Jouroal of Personality and Social Psychology Copyright 1986 by the American Psychological Association, Inc.

1986, Vol. 50, No. 2, 392-402 0022-3514/86/100.75

A Theory and Method of Love

Clyde Hendrick and Susan Hendrick

Texas Tech University

This research was part of a larger research program on love and sex attitudes. Earlier work on love
was reported in Hendrick, Hendrick, Foote, and Slapion-Foote (1984). The work on love extends
Lee's (1973/1976) theory of six basic love styles: Eros (passionate love), Ludus (game-playing love),
Storge (friendship love), Pragma (logical, "shopping list" love), Mania (possessive, dependent love),
and Agape (all-giving, selfless love). Theory development has proceeded concurrently with the devel-
opment of measurement scales. Study I (N = 807) used a 42-item rating questionnaire, with 7 items
measuring each of the love styles. Six love style scales emerged clearly from factor analysis. Internal
reliability was shown for each scale, and the scales had low intercorrelations with each other. Significant
relationships were found between love attitudes and several background variables, including gender,
ethnicity, previous love experiences, current love status, and self-esteem. Confirmatory Study II (N =
567) replicated factor structure, factor loadings, and reliability analyses of the first study. In addition,
the significant relationships between love attitudes and gender, previous love experiences, current love
status, and self-esteem were also consistent with the results of Study I. The love scale shows considerable
promise as an instrument for future research on love.

During the past decade, love has become respectable as an ferent. In a recent scholarly attempt to develop a conceptual and
area for study by psychologists (e.g., Rubin, 1984). Several the- methodological framework within which to examine love, Kelley
ories oflove have been proposed (e.g., readings by Cook & Wilson, (1983) introduced a model for what he called "pragmatic love"
1979; Kelley, 1983). Early theories that used global concepts of (p. 283) while still recognizing passionate love (e.g., Berscheid
love are being replaced by theories that use multidimensional & Walster, 1978) and altruistic love (e.g., Clark & Mills, 1979).
constructs that promise greater yields in knowledge. Pragmatic love emphasizes trust and tolerance and develops with
greater deliberation and self-control than do other types of love.
As Kelley concluded, "love is typically a blend of the different
Theories of Love
forms described by the preceding models" (p. 186). Unfortu-
Blau (1964) proposed an exchange theory oflove that char- nately, a comprehensive model that relates and integrates these
acterized the development oflove as requiring a nicely balanced different forms oflove has not been available.
degree of mutuality and the consistent exchange of rewards be-
tween partners. More recently, Clark and Mills (1979) attempted Scale Development
to differentiate "exchange" from "communal" (e.g., altruistic)
relationships by showing that a "tit for tat" approach may be The primary thrust of psychological work on love has been
accepted in an exchange relationship, but such an approach may toward theory building, with a limited secondary thrust toward
actually damage a communal relationship. scale construction. Considerable work in the area has been done
Berscheid and Walster (1974) proposed an approach to love by Rubin (1970, 1973, 1974), who was one of the first to study
which described romantic, passionate love as physiological the similarities and differences between loving and liking. Rubin
arousal accompanied by appropriate cognitive cues such that viewed the two as conceptually distinct though linked phenom-
"passionate love" is the appropriate label for the arousal. From ena, and he developed two scales to measure the two constructs.
this essentially labeling approach, Walster and Walster (1978) Although the scales have been widely used, and the distinction
proposed two general kinds of love: passionate love and com- between the two concepts was clear, the correlation between the
panionate love, with the former nearly always evolving to the two scales was higher than desirable for measures of two inde-
latter in an enduring close relationship. pendent constructs. However, some positive evidence for the va-
Moving counter to the increasing emphasis on love as multi- lidity of the distinction has been obtained (Dermer & Pyszczyn-
dimensional, Sternberg and Grajek (1984) proposed that there ski, 1978; Hill, Rubin, & Peplau, 1976). In contrast, in a con-
is a "general factor" of love which is quite consistent across ro- ceptual analysis, Kelley (1983) identified four components of
mantic, familial, and friendship relationships. However, the au- Rubin's love scale (needing, caring, trust, tolerance) and suggested
thors note that although the various love experiences may be that the liking scale might better have been named as a measure
similar, the "concomitants" of the experiences may be quite dif- of respect. Research by Steck, Levitan, McLane, and Kelley
(1982) attempted to manipulate these presumed components of
the love scale and found quite different responses to different
Correspondence concerning this article should be addressed to Clyde stimulus profiles, even though the total love score attributed to
Hendrick, Department of Psychology, Texas Tech University, Lubbock, stimulus strangers was constant across stimulus profiles. For ex-
Texas 79409. ample, a "caring" profile connoted a greater degree oflove than

