104HW

Journal of Personality and Social Psychology 1973, Vol. 26, No. 1, 113-121

"PLAYING HARD TO GET": UNDERSTANDING AN ELUSIVE PHENOMENON 1

ELAINE WALSTER,2 G. WILLIAM WALSTER, JANE PILIAVIN,

AND LYNN SCHMIDT

University of Wisconsin

According to folklore, the woman who is hard to get is a more desirable catch than the woman who is too eager for an alliance. Five experiments were con- ducted to demonstrate that individuals value hard-to-get dates more than easy- to-get ones. All five experiments failed. In Experiment VI, we finally gained an understanding of this elusive phenomenon. We proposed that two com- ponents contribute to a woman's desirability: (a) how hard the woman is for the subject to get and (6) how hard she is for other men to get. We predicted that the selectively hard-to-get woman (i.e., a woman who is easy for the subject to get but hard for all other men to get) would be preferred to either a uniformly hard-to-get woman, a uniformly easy-to-get woman, or a woman about which the subject has no information. This hypothesis received strong support. The reason for the popularity of the selective woman was evident. Men ascribe to her all of the assets of uniformly hard-to-get and the uniformly easy-to-get women and none of their liabilities.

According to folklore, the woman who is hard to get is a more desirable catch than is the woman who is overly eager for alliance. Socrates, Ovid, Terence, the Kama Sutra, and Dear Abby all agree that the person whose affection is easily won is unlikely to inspire passion in another. Ovid (1963), for example, argued:

Fool, if you feel no need to guard your girl for her own sake, see that you guard her for mine, so I may want her the more. Easy things nobody wants, but what is forbidden is tempting. . . . Anyone who can love the wife of an indolant cuckold, I should sup- pose, would steal buckets of sand from the shore [pp. 65-66].

When we first began our investigation, we accepted cultural lore. We assumed that men would prefer a hard-to-get woman. Thus, we began our research by interviewing college men as to why they preferred hard-to-get women. Predictably, the men responded to experimenter demands. They explained that they preferred hard-to-get women because the elusive woman is almost inevitably a valuable

1 This research was supported in part by National Science Foundation Grants GS 2932 and GS 30822X and in part by National Institute for Mental Health Grant MH 16661.

2 Requests for reprints should be sent to Elaine Walster, Department of Sociology, University of Wis- consin, Madison, Wisconsin S3 706.

woman. They pointed out that a woman can only afford to be "choosy" if she is popular— and a woman is popular for some reason. When a woman is hard to get, it is usually a tip-off that she is especially pretty, has a good personality, is sexy, etc. Men also were in- trigued by the challenge that the elusive woman offered. One can spend a great deal of time fantasizing about what it would be like to date such a woman. Since the hard-to- get woman's desirability is well recognized, a man can gain prestige if he is seen with her.

An easy-to-get woman, on the other hand, spells trouble. She is probably desperate for a date. She is probably the kind of woman who will make too many demands on a per- son; she might want to get serious right away. Even worse, she might have a "disease."

In brief, nearly all interviewees agreed with our hypothesis that a hard-to-get woman is a valuable woman, and they could supply abundant justification for their prejudice. A few isolated men refused to cooperate. These dissenters noted that an elusive woman is not always more desirable than an available woman. Sometimes the hard-to-get woman is not only hard to get—she is impossible to get, because she is misanthropic and cold. Sometimes a woman is easy to get because she is a friendly, outgoing woman who boosts

113

114 WALSTER, WALSTER, PILIAVIN, AND SCHMIDT

one's ego and insures that dates are "no hassle." We ignored the testimony of these deviant types.

We then conducted five experiments de- signed to demonstrate that an individual values a hard-to-get date more highly than an easy-to-get date. All five experiments failed.

Theoretical Rationale

Let us first review the theoretical rationale underlying these experiments.

