UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Premenstrual syndrome in context McFarlane, Jessica 1992

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


831-ubc_1992_spring_mcfarlane_jessica.pdf [ 10.4MB ]
JSON: 831-1.0098916.json
JSON-LD: 831-1.0098916-ld.json
RDF/XML (Pretty): 831-1.0098916-rdf.xml
RDF/JSON: 831-1.0098916-rdf.json
Turtle: 831-1.0098916-turtle.txt
N-Triples: 831-1.0098916-rdf-ntriples.txt
Original Record: 831-1.0098916-source.json
Full Text

Full Text

PREMENSTRUAL SYNDROME IN CONTEXTbyJESSICA MCFARLANEB.A., The University of British Columbia, 1981M.A., The University of British Columbia, 1985A DISSERTATION SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFDOCTORATE OF PHILOSOPHYinTHE FACULTY OF GRADUATE STUDIES(Psychology)We accept this dissertation as conformingto the required standardTHE UNIVERSITY OF BRITISH COLUMBIAApril, 1992© Jessica McFarlaneIn presenting this thesis in partial fulfilment of the requirements for an advanceddegree at the University of British Columbia, I agree that the Library shall make itfreely available for reference and study. I further agree that permission for extensivecopying of this thesis for scholarly purposes may be granted by the head of mydepartment or by his or her representatives. It is understood that copying orpublication of this thesis for financial gain shall not be allowed without my writtenpermission.Department of The University of British ColumbiaVancouver, Canadaiv 0-r( ( 1619)-,DateDE-6 (2/88)AbstractThis study was designed to address several issues related to PMS. One purposewas to compare women who say they have PMS with those who say they do nothave PMS and to compare women diagnosed with PMS with those not diagnosed ashaving it. Forty-eight women and 11 men (included for comparison) who had amean age of 34 years, were not students, and met other screening criteria,volunteered to keep daily charts for 120 days (prospective daily data). They did notknow the menstrual purpose of the study. Each participant's daily reports wereexamined individually for PMS patterns according to strict criteria, and they wereaccordingly assigned to one of five groups. Only six women (12.5%) met thediagnostic criteria for premenstrual syndrome, but 62.5% said they had PMS.Fourteen women and 2 men (randomly assigned to menstrual cycles), 28% of thetotal sample, had diagnosable "downs" in other phases. The greater proportion ofdiagnosabte downs in phases other than the premenstrual phase calls into questionthe appropriateness of a singular focus on PMS rather than on general cyclicity inadults' day-to-day experiences. Indeed, 74% of all participants in this study(including 73% of the men) had one or more diagnosed cyclic patterns in at least oneof the three (menstrual, day of week, lunar) cycles studied. Participants alsorecalled (retrospective data) their menstrual (women only), weekday, and lunarmoods. When prospective and retrospective data were compared, analysesrevealed that participants may have used menstrual and day of week stereotypes toassist in their recall. Both parametric (normative) and nonparametric (idiographic)analyses were conducted, with sometimes contrasting results. These contradictionsand their implications are discussed. It was concluded that it may be inappropriateto refer to a premenstrual syndrome, that the proposed inclusion of Late LutealPhase Dysphoric Disorder (LLPDD) in the Psychiatric Diagnostic and StatisticsManual (DSM) is questionable, and that more needs to be known about healthycyclical changes before conclusions about unhealthy cyclic changes can be drawn.TABLE OF CONTENTSAbstract^  iiList of Tables ^List of Figures  viAcknowledgements viiiIntroduction ^1Classic menstrual mood pattern ^1The enigma of premenstrual syndrome^  1Theories of PMS^  5Overview of our previous and current research^  1 1Method, Procedures  1 3Sample^  1 7Group assignment^  1 7Scoring  19Results, knowledge of the menstrual purpose of theinvestigation^  21Remembered versus forgotten days^  21Demographic results  2 2Misdiagnosable patterns^  2 2Overview of parametric analyses^  2 5Nonparametric results  5 6Cycle patterns in "terrific" and "miserable" episodes^ 5 8Group results for "terrific" and "miserable" episodes  6 0Non-paramentric vs. parametric results ^  6 4General Conclusions^  6 6References^  8 0Appendix A: A literature review of premenstrual syndrome ^ 81Appendix B: Detailed methodology^ 1 74Appendix C: Examples of recruiting advertisements and posters^1 83Appendix D: Package of forms and questionnaires given toparticipants in the initial interview ^  18 5Appendix E: Daily chart^  19 9Appendix F: Final questionnaire  2 0 2Appendix G: Package of forms given to participants to takehome after final interview^  2 4 9Appendix H: Participant attrition and screening ^ 259Appendix I: Discussion of results of group analyses on initialinterview items^ 2 6 3iiiTABLE OF CONTENTS (continued)Appendix J: Comparison of participants who completed, partiallycompleted, and dropped from the study^  269Appendix K: Diagnostic Manual^ 276Appendix L: Menstrual patterns that potentially could bemisdiagnosed as PMS 298Appendix M: Women's explanations of the causes of PMS^ 303Appendix N: Tables of F, df, p, and power values for all MANOVAsand subsequent ANOVAs^ 315Appendix 0: Figures for parametric results^  331Appendix P: Item analyses of selected questionnaires anddiscriminant analyses^  350Appendix Q: Final questionnaire results and discussion^ 357Appendix R: Discussion of results from idiographic analyses ofpatterns in: (1) positive and negative items, (2) terrific andmiserable episodes, and (3) specific dependent variables^ 372Appendix S: Average frequency of significant positive andnegative items^  398Appendix T: Tables of frequencies of "terrific" and "miserable"episodes  403Appendix U: Tables of frequency of individuals with significantchanges in each dependent variable^  416ivLIST OF TABLESTable H1: Participants remaining in the study at each point ^ 260Table H2: Women and men screened from the Ph.D. study^ 262Table I: Results of group analyses on initial interview items^ 266Table J1: Comparison of participants who completed, partiallycompleted, and dropped from the study^  269Table L1: Significant patterns in prospective menstrual datawhich could be misdiagnosed as PMS  302Table L2: Participants in each group who had a PMSmisdiagnosable pattern^  302Table M1: Quotes of women's explanations of the causes of PMS ^ 306Table M2: Women's explanations of the causes of their PMS^ 314Tables N1-25: Tables of F, df, p, and power values for allMANOVAs and subsequent ANOVAs^  315Table P1: Item analyses of Constructive Anger scale^ 354Table P2: Constructive Anger Scale  354Table P3: Item analyses of Body Wisdom scale^  355Table P4: Body Wisdom scale^  355Table P5: Item analyses of Memories of Abuse scale^ 356Table P6: Memories of Abuse History scale^  356Table Q1: Final questionnaire results  364Tables S1-9: Average frequency of significant positive andnegative items, broken down by group and cycle of interest^ 398Tables T1-15: Frequencies of "terrific" and "miserable" episodesbroken down by group and cycle of interest^  403Tables U1-50: Frequency of individuals with significantincreases and decreases in each dependent variable,broken down by group and cycle of interest^  416LIST OF FIGURESFigure 1: Prospective menstrual reports of patience for eachgroup^  332Figure 2: Prospective reports of lack of hurt feelings acrossmenstrual phases^  332Figure 3: Prospective reports of body balance across menstrualphases for each group  333Figure 4: Prospective reports of patience across menstrual phasesfor each group^  333Figure 5: Prospective reports of lack of hurt feelings acrossmenstrual phases for each group^Figure 6: Retrospective reports of body balance across menstrualphases^Figure 7: Retrospective reports of patience across menstrualphasesFigure 8: Retrospective reports of pleasantness across menstrualphases^Figure 9: Retrospective reports of mood stability acrossmenstrual phases^Figure 10: Retrospective reports of lack of anger acrossmenstrual phasesFigure 11: Retrospective reports of lack of hurt feelings acrossmenstrual phases^retrospective reports of body balanceretrospective reports of patienceretrospective reports of pleasantnessretrospective reports of mood stabilityacross menstrual phases^Figure 16: Prospective vs retrospective reports of lack of angeracross menstrual phasesFigure 17: Prospective vs retrospective reports of lack of hurtfeelings across menstrual phases^Figure 18: Prospective reports of body balance across weekdaysFigure 19: Prospective reports of patience across weekdaysFigure 20: Prospective reports of pleasantness across weekdaysviFigure 12: Prospective vsacross menstrual phases^Figure 13: Prospective vsacross menstrual phases^Figure 14: Prospective vsacross menstrual phases^Figure 15: Prospective vs334334335335336336337337338338339339340340341341LIST OF FIGURES (continued)Figure 21: Retrospective weekday reports of body balance foreach group^  342Figure 22: Retrospective weekday reports of patience for eachgroup  342Figure 23: Retrospective weekday reports of pleasantness foreach group^  343Figure 24: Retrospective weekday reports of mood stability foreach group  343Figure 25: Retrospective reports of body balance across weekdays^  344Figure 26: Retrospective reports of patience across weekdays 344Figure 27: Retrospective reports of pleasantness across weekdays^  345Figure 28: Retrospective reports of lack of anger across weekdays 345Figure 29: Prospective vs retrospective weekday reports ofpleasantness for each group^  346Figure 30: Prospective vs retrospective weekday reports of moodstability for each group  346Figure 31: Prospective vs retrospective reports of body balanceacross weekdays^  347Figure 32: Prospective vs retrospective reports of patienceacross weekdays  347Figure 33: Prospective vs retrospective reports of pleasantnessacross weekdays^  348Figure 34: Prospective vs retrospective reports of lack of angeracross weekdays  348Figure 35: Prospective reports of body balance across lunar^phases   349Figure 36: Prospective lunar reports of body balance for eachgroup^  349viiAcknowledgementsI briefly have listed below those whose help has made my completing this dissertationpossble. I hope that each person knows that I am deeply appreciative of her or his support.First, I would like to thank the research assistants, directed study students, andvolunteers who generously gave of their time and energy. It has been my great pleasure andprivilege to have worked with this research team. In alphabetical order these women and men are:Maria Buziashrili, Judy Chai, Marla Crittenden, Evelyn Dalian, Victoria Desroches, Alix Gilliland, PatDubberley-Habich, Karla Dye, Sharon Eng, Leanne Hammond, Yvette 1psaralexi, Susan Kam,Karen Kwinter, liana Katz, Karen Lyons, Shelagh Lytle, Beth Miller, Shelley Moore, FraserMulholland, Carmen Nicholson, Tami Nicholson, Nadene Rehnby, Mojdeh Shahriari, Dana Tadla,Sonia Usmiani, Rhonda Vanderfluit, Eva Weclaw, Sasza Zawisza.Second, I would like to thank the hundreds of participants who were either contacted byphone and/or provided four months of data and came to two interviews. Their dedication andenergy has made it possible for us to continue unravelling the enigma of premenstrual syndrome.Third, I would like to thank my committee members: Tannis MacBeth Williams who hasbeen an enthusiastic and encouraging supervisor, as well as Anita DeLongis and Janet Welkerwho have given generously of their time and energy in serving on my committee.Fourth, I would like to acknowledge the kind help of Michael Chapman in instructing meon prediction analyses and providing technical feedback on matters of internal consistency; KatyStrachan who assisted me in writing the Power in Relationship scale. I am also grateful for theadvice on internal consistency from Jim Russell and Lawrence Ward. I would also like to thankElizabeth McCririck for her excellent work in the monolithic task of word processing.Fah, I would like to thank my loving life partner Bill Richardson, my dearest friends SandraParker and Karen Lyons, and my counsellor Caren Durante, for their love, laughs, shoulders to cryon, and endless support. I would also like to thank Pat Ward for always believing in me (and tellingme so!) when I needed it most.Finally, I am grateful for the financial support of a Social Science and Humanities ResearchCouncil (SSHRC) Doctoral Fellowship and the generous funding of this research by SSHRC grantnumber 410-90-1307.DedicationThis dissertation is dedicated to the memory of Florence Hodkin McFarlane, bornSeptember 28, 1904. She was a wife, mother of two sons, grandmother of 13 grandchildren andabove all a woman with a gardener's soul. She lived a