, Study 4). The magnitude
other are poor signs toward indicating a of these ICCs is similar to those obtained in
long and stable love relationship”; “It is studies where subjects are asked to provide
more important to feel calm and relaxed prototypicality ratings for different kinds of
with the one you love, rather than excited emotions or kinds of love (e.g., Fehr &
and romantic”). Ratings of the passionate Russell, 1984,1991). In addition, mean rat-
love prototype were also negatively corre- ings at both times in Studies 1 and 2, in
lated (for women) with the Storge love Study 4,and in Fehr and Broughton’s (1994)
style, r = -.16,p < .001 (.01 for men; -.18 Study 4 are highly intercorrelated, ranging
for women). from .96 to .99.
326 B. Fehr

If one wishes to demonstrate reliability in conceptualization. One prominent example


the form of internal consistency, the features is the work of Davis and his colleagues (e.g.,
of each prototype can be rated. As men- Davis & Latty-Mann, 1987; Davis &
tioned earlier, when this is done, the alpha Roberts, 1985; Davis & Todd, 1982, 1985;
coefficients reach acceptable levels. The dis- Davis, Todd, & Denneny, 1988), who take a
advantage, however, is a dramatic increase paradigm case formulation approach. In this
in the number of ratings that subjects are re- approach, experts choose a genuine case of
quired t o make. A very attractive feature of the phenomenon under study and list its
the single-item format was that subjects characteristics (see Davis & Todd, 1985, pp.
were exposed to the entire content of each 18-19). The archetypal paradigm case that is
prototype, but by requiring only one rating, derived is an ideal case to which actual in-
a wide range of views of love could be meas- stances of the phenomenon can be com-
ured. Thus, future psychometric work aimed pared. For example, Davis and Todd (1985)
at increasing reliability will come at the cost generated a paradigm case of friendship. Af-
of reducing the bandwidth of the love do- ter the authors had produced a set of charac-
main that can be assessed. teristics, laypeople were given an opportu-
The prototype measures fared well in nity to contribute additional attributes.
terms of validity. In Study 3, subjects were Davis and Todd then developed a scale, the
able to determine the kind of love depicted Relationship Rating Form, to assess the ex-
in each feature list. Study 4 revealed a rich tent to which the characteristics of the para-
and meaningful network of associations be- digm case are present in a given relationship.
tween the prototype measures and extant The scale measures aspects of relationships
measures of love. This study also high- such as intimacy, caring, satisfaction, con-
lighted areas of greatest overlap between flict, and so on.
laypeople’s and experts’ conceptions. In ad- In the attachment area, a related ap-
dition, there is now evidence that these proach has been taken that can be described
measures are well-suited for exploring gen- as expert-driven in terms of conceptualiza-
der and personality differences in views of tion, but prototype-like in methodology.
love (see Fehr & Broughton, 1994). Specifically, descriptions of three (or four)
The studies conducted represent only attachment styles have been generated by
the first steps toward the development of a experts (e.g., Bartholomew, 1990; Hazan &
prototype-based assessment of views of Shaver, 1987). Subjects typically have been
love. A number of issues will have to be presented with these descriptions and asked
addressed in future research before a thor- to indicate which description best repre-
ough evaluation of the merits and pitfalls of sents their orientation to relationships.
this approach can be made. Two such issues These depictions of the various attachment
are discussed next. styles have explicitly been referred to as
prototypes (e.g., Bartholomew, 1990; Levy
Correlation of the prototype scales with other & Davis, 1988).