In Walster, Walster, and Berscheid (1971) we argued that if playing hard to get does increase one's desirability, several psy- chological theories could account for this phenomenon:

1. Dissonance theory predicts that if one must expend great energy to attain a goal, he is unusually appreciative of the goal (see Aronson & Mills, 1959; Gerard & Mathewson, 1966; Zimbardo, 1965). The hard-to-get date requires a suitor to expend more effort in her pursuit than he would normally expend. One way for the suitor to justify such unusual effort is by aggrandizing her.

2. According to learning theory, an elusive person should have two distinct advantages: (a) Frustration may increase drive—by wait- ing until the suitor has achieved a high sexual drive state, heightening his drive level by introducing momentary frustration, and then finally rewarding him, the hard-to-get woman can maximize the impact of the sexual reward she provides (see Kimball, 1961, for evidence that frustration does energize behavior and does increase the impact of appropriate re- wards), (b) Elusiveness and value may be associated—individuals may have discovered through frequent experience that there is more competition for socially desirable dates than for undesirable partners. Thus, being "hard to get" comes to be associated with "value." As a consequence, the conditioned stimulus (CS) of being hard to get generates a fractional antedating goal response and a fractional goal response which leads to the conditioned response of liking.

3. In an extension of Schachterian theory, Walster (1971) argued that two components are necessary before an individual can experi-

ence passionate love: (a) He must be physio- logically aroused; and (b) the setting must make it appropriate for him to conclude that his aroused feelings are due to love. On both counts, the person who plays hard to get might be expected to generate unusual pas- sion. Frustration should increase the suitor's physiological arousal, and the association of "elusiveness" with "value" should increase the probability that the suitor will label his reaction to the other as "love."

From the preceding discussion, it is evi- dent that several conceptually distinct vari- ables may account for the hard-to-get phe- nomenon. In spite of the fact that we can suggest a plethora of reasons as to why play- ing hard-to-get strategy might be an effective strategy, all five studies failed to provide any support for the contention that an elusive woman is a desirable woman. Two experi- ments failed to demonstrate that outside ob- servers perceive a hard-to-get individual as especially "valuable." Three experiments failed to demonstrate that a suitor perceives a hard-to-get date as especially valuable.

Walster, Walster, and Berscheid (1971) conducted two experiments to test the hy- pothesis that teenagers would deduce that a hard-to-get boy or girl was more socially desirable than was a teenager whose affection could be easily obtained. In these experiments high school juniors and seniors were told that we were interested in finding out what kind of first impression various teenagers made on others. They were shown pictures and biogra- phies of a couple. They were told how ro- mantically interested the stimulus person (a boy or girl) was in his partner after they had met only four times. The stimulus person was said to have liked the partner "extremely much," to have provided no information to us, or to like her "not particularly much." The teenagers were then asked how socially desirable both teenagers seemed (i.e., how likable, how physically attractive, etc., both teenagers seemed). Walster, Walster, and Berscheid, of course, predicted that the more romantic interest the stimulus person ex- pressed in a slight acquaintance, the less so- cially desirable that stimulus person would appear to an outside observer. The results were diametrically opposed to those pre-

"PLAYING HARD To GET" 115

dieted. The more romantic interest the stimu- lus person expressed in an acquaintance, the more socially desirable teenagers judged him to be. Restraint does not appear to buy re- spect. Instead, it appears that "All the world does love a lover."

Lyons, Walster, and Walster (1971) con- ducted a field study and a laboratory experi- ment in an attempt to demonstrate that men prefer a date who plays hard to get. Both ex- periments were conducted in the context of a computer matching service. Experiment III was a field experiment. Women who signed up for the computer matching program were con- tacted and hired as experimenters. They were then given precise instructions as to how to respond when their computer match called them for a date. Half of the time they were told to pause and think for 3 seconds before accepting the date. (These women were la- beled "hard to get.") Half of the time they were told to accept the date immediately. (These women were labeled "easy to get.") The data indicated that elusiveness had no impact on the man's liking for his computer date.