long life with many pleasures and sorrows. Atthe age of 81 while living alone and confined to a wheel chair, Florence was raped. She decidedto leave this world 13 days later on September 18, 1985. Florence was my grandmother andtaught me about fierce independence, great courage, and personal integrity and dignity. She alsotaught me that the ocean is my home and that I, too, have a gardener's soul. May her spirit live onin all those who know her story.viii1Introduction"Classic" Menstrual Mood Pattern Up to the mid-1970s, it was common for researchers to find the "classic"menstrual mood pattern (Altman, Knowles, & Bull, 1941; Benedek & Rubenstein,1939a, 1939b; Golub, 1976; Ivey & Bardwick, 1968; Janowsky, Berens, & Davis,1973; Luschen & Pierce, 1972; Moos et al., 1969; Patkai, Johannson, & Post, 1974;Rossi & Rossi, 1977). It is characterized by pleasant affect in the ovulatory phase(when ovarian hormones are typically high), followed by negative affect in thepremenstrual phase (when ovarian hormone concentrations are dropping) andmenstrual phase (when hormones are at their lowest).The Enigma of Premenstrual Sydrome As reviews of the literature (McFarlane & Wiliams, 1990; see Appendix A for acomprehensive up-to-date review) indicate, premenstrual syndrome (PMS) is morepuzzling than discussions of the "classic" menstrual mood pattern suggest.Whereas most adult women (60-70%) report experiencing some symptoms overtheir menstrual cycle, only 20-50% of those who retrospectively reported havingPMS showed a PMS pattern in prospective reports collected over one cycle(Hamilton, Parry, Alagna, Blumenthal & Herz, 1984; Kessler, DeLongis, Haskett, &Tal, 1988) and only about 5% did so in two consecutive cycles (Kessler et al., 1988).Shedding light on this enigma is difficult, however, for a number of reasons:Methodological Problems. Reviewers have identified several methodologicalproblems contributing to inconsistencies in menstrual cycle research (e.g.,McFarlane, Martin, & Williams, 1988; Parlee, 1974; Ruble, 1977; Ruble & Brooks-Gunn, 1979), including: (a) when women know they are in a study of the menstrualcycle they tend to report mood variations, but when the purpose is concealed theydo not (see Feldman, 1989; Parlee, 1974; and Ruble, 1977 for evidence); (b)negative bias in checklists: most consist mainly or entirely of negative "symptoms"2(e.g., the Moos Menstrual Distress Questionnaire, MDQ, Moos, 1968) which mayyield a more negative pattern than if positive options were included; (c) retrospectivereports, which differ from concurrent daily reports obtained in prospective studies(e.g., AuBuchon & Calhoun, 1985; Englander-Golden, Chang, Whitmore, &Dienstber, 1980; McFarland, Ross, & DeCourvill, 1989; McFarlane et al., 1988;Olasov, & Jackson, 1987; Parlee, 1974; Ruble & Brooks-Gunn, 1979; Slade, 1984)and which seem to reflect participants' stereotypes more than recall of actualexperiences; (d) failure to examine women's menstrual mood fluctuations in thecontext of other comparison groups (e.g., men) and other cycles (e.g., day of theweek) (see McFarlane et al., 1988 for evidence). This study was designed withthese problems in mind.Menstrual cycle as a dependent variable. Many researchers treat themenstrual cycle as only an independent variable that can influence women's moodsand/or behaviour. The success of using PMS as a legal defense relies on this one-way interpretation of influence. Data indicating that the menstrual cycle can beinfluenced by psychosocial factors suggest that it should also be conceptualized as apotential dependent variable. For example, the menstrual cycle length can beinfluenced by stress (Osofsky & Fisher, 1967; Russell, 1972; Parlee, 1976; Koeske,1980). The effect of stress on the menstrual cycle provides an importantreinterpretation of Dalton's (1964) claim that the menstrual cycle, particularly thepremenstrual and menstrual phases, plays an important role in the timing ofaccidents, suicides, admissions to psychiatric hospitals, and crimes by women.In sum, investigations of the psychosocial factors that may influence themenstrual cycle are relatively rare compared with those in which the menstrual cycleis treated as an independent variable that affects women's physical andpsychosocial well-being. The danger in the predominance of research on the effectsof the menstrual cycle on women is that it may be assumed that this direction of3influence is the only one possible. When interpreting their data, researchers mayforget to ask how the events that occurred during their study could have affectedwomen's menstrual cycles. Sensitivity to the two-way direction of influence betweenthe menstrual cycle and the events in women's lives is important.Premenstrual Signs, Symptoms versus Syndrome. There is confusion aboutwhat constitutes PMS; over 150 symptoms have been reported (Rubinow & Roy-Byrne, 1984). The importance of distinguishing between premenstrual symptoms and syndrome (Brooks-Gunn, 1986) is now recognized. Endicott et al. (Endicott,Nee, Cohen, & Halbreich, 1986) argued that premenstrual change is more accuratethan premenstrual syndrome. The distinction between symptoms and normalchanges or "signs" also should be made clearer. Prior and Vigna (Prior & Vigna,1987) defined molimina as the healthy, mild, physical and emotional, PMS-likechanges that occur in ovulatory cycles. Prior (1985) contended that molimina andPMS may differ in intensity which may be magnified by such factors as increasedfood intake and a sedentary lifestyle.When one classifies an experience as a sign one implies that the experience isa healthy indicator of an event. For example, menstrual flow and moderate crampsmay both be signs of a healthy menstrual physiology but dysmenorrhea or severecramps may be a symptom of menstrual physiological disorder (i.e., very highprostaglandin concentration). Evidence is still needed to indicate whetherpremenstrual changes are indeed signs of healthy molimina or symptoms of an asyet unidentified disorder.Problems in Diagnosing PMS. There is growing evidence (cited earlier) againstdiagnoses based on retrospective reports; prospective reports are required todemonstrate relationships among mood, behavioural changes, and the menstrualcycle. Definitions also vary (e.g., Dalton, 1964; May, 1976; Taylor, 1979); somewomen have only premenstrual symptoms (e.g., PMS), some only menstrual4symptoms (e.g., dysmennorhea), and some both, but because the distinction is notmade clear, most may say they have PMS when asked. Psychiatric illness may beexacerbated in the premenstrual phase and may or may not be distinct from PMS.Selecting among symptoms poses a problem (e.g., Ascher-Svanum, & Miller, 1990;Blumenthal & Nadelson, 1988; Chisholm, Jung, Cumming, & Fox, 1990; Graze, Nee,& Endicott, 1990; Hamilton, Gallant, & Lloyd, 1989; Haskett, Steiner, Osmun, &Carroll, 1980; O'Boyle, Severino, Hurt, 1988; Severino, Hurt, Shindledecker, 1989;Siegel, 1986). Rubinow et al. (1984) and Woods (1985) found that negative affectwas the most important cause of perimenstrual disability. Women's greater difficultywith mood changes than with somatic changes does not rule out a biologicalinterpretation but underscores the need for research on psychological and otherfactors. A woman may be classified by one set of criteria but not by another ashaving PMS, e.g., prospective reports might not yield a PMS diagnosis because thewoman did not have PMS, because the particular month(s) studied were atypicaland other cycles might have revealed PMS patterns (O'Boyle et al., 1988), orbecause the woman did have PMS but the measures did not tap the relevantdomains.Diagnostic Measures. The criteria for the two major diagnostic proceduresusing prospective data, the U.S. National Institute for Mental Health (NIMH)(Hamilton et al., 1984) and the U.S. Psychiatric DSM-IIIR carry several potentialproblems for both researchers and clinicians. One problem with both is use ofstandard rather than individualized criteria regarding magnitude of change overphases of the cycle. Instead, one standard deviation above or below a woman'saverage response in other phases could be considered uncharacteristicpremenstrually, thus controlling for individual variations in "usual" fluctuations. Oneof the goals of this study was to develop such a method and to use this diagnosticmethod in idiographic analyses of patterns over phases of the menstrual, day of5week, and lunar cycles for each participant.A direct comparison of each of the three diagnostic techniques, which wouldrequire comparisons for each individual participant, was beyond the scope of thisthesis but is planned as another aspect of the larger project.Medical Treatment of PMS. An unusually high placebo response rate (19-88%,with a 60% average; Hamilton et al., 1984), makes it difficult to demonstrate theefficacy of any treatment. Moreover, temporary improvement in symptoms has beenobserved when women begin to complete daily diaries without any other treatment.To date, no pharmacological treatment has been found to be more effective thanplacebo in double-blind placebo-controlled cross-over studies (Norris, 1987;Rubinow, & Roy-Byrne, 1984).In sum, diagnostic techniques and treatment studies based on medical andbiological models of PMS have been fraught with difficulties. When carefulmethodologies have been used, neither the cause of PMS nor treatment moreeffective than placebo has been found, so theories based on alternative perspectives(discussed in the next section) are receiving increasing attention.Theories of PMS Biological Theories of PMS. Reviewers have concluded that to date, clearevidence supporting any biological or medical theory has not been found (e.g.,Halbreich & Endicott, 1985; Halbreich, Alt, & Paul, 1988; Parry & Rausch, 1988;Reid, 1985; Rubinow & Roy-Byrne, 1984).Attribution Theories of PMS. Some researchers (e.g., Bains & Slade, 1988;Koeske, 1975; Parlee, 1974; Rodin, 1976) have argued that situational factors andsubsequent attributions, not physiology, determine how moods are experienced andlabelled. Koeske (1975) found that biology was used to explain negativepremenstrual moods but not positive emotions. The evidence supported thehypothesis that hostility and depression were seen as out-of-role for women and that6it was assumed a biological imbalance in the woman must be present. Whennegative behaviour was attributed to the premenstrual phase, the social factorssimultaneously influencing it were discounted and behaviour was rated as moreextreme.PMS as a Heightened Sense of Reality. A woman's perceptions may be moreaccurate in the premenstrual phase, albeit sometimes unpleasant to those aroundher (e.g., Laws, 1985; Shuttle & Redgrove, 1978; Van den Akker, & Steptoe, 1989;Van der Malen, Merckelbach, & Van den Hout, 1988). Hamilton et al. (1989)contended the question should be how and why some women keep the lid on theiranger for 3 out of 4 weeks rather than why they express anger premenstrually.Women have been socialized not to express anger, so some may conceptualize it asoccurring when they are "not themselves", or may see PMS as an alternate self.This may be especially true of women who fear that expressing their dissatisfactionor anger will provoke abuse from men.Premenstrual Behaviour as "Abnormal". Goodman (1986) contended thatwomen's premenstrual behaviour is not abnormal by comparison with men'sbehaviour, but a woman's anger is considered to be a problem when its expressiondisrupts family harmony or affects others (Lever & Brush, 1981). Others (e.g.,Martin, 1987; Taylor, 1988) have argued that there is a double standard. Maleviolence may be accepted as part of men's nature but women's mood changes maybe labelled PMS, with the source of the frustration ignored. It has yet to bedemonstrated empirically that some women experience or express abnormal levelsof anger premenstrually, but there is some evidence that women's premenstrualmood experiences are not abnormal. In previous research, we (McFarlane et al.,1988) found that prospective daily mood reports of arousal, pleasantness, andstability by university-aged women who were normally cycling (womenNC) or takingoral contraceptives (womenOC) did not differ over the phases of their menstrual7cycles from men's reports, with two exceptions. During their menstrual and follicularphases womenNC reported more pleasant moods than did men (randomly assignedto cycles) and womenOC. Retrospectively, however, both groups of womenreported negative moods in the premenstrual and menstrual cycles and enhancedpositive moods in the follicular phase. We hypothesized that a contrast effect maylead women (and perhaps men) to think that women's increased pleasantnessfollowing the premenstrual phase is normal and that their premenstrual moods aredepressed below normal. Our results for