392
 A THEORY AND METHOD OF LOVE 393

a "needing" profile. Thus it appears that the initial simplicity on each of the six subscales by counting the number of true
of Rubin's scales masks a more complex multidimensional reality. responses for the items in a given subscale. Thus, each subject
An attempt to examine the relationship between romantic could be profiled on all six subscales. It was reported that a
love and locus of control (Dion & Dion, 1973) produced inter- Gutman-Lingoes Smallest Space analysis produced the six love
esting results with explicit application to scaling. In the course types as conceptually distinct (although none of the details of
of their study, the authors asked 255 subjects to rate their sub- the data analyses were reported). Gender differences in love styles
jective experience of romantic love on 23 bipolar adjective items. were also reported. Rosenman (1978) correlated the Lasswells'
The results of a factor analysis of the items suggested at least love scale with Rubin's liking and loving scales. Rubin's love
five different approaches (or styles) of experiencing love. These scale correlated positively with the subscales representing Storge,
were labeled volatile, circumspect, rational, passionate, and im- Agape, and Mania, but not with Pragma, Ludus, and Eros. These
petuous. results fit Kelley's (1983) analysis of Rubin's scale, with Kelley's
There have been several other approaches to the scaling of needing component equaling Mania, whereas caring, trust, and
love (e.g., Hinkle & Sporakowski, 1975; Swenson, 1972), though tolerance equaled Agape and Storge. What was missing from
scale development and theory development were not tightly Rubin's scale was the passion of Eros.
linked. An exception was Munro and Adams (1978), who dif- Although Lee's typology offers an intriguing combination of
ferentiated romantic from conjugal love and related both to de- conceptual richness and clarity, no sustained work had been done
velopmental changes in individuals' role structures. with either Lee's theory or the Lasswells' scale until the current
research program was initiated.
A previous study (Hendrick, Hendrick, Foote, & Slapion-
Colors of Love
Foote, 1984) built on the Lasswells' work, using their items and
One of the more interesting theories of love was proposed by new ones in a Likert format. Approximately 800 students com-
Lee (1973/1976) who forged a classification of several different pleted the revised love scale. Factor analysis of the items provided
approaches to love. After an extensive interview procedure and partial support for Lee's theory. The secondary styles of Mania,
complex data reduction techniques, Lee proposed a typology of Pragma, and Agape emerged clearly as separate factors. However,
love styles that formed a closed circle. Lee identified three pri- each of the primary styles (Eros, Ludus, and Storge) tended to
mary types of love styles: Eros (romantic, passionate love), Ludus combine with another style instead of emerging as independent
(game-playing love), Storge (friendship love), and three main factors, and it was unclear whether Eros existed at all.
secondary styles: Mania (possessive, dependent love), Pragma Theory building and construct measurement are joint boot-
(logical, "shopping list" love), and Agape (all-giving, selfless love). strap operations. From the previous results it was unclear whether
These secondary styles were conceived as compounds of pairs of the theory was partially wrong, or whether the mixed results
primary styles. Analogous to chemical compounds, the secondary were due to poor measurement scales. After much consideration
styles are qualitative transformations of the "base primary ele- of items, we concluded that there were enough questionable
ments." Thus, Mania is a compound of Eros and Ludus, but items to warrant scale revision. Consequently, the love scale was
Mania is qualitatively very different from either primary. In the revised substantially, and the current research was based on this
same fashion, Pragma is a compound of Storge and Ludus, but revision.
has very different properties. The same holds true for Agape, a In addition, our previous study found fascinating gender dif-
compound of Eros and Storge. One implication of the analogy ferences on several of the love subscales. Males were clearly more
to chemical compounds is that although the six love styles are ludic than females, but females were more pragmatic, storgic,
logically interrelated, each style has qualitative properties inde- and manic in love attitudes than males.
pendent of all of the other styles. Empirically, measures of these The intent of Study I was to devise a measurement instrument
six love styles should be orthogonal to each other. In sum, the that would measure the six love styles/attitudes clearly, thereby
love styles are all equally valid ways of loving. There is no one providing evidence that the six different conceptions of love truly
type of love, but rather many different types. exist. Assuming that the love styles exist in fact, we were also
Lee's typology is exceedingly rich theoretically, both because interested in the general process of examining the domain of the
of its multidimensionality and grounding in research, and because theory of love styles. Toward that end, several background mea-
it encompasses less extensive love theories that have been pro- sures were taken that might be related conceptually to the love
posed. For instance, exchange theory is probably a basis for Lee's styles, including gender, ethnic differences, age, effects of previous
(1973/1976) Pragma (logical), whereas Clark and Mills' (1979) and current love experiences, and level of self-esteem. This study
communal love is exemplified by Agape (selfless). Berscheid and was part of a larger study that also measured sexual attitudes.
Walster (1978) would recognize Eros as their passionate love, Only the love attitude data will be considered in the current
whereas companionate love is probably best represented by Storge report. Study II was a replication of Study I in a different geo-
(friendship). Kelley's (1983) pragmatic love would seem to equal graphical area with slightly revised scales.
Pragma. Even Dion and Dion's (1973) factors appear very similar
Study I
to Lee's (1973) constructs: Volatile - Mania, Circumspect =
Storge, Rational = Pragma, and Passionate = Eros. Thus, Lee Method
offers multidimensionality within a coherent theory. A revised questionnaire entitled Attitudes About Love and Sex was
Lee's research inspired the development of a 50-item true- developed based on the instrument used in our previous study (Hendrick
false scale to measure the six love styles (Hatkoff & Lasswell, et al., 1984). The questionnaire included a brief explanation about the
1979; Lasswell & Lasswell, 1976). Each subject received a score study of attitudes, an 11-item Background Inventory, a section entitled
394 CLYDE HENDRICK AND SUSAN HENDRICK

Love Attitudes Scale that contained 42 love items, and a section entitled (41%) were age 18 or less, and some 235 students (29%) were
Sexual Attitudes Scale that contained 58 sexual attitude items. The items age 19. The remaining 30% of the students were 20 or older.
in the attitude sections were rated on a 5-category basis that was trans- Most students were single and had never been married (96%).
formed into a 5-point numerical basis for data analyses: 1 = strongly
However, 16% stated they were now, or had in the past, lived
agree, 1 = moderately agree, 3 = neutral, 4 = moderately disagree, and
with someone of the opposite sex. Males and females did not
5 = strongly disagree. The responses to all items were made on machine-
scorable answer sheets.
differ on this item. The ethnic heritage of the sample was as
It should be noted that following the scaling tradition in this research, follows: Black (5.3%), White-non-Hispanic (50.4%), White-His-
subjects were asked to complete the scale with the current love partner panic (29.0%), Oriental (7.7%), Other (7.6%). A substantial 161
in mind, in so far as possible. To accommodate subjects not currently in students (20%) indicated that they were international students.
love, the following instructions were included for completion of the love The religious heritage of the sample was as follows: Protestant
scale: (13.3%), Catholic (47.6%), Jewish (16.4%), None (4.7%), Other
(18.1%).
Some of the items refer to a specific love relationship, while others
refer to general attitudes and beliefs about love. Whenever possible, One item asked students "How many times have you been in
answer the questions with your current partner in mind. If you are love?" Results were none (15.1%), one (36.8%), two (26.1%),
not currently dating anyone, answer the questions with your most recent three to five (17.7%), more than five (4.2%). Males and females
partner in mind. If you have never been in love, answer in terms of differed on this item, \\4, N = 807) = 26.6, p < .01, with the
what you think your responses would most likely be. difference showing up as greater extremes for males. By more
than a 2:1 margin, males had either never been in love or had
The questionnaire was administered during the fall semester of 1983
and early spring semester of 1984 to groups of students taking introductory been in love three or more times.
psychology at the University of Miami. A total of 807 students completed Males and females also differed on the question "Are you in
the entire questionnaire and were included for data analysis. In addition, love now?" For males, 54.5% said no and 45.5% said yes. For
during the spring semester, 112 of these students completed the ques- females, 36.1% said no and 63.9% said yes, x 2 (l, N = 807) =
tionnaire a second time at 4 to 6 week intervals in order to gather data 26.2, p < .01. The same results occurred in a previous study
for a test-retest reliability analysis. The test-retest subjects were told (Hendrick et al., 1984).
during the second session that we were interested in whether love and A final background question attempted to measure self-esteem:
sex attitudes change over time and to complete the questionnaire in terms
"The way I feel about myself generally is." The great majority
of their current feelings.
(84.6%) rated their esteem very positive or positive, and 12.4%
rated themselves neutral. Only 24 students (3.0% of the sample)
rated their esteem as negative or very negative.
Results