Experiment IV was a laboratory experi- ment. In this experiment, Lyons et al. hy- pothesized that the knowledge that a woman is elusive gives one indirect evidence that she is socially desirable. Such indirect evidence should have the biggest impact when a man has no way of acquiring direct evidence about a coed's value or when he has little confidence in his own ability to assess value. When di- rect evidence is available, and the man pos- sesses supreme confidence in his ability to make correct judgments, information about a woman's elusiveness should have little im- pact on a man's reaction to her. Lyons et al. thus predicted that when men lacked direct evidence as to a woman's desirability, a man's self-esteem and the woman's elusiveness should interact in determining his respect and liking for her. Lyons et al. measured males' self-esteem via Rosenberg's (196S) measure of self-esteem, Rosenfeld's (1964) measure of fear or rejection, and Berger's (1952) mea- sure of self-acceptance.

The dating counselor then told subjects that the computer had assigned them a date. They were asked to telephone her from the

office phone, invite her out, and then report their first impression of her. Presumably the pair would then go out on a date, and eventu- ally give us further information about how successful our computer matching techniques had been. Actually, all men were assigned a confederate as a date. Half of the time the woman played hard to get. When the man asked her out she replied:

Mmm [slight pause] No, I've got a date then. It seems like I signed up for that Date Match thing a long time ago and I've met more people since then—I'm really pretty busy all this week.

She paused again. If the subject suggested another time, the confederate hesitated only slightly, then accepted. If he did not suggest another time, the confederate would take the initiative of suggesting: "How about some time next week—or just meeting for coffee in the Union some afternoon?" And again, she accepted the next invitation. Half of the time, in the easy-to-get condition, the confederate eagerly accepted the man's offer of a date.

Lyons et al. predicted that since men in this blind date setting lacked direct evidence as to a woman's desirability, low-self-esteem men should be more receptive to the hard-to- get woman than were high-self-esteem men. Although Lyons et al.'s manipulation checks indicate that their manipulations were suc- cessful and their self-esteem measure was reli- able, their hypothesis was not confirmed. Elusiveness had no impact on liking, regard- less of subject's self-esteem level.

Did we give up our hypothesis? Heavens no. After all, it had only been discontinued four times.

By Experiment V, we had decided that per- haps the hard-to-get hypothesis must be tested in a sexual setting. After all, the first theorist who advised a woman to play hard to get was Socrates; his pupil was Theodota, a prosti- tute. He advised:

They will appreciate your favors most highly if you wait till they ask for them. The sweetest meats, you see, if served before they are wanted seem sour, and to those who had enough they are positively nauseating; but even poor fare is very welcome when offered to a hungry man. [Theodota inquired] And how can I make them hungry for my fare? [Socrates' reply] Why, in the first place, you must not offer it to them when

116 WALSTER, WALSTER, PILIAVIN, AND SCHMIDT

they have had enough—but prompt them by be- having as a model of Propriety, by a show of re- luctance to yield, and by holding back until they are as keen as can be; and then the same gifts are much more to the recipient than when they're offered before they are desired [see Xenophon, 1923, p. 48].

Walster, Walster, and Lambert (1971) thus proposed that a prostitute who states that she is selective in her choice of customers will be held in higher regard than will be the prosti- tute who admits that she is completely unse- lective in her choice of partners.

In this experiment, a prostitute served as the experimenter. When the customer arrived, she mixed a drink for him; then she delivered the experimental manipulation. Half of the time, in the hard-to-get condition, she stated, "Just because I see you this time it doesn't mean that you can have my phone number or see me again. I'm going to start school soon, so I won't have much time, so I'll only be able to see the people that I like the best." Half of the time, in the easy-to-get condition, she did not communicate this information. From this point on, the prostitute and the customer interacted in conventional ways.

The client's liking for the prostitute was de- termined in two ways: First, the prostitute estimated how much the client had seemed to like her. (i.e., How much did he seem to like you? Did he make arrangements to return? How much did he pay you?) Second, the ex- perimenter recorded how many times within the next 30 days the client arranged to have subsequent sexual relations with her.

Once again we failed to confirm the hard- to-get hypothesis. If anything, those clients who were told that the prostitute did not take just anyone were less likely to call back and liked the prostitute less than did other clients.