changes over days of the week also wereconsistent with this contrast hypothesis, which is explored further in this currentstudy.Role of Stereotypes in PMS Reports. A number of researchers (e.g.,McFarland et al., 1989; McFarlane et al., 1988; Parlee, 1974; Ruble, 1977) havefound that women's recollections of their menstrual moods may be influenced bymenstrual stereotypes. McFarland et al. (1989) found that the more women believedthat menstruation was distressing the greater they biased their recollections of theirmenstrual experiences in the direction of their negative menstrual theory. McFarlaneet al. (1988) found that young adult women, who were not aware of the researchers'interest in the menstrual cycle, recalled experiencing a "classic" menstrual moodpattern over the previous three months even though their prospective, daily reportsfor the same time period did not reveal a "classic" pattern.Why would there be a difference between what women reported on a day-to-day basis and what they remembered? The most likely explanation is that they wereinfluenced by stereotypes. When they did not know the researchers were interestedin the menstrual cycle they did not feel as if they were experiencing PMS symptoms,but these same women believed the stereotype that most women, including them,have menstrual mood changes like PMS. When asked to remember their menstrualmoods they reported what they believed was their typical menstrual pattern, i.e., the8stereotype, even though it was inaccurate for them.How might stereotypes about PMS arise? Our previous finding (McFarlane etal., 1988) of more positive moods in the follicular phase and exaggerated recall ofthese positive moods led us to propose the following hypothesis. Perhaps women(and their partners) believe these especially positive moods are "normal" and that,by contrast, the lower (but still positive) moods that follow them are "abnormal". Theprospective or concurrent data indicate instead, however, that premenstrual moodsare as positive as moods in all other phases except the follicular phase, when themoods of normally cycling women are even more positive.The second finding that lends credence to our interpretation is that a similarpattern occurred for weekly cycles. When men and women were asked toremember their average moods for each day of the week, over the last few months,they exaggerated their day-to-day moods in keeping with the stereotype of "Mondayblues" and "Thank-God-Its-Friday" feelings. They did not report Monday Blues intheir daily records, but they recalled Mondays as bluer than other days, just asStone, Hedges, Neale, and Satin (1985) found. They did prospectively reportFriday/Saturday highs in their daily records but they recalled the weekends as evenmore positive than they had reported concurrently. Thus the contrast effect found forthe premenstrual/menstrual versus follicular phases also seems to apply to days ofthe week. The consistency of the findings for two different kinds of cycles, menstrualand day of week, lends strength to this contrast hypothesis.The difficulty with all stereotypes is their resistance to change (Martin &Halverson, 1981, 1983). People tend to notice things that are stereotype-consistentand ignore inconsistent evidence. Moreover, even if inconsistent evidence isnoticed, it is likely to be forgotten, or distorted in memory. In the case of PMS andstereotypes about mood fluctuations over the menstrual cycle, a similar process mayoccur. If a woman feels "down" on a particular day she will wonder why and look for9a reason. She usually knows if her period is due, and does know if she ismenstruating, so if either is true (and they will be true 25-35% of the time), she maydecide, "I'm down because of my period." Most people are more likely to search forbiological explanations than for environmental ones for their negative feelings, e.g.,work-related or interpersonal events (e.g., Bains & Slade, 1988; Koeske, 1975;Koeske & Koeske, 1975; Slade,1984). Moreover, the woman's partner or friendsmay reinforce this process. If they are having an argument, her partner or friendmay say, "What is the matter with you today? Are you getting your period?" Neitherthe woman nor her partner is likely to say on other occasions, "You are in a terrificmood today," and attribute that to her menstrual cycle. Evidence that contradicts thestereotype is likely to be ignored, so all that is required to maintain the stereotype isan occasional piece of consistent evidence. The woman may only feel "down" onceor twice a year during her premenstrual phase but she is more likely to notice thatand to ignore the more frequent occasions when she feels fine.It may well be that some or even most women do on occasion experience an"out-of-character-for-me" mood or reaction during the premenstrual phase(McFarlane et al., 1988). For example, if they are over-tired, and in a very frustratingsituation over which they have no control (and maybe x and y other things co-occur),and they are premenstrual, they may be more likely to cry, or express anger, or actirritable, etc. The important question is how often all these events co-occur toproduce this reaction and whether it is debilitating for the woman. If it only happensonce every year or so, how much attention does the premenstrual experiencedeserve? Maybe the same woman has that or some other "out-of-character-for me"reaction when another set of events not involving the menstrual cycle co-occur, andfor that matter, maybe this is true for many men.PMS and Stress. Several researchers have linked PMS to stress, but theprecise nature of the relationship is not yet clear (e.g., Beck, 1989; Bruto, 1988;10Chuong, Colligan, Coulam, & Bergstralh, 1988; Hanson, 1987; Heilbrun & Frank,1989; Kerstner, 1987; Meyer, 1989; Shapiro, 1990). Some evidence indicates thatstress may exacerbate PMS symptoms, other evidence indicates that women withPMS may be more physiologically reactive to stress premenstrually, and still otherevidence suggests that stress and other variables may interact to exacerbatesymptoms.PMS as a Culture Bound Syndrome. Johnson (1987) reviewed evidence thatpremenstrual symptoms have been reported since Aristotle but premenstrual syndrome is specific to recent Western culture. He contended that PMS is a culturebound syndrome (Prince & Tcheng-Laroche, 1987) that is a "barometer" of rolestrain, which has increased as women have entered the labour force in largenumbers while continuing to carry primary responsibility for their households andchildren. He hypothesized that as the issues around women's multiple roles areresolved, reports of PMS will decline. Other researchers are studying cross-culturalaspects of PMS (e.g., Buckley & Gottlieb, 1988; Hamilton et al., 1989; Dye, 1991;Monagle, 1991a; Monagle, 1991b).PMS and abuse. The theme of an association between a history of abuse andPMS was discussed at the 9th National Conference of the Society for MenstrualCycle Research. Dekker (1986) found a significant positive association betweenwomen's self-reported premenstrual symptomatology and the degree of distressexperienced after unwanted sexual experiences. Unfortunately, however, this wasthe only reference to research that examined this association. Given the manydifferent approaches in the literature to understanding PMS, it is surprising that moreresearchers have not systematically explored links between history of physical,sexual, and emotional abuses and PMS.Other Theories of PMS. Lander (1988) concluded that feeling out of control isone of the most disturbing premenstrual symptoms for women, and suggested that11PMS may be a "metaphor for women's common position of not being in control oftheir situation" (p. 97). Blechman and Clay (1987) proposed a danger-signalhypothesis of PMS. Women who have very painful periods may search for anyevidence of an oncoming period, and their vigilance may cause them to be irritableand hence manifest PMS. Several authors (Cohen, 1987; Hamilton, 1988; Kattray,1987) have suggested that PMS may be a symptom of marital dissatisfaction or maybe exacerbated by it. Morse and Dennerstein (1988) suggested that family historyand past psychiatric disorders may predispose women to PMS, that biochemicalchanges or stress may precipitate it, and that lifestyle habits may sustain it. Schmidtet al. (1990) suggested that women with PMS have state-related alterations in moodand perception or an interaction between their affective state and their environment.Overview of our Previous and Current Research Methodological and diagnostic issues point clearly to the need for longitudinalresearch involving prospective data collection over several months. In our initialstudy (McFarlane et al., 1988), designed to remedy such methodological problems,we collected daily data for three months from 42 university-aged women and men.As mentioned earlier, the women retrospectively reported they had experiencedstereotypical PMS mood fluctuations but their prospective data over the same perioddid not show that pattern. A similar result occurred for days of the week: men andwomen retrospectively reported Monday blues but this did not occur in theirprospective daily reports. These interesting initial results were encouraging butpointed to the need for an even larger, more complex, multifaceted project.In contrast to McFarlane et al. (1988), this investigation was designed to: (1)study older people, as there is some evidence that mood and other difficulties overthe menstrual cycle may increase with age and/or with multiple roles; (2) recruitpeople from the community in general rather than the university; (3) collect daily datafor 4 rather than 3 months (to cover more cycles); (4) conduct individual analyses for12each participant for the menstrual, weekday, and lunar cycles, as well as groupanalyses (the groups are described in the next paragraph); and (5) apply a newlydeveloped method of PMS diagnosis that uses each woman's own means andstandard errors to evaluate the unusualness (1 standard deviation or more) of highand low scores for each variable in each phase of each cycle.After completion of data collection, participants were assigned to one of fivegroups: women diagnosed with PMS by conservative criteria (conPMS); womendiagnosed with PMS by liberal criteria (IibPMS); women who said they had PMS butdid not meet diagnostic criteria (nodxsay); women who did not say they had PMSand did not meet PMS criteria (nodxNsay); and men.Hypotheses. The prospective daily data and analyses were the focus ofseveral hypotheses regarding the menstrual cycle. Some hypotheses addressedvarious aspects of PMS. For example, it was expected that more nodxsay womenwould have "down" phases that were not premenstrual than would be true of theother groups. NodxNsay women as compared with other groups were expected tohave the fewest number of diagnosable downs, as they were expected to besymptom free. Group analyses of the prospective data were expected to reveal thatby comparison with the other groups, conPMS women would have more negativereports overall and in particular, a more negative premenstrual phase (confirming thegroup selection process). With regard to the retrospective data, conPMS, IibPMS,and nodxsay women were expected to recall more negative premenstrual moodsthan nodxNsay women.Other hypotheses addressed the issue of placing PMS in context. Forexample, with regard to the day-of-week cycle, all groups were expected to showweekend highs and possibly Monday lows in their prospective data, and toexaggerate those highs and lows in their recollections of weekday moods (based onour previous work, McFarlane et al., 1988). It was difficult to generate specific13hypotheses about lunar patterns, given the paucity of systematic prospectiveresearch on lunar cycles, but in keeping with the Transylvania effect, it washypothesized that the full moon phase would be associated with more distinctpatterns than the other lunar phases in both the prospective and retrospective data.MethodProceduresA more detailed methodology is provided in Appendix B.Participants were recruited via advertisements in community newspapers andposters (see Appendix C) placed in public buildings. They were asked to call ifinterested. During the initial telephone contact, the general purpose and length ofthe study were described. To conceal our interest in the menstrual cycle,participants were told that the investigation concerned emotional, behavioural, andhealth patterns. In the initial 1-hour individual interview, demographic and otherinformation was obtained. Participants were asked about typical patterns ofexercise, sleep, health, diet, menstrual cycle flow, discomfort, menstrual moodchanges, and whether or not the menstrual discomfort and menstrual mood changesposed significant problems. Other initial