nieasiires. Study 4 focused exclusively on ex- Recently, several researchers have be-
perts’ measures of love (see footnote 4), gun to examine how the Relationship Rat-
given that this construct was what the proto- ing Form and the attachment style descrip-
types were intended to assess. In doing so, tions relate to various love scales. For
Study 4 enabled a comparison of experts’ example, Hendrick and Hendrick (1989)
and laypeople’s conceptions. However, it published a factor analysis of a group of
also would be useful to consider the proto- scales including their love style scales (Love
type measures in relation to several current Attitudes Scale), Hatfield and Sprecher’s
approaches to the study of relationship phe- (1986) Passionate Love scale, Sternberg’s
nomena that fail somewhere between a Triangular Theory of Love scale, Davis’s
strict expert and a strict layperson-based Relationship Rating Form, and Hazan and
Assessment of views of love 327

Shaver’s (1987) attachment style descrip- across terms such as passionate, romantic,
tions. Their five-factor solution has since and puppy love), and 92% correctly labeled
been replicated by Davis (personal commu- the puppy love prototype. In Study 4, the
nication, 1994) and Tzeng (1993). It would infatuation prototype was significantly cor-
be helpful to extend this line of work by related with the Passionate Experience fac-
including the prototype scales. An examina- tor, whereas the puppy love prototype was
tion of the correlations between the proto- not. The same case could be made for other
types and the Relationship Rating Form conceptually related prototypes such as sib-
and attachment scales would illuminate ling, sisterly, and brotherly love. Again, even
how these approaches, which represent though there was substantial overlap in the
blends of expert and prototype approaches, features of those prototypes, subjects did
relate to a strict layperson-based assess- not treat them as synonymous. In Study 4,
ment. A complete mapping of the relations for example, scores on the sisterly love pro-
among expert, lay, and the combination of totype scale were negatively correlated
expert and lay approaches to the concep- with the Pragmatic Love factor (especially
tion and measurement of love (and related for women), whereas the correlations be-
constructs) would be of great benefit to tween the sibling love prototype and this
close-relationship researchers. factor were positive. Results such as these
present interesting material for future in-
How many prototypes should there be? Al- vestigations and probably would not have
though it is generally acknowledged that been found had the prototypes been col-
love exists in a myriad of forms, social psy- lapsed into smaller sets.
chologists usually focus on only two or An analogy can be drawn with the work
three-romantic/passionate and compan- that has been done on identifying the di-
ionate (e.g., Hatfield & Walster, 1978). A mensions underlying types of relationships.
central purpose of the present research was Davis and Roberts (1985) noted that such
to study a broader range of types of love analyses generally yield a small number of
than is usually considered in the literature. dimensions and commented that:
Therefore, the mean ratings, choices of pro-
totype label, and correlations with other For some limited purposes such accounting
scales were presented for all 15 prototypes. schemes may be useful, but if one aspires to rep-
Obviously, these data could have been re- resent the social knowledge that persons have of
duced to a smaller number of clusters or to relationships (or for that matter of emotions...),
a few underlying dimensions. As is dis- it is clearly a mistake to impose a methodological
cussed later, it may be beneficial to move in prejudice upon the problem before one has even
that direction in the future. However, at this explored what the phenomena are. (p. 155)
early stage, the loss of information that
would result from data reduction was diffi- The present studies were conducted in
cult to justify in an area where there is so this spirit of exploring “what the phenom-
little information. ena are.” Nevertheless, as acknowledged
For example, at first glance, it might earlier, from a measurement perspective, a
seem as though prototypes such as infatu- reduction in the number of prototype scales
ation and puppy love are so similar (if not may be useful in the long run. This reduc-
redundant) that subjects would respond to tion could be accomplished through a vari-
these prototypes in identical ways. How- ety of methods, including cluster or multidi-
ever, Study 3 revealed that subjects rarely mensional scaling analyses. In fact, a
confuse the prototypes of these two kinds preliminary series of cluster analyses were
of love--85% of subjects correctly identi- performed-not so much with the intent of
fied the infatuation prototype (and incor- reducing the prototypes to a smaller num-
rect responses were evenly distributed ber (although the results suggest some of
328 B. Fehr

the forms that data reduction might take), mental to both laypeople's and experts'
but rather to determine whether the proto- conceptions of love (see, e.g., Fehr, 1988,
types would cluster in explicable ways. If so, 1993, in press) and surfaces again here.