At this point, we ruefully decided that we had been on the wrong track. We decided that perhaps all those practitioners who advise women to play hard to get are wrong. Or per- haps it is only under very special circum- stances that it will benefit one to play hard to get.

Thus, we began again. We reinterviewed students—this time with an open mind. This time we asked men to tell us about the ad- vantages and disadvantages of hard-to-get and

easy-to-get women. This time replies were more informative. According to reports, choosing between a hard-to-get woman and an easy-to-get woman was like choosing between Scylla and Charybdis—each woman was uniquely desirable and uniquely frightening.

Although the elusive woman was likely to be a popular prestige date, she presented cer- tain problems. Since she was not particularly enthusiastic about you, she might stand you up or humiliate you in front of your friends. She was likely to be unfriendly, cold, and to possess inflexible standards.

The easy-to-get woman was certain to boost one's ego and to make a date a relaxing, en- joyable experience, b u t . . . . Unfortunately, dating an easy woman was a risky business. Such a woman might be easy to get, but hard to get rid of. She might "get serious." Per- haps she would be so oversexed or overaffec- tionate in public that she would embarrass you. Your buddies might snicker when they saw you together. After all, they would know perfectly well why you were dating her.

The interlocking assets and difficulties en- visioned when they attempted to decide which was better—a hard-to-get or an easy-to-get woman—gave us a clue as to why our previ- ous experiments had not worked out. The as- sets and liabilities of the elusive and the easy dates had evidently generally balanced out. On the average, then, both types of women tended to be equally well liked. When a slight difference in liking did appear, it favored the easy-to-get woman.

It finally impinged on us that there are two components that are important determi- nants of how much a man likes a woman: (a) How hard or easy she is for him to get; (b) how hard or easy she is for other men to get. So long as we were examining the desira- bility of women who were hard or easy for everyone to get, things balanced out. The minute we examined other possible configura- tions, it becomes evident that there is one type of woman who can transcend the limita- tions of the uniformly hard-to-get or the uni- formly easy-to-get woman. If a woman has a reputation for being hard to get, but for some reason she is easy for the subject to get, she should be maximally appealing. Dating such a woman should insure one of great prestige;

TLAYING HARD To GET" 117

she is, after all, hard to get. Yet, since she is exceedingly available to the subject, the dat- ing situation should be a relaxed, rewarding experience. Such a selectively hard-to-get woman possesses the assets of both the easy- to-get and the hard-to-get women, while avoiding all of their liabilities.

Thus, in Experiment VI, we hypothesized that a selectively hard-to-get woman (i.e., a woman who is easy for the subject to get but very hard for any other man to get) will be especially liked by her date. Women who are hard for everyone—including the subject—to get, or who are easy for everyone to get—or control women, about whom the subject had no information—will be liked a lesser amount.

METHOD

Subjects were 71 male summer students at the University of Wisconsin. They were recruited for a dating research project. This project was ostensibly designed to determine whether computer matching techniques are in fact more effective than is random matching. AH participants were invited to come into the dating center in order to choose a date from a set of five potential dates.

When the subject arrived at the computer match office, he was handed folders containing background information on five women. Some of these women had supposedly been "randomly" matched with him; others had been "computer matched" with him. (He was not told which women were which.)

In reality, all five folders contained information about fictitious women. The first item in the folder was a "background questionnaire" on which the woman had presumably described herself. This ques- tionnaire was similar to one the subject had com- pleted when signing up for the match program. We attempted to make the five women's descriptions dif- ferent enough to be believable, yet similar enough to minimize variance. Therefore, the way the five women described themselves was systematically var- ied. They claimed to be 18 or 19 years old; fresh- men or sophomores; from a Wisconsin city, ranging in size from over 500,000 to under 50,000; 5 feet 2 inches to 5 feet 4 inches tall; Protestant, Catholic, Jewish, or had no preference; graduated in the upper 10%-50% of their high school class; and Caucasians who did not object to being matched with a person of another race. The women claimed to vary on a political spectrum from "left of center" through "moderate" to "near right of center"; to place little or no importance on politics and religion; and to like recent popular movies. Each woman listed four or five activities she liked to do on a first date (i.e., go to a movie, talk in a quiet place, etc.).