interview questions covered issues such ascomplexion problems, sexual orientation, libido and activity, birth control, andemployment and student status (see Appendix D).Each participant was asked to complete a two-sided 8" x 14" chart (seeAppendix E) each day over the next 120 days, and was instructed to fill out this dailychart at approximately the same time each day. The first section of the chartindicated the time and date of completion. Participants were instructed to be sure tofill out a chart every day but if they forgot, to "catch-up" by filling in two charts thenext day and indicating which one was for the missed day. For forgotten days, theywere asked to provide only information about which they were reasonably sure andto leave a question blank rather than to guess. The second section, which dealt with14body awareness, mental state (e.g., patience), and emotions, in part, was designedto distract the participants from our interest in the menstrual cycle. This section alsoincluded several items suggested in the DSM-111-R as being related to PMS. Aquestion asking women whether or not they were menstruating that day wasembedded in several other questions about general health. All questions in thissection were answered by placing a mark in a box on a nine-point scale.The daily charts took 10-15 minutes to complete each day and were returned,via mail, each week. Participants received a personal telephone contact every 2weeks to help maintain interest and motivation.After providing approximately 4 months of daily data (ranging from 12 to 18weeks of data), participants returned for a 3-hour final interview. They firstcompleted a set of questionnaires (see Appendix F). Some questions weredesigned to ascertain whether they had guessed the purpose of the study, and theirwillingness to continue in the study. They then provided retrospective data byrecalling their moods since the study began. Women (only) indicated their averagemoods since beginning the study during each menstrual phase. Both women andmen did the same thing for each day of the week and each lunar phase. For each ofthese 16 phases (5 menstrual, 7 weekday, 4 lunar), participants recalled theiraverage body experience (positive to negative), patience, pleasantness, moodstability, anger, and hurt feelings.Next, participants completed a series of questionnaires (all included inAppendix F). The Hassles and Uplifts Scale (Delongis, Coyne, Dakof, Folkman, &Lazarus, 1982) consisted of 52 items scored from 0 (none) to 3 (a great deal),indicating the extent to which each item had been a hassle and an uplift over the last4 months.An anger questionnaire developed for this study, the last section of whichincluded ten items rated from 1 (strongly disagree) to 7 (strongly agree) was15designed to ascertain participants' attitudes toward their expression and experienceof anger.A 10-item questionnaire was designed to ascertain attitudes about one's body,ranging from "body as a source of wisdom" to "body as betrayer". Items werescored on a scale from 1 (disagree strongly) to 7 (agree strongly).A feminist attitude questionnaire (Smith, Ferree, & Miller, 1975) consisted ofthe original 26 items plus two added for this study. Several items also werereworded to balance the number worded positively and negatively. Each item wasscored from 1 (disagree strongly) to 7 (agree strongly).All participants then completed a questionnaire about memories of adult andchildhood physical, sexual, and emotional abuse and physical and emotionalneglect. These questions were answered yes or no. During the subsequentinterview, participants who wanted to describe the nature of the abuse theyexperienced were compassionately listened to and their pain validated by theinterviewer. Detailed information about counselling was made available.At the end of the final questionnaire, participants completed a lottery entry formfor a prize of $250.After the questionnaires had been completed (approximately 2 hours), theinterviewer clarified any questions, with special attention to issues that may havebeen disturbing the participant, such as the abuse history. Once this discussionreached its natural conclusion, the participant was given a summary of the computeranalyses of her or his own individual daily charts. This included means and standarddeviations for each daily chart item, prediction analysis results showing weekday,menstrual, and lunar patterns for selected items, and the items from the daily chartthat were most highly correlated (using Pearson correlations) with the item regardingoverall feelings for the day (see Appendix G). In an attempt to control Type I erroronly correlations with p < .009 were reported. In addition, the r coefficients were16ranked into high (p. > .5), medium (.3 < p < .5) and low (p < .3) categories, andparticipants were instructed to place emphasis only on high correlations and to lookfor patterns rather than to focus on a single correlation. For some participants thepresentation of the statistical results was daunting, so we developed several usefultechniques to translate the technical/mathematical information into lay language. Atthe end of the interview participants received an interim debriefing with a promisethat they would be fully debriefed when the last participant had completed the study.An important aspect to the method of this study's methodology was the rapportbetween participants and the research team. Interviewers and telephoners werefriendly and compassionate with highly developed social skills. Several werecounsellors or clinicians in training. The participants consistently remarked toindividual members of the team how much they enjoyed the project and appreciatedthe care they received. Part of the care given participants included: "chats" duringcontact calls, referrals to counselling professionals during distressing times, coffee ortea during interview visits, and extra data collection and analyses if participants hadtheir own questions they wanted to investigate (e.g., is there a relationship betweenautogenic or meditative practices and overall feelings for the day, or sugarconsumption and well being, etc.). Not only were the participants affected by theresearch team's involvement but the team members often were affected by thepoignant, happy, and/or sometimes tragic aspects of the participants' lives. A veryimportant behind the scenes component to the method was the debriefing andmutual support network amongst team members. The recounting of abuse historieswas at times very distressing not only for the participants but also for the teammembers. After caring for the participant, team members were encouraged todebrief with another team member and to do what was necessary to care forthemselves. All participants' stories were kept strictly confidential outside theresearch group, but the research team members counted on each other for support17whenever necessary so that they didn't carry alone the burden of the sometimestraumatic details of abuse and violence. In sum, there was an ethic of responsibilityto both participants and researchers and a strong general emphasis on ethics in thisstudy.Sample A total of 382 people telephoned after seeing ads or posters to say they wereinterested in learning more about the study (295 women, 99 men, and 24 ofunknown gender). A total of 89 people (74 women and 15 men) completed the studyin time for inclusion in this report. After screening (see Appendix H) for depression,chronic illness, etc., data from 48 women and 11 men were available for analyses.Participants were screened for depression in two ways, as measured by the Beckinventory and according to their responses to a history of depression question.Women were screened for cycle length (shorter than 21 orlonger than 35 days).The participants were predominantly white, middle class, and employed. Their meanage was 34 and ranged from 20 to 48 years (for more detailed demographicinformation see Appendix I; analyses comparing the demographic characteristics ofparticipants who completed, partially completed, and dropped from the study arepresented in Appendix J).Group Assignment The procedure for assigning participants to groups is fully described inAppendix K, with a brief description provided here. After each participant hadcompleted all prospective data collection, his or her own mean and standarddeviation were used to construct an upper and lower bound. Any datum fallingbeyond these bounds would be considered positive and non-average or negativeand non-average, respectively. All non-average data were tallied according to howmany fell during each phase. Prediction analyses were then used to determinewhether the non-average days occurred statistically more often during one phase18than the others. Variables with a statistically significant phase pattern were thenexamined for a PMS pattern.Each participant's data were examined for the absence or presence of PMSaccording to strict criteria. Briefly, these criteria can be summarized as: (1) asignificant (according to prediction analyses procedures) cluster of three or morenegative dependent measures (this negative cluster was called a "miserable"episode) in the premenstrual phase; (2) no other "miserable" episodes at any othertime during the menstrual cycle; and (3) at least 75% of all significant variables in thepremenstrual phase were on the negative side of the scale. That is, there was nosubstantial number of positive variables co-existing with negative variables in thepremenstrual phase.Two researchers independently assessed each participant for the existence ofany diagnosable episodes. The interrater reliability was 94%. All discrepancieswere discussed and resolved.Women meeting these criteria were diagnosed with PMS by conservativecriteria and called conPMS women. If women did not meet conPMS criteria but hadone significant negative variable significantly more often in the premenstrual phasethan any other phase, they were diagnosed with PMS by liberal criteria and calledlibPMS women. Unfortunately, the conPMS and libPMS groups were too small tofurther divide the women by whether or not they said they had PMS, and there weresome of each in both groups.Women who said they had PMS but did not have more negative variables intheir prementrual phase than in other phases, that is, no PMS, were referred to asnodxsay women. Women not diagnosed with PMS and who did not say they hadPMS were referred to as nodxNsay women. Finally, men's data were examined forPMS-like patterns but regardless of those patterns all men formed the fifth and finalgroup.19After screening and group assignments 6 women formed the conPMS group,4 of whom said they had PMS and 2 of whom said they did not have PMS; 10women formed the IibPMS group, 6 of whom said they had PMS and 4 of whom saidthey did not have PMS; 20 women formed the nodxsay and 12 women in thenodxNsay groups. In addition, there were 11 men.Scoring Menstrual cycle. The scoring procedure will briefly be described here butmore details are available in Appendix K. The menstrual cycle was divided into fivephases: Menstrual - the actual days of menstrual flow; Post Menstrual - the daysbetween Menstrual and Mid Cycle phases; Mid Cycle - the middle five days betweenfirst and last days of the cycle; Post Mid Cycle - the days between Mid cycle andPremenstrual Phases; and Premenstrual - the seven days before the first day offlow.Men were randomly assigned to 28-day pseudo-menstrual cycles. Day 1varied so that men's "cycles" began on different dates and simulated the startingdate variations of women's cycles. Men were assigned to cycles so they could beincluded in the analyses of prospective reports.The means for the prospective analyses were calculated by averaging foreach individual across the days within each menstrual phase for all menstrualcycles. For example, the mean for premenstrual pleasantness was based on thepleasantness scores for the seven days in each of (usually) three premenstrualphases (21 days total). Means were calculated for each menstrual phase for bodybalance, patience, pleasantness, mood stability, lack of anger and hurt feelings andoverall feelings. The women's retrospective menstrual cycle scores were the singlescores recalled on all the dependent measures listed except overall feelings for eachmenstrual phase. Retrospective menstrual cycle data could not be obtained frommen.20It was not possible to measure for biological indicators of ovulation becausethese techniques (keeping daily oral temperature or blood or urine tests) would havebeen likely to alert women to our menstrual interest. We did ask women to give dailyreports of breast tenderness, as this experience is thought to be a signal of ovulation(Prior, 1990). More recently, however, Prior, 1992, theorized that only side breasttenderness is an