Each of the six love styles was measured by 7 items. The items, Factor Structure of the Love Items
grand means, and standard deviations are shown in Table 1.
Because of the way the items were scored (i.e., 1 = strongly agree, The 42 items were intercorrelated and factored. Several pos-
sible principal component solutions were examined. The best
5 = strongly disagree), the lower the score the more a subject
subscribed to the love style measured by a given item. A perusal solution extracted six factors, with unities in the diagonal, using
of the item means indicates that subjects tended to endorse the varimax rotation of the factors. The factor loading for each item
items positively. A large majority of the item means (except for (on its factor) is shown in the third column of Table 1. Before
Ludus) fell on the agreement side of the scale neutral point (i.e., rotation, the percent of total variance accounted for by each of
3). The fact that the grand means tended toward the center of the factors was as follows: Eros (6.2), Ludus (6.8), Storge (4.3),
the scale suggested that there were no problems of end effects or Pragma (9.3), Mania (4.8), Agape (12.9). Agape was the first
scale restriction. Also, the size of the standard deviations indicated factor extracted and (by definition) it accounted for the most
that degree of agreement with a given item varied widely across variance. However, it is worth noting that this first factor was
subjects. not very general and did not overshadow the other factors. In-
stead, the factors were roughly comparable in size, with only
These descriptive data suggested that the items were quite
modest decreases in variance per factor. Further, all six factors
suitable for intercorrelation and subsequent factor analysis.
accounted for a healthy 44.2% of the total variance.
The factor loadings shown in Table 1 were substantial. For
Sample Characteristics the critical seven items that defined a love scale, in two cases
(Ludus, Pragma), no loading was less than .50. For two others
The University of Miami is a large, urban, private university. (Agape, Mania), only one of the seven items had a loading less
It enrolls students from many states and has a large contingent than .50. Eros had two items loading at less than .50, and Storge
of international students. Many Hispanic students, mostly Cuban, had three such items. However, inspection of Storge suggests that
attend the University. Because large numbers of students take it was a substantial factor.
introductory psychology, the sample was reasonably represen- The full factor matrix showed quite remarkable results. The
tative of the student population. The great diversity of the student 7 critical variables for a scale showed strong loadings on a factor,
population was especially valuable in developing an instrument and the loadings for the remaining 35 variables were low on that
to measure love attitudes. factor, often approaching zero. Only two items were questionable.
Selected sample characteristics were as follows. There were Item 5 (Eros) showed loadings of .25 to .35 on three other factors
466 males (58%) and 341 females (42%). Some 330 students in addition to Eros. Item 36 (Agape) also showed a loading of
 A THEORY AND METHOD OF LOVE 395

Table 1
Means, Standard Deviations, and Factor Loadings for the Love Attitude Items

Study I Study II

Factor Factor
Item M SD loadings M SD loadings

Eros
1 . My lover and I were attracted to each other
immediately after we first met. 2.4 1.2 .48 2.3 1.2 .53
2. My lover and I have the right physical
"chemistry" between us. 2.0 1.0 .76 2.0 .9 .78
3. Our lovemaking is very intense and
satisfying. 2.1 1.1 .68 2.1 1.1 .61
4. I feel that my lover and I were meant for
each other. 2.2 1.1 .65 2.1 1.1 .64
5. My lover and I became physically involved
very quickly. (My lover and I became
emotionally involved rather quickly.) 3.0 1.4 .36 2.4 1.2 .37
6. My lover and I really understand each
other. 2.1 1.1 .57 2.0 1.0 .56
7. My lover fits my ideal standards of physical
beauty/handsomeness. 2.1 1.0 .59 1.9 1.0 .57
Ludus
8. I try to keep my lover a little uncertain
about my commitment to him/her. 3.2 1.3 .70 3.4 1.4 .61
9. I believe that what my lover doesn't know
about me won't hurt him/her. 3.2 1.4 .67 3.6 1.4 .66
10. I have sometimes had to keep two of my
lovers from finding out about each other. 3.6 1.5 .69 3.8 1.4 .68
11. I can get over love affairs pretty easily and
quickly. 3.5 1.3 .55 3.9 1.2 .52
12. My lover would get upset if he/she knew of
some of the things I've done with other
people. 3.0 1.4 .57 3.1 1.4 .47
13. When my lover gets too dependent on me,
I want to back off a little. 2.9 1.2 .50 3.0 1.3 .52
14. I enjoy playing the "game of love" with a
number of different partners. 3.8 1.3 .72 4.2 1.2 .66
Storge
15. I did not realize that 1 was in love until 1
actually had been for some time. (It is
hard to say exactly where friendship
ends and love begins.) 2.7 1.2 .36 2.5 1.2 .33
16. I cannot love unless I first had caring for
awhile. (Genuine love first requires
caring for awhile.) 2.2 1.1 .50 1.6 .9 .33
17. I still have good friendships with almost
everyone with whom 1 have ever been
involved in a love relationship. (I expect
to always be friends with the one I love.) 2.5 1.2 .38 1.5 .9 .32
1 8. The best kind of love grows out of a long
friendship. 2.3 1.2 .69 2.3 1.1 .76
19. It is hard to say exactly when my lover and
I fell in love. (Our friendship merged
gradually into love over time.) 2.6 1.1 .49 2.6 1.3 .77
20. Love is really a deep friendship, not a
mysterious, mystical emotion. 2.5 1.3 .57 2.5 1.3 .56
21. My most satisfying love relationships have
developed from good friendships. 2.5 1.2 .69 2.7 1.3 .80
Pragma
22. I consider what a person is going to
become in life before I commit myself
to him/her. 3.1 1.3 .69 2.9 1.3 .58
23. I try to plan my life carefully before
choosing a lover. 2.9 1.2 .68 2.8 1.2 .52
24. It is best to love someone with a similar
background. 2.7 1.2 .54 2.6 1.1 .57
25. A main consideration in choosing a lover is
how he/she reflects on my family. 3.2 1.2 .69 2.8 1.2 .73
26. An important factor in choosing a partner
is whether or not he/she will be a good
parent. 2.5 1.2 .69 2.0 1.0 .66
(Table 1 continues on next page.)
396 CLYDE HENDRICK AND SUSAN HENDRICK

Table 1 (continued)