this finding would lend further credence to Finally, the robustness of the Study 1 re-
the claim that these prototype measures are sults was investigated by conducting a simi-
valid. First, a cluster analysis using an aver- lar analysis on the data from Study 2 (as in
age linkage rule was performed on the cor- Study 1,Time 1 data were used). The pat-
relation matrix containing subjects' similar- tern of clustering was quite similar, with
ity ratings in Study 1.(Time 1 ratings were prototypes depicting passionate varieties of
used to maximize the number of subjects.) love clustering in the first few steps and
The clustering algorithm began with each companionate love prototypes clustering in
prototype as its own cluster and iteratively the subsequent stages. The one exception
combined prototypes (and sets of proto- was that, in this sample, committed love
types) with the greatest similarity until one joined the companionate, not the passion-
large cluster consisting of all 15 prototypes ate, cluster as in the dating sample. Perhaps
was formed. As can be seen in Figure 1,the the daters were more inclined to equate
prototypes clustered in a meaningful way. commitment with passionate (romantic)
A t the first step, puppy love and infatuation love than companionate love, although any
were joined. Next, the sexual love and pas- conclusions along this line should be post-
sionate love prototypes were combined. poned until this finding is replicated. Im-
Then, these two clusters were amalga- portantly, the various groupings again ap-
peared to result in a distinct passionate
mated. Committed and romantic love clus-
cluster and a distinct companionate cluster,
tered together next. (This grouping is not
thereby largely replicating Study 1 and con-
surprising, given that, as was pointed out in
firming extant literature.
Study 4,scales to measure these two con-
In conclusion, there is no single, correct
structs are highly correlated.) In the sub-
answer to the question, How many proto-
sequent steps, the prototypes depicting
types should there be? It is informative,
various kinds of nonpassionate love (e.g.,
however, to consider this issue in light of
familial and friendship; sibling and paren-
the literature on the hierarchical structure
tal) clustered. Interestingly, this process ul-
of everyday concepts. Rosch et al. (1976)
timately produced two final clusters that
demonstrated that natural language catego-
quite clearly depict a passionate love
ries are vertically organized as hierarchies
grouping and a companionate love group- with different levels of abstraction. Catego-
ing. This distinction has proven to be funda- ries at higher levels subsume those at lower
levels. For example, the concept of fruit is a
superordinate-level category, with apples,
oranges, and figs at the next, lower level
(the basic level). Subcategories of each of
these (e.g., Granny Smith apple, crab apple,
1 +xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
2 +xxxxxxxxxxxxxxxx xxxxxxxxxxxxxxxxxxxxxxxxx and McIntosh apple) are situated at the
3 +xxxxxxxxxxxxxxxx xxxxxxxxxxxxxxxx xxxxxxx
4 +xxxxxxxxxxxxxxxx xxxx xxxxxxxxxx xxxxxxx lowest (subordinate) level. If one thinks of
5 +xxxxxxxxxxxxxxxx xxxx xxxxxxxxxx x xxxx
6 +XXXXXXXXXXXXXXXX XXXX XXXX XXXX X XXXX these various love categories as a hierarchy
7 txxxxxxxxxxxxxxxx xxxx xxxx xxxx x x x
8 +XXXXXXXXXXXXXXXX XXXX XXXX x x x x x
(as has been suggested in the literature; see
9 +xxxx xxxxxxxxxx xxxx xxxx x x x x x Fehr & Russell, 1991), then love itself would
10 txxxx xxxxxxxxxx x x xxxx x x x x x
11 +xxxx xxxxxxxxxx x x x x x x x x x form the apex, with companionate and pas-
1 2 +x x xxxxxxxxxx x x x x x x x x x
1 3 +X X XXXX XXXX X X X X X X X X X sionate love at the next lower level. Each of
1 4 + X X X X XXXX X X X X X X X X X
these kinds of love could then be further
Figure 1. Cluster analysis of 15 prototypes of love:
divided into subcategories. For example,
Study 1data. types
. -
of love such as friendship, sisterly
Assessment of views of love 329

love, and maternal love might be consid- ceived the highest typicality ratings. Thus,
ered varieties of companionate love, there is a convergence of opinion, regard-
whereas infatuation, sexual love, and puppy less of whether subjects are asked to rate
love could be taken as subcategories of pas- the word “friendship love” in terms of how
sionate love. prototypical it is as a kind of love, rate
The current data do not address whether friendship-like features of love in terms of
laypeople actually organize these kinds of how typical they are of love (or various
love in this way. In fact, recent evidence kinds of love) (see Fehr, 1993, in press), or,
suggests that, for natural-language catego- as in the present studies, rate a feature list
ries, laypeople’s hierarchies are ill-defined depicting friendship love in terms of sub-
and fuzzy, rather than neat taxonomies with jects’ own view of love.