indication of ovulation and that front breast tenderness is more anindicator of estrogen concentration. Our procedure was not successful. Becausewe embedded the breast tenderness item amongst other muscle and bodytenderness items, we could not specifically train women to report menstrual breastchanges. As a result of the lack of specific training women seemed to report moreabout muscle tenderness in the breast area than tenderness in the breast itself. Forexample, one women reported no breast tenderness in the "tenderness" section ofher daily charts but did report "PMS breasts" in the open ended negative bodyexperience section. Thus, we do not know what proportion of the observedmenstrual cycles were ovulatory. Some researchers contend than true PMS is onlypossible in ovulatory cycles (Prior, 1990). Unfortunately, we were not able to testthis hypothesis.With no acceptable indicators of ovulation we could not be certain that the midcycle phase reflects an ovulatory phase. We could, however, be certain of themenstrual and premenstrual phase. For the other phases, therefore, labels such aspost menstrual, mid cycle, and post mid cycle are used rather than follicular,ovulatory, and luteal phases because of the lack of biological validation.Weekday cycle. For each measure, the prospective means were calculatedby averaging across all observed Sundays (for Sunday's mean), and so on for eachday of the week. The retrospective scores were the single dependent measure(except overall feelings) scores recalled for each day of the week.Lunar cycle. The lunar cycle was divided into four phases (new moon, first21quarter, full moon, last quarter), based on lunar calendar data. For each measure,the means for the prospective analyses were calculated by averaging across all daysof each lunar phase. The retrospective scores were the single dependent measure(except overall feelings) scores recalled for each day of the week.ResultsKnowledge of the Menstrual Purpose of the Investigation Only one woman guessed the menstrual purpose of the investigation.Unfortunately, with only one person guessing, it was not possible to statisticallydetermine whether there were differences due to guessing versus not guessing themenstrual purpose. This woman was excluded from further analyses.Remembered versus Forgotten DaysThe forgotten days constituted 23.9% of the total data. A MANOVA on allvariables (i.e., body balance7 patience, pleasantness, mood stability, lack of anger,lack of hurt feelings, and overall feelings) comparing remembered with forgottendays was significant, F(7, 8768) = 4.18; p = .000. Follow-up univariate analysesrevealed that participants reported higher mean positive body experiences onforgotten days than remembered days (f(1, 10,369) = 13.205; p = .0003; forgottendays mean = 5.803; remembered days mean = 5.6513). In addition, participantsreported more pleasantness on forgotten days than on remembered days (f(1,11,073) = 6.163, p = .013; forgotten days mean = 6.084; remembered days mean =5.979). Univariate analyses on reports of patience, F(1, 11,027) = 1.363, p = .243;mood stability, F(1, 11,025) = .0645, p = .800; lack of anger, F(1, 10,837) = 2.786;= .0952; lack of hurt feelings, F(1, 10,744) = .0318, .2 = .859; and overall feelings,F(1, 9,604) = 2.438, p = .118; were not significant.Three tests, one for each cycle, were conducted to determine whether therewas a pattern to the frequency of forgotten days. Chi square analyses revealed nosignificant pattern for menstrual (x 2 (4) = 6.147; p = .1885) or lunar (x 2 (3) = 2.528; p22= .4703) cycles. The frequency of forgotten days differed across days of the week,however (x2 (6) = 150.868; p = .0000). The frequencies of forgotten days for all daysin the study across all participants were: 409, 343, 312, 350, 329, 496, 527 forSunday, through Saturday respectively (Note: for comparison, there were 11,199participant-days in the study). Follow-up analyses revealed that Friday and Saturdaywas associated with significantly more forgotten days than Sunday through Thursday(p < .05 for all ten). In addition, Sunday was associated with more forgotten daysthan Tuesday and Thursday (2 < .05 for both).Recall that forgotten days had a significantly higher mean body experiencesand pleasantness. It is possible that rather than reflecting a general positive bias inrecall on these variables, participants forgetting to complete their charts on theweekends were influenced by their weekday stereotypes when recalling the missedweekend day. Thus, the bias could be due to the type of day forgotten (i.e.,weekend) rather than a general positive halo when recalling body experiences andpleasantness.Demographic ResultsThere were no differences amongst the five groups on any of the 23demographic and health variables (from the initial interview) tested. These includedmarital status, highest level of education obtained, living with partner or alone, race,spirituality, health status, history of depression, number of children, and age ofyoungest child. Details for these 23 variables and an additional 15 dealing with themenstrual cycle are provided in Appendix I.Misdiagnosable Patterns Why do so many women say they have PMS but so few actually meetdiagnostic criteria? One possibility is that they may experience demonstrable cyclesthat are not strictly premenstrual and mistakenly>label their experience as PMS. Totest this hypothesis, patterns of "terrific" and "miserable" episodes (i.e., a cluster of23positive or negative items in a given phase; see nonparametric results section belowand Appendix K for more details) for each individual were inspected for the following:a menstrual phase down; a post menstrual down; a mid cycle down; a post mid cycledown; and a pattern in which a high at some phase of the menstrual cycle mightmake the premenstrual phase seem down by comparison (see Appendix L for detailsof these analyses, and Appendix M for women's explanations of the causes of theirPMS).The types of misdiagnosable patterns for each group are displayed in TableL1. The most common misdiagnosable pattern was the "in-contrast-to-positive-PM-looks-down" pattern, which occurred for 19% of the total sample. Next mostcommon was a menstrual down, with 12% showing this pattern, followed by postmenstrual, mid cycle, and post mid cycle downs (3%, 5%, and 10%, respectively). Itis interesting that the most common pattern involved a contrast to a positive orterrific episode, which was one of the hypotheses we proposed to explain some ofour previous results (McFarlane et al., 1988). The majority of researchers studyingPMS do not measure women's positive experiences, so would not be able to detectsuch a pattern. It seems plausible that if some women who believe they have PMSwere told they actually had prospectively reported experiencing an "in-contrast-to-positive..." pattern, they might have their experience of fluctuations over themenstrual cycle validated and be provided with information that would help themreframe their experience. That is, instead of understanding their experiences aspremenstrual distress, these women could understand that their premenstrual phaseis their "ordinary" time and that there is another phase in which they are"extraordinarily" positive.Another way to consider the issue of misdiagnosis is to ask which group hadthe greatest number of misdiagnosable patterns (see Table L2). It was expectedthat women who were nodxsay would show more of these patterns than would24women who were nodxNsay. The proportion of each group with any misdiagnosablepattern was calculated (Note that if a participant had two patterns, she or henevertheless entered only once into the calculation. If anything, therefore, theseproportions conservatively estimate the misdiagnosable patterns observed). Toestablish an unbiased criterion for judging whether one group had a higherproportion of misdiagnosable patterns, the means and standard deviations of theproportion across the five groups were calculated (mean = 45%; sd = 12.02%).Then a confidence interval (CI = 32.98%, 57.02%) was constructed by adding theSD to the mean (upper bound) and subtracting the SD from the mean (lower bound).Only one group exceeded the confidence interval. As expected, nodxsay womenhad the highest proportion (65%) of patterns which could be potentiallymisdiagnosed as PMS. Women who were nodxNsay did not, however, have a lownumber of misdiagnosable patterns by comparison with the other groups.In summary, it is possible that some nodxsay women experience significantmood or other fluctuations over their cycle, some of which may be as distressing asPMS. Because these patterns are not the classic premenstrual cycle pattern,however, these women's experiences may be ignored. It seems important thatclinicians and researchers examine their possible biases. Why should a woman withthe same degree of distress and demonstrable cyclicity as a women with "classic"PMS not get serious attention and validation merely because her down time falls inthe post menstrual or mid cycle phases? The existence of these "down" phases,which meet the same stringent criteria used in diagnosing premenstrual "downs,"clearly suggests that a purely hormonal or biological etiology is unlikely. It is difficultto understand how the same hormonal milieu could explain the "downs" observed inevery phase of the menstrual cycle. To put this dilemma in perspective, it may behelpful to consider the proportions of women (and men randomly assigned tomenstrual cycles) who experienced "down" or "miserable" phases at each point in25the cycle. A total of 10% of the sample met the criteria for a premenstrual "down" or"syndrome". The same proportion of all participants (10%) met the criteria for a postmid cycle down, and slightly more, 12%, met the criteria for a menstrual down. Asmall proportion also met criteria for a post menstrual down (3%) and/or mid cycledown (5%). As this was a volunteer sample there is no way of knowing whether inthe population of adult women and men (randomly assigned to cycles) similarproportions would be found. Nevertheless, these results do suggest that focusingonly on the classic PMS pattern instead of all possible "down" phases could lead tobiased or incomplete conclusions for some adult women and men.To restate and broaden the implications of the diagnostic results, summingacross all 59 individuals in the study, there were 68 diagnosable terrific andmiserable episodes: 6 premenstrual miserable episodes; 18 miserable episodes inother menstrual phases; 11 terrific menstrual episodes (one occurring in thepremenstrual phase); 10 miserable day of week episodes; 4 terrific weekdayepisodes; 12 miserable lunar episodes; and 7 terrific lunar episodes (see AppendixT, Tables T1, T4, and T7). A focus on the 6 premenstrual miserable episodes, 8.8%of the total diagnosable patterns, seems misdirected. If proportions of diagnosedepisodes in the population are close to those found in this sample, it would bedifficult for researchers and diagnosticians to justify such a narrow focus on PMSwhen nonPMS mood cyclicity appears to be more common.Overview of Parametric Analyses Three similar sets of analyses were performed, one for each cycle (menstrual,weekday, lunar). In the interest of brevity, the complete set of analyses for themenstrual cycle will be outlined in this overview. The necessary changes foranalyses of the weekday and lunar cycles will be noted.First, a repeated measures multivariate analysis of variance (MANOVA) wasused to compare the prospective reports of the five groups, that is, conPMS, IibPMS,26nodxsay, and nodxNsay women and men (who were assigned to pseudo-menstrualcycles), across the five menstrual phases (menstrual, post menstrual, mid cycle,post mid cycle, and premenstrual) using the means for each dependent measure(body balance, patience, pleasantness, mood stability, lack of anger and hurtfeelings, overall feelings). Significant MANOVAs were followed up with therespective univariate analyses. The source of any significant interactions wasassessed with simple main effects analyses. Differences among pairs of meanswere tested with the Tukey A range test.After the prospective data were analyzed, the same statistical procedure wasused on retrospective menstrual means except that four groups were compared,since men could not provide these data. Finally, prospective means were comparedto retrospective means using the same basic procedure, again without men.The analyses were repeated forthe weekday cycle using seven days and forthe lunar cycle using the four phases. The major differences between theseanalyses and those for the menstrual cycle were that (1) men were included in theretrospective weekday analyses but (2) men chose not to provide any recollection oftheir lunar cycle moods and could not be included in