Study I Study II

Factor Factor
Item M SD loadings M SD loadings

Pragma (continued)
27. One consideration in choosing a partner is
how he/she will reflect on my career. 2.8 1.2 .72 2.5 1.1 .67
28. Before getting very involved with anyone, I
try to figure out how compatible his/her
hereditary background is with mine in
case we ever have children. 3.3 1.2 .71 3.2 1.2 .56
Mania
29. When things aren't right with my lover and
me, my stomach gets upset. 3.0 1.3 .54 2.5 1.3 .54
30. When my love affairs break up, I get so
depressed that I have even thought of
suicide. 4.3 1.1 .45 4.1 1.2 .46
31. Sometimes I get so excited about being in
love that I can't sleep. 2.7 1.2 .63 2.2 1.2 .64
32. When my lover doesn't pay attention to
me, I feel sick all over. 3.0 1.2 .76 2.8 1.2 .74
33. When I am in love, I have trouble
concentrating on anything else. 2.9 1.2 .67 2.7 1.2 .72
34. I cannot relax if I suspect that my lover is
with someone else. 2.3 1.2 .58 2.2 1.2 .57
35. If my lover ignores me for a while, I
sometimes do stupid things to get his/
her attention back. 3.0 1.3 .59 2.8 1.2 .50
Agape
36. I try to use my own strength to help my
lover through difficult times. (I try to
always help my lover through difficult
times). 1.7 .8 .30 1.3 .6 .30
37. I would rather suffer myself than let my
lover suffer. 2.2 1.0 .74 1.9 .9 .68
38. I cannot be happy unless I place my lover's
happiness before my own. 2.6 1.1 .79 2.5 1.1 .83
39. I am usually willing to sacrifice my own
wishes to let my lover achieve his/hers. 2.7 1.1 .77 2.4 1.0 .77
40. Whatever I own is my lover's to use as he/
she chooses. 2.4 1.2 .67 2.4 1.2. .64
41. When my lover gets angry with me, I still
love him/her fully and unconditionally. 2.1 1.0 .56 1.9 1.0 .52
42. I would endure all things for the sake of
my lover. 2.8 1.2 .77 2.4 1.1 .69

Note. Items 5, 15, 16, 17, 19, 36 were revised from Study I to Study II. The revision is shown in parentheses. Under Study I, the data shown are for
the original version of the item; under Study II, the data shown are for the revised item.

.39 on the Eros factor. After careful consideration, it was decided ing of the love style scores on a short-term basis. One very ten-
to retain these two items on their scales for further analyses. tative conclusion is that the love style scales are measures of
relatively changeable attitudes, rather than indices of enduring
personality traits.
Formal Scale Analyses
Because of the nature of principal components factor analysis,
Each of the love style scales was subjected to the standard the factors were orthogonal to each other. Thus, items with high
reliability analysis of the Statistical Package for the Social Sciences loadings on one factor and low loadings on other factors would
(SPSS) program. In addition, test-retest correlations were ob- necessarily be relatively independent of each other. However, it
tained for a subsample of 112 subjects. The results are shown does not follow that scale scores based on a sum of items (for
in the top panel of Table 2. specific factors) will be independent of each other. To assess degree
The alpha coefficients were substantial. All were .70+ except of love scale independence, sum scores were computed for each
Storge, which was .62, a result consistent with the relatively lower of the six scales and intercorrelated. The results are shown in
factor loadings for the items on this scale (see Table 1). The test- the top panel of Table 3.
retest correlations ranged from a low of .60 for Eros to a high Because of the large N, very small correlations were significant.
of .78 for Pragma. Although based on a smaller N, which might Several of the significant correlations shown in Table 3 were
have affected stability, these results suggested some relative shift- trivial in size. The only scale with possible problems was Agape.
 A THEORY AND METHOD OF LOVE 397

Table 2
Reliability Analyses of the Love Scales

Measure N Eros Ludus Storge Pragma Mania Agape

Study I
Mean interitem correlation 807 .27 .31 .19 .38 .28 .43
Alpha 807 .70 .76 .62 .81 .73 .84
Standardized item alpha 807 .72 .76 .62 .81 .73 .84
Test-Retest correlations 112 .60 .72 .72 .78 .75 .73

Study II
Mean interitem correlation 567 .26 .29 .24 .30 .27 .41
Alpha 567 .70 .74 .69 .74 .72 .83
Standardized item alpha 567 .71 .74 .68 .75 .72 .83
Test-Retest correlations 55 .74 .82 .74 .71 .70 .81

Note. There were seven items in each subscale. Test-retest correlations were based on the sum of the seven items.

It was significantly correlated with four of the other scales. How- The means and F ratios for the six love styles are shown in
ever, the largest of these correlations (.30) accounted for only 9% Table 4 for several background variables. The F ratio for each
of the variance in scale scores. It seems likely that the modest one-way analysis is shown at the top of the relevant column of
level of the correlations reflected the common method variance means. The means reported in Table 4 were derived by summing
of the rating instrument. the seven items on a scale, and taking the average. Thus, means
In summary, the analyses suggested a viable set of scales to could vary from 1.0 to 5.0, and the lower the mean the stronger
measure the six love styles. The scales emerged nicely from factor the particular love style.
analysis of the items, demonstrated suitable internal reliability, It will be useful to consider the results in terms of the back-
and reasonable independence from each other when considered ground variables.
as additive scales. Age of subjects. There were no significant main effects of age
for any of the love styles. Therefore, the means of the love styles
Background Variables and Attitudes as a function of age are not reported in Table 4. There was of
course not much variation in age for this college sample. The
A number of specific questions pertaining to the subjects' item was included to rule out shifts in preferred love styles from
background were included to assess in an orderly way some of freshman to junior years. The lack of results suggests that such
the theoretical suppositions about how different love styles func- shifts did not occur.
tion. The approach used treated each background variable as an Gender differences. Males were significantly more ludic than
independent variable, using subjects' sum scores on a given love females. Females were significantly more storgic, pragmatic, and
scale as a dependent variable. One-way analyses of variance were manic than males. Males and females did not differ on Eros and
performed on the data for each love style. (Attempts at two-way Agape. The pattern of these results was very similar to the out-
analyses by crossing two independent variables yielded nonor- come of a previous study (Hendrick et al., 1984).
thogonal effects because of the wide disparity in cell frequencies.) Ethnic background. Two questions tapped ethnic/cultural
background. One item stated "My ethnic heritage is," with five
response categories. The second item stated "I am an interna-
Table 3 tional student" with yes or no response categories. For ethnic
Interconelations Among Love Scale Sum Scores heritage, there were no differences on Ludus and Mania. For
Eros, Black, White-non-Hispanic, and White-Hispanic students
Scale Ludus Storge Pragma Mania Agape
were at about the same level, whereas Oriental students were
Study I significantly less erotic in orientation. The omnibus category of
Eros .00 -.05 -.05 .07 .27* "Other" yielded a mean similar to the Oriental mean, although
Ludus — .03 .12* -.05 -.28* due to small sample size for the Other category, its mean did not
Storge — .23* .06 .13*
differ from the Black mean for Eros. As might be expected, there
Pragma .11* .05
Mania — .30* was also a tendency for Oriental students to be more storgic and
—
pragmatic than were the other three ethnic groups. The tendency
Study II
.32* was strongest for Pragma. Finally, Black students were least
Eros -.22" -.04 -.04 .13*
Ludus — -.05 .09* -.03 -.42' agapic, although not all mean comparisons were significant due
Storge .25* .01 .15* to the small number of Black subjects (N = 43).
—
Pragma .13* .04 International students (a diverse grouping) rated themselves
—
Mania — .23*
as less erotic, more ludic, storgic, and pragmatic than U.S. citi-
Nate. N = 567 in Study II; N = 807 in Study I. zens. The two groupings did not differ on Mania or Agape.
*p<.01. Number of times in love. This question asked directly "How
398 CLYDE HENDRICK AND SUSAN HENDRICK