clear, discrete levels (Russell & Fehr, 1994). The family kinds of love also garnered high
Nevertheless, it seems safe to assume that a similarity-to-my-own-view ratings in all stud-
concept such as love exists at a higher level ies. These companionate kinds of love have
of abstraction than a concept such as puppy been largely neglected by social psychologists.
love or maternal love. Just which level However, results of these studies, coupled
should be targeted in research depends on with those of Fehr and Russell (1991), suggest
how fine-grained an analysis is required. that these types of love also are highly repre-
Sometimes a category at a high level of ab- sentative of the meaning of the construct to
straction will do. The work by Hatfield and laypeople. Given their importance, research
her colleagues is a compelling demonstra- on the influence of various conceptions of fa-
tion that it can be sufficient simply to assess milial love on actual relationships would be
companionate and passionate love. Other valuable.
times it is desirable or even necessary to Finally, the two kinds of love that have
conduct a more detailed analysis of mem- received the most research attention, ro-
bers of a category. In such cases, the focus mantic love and passionate love, were not
would be on categories at lower hierarchi- as central to subjects’ views of love as were
cal levels. This approach was used in the friendship and familial kinds of love. Ro-
present research where the intent was to mantic love tended to receive intermediate
explore the many varieties of love. ratings. Passionate-like prototypes (i.e., pas-
sionate, infatuation, puppy love, and sexual
love) received the lowest similarity ratings
Lay conceptions of love
in all of these studies. Similarly, Fehr (1988)
These studies further elucidated how lay- found that passionate-like features of love
people view love. The data provide addi- received the lowest typicality ratings. These
tional support for the conclusion that, for terms also received the lowest typicality
laypeople, companionate varieties of love, ratings in Fehr and Russell’s (1991) re-
such as friendship love or familial kinds of search. This does not imply that relation-
love, capture the meaning of the concept. ship researchers should not study these
For example, friendship love received the kinds of love. They are clearly part of the
highest mean rating at each testing session layperson’s conception. However, they are
in each study (see Table 2). Similar findings seen as more peripheral to the core mean-
were obtained in Study 4 and in data col- ing of love than are friendship-based and
lected by Fehr and Broughton (1994, Study familial types of love.
4). Fehr and Russell (1991) found that the
term “friendship love” ranked near the top
Conclusion
when subjects provided prototypicality rat-
ings for 20 love terms. Similarly, Fehr (1988) In conclusion, these prototype-based meas-
found that companionate or friendship-like ures worked well in further uncovering
features of the generic concept of love re- laypeople’s conceptions of love. Moreover,
330 B. Fehr

a meaningful pattern of relations was re- ference constructs. As such, they represent
vealed between the prototype measures a first step toward the goal of accumulating
and existing love scales. These measures a body of knowledge for the other varieties
should prove useful for exploring the rela- of love similar to that which now exists for
tion between views of many kinds of love romantic love.
and other relationship and individual dif-

References
Aron, A,, & Westbay, L. (1991). Latent structure of description. In S. W. Duck & D. Perlman (Eds.),
love-prototype features and its relation to Erikson’s Understanding personal relationships (pp. 17-38).
intimacy stage. Unpublished manuscript, Univer- Beverly Hills, CA: Sage.
sity of California, Santa Cruz. Davis, K. E., Todd, M. J., Denneny, J. B. (1988). Per-
Bartholomew, K. (1990). Avoidance of intimacy: An sonal networks, friendship, and love relationships
attachment perspective. Journal of Social and Per- over the life cycle. Social and Behavioral Science
sonal Relationships, 7, 147-178. Documents, I7(2), 1-18.
Brchm, S. S. (1985). Intimate relationships. Ncw York: Dion, K. L., & Dion, K. K. (1973). Correlates of roman-
Random Housc. tic lovc. Journal of Consulting and Clinical Psychol-
Broughthn, R. (1984). A prototype strategy for con- ogy, 41,51-56.
struction of personality scales. Journal of Personal- Dion, K. K., & Dion, K. L. (1985). Personality, gender,
ity arid Social Psychology, 47,1334-1346. and the phenomenology of romantic love. In P.