those analyses. Anecdotally,many participants reported that they could not guess what their lunar mood patternscould be and did not provide lunar mood recollections. As a result of the reducedsample size in the retrospective lunar analyses, power to detect significantdifferences was reduced.Tables of all MANOVA and repeated measures ANOVA F values are providedin Appendix N. All Figures are shown in Appendix 0. To assist the reader throughthe large number of analyses, a summary and discussion is provided at the end ofeach section and overall summaries for menstrual, weekday, and lunar results arealso provided.Prospective menstrual results. In the overall MANOVA the main effects for27group, F(24, 178) = 2.465, p =.000, and phase, F(24, 790) = 53.895; p = .000, andthe group by phase interaction, F(96, 1184) = 1.513, p = .001, were significant.Follow-up repeated measures ANOVAs to the MANOVAs revealed that thegroups varied significantly on the amount of patience reported, F(4, 52) = 3.460, p =.014 (see Figure 1). The means for patience were 5.773, 7.092, 6.270, 6.244, and6.315 for conPMS, IibPMS, nodxsay, and nodxNsay women and men, respectively.Tukey analyses revealed that IibPMS women reported more patience prospectivelythan did conPMS women (p <.01).Univariate tests revealed that there was one marginally significant main effectfor menstrual phase in the prospective reports of hurt feelings, F(4, 208) = 2.306, p =.059 (see Figure 2). The means were: 8.152, 8.184, 8.011, 7.824, and 8.007 for themenstrual, premenstrual, mid cycle, post mid cycle, and premenstrual phases,respectively. Tukey analyses revealed that the post mid cycle phase was associatedwith significantly more hurt feelings than the menstrual, post menstrual, mid cycle,and premenstrual phases (p < .01, for all comparisons). In addition, thepremenstrual phase was associated with more hurt feelings than the menstrual andpost menstrual phases (p < .01 for both). Finally, the mid cycle phase wasassociated with more hurt feelings than the menstrual and post menstrual phases (p< .01 for both).There also was a significant group by phase interaction for reports of bodybalance, F(16, 204) = 2.181, p = .007 (see Figure 3). Simple main effects analyseswhich examined the pattern of menstrual phase means within each group revealedthat only IibPMS women prospectively reported significant fluctuations in their bodybalance across menstrual phases, F(4, 204) = 6.874, p < .01. Tukeys revealed thatthe post-menstrual phase was associated with more positive body experiences thanwere the post mid cycle, premenstrual, and mid cycle phases (p < .01, for all).When simple main effect analyses were used to examine group differences in28each menstrual phase, significant differences were found in all but the mid cyclephase. In the menstrual phase, F(4, 204) = 4.165, p < .01), mean prospectivereports of body balance were 5.088, 6.576, 5.751, 5.437, and 5.442 for conPMS,IibPMS, nodxsay and nodxNsay women and men, respectively. Tukey analysesrevealed that IibPMS women reported more positive mean body experiences in themenstrual phase than did conPMS and nodxNsay women and men.Mean reports of body balance also differed in the post menstrual phase, F(4,204) = 6.518, .p < .01, with mean values of 5.133, 6.827, 5.983, 5.270, and 5.530 forconPMS, IibPMS, nodxsay, and nodxNsay women, and men, respectively. Tukeysrevealed that IibPMS women reported more positive mean body experiences in thepost menstrual phase than did conPMS and nodxNsay women, and men (p < .01 forall three), and nodxNsay women and men (p < .05). In addition, nodxsay womenreported more positive mean body experiences than did conPMS women (p < .05).Reports of body balance also showed a significant post mid cycle pattern,F(4, 204) = 2.702, p < .01, with mean values of 4.808, 5.382, 5.964, 5.579, and5.503 for conPMS, IibPMS, nodxsay, and nodxNsay women and men, respectively.Tukeys revealed that nodxsay women reported more positive mean bodyexperiences in the post mid cycle phase than did conPMS women (p < .01).Finally, reports of body balance showed a significant premenstrual pattern,F(4, 204) = 2.883, p < .05, with mean values of 4.943, 5.216, 5.735, 5.134, and5.958 for conPMS, IibPMS, nodxsay, and nodxNsay women and men, respectively.Tukeys revealed that men reported more positive mean body experiences in their(randomly assigned) premenstrual phase than did conPMS women (p < .01).Participants' reports of patience showed a significant group by phaseinteraction, F(16, 208) = 1.897, p = .022 (see Figure 4). When simple main effectsanalyses were conducted so that the pattern of menstrual phase means wasexamined within each group, two significant effects were observed. First, reports of29patience for conPMS women varied significantly across the menstrual cycle, F(4,208) = 4.406; p < .01. The means were: 5.878, 6.240, 5.542, 5.378, and 5.828 forthe menstrual, post menstrual, mid and post mid cycle and premenstrual phases,respectively, but Tukey analyses revealed that no pair of means was statisticallydifferent.Second, prospective reports of patience for IibPMS women varied significantlyacross the menstrual cycle, F(4, 208) = 7.599, p < .01. The means were: 7.461,7.338, 7.169, 7.309, and 6.184 for the menstrual, post menstrual, mid and post midcycle and premenstrual phases, respectively. Tukey analyses revealed that forIibPMS women the menstrual and post menstrual phases (g < .01 for both), and themid and post mid cycle phases (p < .05 for both) all were associated with morepatience than was the premenstrual phase.When simple main effect analyses were conducted such that groupdifferences were examined for each menstrual phase, significant differences werefound for all but the premenstrual phase. First, group reports of patience varied inthe menstrual phase, F(4, 208) = 5.232, p < .01. The means were: 5.878, 7.461,6.421, 6.277, and 6.129 for conPMS, IibPMS, nodxsay, and nodxNsay women andmen, respectively. Tukey analyses revealed that libPMS women reported morepatience in the menstrual phase than did conPMS (p < .01), nodxsay (p < .05), andnodxNsay women and men (2 < .01 for both).Second, group reports of patience varied in the post menstrual phase, F(4,208) = 4.208, p < .01. The means were: 6.240, 7.338, 6.271, 6.087, and 6.144 forconPMS, libPMS, nodxsay, and nodxNsay women and men. Tukey analysesrevealed that libPMS women reported more patience in the post menstrual phasethan did conPMS (2 < .01), nodxsay < .05), and nodxNsay women and men (2 <.01 for both).Third, group reports of patience varied in the mid cycle phase, F(4, 208) =304.467, .2 < .01. The means were: 5.542, 7.169, 6.237, 6.174, and 6.369, forconPMS, IibPMS, nodxsay and nodxNsay women and men. Tukey analysesrevealed that libPMS women reported more patience inthe mid cycle phase than didconPMS (p < .01), nodxsay and nodxNsay women (p < .05 for both).Finally, group reports of patience varied in the post mid cycle phase, F(4, 208)= 6.260, .2 < .01. The means were: 5.378, 7.309, 6.164, 6.355, and 6.092 forconPMS, IibPMS, nodxsay, and nodxNsay women and men. Tukey analysesrevealed that IibPMS women reported more patience in the post mid cycle phasethan did conPMS, nodxsay (p < .01 for both) and nodxNsay (p, < .05) women andmen (p < .01). In addition, nodxNsay women reported more patience than didconPMS women (p < .05).Participants' prospective reports of lack of hurt feelings showed a significantgroup by phase interaction, F(16, 208) = 1.974, p < .016 (see Figure 5). Whensimple main effects analyses were conducted so that the pattern of menstrual cyclemeans was examined within each group all but one of the analyses, menstrual cyclemeans for nodxsay women, was significant. Reports of lack of hurt feelings forconPMS women showed a significant menstrual cycle pattern, F(4, 208) =_3.546, p <.01 (the means were: 8.535, 8.538, 8.230, 7.772, 7.853, for menstrual throughpremenstrual phases, respectively), as did the prospective reports of IibPMS, F(4,208) = 5.977, p < .01 (the means were: 7.838, 7.888, 7.469, 7.392, 8.05, formenstrual through premenstrual phases, respectively), and nodxNsay women, F(4,208) = 3.206, p < .05 (the means were: 8.147, 8.372, 8.336, 8.247, 7.724, formenstrual through premenstrual phases, respectively). Tukey analyses revealed,however, that no pair of means from any of these three analyses differedsignificantly. The men's prospective lack of hurt feeling reports also showed asignificant menstrual cycle pattern, F(4, 208) = 11.122, p < .01, with means of:7.727, 7.572, 7.791, 7.321 and 8.315 for the menstrual, post menstrual, mid and31post mid cycle and premenstrual phases, respectively. Tukeys revealed that menreported less hurt feelings in their mean (randomly assigned) premenstrual than theirpost mid cycle phase.When simple main effects were conducted to examine group differences ateach menstrual phase all but one analysis, for the premenstrual phase, weresignificant. There were significant group patterns at the menstrual, F(4, 208) =3.619, p < .01, and mid cycle phases, F(4, 208) = 3.157, p < .05, but in Tukeyanalyses no pair of means from these two analyses differed significantly. Grouppatterns were significant at the post menstrual phase, F(4, 208) = 4.974, p < .01,with means of 8.538, 7.888, 8.552, 8.372, and 7.572 for IibPMS, conPMS, nodxsayand nodxNsay women and men, respectively. Tukey analyses revealed that menreported more hurt feelings in the post menstrual phase than did conPMS andnodxsay women (p < .05).Finally, there was a significant group pattern in the post mid cycle phase, F(4,208) = 6.959, p, < .01, with means of: 7.772, 7.392, 8.388, 8.247, and 7.321 forconPMS, IibPMS, nodxsay and nodxNsay women and men, respectively. Tukeysrevealed that nodxsay women reported less hurt feelings in the post mid cycle phasethan did IibPMS women and men (p, < .05 for both). In addition, nodxNsay womenreported less hurt feelings than men in this phase (p < .05).Prospective menstrual summary. Analyses of the prospectively obtained dataover the menstrual cycle phases, with men randomly assigned to cycles, revealedseveral specific differences. Across the menstrual cycle, IibPMS women reportedbeing more patient prospectively than did conPMS women. Within their cycles,IibPMS women reported more positive body experiences in their post-menstrualphase than in their mid and post mid cycle and premenstrual phases. Bycomparison with the other groups, IibPMS women prospectively reported morepositive mean body experiences in their menstrual phases than did conPMS and32nodxNsay women and men. In the post menstrual phase, IibPMS women's positivebody experiences mean was higher than that for all other groups. The negativebody experience mean of conPMS women was lower than that for nodxsay womenin the post mid cycle phase and men in the premenstrual phase. There also weresignificant patterns in reports of lack of hurt feelings. Within their cycles, the meanfor men was lower (less hurt feelings) in their premenstrual than in their post midcycle phase. By comparison with the other groups, men reported more hurt feelingsthan: conPMS and nodxsay women in their post menstrual phase; and nodxsay andnodxNsay women in their post mid cycle phase. Finally, nodxsay women reportedless hurt feelings than did IibPMS women in their post mid cycle phase.It is difficult to interpret the findings that men reported less hurt feelings intheir premenstrual than in their post mid cycle phase, and that they tended to reportmore hurt feelings than some of the women's groups in their post menstrual and postmid cycle phases. That men were found, as a group, to have menstrual fluctuationssuggests that these results may be spurious. Since each man was randomlyassigned to a different menstrual cycle, these random cycles are highly unlikely toreflect the same cyclic experience. For example, one man may receive a paycheque during each of his four phases that were randomly designated aspremenstrual, but it is very unlikely that this coincidence would occur for the othermen. The possibility that the men's group by phase interaction for prospectivereports of lack of hurt feelings may be spurious in turn calls into question the bodybalance results, since the respective values differ by only .064 and p values by .003.Both the body balance and the hurt feelings group by phase interactions, therefore,will be interpreted cautiously.Retrospective menstrual results. The MANOVAs on retrospective menstrualmeans revealed that the main effect for group, F(18, 104) = 1.031, p < .433, and theinteraction of group with menstrual cycle, F(72, 944) = .090, p < .688, were not33significant, but the main effect for menstrual cycle was significant for all thedependent measures, F(24, 630) = 7.373, p = .000.Women's (no men were in these analyses