In addition to the background questionnaire, three of the five folders contained five "date selection

forms." The experimenter explained that some of the women had already been able to come in, examine the background information of their matches, and indicate their first impression of them. Two of the subject's matches had not yet come in. Three of the women had already come in and evaluated the subject along with her f o u r other matches. These women would have five date selection forms in their folders. The subject was shown the forms, which consisted of a scale ranging from "definitely do not want to date" (—10) to "definitely want to date" (+10). A check appeared on each scale. Presumably the check indicated how much the woman had liked a given date. (At this point, the subject was told his identification dating number. Since all dates were identified by numbers on the forms, this identification number enabled him to ascertain how each date had evaluated both him and her four other matches.)

The date selection forms allowed us to manipulate the elusiveness of the woman. One woman appeared to be uniformly hard to get. She indicated that though she was willing to date any of the men assigned to her, she was not enthusiastic about any of them. She rated all five of her date choices from +1 to +2, including the subject (who was rated 1.75).

One woman appeared to be uniformly easy to get. She indicated that she was enthusiastic about dating all five of the men assigned to her. She rated her desire to date all five of her date choices +7 to +9. This included the subject, who was rated 8.

One woman appeared to be easy for the subject to get but hard for anyone else to get (i.e., the selectively hard-to-get woman). She indicated mini- mal enthusiasm for four of her date choices, rating them from +2 to +3, and extreme enthusiasm (+8) for the subject.

Two women had no date selection forms in their folders (i.e., no information women).

Naturally, each woman appears in each of the five conditions.

The experimenter asked the man to consider the folders, complete a "first impression questionnaire" for each woman, and then decide which one of the women he wished to date. (The subject's rating of the dates constitute our verbal measure of liking; his choice in a date constitutes our behavioral mea- sure of liking.)

The experimenter explained that she was conduct- ing a study of first impressions in conjuction with the dating research project. The study, she continued, was designed to learn more about how good people are at forming first impressions of others on the basis of rather limited information. She explained that filling out the forms would probably make it easier for the man to decide which one of the five women he wished to date.

The first impression questionnaire consisted of three sections:

1. Liking for various dates. Two questions assessed subject's liking for each woman: "If you went out with this girl, how well do you think you would

118 WALSTEE, WALSTER, PILIAVIN, AND SCHMIDT

TABLE 1

MEN'S CHOICES IN A DATK

Item

Number of men choosing to date each woman

Selec- tively hard to get

42

Uni- formly hard to get

6

Uni- formly

ca*y to get

5

No infor-

mation for

No. 1

11

No infor-

mation for

No. 2

7

get along?"—with possible responses ranging from "get along extremely well" (S) to "not get along at all" (1)—and "What was your overall impression of the girl?"—with possible responses ranging from "extremely favorable" (7) to "extremely unfavor- able" (1). Scores on these two questions were summed to form an index of expressed liking. This index enables us to compare subject's liking for each of the women.

2. Assets and liabilities ascribed to various dates. We predicted that subjects would prefer the selective woman, because they would expect her to possess the good qualities of both the uniformly hard-to-get and the uniformly easy-to-get woman, while avoid- ing the bad qualities of both her rivals. Thus, the second section was designed to determine the extent to which subjects imputed good and bad qualities to the various dates.

This section was comprised of 10 pairs of polar opposites. Subjects were asked to rate how friendly- unfriendly, cold-warm, attractive-unattractive, easy- going-rigid, exciting-boring, shy-outgoing, fun- loving-dull, popular-unpopular, aggressive-passive, selective-nonselective each woman was. Ratings were made on a 7-point scale. The more desirable the trait ascribed to a woman, the higher the score she was given.