many times have you been in love," and provided five response ludic if they had never been in love or had been in love five or
categories ranging from noneio more than five. This last category more times, than were subjects in the intermediate categories.
was endorsed by only 34 of 807 subjects, and mean comparisons There were no differences due to number of times in love for
involving it were not very reliable. As expected, students who Pragma. The main effect of Storge, though significant, did not
had never been in love were least erotic, and there were no dif- suggest any particular interpretation of the means. There were
ferences between the Eros means for one, two, three to five, or significant main effects for both Mania and Agape. Subjects who
five or more times in love. The critical cut was none versus all had never been in love or who had been in love five or more
other categories. times were less manic than the other groups. Also, as expected,
Because Eros is supposed to value the intensity of love, then subjects who had never been in love were less agapic than the
never to have been in love is never to have been an Eros. It was other groups.
noted previously that relatively more males than females had Are you in love now? This item was important for theory
never been in love before. However, because the difference be- testing. It was expected that Eros types would tend to be in love,
tween males and females for Eros was not significant, the results Ludus types would not, and because Storge is friendship and
for number of times in love cannot be attributed to a confound Pragma is practical, these two types would not differ. Predictions
with gender. for Mania and Agape were uncertain. Turning the situation
Ludus is theoretically conceived as "love as a game." Love around in terms of the current independent-dependent variable
affairs for Ludus types should not have the fire of passion that convention, the following results emerged. Subjects "in love now"
is true for Eros. In fact, depending on how a ludic person defines were more erotic, more storgic, more manic, more agapic, and
love, he or she should either have been in love many times (each less ludic than subjects "not in love now." Clearly, the perception
casual affair is "love") or none (each casual affair is defined as of being currently in love (or not) cued off an entire response
a "casual affair"). The means for number of times in love showed pattern that involved five of the six love styles. These results pose
precisely this result for Ludus. Subjects were significantly more certain interesting theoretical issues to which we will return.

Table 4
Study I: Means and F Ratios for Each Love Style as a Function Of Selected Background Variables
Love styles

Variable N Eros Ludus Storge Pragma Mania Agape

Gender F= .7 F = 48.3* F = 4.4* F = 7.8* F = 3.8* F= 1.8

Males 466 2.3 3.1. 2.6. 3.0. 3.1. 2.3
Females 341 2.3 3.6b 2.5b 2.8b 3.0, 2.4

Ethnicity F = 6.8* F = 1.0 F=5.5* F= 11.9* F = 2.1 F = 2.6*

Black 43 2.3* 3.2 2.3,, 3.0.b 3.1 2.6.
White-non-Hispanic 407 2.2. 3.4 • 2.6, 3-1. 3.0 2.4*
White-Hispanic 234 2.3, 3.3 2.5* 2.9C 3.1 2.3b
Oriental 62 2.6b 3.2 2.3b 2.4b 2.8 2.3.,
Other 61 2.5* 3.3 2.3b 2.6* 3.1 2.3*
International student F= 14.5* F = 13.0* F = 7.9* F=31.9* F=2.3 F=.2
Yes 161 2.5, 3.1. 2.4. 2.6, 3.1 2.4
No 646 2.2b 3.4,, 2.5b 3.0b 3.0 2.4

No. times in love F =5.5* F= 10.7* F = 3.8* F= 1.4 F = 3.4* F = 11.9*

None 122 2.5. 3.1. 2.6. 2.8 3.2, 2.8.
One 297 2.2b 3.5b 2.4, 2.9 3.0b 2.2b
Two 211 2.3b 3A 2A 2.9 3.0b 2.3b
3-5 143 2.1b 3.4b 2.6., 2.9 3.0b 2.3b
5 &up 34 2-2ab 2.6. 2-7* 3.1 3.2* 2.5*
In love now F = 79.7* F = 44.7* F = 4.2* F= .1 F=7.1* F = 80.4*
No 377 2.5. 3.1. 2.6. 2.9 3.1. 2-7.
Yes 430 2.1b 3.5b 2.5b 2.9 10b 2.1b
Self-esteem F = 17.5* F=5.1* F=.9 F= 1.0 F = 15.7* F= 1.9
Very positive 251 2-1. 3.2. 2.5 2.9 3.2. 2.4
Positive 432 2.3b 3.4,, 2.5 2.9 3-Ob 2.4
Neutral or lower 124 2.5C 3.5b 2.5 3.0 2.8C 2.2

Note. Means could vary from 1.0 to 5.0. The lower the mean, the greater the agreement with the given love style. Within each column, for each
variable, means with no subscripts in common differed at the .05 level, either by the Ftest directly for a pair of means or by the Multiple Range Test
for three or more means.
* p < .05.
 A THEORY AND METHOD OF LOVE 399