since they could not recall theirmenstrual cycles) recollections of their body experiences varied significantly byphase of the menstrual cycle, F(4, 168) = 26.721, p < .001 (see Figure 6). Theirmeans were: 4.548, 6.769, 6.204, 6.122, and 3.774 for the menstrual, postmenstrual, mid and post mid cycle, and premenstrual phases, respectively. Tukeysrevealed that women recalled a more negative body experience mean in thepremenstrual and menstrual phases than in the post menstrual, mid cycle and postmid cycle phases (p < .01 for all comparisons).Recollections of patience varied significantly by menstrual phase, F(4, 168) =38.255, p. < .001 (see Figure 7). The means were: 4.924, 6.940, 6.274, 6.134, and3.660 for menstrual, post menstrual, mid and post mid cycle, and premenstrualphases, respectively. Tukeys revealed that women recalled being more irritable intheir premenstrual phase than in their post menstrual, mid cycle and post mid cyclephases (p < .01 for all three). They also recalled being more irritable in theirmenstrual phase than in their mid and post mid cycle and post menstrual phases (p< .01 for all three). Finally, they recalled being more irritable in their premenstrualphase than in their post menstrual phase (p. < .05).Women's recollections of pleasantness varied significantly by menstrualphase, F(4, 168) = 22.073, p < .001 (see Figure 8). The means were: 5.169, 7.042,6.399, 6.264, and 4.284 for menstrual, post menstrual, mid and post mid cycle, andpremenstrual phases, respectively. Tukeys revealed that the women recalled theirpremenstrual phase as associated with less pleasantness than any other phase (p. <.01 for all four). They also recalled their menstrual phase as associated with lesspleasantness than their post menstrual, mid and post mid cycle phases. Finally,they recalled their post mid cycle phase as associated with less pleasantness than34their post menstrual phase (p < .05).There was significant menstrual phase variation in recollections of moodstability, F(4, 168) = 29.061, p < .001 (see Figure 9). The means were: 4.951, 6.942,6.386, 6.039, and 3.783 for the menstrual, post menstrual, mid cycle and post midcycle, and premenstrual phases, respectively. Tukeys revealed that the womenremembered their premenstrual phase as characterized by more unstable moodsthan any other phase (2 < .01 for all four). Women also recalled that their moodswere more unstable in their menstrual phases than in their post menstrual, and midand post mid cycle phases (p < .01 for all three). Finally, the post mid cycle phasewas associated with recollections of more unstable moods than was the postmenstrual phase (p< .05).Women's recollections of their lack of anger varied by menstrual phase, F(4,164) = 14.628, p < .001 (see Figure 10). The means were: 5.625, 7.163, 6.700,6.863, and 5.138 for the menstrual, post menstrual, mid and post mid cycle andpremenstrual phase, respectively. Tukeys revealed that women recalled that theywere more angry in their premenstrual and menstrual phases than during their postmenstrual and mid and post mid cycle phases (p. < .01 for all).Finally, women's recollections of their lack of hurt feelings varied by menstrualphase, F(4, 164) = 17.096, p < .001 (see Figure 11). The means were: 5.900, 7.438,7.225, 7.350 and 5.400 for the menstrual, post menstrual, mid and post mid cycle,and premenstrual phase, respectively. Tukeys revealed that women recalled morehurt feelings in their premenstrual and menstrual phases than in their post menstrualand mid and post mid cycle phases (p < .01 for all).Retrospective menstrual summary. In sum, women recalled theirpremenstrual and menstrual phases during the study as characterized by morenegative body experiences and irritability, less pleasantness, more unstable moods,and more anger and hurt feelings than most other phases. Their average menstrual35phase was recalled as having been associated with more irritability than thepremenstrual phase. Conversely, they recalled their premenstrual phase as lesspleasant and associated with more unstable moods than their menstrual phase.Finally, they recalled their post mid cycle phase as associated with more irritability,less pleasantness and more unstable moods than their post menstrual phase.More of the retrospective than the prospective findings were statisticallysignificant, and whereas all of the significant prospective results were group byphase interactions, all of the retrospective ones were significant phase main effects.The clear pattern in the retrospective menstrual results was that women recalled thatthey had experienced classic menstrual mood patterns, and this did not vary bygroup. Their recall for the specific variables (e.g., patience) varied in terms ofwhether they ranked the premenstrual phase as more negative, tied with, or morepositive than the menstrual phase, but they clearly reported that these two phaseshad been more negative than the other phases.It is interesting to note that there were no group differences in menstrualrecall. It was expected that nodxsay women would be more likely to report theclassic menstrual mood pattern than would nodxNsay women. That nodxNsaywomen as a group also recalled a classic menstrual mood pattern provides furtherevidence that recollections of menstrual experiences have less to do with women'sown experiences and more to do with menstrual stereotypes.Prospective versus retrospective menstrual results. Multivariate analysesrevealed that the main effect for report (prospective versus retrospective; F(6, 33) =9.810, p < .000), and the report by group, F(18, 95) = 1.881, p < .027, and report byphase F(24, 582) = 4.699, p < .000, interactions were significant. (These analyseswere restricted to women because men could not recall menstrual cycles.)Follow-up univariate tests revealed that the main effect for type of report (i.e.,pros vs. retro) was significant for women's reports of patience, F(1, 40) = 17.680, p <36.001, lack of anger F(1, 39) = 29.999, p < .001, and lack of hurt feelings, F(1, 39) =41.135, p < .001. Overall, women recalled more anger, hurt feelings, and irritabilitythan they had reported prospectively.Follow-up univariate tests on women's reports of body experiences revealed asignificant report by phase interaction, F(4, 156) = 15.724, p < .001 (see Figure 12).Simple main effects analyses comparing reports at each menstrual phase revealedthat women recalled that they had more negative body experiences than they hadreported prospectively both in the premenstrual, F(1, 156) = 40.397, p < .01, andmenstrual, F(1, 156) = 23.870, p < .01, phases. The pairs of means for prospectiveversus retrospective reports of body experiences were: 5.371 vs. 3.773 and 5.748vs. 4.533 for the premenstrual and menstrual phase, respectively. In the postmenstrual, F(1, 156) = 11.395, p < .01, mid cycle, F(1, 156) = 4.836, p < .05, andpost mid cycle, F(1, 156) = 4.772, p < .05, phases, women recalled more positivebody experiences than they had reported prospectively. The pairs of means forprospective versus retrospective reports of body experience were: 5.872 vs. 6.711,5.606 vs. 6.159, and 5.655 vs. 6.205 for the post menstrual, mid and post midcycles, respectively.There was a significant report by phase interaction for women's reports ofpatience, F(4, 160) = 23.853, p < .001 (see Figure 13). Simple main effect analysesrevealed that women recalled more irritability than they prospectively reported intheir premenstrual, F(1, 160) = 129.24, p < .01, and menstrual, F(1, 160) = 51.465,< .01, phases. The pairs of means for prospective versus retrospective reportswere: 6.205 vs. 3.667 and 6.535 vs. 4.933 for the premenstrual and menstrualphases, respectively. Women's prospective and retrospective reports of patience didnot significantly differ in the other three phases.There was a significant report by phase interaction for women's reports ofpleasantness, F(4, 160) = 14.667, p < .001 (see Figure 14). Simple main effects37analyses revealed that women recalled less pleasantness in their premenstrual, F(1,160) = 50.448, p < .01, and menstrual, F(1, 160) = 13.806, p < .01 phases than theyhad reported prospectively. The pairs of means for prospective versus retrospectivereports were: 5.860 vs. 4.222 and 6.101 vs. 5.244 for the premenstrual andmenstrual phases, respectively. Women also recalled more pleasantness in theirpost menstrual phase than they had reported prospectively, F(1, 160) = 9.753, p <.01. The means were: 6.191 and 6.911 for their prospective and retrospectivereports, respectively.There was another significant report by phase interaction for reports of moodstability, F(4, 160) = 18.176, p < .001 (see Figure 15). Simple main effect analysesrevealed that women recalled having more unstable moods in their premenstrual,F(1, 160) = 61.347, p < .01, and menstrual, F(1, 160) = 31.571, p < .01, phases thanthey reported prospectively. The pairs of means for prospective versus retrospectivereports were: 5.672 vs. 3.778 and 6.003 and 4.644 for the premenstrual andmenstrual phases, respectively. Women also recalled having more stable moods intheir mean post menstrual phase than they had reported prospectively, F(1, 160) =7.086, p < .01. The means for prospective and retrospective reports were 6.023 and6.667, respectively.There was a significant report by phase interaction for lack of anger, F(4, 156)= 9.793, p < .001 (see Figure 16). Simple main effect analyses revealed that womenrecalled experiencing more anger than they had reported prospectively in theirmenstrual, F(1, 156) = 98.143, p < .01, post menstrual, F(1, 156) = 9.984, p < .01,mid, F(1, 156) = 22.564, p < .01, post mid cycle, F(1, 156) = 12.321, p < .01, andpremenstrual phases, F(1, 156) = 101.064, p < .01. The pairs of means forprospective versus retrospective reports of lack of anger were: 7.793 vs. 5.455,7.860 vs. 7.114, 7.599 vs. 6.477, 7.579 vs. 6.750, and 7.480 vs. 5.133, for themenstrual, post menstrual, mid and post mid cycle, and premenstrual phases,38respectively.A similar report by phase interaction was found for reports of lack of hurtfeelings, F(4, 156) = 13.171, p < .001 (see Figure 17). Simple main effects analysesrevealed that women recalled more hurt feelings than they had reportedprospectively in their menstrual, F(1, 156) = 123.977, p < .01, post menstrual, F(1,156) = 25.124, p < .01, mid, F(1, 156) = 26.978, p < .01, post mid cycle, F(1, 156) =17.568, p < .01, and premenstrual phases, F(1, 156) = 134.506, p < .01. The pairsof means for prospective versus retrospective reports of lack of hurt feelings were:8.342 vs. 5.864, 8.411 vs. 7.295, 8.088 vs. 6.932, 8.115 vs. 7.182 and 7.953 vs.5.400 for the menstrual, post menstrual, mid and post mid cycle and premenstrualphases, respectively.Prospective versus retrospective menstrual summary. Overall, averagingacross the menstrual phases, women recalled more anger, hurt feelings, andirritability than they had reported prospectively. Specifically, they recalled morenegative mean body experiences, irritability, unpleasantness, unstable moods,anger, and hurt feelings in their premenstrual and menstrual phases. They recalledmore positive body experiences, patience, pleasantness, and stable moods but moreanger and hurt feelings in their post menstrual phases than they had reportedprospectively. In addition, women also recalled more positive body experiences butmore anger and hurt feelings than they had reported prospectively in their mid cyclephase. Finally, they recalled more anger and hurt feelings than they had reportedprospectively in their post mid cycle phase.The pattern in these findings that women retrospectively reported morenegatively than they had prospectively reported their premenstrual and menstrualphase experiences, and reported more positively their post menstrual phaseexperiences, suggests that they may have been using the classic menstrual cyclestereotypes to assist them in their recall. These stereotypes appear to be ill-suited39to the prospectively reported experiences of most of these particular women.It is noteworthy that overall, women's retrospective reports of their anger andhurt feelings were much more negative than their prospective reports, especially inthe premenstrual and menstrual phases, for which the F values approached orexceeded 100. It seems possible that just as schemata or stereotypes about themenstrual cycle may influence recall, schemata about the amount of anger and hurtfeelings one experiences over time may operate similarly. Prospectively, averagingacross the women, anger and hurt feelings were relatively rarely reported; themeans were 7.704 for lack of anger (i.e., 1.3 for anger) and 8.203 for lack of hurtfeelings (i.e., .8 for hurt feelings) on a 9-point scale where 9 was no anger or hurtfeelings. Retrospectively, these same women recalled higher