3. Liabilities attributed to easy-to-get women: The third scale was designed to assess the extent to which subjects attributed selected negative attributes to each woman. The third scale consisted of six statements:

She would more than likely do something to em- barrass me in public.

She probably would demand too much attention and affection from me.

She seems like the type who would be too de- pendent on me.

She might turn out to be too sexually promiscuous. She probably would make me feel uneasy when

I'm with her in a group. She seems like the type who doesn't distinguish

between the boys she dates. I probably would be "just another date."

Subjects were asked whether they anticipated any of the above difficulties in their relationship with each woman. They indicated their misgivings on a

scale ranging from "certainly true of her" (1) to "certainly not true of her" ( 7 ) .

The experimenter suggested that the subject care- fully examine both the background questionnaires and the date selection forms of all potential dates in order to decide whom he wanted to date. Then she left the subject. (The experimenter was, of course, unaware of what date was in what folder.)

The experimenter did not return until the subject had completed the first impression questionnaires. Then she asked him which woman he had decided to date.

After his choice had been made, the experimenter questioned him as to what factors influenced his choice. Frequently men who chose the selectively easy-to-get woman said that "She chose me, and that made me feel really good" or "She seemed more selective than the others." The uniformly easy-to-get woman was often rejected by subjects who com- plained "She must be awfully hard up for a date— she really would take anyone." The uniformly hard- to-get woman was once described as a "challenge," but more often rejected as being "snotty" or "too picky."

At the end of the session, the experimenter de- briefed the subject and then gave him the names of five actual dates who had been matched with him.

RESULTS

We predicted that the selectively hard-to- get woman (easy for me to get but hard for everyone else to get) would be liked more than women who were uniformly hard to get, uniformly easy to get, or neutral (the no information women). We had no prediction as to whether or not her three rivals would differ in attractiveness. The results strongly support our hypothesis.

Dating Choices

When we examine the men's choices in dates, we see that the selective woman is far more popular than any of her rivals. (See Table 1.) We conducted a chi-square test to determine whether or not men's choices in dates were randomly distributed. They were not ( X

2 = 6 9 . S , df-4, p < .001). Nearly all subjects preferred to date the selective woman. When we compare the frequency with which her four rivals (combined) are chosen, we see that the selective woman does get far more than her share of dates (x2 = 68.03, df= l,p< .001).

We also conducted an analysis to determine whether or not the women who are uniformly hard to get, uniformly easy to get, or whose

"PLAYING HARD To GET" 119

popularity is unknown, differed in popularity. We see that they did not (x

2 = 2.86, dj = 3).

Liking for the Various Dates

Two questions tapped the men's romantic liking for the various dates: (a) "If you went out with this woman, how well do you think you'd get along?" and (b) "What was your overall impression of the woman?" Scores on these two indexes were summed to form an index of liking. Possible scores ranged from 2 to 12.

A contrast was then set up to test our hypothesis that the selective woman will be preferred to her rivals. The contrast that tests this hypothesis is of the form l\ — 4/n (selec- tively hard to get) — 1/x (uniformly hard to get) — 2ft, (neutral). We tested the hypothe- sis I\ = 0 against the alternative hypothe- sis T! 7^= 0. An explanation of this basically simple procedure may be found in Hays (1963). If our hypothesis is true, the preced- ing contrast should be large. If our hypothesis is false, the resulting contrast should not differ significantly from 0. The data again provide strong support for the hypothesis that the selective woman is better liked than her rivals (F = 23.92, dj = 1/70, p < .001).

Additional Data Snooping

We also conducted a second set of contrasts to determine whether the rivals (i.e., the uni- formly hard-to-get woman, the uniformly easy-to-get woman, and the control woman) were differentially liked. Using the procedure presented by Morrison (1967) in Chapter 4, the data indicate that the rivals are differ- entially liked (F = 4.43, rf/=2/69). As Table 2 indicates, the uniformly hard-to-get woman seems to be liked slightly less than the easy-to-get or control women.