Self-esteem. Because of the lew frequency of the subjects (as in Study I), and six factors were extracted. Before rotation,
rating themselves as negative or very negative in self-esteem, these the percent of total variance accounted for by each factor was
two categories were collapsed with neutral to form a category of Eros (4.1), Ludus (8.5), Storge (4.5), Pragma (7.1), Mania (5.2),
neutral or lower. Clear predictions could be made for Eros and and Agape (13.6), accounting for 43.1% of the total variance
Mania. Eros gives fully, intensely, and takes risks in love—it re- (structure almost identical to that in Study I). The factor loadings
quires substantial ego strength. Conversely, people high in self- nearly replicated those of Study I, with no loading on Ludus,
esteem should more likely be erotic than people low in self-es- Pragma, or Mania less than .45, and one Eros item, one Agape
teem. The same reasoning should apply in reverse for Mania. In item, and three Storge items loading less than .45 (but above
fact, one reason manic lovers are manic is because of uncertainty .30). The scale structure was very clear, with only four items
of self in the relationship. loading on more than one factor. In each of the four cases, the
The results in Table 4 support these predictions. There was a item loaded most highly on its appropriate factor and had a
clear positive relation between self-esteem and agreement with negative loading on a second factor. These loading patterns were
the Eros scale. There was a clear negative relation between self- all conceptually congruent with Lee's theory.
esteem and agreement with the Mania scale. It is of interest to Reliability analysis yielded standardized item alpha coefficients
note that subjects with very positive self-esteem were more ludic varying from .68 for Storge to .83 for Agape (similar to Study I,
than were either of the other two self-esteem groups. This result see Table 2). Test-retest reliability values (bottom panel of Table
makes sense, but was not predicted. Apparently, it takes good 2) indicated coefficients of .70 or above for all scales. Intercor-
ego strength to play seriously at love as a game. relations of summed scale scores revealed suitable independence
There were no differences due to self-esteem on Storge, and a pattern of correlations similar to that shown previously
Pragma, or Agape. None was expected for Storge or Pragma; (see Table 3), although the correlation of Agape with Ludus was
however, we rather expected people with very positive self-esteem somewhat higher, probably due to the stronger rejection of Ludus
to be more agapic. But such was not the case. by the sample.
Although the results revealed success in the technical aspects One-way analyses of variance were performed on the back-
of scale development as well as some theoretical confirmation of ground variable data for each love style, similar to the analyses
the love styles, it was felt that a confirmatory study was needed of Study I. The results are shown in Table 5. Consistent with
to substantiate Lee's theory and the Love Attitudes Scale as a Study I, there were no meaningful age differences in love styles.
major new development in research on love. There were gender differences on a number of background vari-
ables (including times in love, in love now, and self-esteem), as
Study II well as on five of the six love scales. Males were significantly
more ludic (though both genders were above the mean of 3.0
Method
and thus relatively rejected the ludic style), whereas females were
The Love Attitudes Scale was subjected to limited revision. One Hem more erotic, storgic, pragmatic, and manic than were males. The
each from Eros and Agape was revised and four items from Storge were pattern of results replicated almost exactly the results of the pre-
either partially or fully revised. vious studies. Although ethnic homogeneity precluded the
The new revision of the Attitudes About Love and Sex scale included broader analysis of ethnic groups presented in Study I, exami-
a brief explanation about the study of attitudes, a 17-item Background
nation of White-Hispanic and White-non-Hispanic subjects
Inventory (6 items added), a section entitled Love Attitudes Scale that
showed the Hispanics to be significantly more ludic than the
contained 42 items, and a section entitled Sexual Attitudes Scale that
contained 46 items. It was administered to students taking introductory
non-Hispanics.
psychology courses at Texas Tech University during the fall of 1984. The Subjects who reported that they had never been in love were
Anal sample consisted of 368 females and 199 males, 567 subjects in all. least endorsing of Eros and Agape, whereas subjects who had
Selected sample characteristics showed differences from the two previous been in love twice were most endorsing of these two styles. There
South Florida samples. Fewer students were Jewish (.7%) and more were were also significant differences on Ludus, with those who had
Protestant (49%). More students were White-non-Hispanic (83%) and been in love three or more times most ludic (followed closely by
fewer were White-Hispanic (11%), Black (2.5%), and Oriental (2%). Fewer those who had never been in love); those subjects who had been
of the Texas students had never been in love or had been in love three or in love once were least ludic. Although significant differences did
more times. The samples were relatively similar on other variables such
not appear on Storge or Mania, the results considered supportive
as age and self-esteem.
of Lee's theory in Study 1 (results on Eros, Ludus, Agape) were
replicated almost exactly in Study II.
Results
For the question of whether a subject was in love at the time
Examination of the item means indicated that patterns of en- of testing, subjects "in love now" were more erotic and agapic
dorsement closely replicated those of Study I, though there was and less ludic and pragmatic than subjects not in love. These
slightly more rejection of Ludus and slightly more endorsement results were very similar to those of Study I. For the self-esteem
of the other scales (see Table 1). The grand means, standard variable, whose results in Study I appeared to support Lee's
deviations, and factor loadings of the items are shown in the last theory (Eros high in self-esteem, Mania low in self-esteem), results
three columns of Table 1. Revisions of items are shown in pa- of Study II were fully consistent. Subjects whose self-esteem was
rentheses beneath the original items. Thus the final version of very positive most endorsed Eros (with the neutral category least
the scale is represented in Table 1. endorsing), and those who were neutral in self-esteem most en-
The love scale was subjected to a principal components analysis dorsed Mania (whereas very positive subjects endorsed it least).
400 CLYDE HENDRICK AND SUSAN HENDRICK

Table 5
Study II: Means and F Ratios for Each Love Style as a Function Of Selected Background Variables

Love styles

Variable N Eros Ludus Storge Pragma Mania Agape

Gender F = 5.2* F = 22.9* F-95* F = 9.5* F = 7.8* F= 1.8

Males 199 2.2. 3.3, 2.4. 2.8. 2.9. 2.2
Females 368 2.1b 3.7b 2.2b 2.6b 2.7b 2.1

Ethnicity F= .1 F=7.3* F= .1 JP-.2 F- 1.1 F=.l

White-non-Hispanic 472 2.1 3.6. 2.2 2.7 2.8 2.1
White-Hispanic 63 2.1 3.3b 2.2 2.6 2.7 2.2

No. times in love F = 7.0* F = 4.8* F= .2 f=.5 F=,6 F=8.l*

None 68 2.4. 3.4.b 2.2 2.6 2.8 2.5.
One 232 2.1, 3.7C 2.3 2.7 2.7 2.1.x,
Two 181 2.0, 3.6* 2.2 2.7 2.8 2.0C
3 or more 86 2.3. 3.3. 2.3 2.8 2.8 2.2b

In love now F = 99.7* F=28.7* F=.4 F = 7.2* F=.2 F=78.5*

No 252 2.4. 3.4. 2.2 2.6. 2.8 2.4.
Yes 315 1.9b 3.7b 2.3 2.8, 2.8 1.9,

Self-esteem F = 5.4* F =.1 F=,4 F= .5 F= 8.7* F=.6

Very positive 103 2.0. 3.6 2.3 2.7 3.0. 2.2
Positive 346 2.1. 3.6 2.2 2.7 2.7b 2.1
Neutral or lower 118 2.3b 3.5 2.3 2.7 2.6* 2.2

Note. Means could vary from 1.0 to 5.0. The lower the mean, the greater the agreement with the given love style. Within each column, for each
variable, means with no subscripts in common differed at the .05 level, either by the F test directly for a pair of means or by the Multiple Range Test
for three or more means.
* p < .05.