Broughton, R. (1986). Distance from f h e prototype: A Shaver (Ed.), Review ofpersonality andsocialpsychol-
multidintensional scaling approach to personality ogy (Vol. 6, pp. 209-239). Beverly Hills, C A Sage.
assessment. Unpublished doctoral dissertation, Fchr, B. (1988). Prototype analysis of thc concepts of
University of British Columbia, Vancouver. love and commitment. Journal of Personality and
Broughton, R. (1990). The prototype conccpt in pcr- Social Psychology, 55,557-579.
sonality asscssmcnt. Canadian Psychology, 31,26- Fehr,B. (1993).HowdoIlovethee ...? Letmeconsult
37. my prototype. In S. Duck (Ed.), Individuals in
Buss, D. M. (1988). The evolutionary biology of love. relationships (Vol. 1, pp. 87-120). Newbury Park,
In R. J. Stcrnbcrg & M. L. Barnes (Eds.), The CA: Sage.
psychology of love (pp. 100-118). Ncw Haven, C T Fehr, B. (1994). Unpubiished raw data, University of
Y a k University Press. Winnipcg.
Buss, D. M., & Craik, K. H. (1980). Thc frequency Fchr, B. (in press). Lovc. In D. Levinson (Ed.), Ency-
conccpt of disposition: Dominance and prototypi- clopedia of marriage and the family. New York:
cally dominant acts. Journal of Research in Person- Macmillan.
ality, 43,379-392. Fehr, B., & Broughton, R. H. (1994). Individual differ-
Buss, D. M., & Craik, K. H. (1983). Thc act frequency ences in views of love. Unpublished manuscript,
approach to personality. Psychological Review, 90, University of Winnipeg.
105-126. Fehr, B., & Russell, J. A. (1984). Concept of emotion
Campbell, J. I. D., & Clark, J. M. (1988). An cncoding- viewed from a prototype perspective. Journal of
coinplcx vicw of cognitive number proccssing: Experimental Psychology: General, 113,464486.
Coinincnt on McCloskey, Sokol, and Goodman Fehr, B., & Russell, J. A. (1991). Concept of love
(1986). Jozirriut of Experimental Psychology: Geri- vicwed from a prototype perspective. Journal of
eral, I 17,204-214. Personality and Social Psychology, 60,425-438.
Collins, N. L., & Rcad, S. J. (1990). Adult attachrncnt, Fengler, A. P. (1974). Romantic love in courtship: Di-
working models, and relationship quality in dating vergent paths of male and female students. Journal
couplcs. Jourrial of Personality arid Social Psychol- of Coniporative Family Studies, 5, 134-139.
ogy, 58,644-663. Fitncss, J., & Fletchcr, G. J. 0. (1993). Lovc, hatc,
Davis, K. E., & Latty-Mann, H. (1987). Lovc stylcs and angcr, and jealousy in close relationships: A proto-
relationship quality: A contribution to validation. type and cognitive appraisal analysis. Journal of
Journal of Social and Personal Relationships, 4, Personality arid Social Psychology, 65,942-958.
409428. Hatfield, E. (1982). Passionate love, companionate
Davis, K. E., & Robcrts, M. K. (1985). Relationships in love and intimacy. In M. Fisher & G. Strickcr
the real world: Thc descriptive psychology ap- (Eds.), Intimacy (pp. 267-292). Ncw York: Plenum.
proach to pcrsonal rclationships. In K. J. Gcrgen & Hatficld, E., & Sprecher, S. (1986). Measuring passion-
K. E. Davis (Eds.), The social constructiori of the ate love in intimate relationships. Journal of A d o -
person (pp. 145-1 63). New York: Springer-Vcrlag. lescence, 9,383410.
Davis, K. E., & Todd, M. J. (1982). Fricndship and love Hatfield, E., & Walster, G. W. (1978). A new look at
rclationships. In K. E. Davis & T. 0. Mitchell love. Rcading, MA: Addison-Wcsley.