In any attempt to explore data, one must account for the fact that observing the data permits the researcher to capitalize on chance. Thus, one must use simultaneous testing methods so as not to spuriously inflate the probability of attaining statistical signifi- cance. In the present situation, we are inter- ested in comparing the mean of a number of dependent measures, namely the liking for the different women in the dating situation. To perform post hoc multiple comparisons in

this situation, one can use a transformation of Hotelling's t2 statistic which is distributed as F. The procedure is directly analogous to Scheffe's multiple-comparison procedure for independent groups, except where one com- pares means of a number of dependent measures.

To make it abundantly clear that the main result is that the discriminating woman is better liked than each of the other rivals, we performed an additional post hoc analysis, pitting each of the rivals separately against the discriminating woman. In these analyses, we see that the selective woman is better liked than the woman who is uniformly easy to get (F = 3.99, df = 3/68), than the woman who is uniformly hard to get (F — 9.47, df = 3/68), and finally, than the control women (F = 4.93, df = 3/68).

Thus, it is clear that although there are slight differences in the way rivals are liked, these differences are small, relative to the overwhelming attractiveness of the selective woman.

Assets and Liabilities Attributed to Dates

We can now attempt to ascertain why the selective woman is more popular than her rivals. Earlier, we argued that the selectively hard-to-get woman should occupy a unique position; she should be assumed to possess all of the virtues of her rivals, but none of their flaws.

The virtues and flaws that the subject ascribed to each woman were tapped by the polar-opposite scale. Subjects evaluated each woman on 10 characteristics.

We expected that subjects would associate two assets with a uniformly hard-to-get woman: Such a woman should be perceived to be both "selective" and "popular." Un- fortunately, such a woman should also be assumed to possess three liabilities—she should be perceived to be "unfriendly," "cold," and "rigid." Subjects should ascribe exactly the opposite virtues and liabilities to the easy-to-get woman: Such a woman should possess the assets of "friendliness," "warmth," and "flexibility," and the liabilities of "un- popularity" and "lack of selectivity." The selective woman was expected to possess only assets: She should be perceived to be as

120 WALSTER, WALSTEE, PILIAVIN, AND SCHMIDT

TABLE 2 MEN'S REACTIONS TO VARIOUS DATES

Ilem

Men's liking for dates

Evaluation of women's assets and liabilities

Selective11 Popularb

Friendly" Warm0

Easy-going0

Problems expected in dating

Type of date

Selec- tively hard

to get

9.41°

5.23 4.83 5.58 5.15 4.83

5.23d

Uni- formly

hard to get

7.90

4.39 4.58 5.07 4.51 4.42

4.86

Uni- formly easy

to get

8.53

2.85 4.65 5.52 4.99 4.82

4.77

N -O infor-

mation

8.58

4.30 4.83 5.37 4.79 4.61

4.99

ft The higher the number, the more liking the man is express- ing for the date.

b Traits we expected to be ascribed to the selectively hard- to-get and the uniformly hard-to-gel dates.0 Traits we expected to be ascribed to the selectively hard- to-get and the uniformly easy-to-get dates.d The higher the number the fewer the problems the subject anticipates in dating.

"selective" and "popular" as the uniformly elusive woman, and as "friendly," "warm," and "easy-going" as the uniformly easy woman. A contrast was set up to test this specific hypothesis. (Once again, see Hays for the procedure.) This contrast indicates that our hypothesis is confirmed (F = 62.43, dj = 1/70). The selective woman is rated most like the uniformly hard-to-get woman on the first two positive characteristics; most like the uniformly easy-to-get woman on the last three characteristics.

For the reader's interest, the subjects' rat- ings of all five women's assets and liabilities are presented in Table 2.

Comparing the Selective and the Easy Women

Scale 3 was designed to assess whether or not subjects anticipated fewer problems when they envisioned dating the selective woman than when they envisioned dating the uni- formly easy-to-get woman. On the basis of pretest interviews, we compiled a list of many of the concerns men had about easy women (e.g., "She would more than likely do some- thing to embarrass me in public.").