Thus the results for Study II were almost fully consistent with What the Love Scales Measure
those of Study I for the theoretically significant variables of gender,
Careful perusal of the content of the items indicated that the
number of times in love, whether a subject was in love at the
evolution of item sets has remained faithful to the theoretical
present time, and self-esteem. Several issues merit discussion,
conception of each love style.
including technical aspects of scale development, the nature of
1. Eros: Strong physical preferences, early attraction, and in-
the love style concepts, theoretical issues revolving around the
tensity of emotion are attributes of erotic love, along with strong
love styles, and some notion about where this research will even-
commitment to the lover. Love is highly valued by Eros types.
tually lead. Results from the two studies are so similar that they
The items in Table 1 reflect these attributes.
can for the most part be discussed together.
2. Ludus: Love as an interaction game to be played out with
diverse partners appears to be the main attribute of Ludus types.
General Discussion Deception of the lover is acceptable within proper role limits.
There is not great depth of feeling; indeed, the ludic lover is wary
Scale Construction
of emotional intensity from others. Ludic love has a manipulative
One strong motive for the present studies was the lack of factor quality to it. This aspect results in apparent lower social desir-
clarity in Hendrick et al. (1984). It was unclear whether the theory ability, as reflected in the item means in Table 1. It is important
of love styles was incorrect, or whether the measurement scale to note, however, that there are ludic aspects to many, if not
needed improvement. We opted for the second alternative, and most, love relationships. Lee's (1973/1976) point was that this
the results from the current studies indicate that Lee's (1973/ approach to love reflects an existing reality for many people. The
1976) theory is viable, and that each of the six concepts of love items and their loadings in Table 1 suggest that this style was
can be measured in a clear manner. measured well.
In terms of the various criteria for scale construction and val- 3. Storge: This style reflects an inclination to merge love and
idation, the results were about as good as can be expected from friendship. There is no fire in storgic love; it is solid, down-to-
real data obtained from real subjects. With the changes made earth, and presumably enduring. This "evolutionary" rather than
between Studies I and II, the love scale might be considered in "revolutionary" emphasis is reflected by most of the Storge items,
a nearly final form at this point. Although more work should be several of which have extremely high loadings.
done before it is used clinically, the scale is adequate in its present 4. Pragma: Rational calculation with a focus on desired at-
form as a research instrument for correlation with other scales, tributes of the lover is central to pragmatic love. In fact, "love
preselection of subjects, and so forth. planning" might be an apt description. Because Pragma types
 A THEORY AND METHOD OF LOVE 401

use criteria matching, it is easy to view them as "computer mat- This theory needs much elaboration. But it does point the
ing" people. The items and loadings shown in Table 1 indicate way toward an explanation of why the love styles seem to be
that this love style was also well measured. both personality traits and malleable attitudes. We need not be
5. Mania: Reading the items suggests that Mania is "symptom forced to choose in an either/or fashion.
love," based on uncertainty of self and the lover. It may be most
characteristic of adolescents, but examples of older manic lovers
frequently occur. The items and their factor loadings indicate Further Theoretical Issues
success in construct measurement.
The results in Tables 4 and 5 bear on several issues that merit
6. Agape: Lee did not find this style manifested fully in actual
further consideration.
human beings. However, the factor results in Table 1 suggest that
The mean differences in love styles between males and females
it is a viable style. Clearly it is an all-giving, nondemanding love.
replicated our previous work (Hendrick et ah, 1984). Hatkoff
In fact, the item with a modest loading (Item 36) mentions
and Lasswell (1979) found roughly similar gender differences.
"strength" (in Study I), a theme not occurring in any of the other
To an extent the differences in love attitudes parallel male-female
Agape items. Even the item revision with "strength" deleted was
differences in attitudes toward sexuality (e.g., Ferrell, Tolone, &
somewhat low in its loading. Perhaps any item not denoting self-
Walsh, 1977; Laner, Laner, & Palmer, 1978; Medora & Wood-
abnegation would load less strongly on Agape.
ward, 1982; Mercer & Kohn, 1979). In general, males are more
The conclusion of this detailed scrutiny is that the six love
permissive and instrumental in their sexual attitudes (Hendrick,
scales appear to be content valid as well as technically sound as
Hendrick, Slapion-Foote, & Foote, 1985), a result consistent with
measurement scales.
males being more ludic in their love styles. Traditionally, females
have been more conservative in sexual attitudes, a conservatism
Personality or Attitudes? that partially stems from socialization to view sex as a precious
commodity that must be guarded. Also, women have historically
Do the love styles measure enduring personality traits or more
been socialized to marry both a love partner and a potential
transient attitudes? This interesting question cannot be answered
provider. From such a state of dependence on males, it would
definitively by the present studies. Lee (1973/1976) discussed
be surprising if females were not more pragmatic than males.
the love styles as a typology. To psychologists this approach im-
The same reasoning may possibly account for females being more
plies traits. But Lee also believed that it is possible to be simul-
storgic than males. The same socialized dependency may also
taneously in one type of relationship with one person (e.g., erotic),
account for more manic attitudes by females, although this effect
and in another type (e.g., ludic) with a second person. This pos-
might be due to an artifact, namely that females report more
sibility implies that the cause of the love style lies in the nature
symptoms in general than males. In any event, gender differences
of the relationship with another person.
in love styles is an important topic worthy of more research
Conceptually, it might be argued that love styles partake of
effort.
both trait and state characteristics. One interesting aspect of the
The current studies merely suggest that ethnic differences in
love styles is that they vary in emotional intensity. Eros and Mania
love styles may be a fruitful direction for future research. One
are high in emotion, Agape is average, and Ludus, Storge, and
interesting outcome in Study I was that Oriental students seemed
Pragma are all low. To whatever extent emotional expressivity is
relatively low in affect (low in Eros, high in Storge and Pragma).
a temperamental facet of the person, to that extent there may
It may well be that the six love styles do not capture properly
be a constitutional predisposition toward different love styles.
Oriental conceptions of love. In Study II, Hispanic subjects ap-
At the same time, data from the present study also point in
peared more ludic than non-Hispanic subjects. Future cross-
an attitudinal direction. Ethnic and gender differences suggest
cultural research should be sensitive to the possibility that more
the effects of socialization differences. The fascinating results for
than six styles exist and that different styles are relatively more
"are you in love now" also suggest an attitude conception of the
dominant in different cultures.
love styles. Students in love now were "more" on several of the
It would appear, from this research, that the love styles are
styles than students not in love now. In contrast, the self-esteem
not independent of one's current love situation, or for that matter,
data may perhaps be interpreted as supporting a trait (personality)
the number of past love relationships. The issue merits further
interpretation of the love styles (those endorsing Eros highest in
study. It would be desirable to monitor possible changes in love
self-esteem, those endorsing Mania lowest in self-esteem).
attitudes as a love relationship progresses from first encounter
One theoretical approach is to consider the six love styles as
to binding commitment. The diverse results for "in love now"
reflecting a six-dimensional matrix in each person's psyche. Ev-
versus "not in love now" indicate that something important is
eryone has some location at a given time on each of the dimen-
going on. One theoretical interpretation was noted earlier, other
sions. Constitutional differences may tend to bolster one (or more)
possibilities should be examined.
dimension. However, specific socialization practices also affect
the development of the conceptual love matrix. Possibly some
dimensions are more changeable by experience than other di-
Future Directions
mensions. Relative standing on the six dimensions may vary over
time. Strong experience, such as "being in love now," may cause Research on attitudes toward love can lead in many directions:
a flare-up on several dimensions, perhaps resulting in temporarily More work is needed to assess stability of love styles. If scores
correlated dimensions that are uncorrelated under conditions of are relatively stable, then the scale could become a valuable tool
ordinary, nonintense emotional experience. for preselection of subjects for a wide variety of interaction stud-
402 CLYDE HENDRICK AND SUSAN HENDRICK