(Eds.), Advnnces in descripfivepsycliology (Vol. 2, Hazan, C., & Shavcr, P. (1987). Romantic love concep-
pp. 79-122). Grecnwich, CT: JAI Prcss. tualizcd as an attachment process. Journal of Per-
Davis, K. E., & Todd, M. J. (1985). Assessing friend- sonality arid Social Psychology, 52,511-523.
ship: Prototypcs, paradigm cascs, and rclationship Hcndrick, C., & Hcndrick, S. S. (1986). A theory and
Assessment of views of love 331

method of love. Journal of Personality and Social Roseh, E., Mervis, C. B., Gray, W. D., Johnson, D. M.,
Psychology, 50,392-402. & Boycs-Braem, P. (1976). Basic objects in natural
Hendrick, C., & Hendriek, S. S. (1989). Research on categories. Cognitive Psychology, 8,382-439.
love: Does it measure up? Journal of Personality Rubin, Z. (1970). Measurement of romantic love. Jour-
and Social Psychology, 56,784-794. nal of Personality and Social Psychology, 16, 265-
Hobart, C. W. (1958). Incidence of romanticism during 273.
courtship. Social Forces, 36,363-367. Rubin, Z. (1973). Liking and loving. New York: Holt,
Holliday, S. G., & Chandler, M. (1986). Wisdom: En- Rinehart & Winston.
plorations in adult competence (Human Develop- Russell, J. A., & Fehr, B. (1994). The varieties of anger:
m e n t M o n o g r a p h No. 17). Basel: C a r g e r Fuzzy concepts in a fuzzy hierarchy. Journal of
Publishing. Personality and Social Psychology.
Kanin, E. J., Davidson, K. R., & Scheck, S. R. (1970). Shaver, P., Sehwartz, J., Kirson, D., & O’Connor, C.
A research note on male-female differentials in the (1987). Emotion knowledge: Further exploration of
experience of heterosexual love. Journal of Sex a prototype approach. Journal of Personality and
Research, 6,64-72. Social Psychology, 52,1061-1086.
Kelley, H. H. (1983). Love and commitment. In H. H. Simpson, J. A. (1990). The influence of attachment
Kelley et al. (Eds.), Close relationships (pp. 265- styles on romantic relationships. Journal of Person-
314). New York: Freeman. ality and Social Psychology, 59,971-980.
Knox, D. H., & Sporakowski, M. J. (1968). Attitudes of Spaulding, C. (1970). The romantic love complex in
collcgc students toward love. Joitrnal of Marriage American culture. Sociology and Social Research,
and the Family, 30, 638-642. 55,82-100.
Lee, J. A. (1973). The colours of love. Don Mills, Sprccher, S . , & Metts, S. (1989). Development of
Ontario: Ncw Press. the “Romantic Beliefs Scale” and examination of
Levy, M., &Davis, K. E. (1988). Lovestyles and attach- the effects of gender and gender-role orientation.
ment styles compared: Their relations to each other Journal of Social and Personal Relationships, 6,
and to various relationship characteristics. Journal 385-409.
of Social and Personal Relationships, 5,439471. Sternberg, R. J., & Grajek, S. (1984). The nature of
Longman, R. S., Cota, A. A., Holden, R., & Fekken, love. Journal of Personality and Social Psychology,
G . C. (1989). A regression equation for the parallel 47,312-329.
analysis criterion in principal components analysis: Sternberg, R. J., & Barnes, M. L. (Eds.). (1988). The
Mean and 95th percentile eigenvalues. Multivariate psychology of love. New Haven, C T Yale Univer-
Behavioral Research, 24,59-69. sity Press.
Lund, M. (1985). The development of investment and Tzeng, P. C. S. (1993). Measure of love and intimate
commitnicnt scalcs for predicting continuity of per- relationx Theories, scales, applications for love de-
sonal relationships. Joiirnal of Social and Personal velopment, nzaintenance, and dissolution. Westport,
Relationships, 2, 3-23. CT: Praeger.
McCroskey, J. C., & Young, T. J. (1979). The use and Wiggins, J. S., Trapnell, P., & Phillips, N. (1988). Psy-
abuse of factor analysis in communication research. chometric and geometric characteristics of the
Hitman Cornniunication Research, 5,375-382. Revised Interpersonal Adjective Scales (IAS-R).
Munro, B., & A d a m , G. R. (1978). Love American style: Multivariate Behavioral Research, 23,517-530.
A tcst of role structure theory on changes in attitudes Winer, B. J. (1962). Statisticalprinciples in experimental
toward love. Hiirnan Relations, 31,215-218. design. New York: McGraw-Hill.

Você também pode gostar