We, of course, predicted that subjects would experience more problems when con-

templating dating the uniformly easy woman than when contemplating dating a woman who was easy for them to get, but hard for anyone else to get (i.e., the selective woman).

Men were asked to say whether or not they envisioned each of the difficulties were they to date each of the women. Possible replies varied from 1 (certainly true of her) to 7 (certainly not true of her). The subjects' evaluations of each woman were summed to form an index of anticipated difficulties. Pos- sible scores ranged from 6 to 42.

A contrast was set up to determine whether the selective woman engendered less concern than the uniformly easy-to-get woman. The data indicate that she does (F = 17.50, dj = 1/70). If the reader is interested in compar- ing concern engendered by each woman, these data are available in Table 2.

The data provide clear support for our hypotheses: The selective woman is strongly preferred to any of her rivals. The reason for her popularity is evident. Men ascribe to her all of the assets of the uniformly hard-to-get and the uniformly easy-to-get women, and none of their liabilities.

Thus, after five futile attempts to under- stand the "hard-to-get" phenomenon, it ap- pears that we have finally gained an under- standing of this process. It appears that a woman can intensify her desirability if she acquires a reputation for being hard-to-get and then, by her behavior, makes it clear to a selected romantic partner that she is at- tracted to him.

In retrospect, especially in view of the strongly supportive data, the logic underlying our predictions sounds compelling. In fact, after examining our data, a colleague who had helped design the five ill-fated experiments noted that, "That is exactly what I would have predicted" (given his economic view of man). Unfortunately, we are all better at postdiction than prediction.

REFERENCES

ARONSON, E., & MILLS, J. The effect of severity of initiation on liking for a group. Journal of Abnor- mal and Social Psychology, 1959, 67, 31-36.

BERBER, E. M. The relation between expressed ac- ceptance of self and expressed acceptance of others. Journal of Abnormal and Social Psychology, 19S2, 47, 778-782.

"PLAYING HARD To GET" 121

GERARD, H. B., & MATHEWSON, G. C. The effects of severity of initiation and liking for a group: A replication. Journal of Experimental Social Psy- chology, 1966, 2, 278-287.

HAYS, W. L. Statistics for psychologists. New York: Holt, Rinehart, 1963.

KIMBALL, G. A. Hilgard and Marquis' conditioning and learning. New York: Appleton-Century-Crofts, 1961.

LYONS, J., WALSTER, E., & WALSTER, G. W. Playing hard-to-get: An elusive phenomenon. University of Wisconsin, Madison: Author, 1971. (Mimeo)

MORRISON, D. F. Multivariate statistical methods. New York: McGraw-Hill, 1967.

OVID. The art of love. Bloomington: University of Indiana Press, 1963.

ROSENBERG, M. Society and the adolescent self image. Princeton, N.J.: Princeton University Press, 1965.

ROSENFELD, H. M. Social choice conceived as a level of aspiration. Journal of Abnormal and Social Psy- chology, 1964, 68, 491-499.

WALSTER, E. Passionate love. In B. I. Murstein (Ed.), Theories of attraction and love. New York: Springer, 1971.

WALSTER, E., WALSTER, G. W., & BERSCHEID, E. The efficacy of playing hard-to-get. Journal of Experi- mental Education, 1971, 39, 73-77.

WALSTER, E., WALSTER, G. W., & LAMBERT, P. Play- ing hard-to-get: A field study. University of Wis- consin, Madison: Author, 1971. (Mimeo)

XENOPHON. Memorabilia. London: Heinemann, 1923. ZIMBARDO, P. G. The effect of effort and improvisa-

tion on self persuasion produced by role-playing. Journal of Experimental Social Psychology, 196S, 1, 103-120.

(Received October 5, 1971)

ERRATUM

The receipt date for Timothy P. Meyer's article "Effects of Viewing Justified and Unjustified Real Film Violence on Aggressive Behavior" (Journal of Personality and Social Psychology, July 1972, Volume 23) was printed as February 15, 1941. The date should, of course, have ap- peared as February 15, 1971. We point this out to dispel wild rumors about the publication lag on the Journal.