ies. It is of great interest to investigate how well males and females Hendrick, C, Hendrick, S., Foote, F. H., & Slapion-Foote, M. J. (1984).
with different dominant love styles "mesh," as compared with Do men and women love differently? Journal of Social and Personal
couples with similar styles. Relationships, I, 177-195.
Descriptive work on the love styles held by society would be Hendrick, S., Hendrick, C., Slapion-Fbote, M. J., & Foote, F. H. (1985).
Gender differences in sexual attitudes. Journal of Personality and Social
of great value. It may be that (in Western society at least) people
Psychology, 48, 1630-1642.
go through a kind of modal developmental sequence of love styles.
Hill,C. T., Rubin, Z., & Peplau, L. A. (1976). Breakups before marriage:
As noted previously, manic love may be most characteristic of
The end of 103 affairs. Journal of Social Issues, 32(1), 147-168.
adolescents. In early adulthood the preferred style may evolve
Hinkle, D. E., & Sporakowski, M. 1, (1975). Attitudes toward love: A
toward Eros, which in turn may evolve toward Storge and Pragma reexamination. Journal of Marriage and the Family, 37, 764-767.
during the middle and later years. And we have all known at Kelley, H. H. (1983). Love and commitment. In H. H. Kelley, E. Berscheid,
least one "old" couple who appeared very agapic. Such a devel- A. Christensen, J. H. Harvey, T. L. Huston, G. Levinger, E. McClintock,
opmental sequence would account for the common observation L. A. Peplau, & D. R. Peterson (Eds.), Close relationships. New %rk:
noted by Walster and Walster (1978) that what usually starts as Freeman.

passionate love (Eros) sooner or later settles down to compa- Laner, M. R., Laner, R. H., & Palmer, C. E. (1978). Permissive attitudes
toward sexual behaviors: A clarification of theoretical explanations.
nionate love (Storge, Pragma). If such a "love history" occurs
Journal of Sex Research, 14, 137-144.
for substantial numbers of people, then knowledge of the sequence
Lasswell, T. E., & Lasswell, M. E. (1976). I love you but I'm not in love
could enable people to intervene to change (or come to terms
with you. Journal of Marriage and Family Counseling, 38, 211-224.
with) their histories (e.g., see Gergen, 1973). Lee, J. A. (1973). The colors of love: An exploration of the ways of loving.
Thus, the ramifications for future research appear indefinitely Don Mills, Ontario: New Press. (Popular Edition, 1976).
broad in scope. We believe that all of them are worth pursuing. Medora, N., & Woodward, J. C. (1982). Premarital sexual opinions of
What is more important than love? undergraduate students at a midwestern university. Adolescence, 17,
213-224.
Mercer, G. W, & Kohn, P. M. (1979). Gender differences in the integration
of conservatism, sex urge, and sexual behaviors among college students.
References
Journal of Sex Research, 15, 129-142.
Munro, B., & Adams, G. R. (1978). Love American style: A test of role
Berscheid, E., & Walster, E. (1974). A little bit about love. In T. L. Huston structure theory on changes in attitudes toward love. Human Relations,
(Ed.), Foundations of interpersonal attraction. New York: Academic 31, 215-228.
Press. Rosenman, M. F. (1978). Liking, loving, and styles of loving. Psychological
Berscheid. E., & Walster, E. (1978). Interpersonal attraction (2nd ed.). Reports, 42, 1243-1246.
Reading, MA: Addison-Wesley. Rubin, Z. (1970). Measurement of romantic love. Journal of Personality
Blau, P. M. (1964). Exchange and power in social life. New York: Wiley. and Social Psychology, 16, 265-273.
Clark, M. S., & Mills, I. (1979). Interpersonal attraction in exchange and Rubin, Z. (1973). Liking and loving: An invitation to social psychology.
communal relationships. Journal of Personality and Social Psychology. New York: Holt, Rinehart, and Winston.
37, 12-24. Rubin, Z. (1974). From liking to loving: Patterns of attraction in dating
Cook, M., & Wilson, G. (Eds.). (1979). Love and attraction: An inter- relationships. In T. L. Huston (Ed.), Foundations of interpersonal at-
national conference. Oxford: Pergamon Press. traction. New York: Academic Press.
Dermer, M., & Pyszczynski, T. A. (1978). Effects of erotica upon men's Rubin, Z. (1984). Toward a science of relationships. Contemporary Psy-
loving and liking responses for women they love. Journal of Personality chology, 29, 856-858.
and Social Psychology, 36, 1302-1309. Steck, L., Levitan, D., McLane, D., & Kelley, H. H. (1982). Care, need,
Dion, K. L., & Dion, K. K. (1973). Correlates of romantic love. Journal and conceptions of love. Journal of Personality and Social Psychology,
of Consulting and Clinical Psychology, 41, 51-56. 43, 481-491.
Ferrell, M. A., Tolone, W. L., & Walsh, R. H. (1977). Maturational and Sternberg, R. J., & Grajek, S. (1984). The nature of love. Journal of
societal changes in the sexual double-standard: A panel analysis (1967- Personality and Social Psychology, 47, 312-329.
1971; 1970-1974). Journal of Marriage and the Family, 39, 255-271. Swenson, C. H. (1972). The behavior of love. In H. A. Otto (Ed.), Love
Gergen, K. J. (1973). Social psychology as history. Journal of Personality today. New York: Association Press.
and Social Psychology, 26, 309-320. Walster, E., & Walster, G. W. (1978). A new look at love. Reading, MA:
Hatkoff, T. S., & Lasswell, T. E. (1979). Male-female similarities and Addison-Wesley.
differences in conceptualizing love. In M. Cook & G. Wilson (Eds.),
Love and attraction: An international conference. Oxford: Pergamon Received August 13, 1984
Press. Revision received April 9, 1985 •