levels of anger andhurt feelings (lack of anger = 6.219, i.e., 2.8 for anger; lack of hurt feelings = 6.567,i.e., 2.4 for hurt feelings). One possibility is that because of social sanctions againstexpression of anger and hurt feelings these women's experiences did not getdiscussed and resolved when they occurred. The unresolved anger or hurt feelingsmay remain very salient and thus provide the basis for believing that anger and hurtfeelings occurred more frequently than they were reported prospectively. Anotherpossibility is that women do not recognize at the time that they are angry or hurt(because of social sanctions or whatever), that is, they do not recognize their ownfeelings and body signals at the time they prospectively report, and later whenresentment builds and/or they reflect back they realize they really were angry at thetime. It would be difficult to tease out whether women are accurately reporting theiranger and hurt feelings prospectively and inaccurately reporting their anger and hurtfeelings at the time of recall or vice-versa. Another investigation designedspecifically to assess the source of discrepancies would be necessary to answer thisquestion. Regardless of which interpretation is favoured, however, women'sprospectively reported and recalled concept of their own angry and hurt feelings are40at odds.Review of all menstrual cycle results. The prospective menstrual cycle resultsgave scant or no support for the hypotheses that: conPMS women on average havemore negative life experiences than other groups; the premenstrual and/or menstrualphases are in general associated with more negative experiences than otherphases; conPMS and libPMS women would show a classic menstrual patternprospectively but that nodxsay and nodxNsay women and men would not; menwould show no significant menstrual pattern. There were some mean groupdifferences, e.g., IibPMS women reported being more patient than did the othergroups, but there was only a marginal mean menstrual phase effect, i.e., the postmid cycle phase was associated with the most hurt feelings, followed by thepremenstrual and mid cycle phases. These main effects for group and phase werenot as predicted. Further, although there were three significant interactions betweengroup and phase prospective reports, the finding of significant menstrual patterns formen makes it difficult to interpret the significant patterns for women.In contrast, the retrospective menstrual cycle results did support thehypothesis that women would recall a classic menstrual pattern. The analyses forsome dependent measures revealed that women recalled premenstrual and/ormenstrual downs whereas others revealed that they recalled post menstrual highs.Unexpectedly, however, conPMS women were not distinguishable from any othergroup of women in the extent to which they recalled premenstrual downs.Finally, comparing prospective to retrospective reports revealed thatparticipants may have been using the classic menstrual stereotype to assist them inrecalling of their menstrual cycle experiences. Specifically, women recalled theirpremenstrual and menstrual phases as more negative and their post menstrualphase as more positive than they had reported experiencing prospectively. Thesefindings suggest that, for most of these older women from the community, the classic41menstrual stereotype was ill-founded just as we (McFarlane et al., 1988) foundpreviously for younger university women.Prospective weekday results. In the overall MANOVA for prospectiveweekday reports for the main effect for group was not significant, F(24, 182) = 1.434,< .097, but the main effect for weekday, F(36, 1832) = 44.514, p < .0001, and thegroup by weekday interaction, F(144, 1832) = 1.558, p < .001 were significant.Follow-up repeated measures ANOVAs revealed a significant day of weekpattern in participants' prospective reports of body experience, F(6, 312) = 2.742, p <.013 (see Figure 18). The means were: 5.678, 5.541, 5.509, 5.384, 5.540, 5.576,and 5.737 for Sunday through Saturday, respectively. Tukey analyses revealed thatparticipants prospectively reported a more negative mean body experience onWednesday than on Saturday or Sunday (p < .01 for both).In addition, univariate analyses revealed a significant weekday pattern inparticipants' prospective reports of patience, F(6, 318) = 2.304, p < .034 (see Figure19). The means were: 6.153, 6.326, 6.413, 6.414, 6.318, 6.401, and 6.295 forSunday through Saturday, respectively. Tukeys revealed that participants reportedless patience on Sunday than on Tuesday, Wednesday, or Friday (p < .01 for allthree).There also was a significant pattern over days of the week in prospectivereports of pleasantness, F(6, 318) = 2.133, p < .049 (see Figure 20). The meanswere: 6.010, 5.832, 5.911, 5.886, 5.836, 5.943, and 6.094 for Sunday throughSaturday, respectively. Tukey analyses revealed that participants reported morepleasantness on Saturday than on Monday or Thursday (p < .05 for both). No groupby day of week interactions were significant for any prospective dependent variables.Prospective weekday summary. Overall, participants' prospective reportsvaried significantly by day of the week. Specifically, they reported relatively morepositive body experiences and less patience on Sunday, less pleasantness on42Monday, more patience on Tuesday, more negative body experiences and morepatience on Wednesday, less pleasantness on Thursday, more patience on Friday,and more positive body experiences and more pleasantness on Saturday. Theseresults partially support the hypotheses of weekend highs and Monday lows, in thatFriday was associated with one positive variable and Saturday with two, whereasMonday was associated with one negative item. The difficulty with the evidence fora weekend high is that Tuesday was as positive as Friday, and the difficulty with theMonday low evidence is that Thursday was just as low as Monday. These difficultiessuggest that neither the Friday nor the Monday pattern is unique. It was concluded,therefore, that the prospective weekday reports supported only the hypothesis of aSaturday high pattern. In addition, the hypothesis that conPMS women would,overall report more negative experiences than other groups was not confirmed.Retrospective weekday results. The overall MANOVA on retrospective day ofweek reports revealed significant main effects for group, F(24, 170) = 1.627, p <.040, and weekday, F(36, 1724) = 2.209, p < .0001, but not a significant group byweekday interaction, F(144, 1724) = .991, p < .516.Follow-up univariate tests revealed that participants' recollections of theirbody experiences varied by group, F(4, 50) = 4.019, p < .007 (see Figure 21). Themeans for recalled body experiences were: 5.200, 7.127, 6.143, 6.494, and 5.943 forconPMS, libPMS, nodxsay and nodxNsay women and men respectively (note thatwhereas men did not retrospectively provide menstrual cycle data, they did provideretrospective day of week data). Tukeys revealed that libPMS women recalled amore positive mean body experiences than did conPMS women (p < .05).Similarly, univariate analyses revealed that recollections of patience varied bygroup, F(4, 50) = 3.658,2 < .011 (see Figure 22). The means for recalled patiencewere: 5.286, 7.127, 6.014, 6.649, and 6.071 for conPMS, IibPMS, nodxsay andnodxNsay women and men, respectively. Tukeys revealed that IibPMS women43recalled feeling more patient than did conPMS women (p < .05).Participants' recollections of pleasantness also varied by group, F(4, 50) =3.800, p < .009 (see Figure 23). The means for recalled pleasantness were: 5.457,7.317, 6.321, 6.779, and 6.014 for conPMS, libPMS, nodxsay and nodxNsay womenand men, respectively. Tukeys again revealed that IibPMS women recalled morepleasantness than did conPMS women (p < .05).Finally, participants' recollections of their mood stability also varied by group,F(4, 50) = 4.735, p < .003 (see Figure 24). The means for recalled mood stabilitywere: 5.429, 7.508, 6.029, 6.909, and 6.057 for conPMS, IibPMS, nodxsay andnodxNsay women and men, respectively. Tukeys revealed that IibPMS womenrecalled having more stable moods than did conPMS women (p < .05).Univariate tests also revealed significant day of week main effects for four ofthe six dependent measures. First, participant& recollections of their mean bodyexperiences varied significantly across day of the week, F(6, 300) = 3.868, p < .001(see Figure 25). The means were: 6.447, 5.659, 6.093, 5.954, 5.997, 6.402, and6.718 for Sunday through Saturday, respectively. Tukeys revealed that participantsrecalled a more positive mean body experience on Saturday than on Monday (p <.01), Tuesday (p < .05), Wednesday (p < .01), and Thursday (p < .05). They alsorecalled a less positive mean body experience on Monday than on Friday or Sunday(p < .01 for both).Second, participants' recollections of patience varied significantly acrossweekdays, F(6, 300) = 2.862, p < .010 (see Figure 26). The means for recalledpatience were: 6.320, 5.795, 6.080, 6.117, 6.151, 6.430, and 6.715 for Sundaythrough Saturday, respectively. Tukeys revealed that participants recalled morepatience on Saturday than on Monday (p < .01) and Tuesday (p < .05). In addition,they recalled more patience on Friday than on Monday (p < .05).Third, participants' recollections of pleasantness varied significantly by44weekday, F(6, 300) = 8.014, p < .001 (see Figure 27). The means for recalledpleasantness were: 6.604, 5.689, 6.091, 6.108, 6.262, 6.838, and 7.053 for Sundaythrough Saturday, respectively. Tukeys revealed that participants recalled morepleasantness on Saturday than on Monday, Tuesday, Wednesday, or Thursday (p <.01 for all four). In addition, they recalled more pleasantness on Friday than onMonday, Tuesday, Wednesday (p < .01 for all three), and Thursday (p < .05). Theyalso recalled more pleasantness on Sunday than on Monday (p < .01).Finally, participants' recollections of their lack of anger varied significantly byweekday, F(6, 294) = 2.624, .p < .017 (see Figure 28). The means for recalled lackof anger were: 6.744, 6.519, 7.015, 6.955, 6.912, 7.001, and 7.380 for Sundaythrough Saturday, respectively. Tukeys revealed that they recalled less anger onSaturday than on Sunday (p < .05) or Monday (p < .05).Retrospective weekday summary. Overall, averaging across days of theweek, when participants were asked to recall their average experiences for each dayof the week while in the study, IibPMS women recalled more positive bodyexperiences, patience, pleasantness, and stable moods than did conPMS women.This finding of relatively positive average recollections across days of the week forIibPMS women and relatively negative recollections for conPMS women wasunexpected. Perhaps it indicates that conPMS women have a negative bias andlibPMS women a positive bias in their recollections when asked about days of theweek. Note, however, that this was not true for recollections of menstrual phases,suggesting once again that recollections are context dependent. There also weresignificant overall weekday patterns that did not vary by group. Sunday was recalledas characterized by relatively high positive body experiences, more pleasantness,and more anger. Monday was associated with relatively less positive bodyexperiences, less patience, less pleasantness, and more anger. Tuesday wasassociated with relatively less positive body experiences, less patience, and less45pleasantness. Wednesday was associated with less positive body experiences, andless pleasantness. Thursday was associated with less positive body experiencesand less pleasantness. Friday was associated with more positive body experiences,more patience and more pleasantness. Finally, Saturday was associated with morepositive body experiences, more patience, more pleasantness, and less anger.These results for participants' recollections over days of the week are clearlyconsistent with the Monday low and weekend high stereotypes. Moreover, just asthe four groups of women did not differ in recollections of their experiences over theirmenstrual cycle phases, they did not differ (amongst themselves or from men) intheir recollections of their experiences over days of the week).Prospective versus retrospective weekday results. The overall MANOVArevealed a significant main effect for report, F(6, 43) = 14.937, p < .0001, andsignificant report by group, F(24, 166) =1.589, p < .049, and report by weekday,F(36, 1688) = 1.657, p < .009, interactions. The report by group by weekdayinteraction was not significant. Univariate analyses revealed significant main effectsfor report on five of the six dependent measures: body experiences, F(1, 49) =24.413, .2 < .001, with prospective and retrospective means of 5.631 vs. 6.235,respectively; pleasantness, F(1, 49) = 19.526, .2 < .001, with means of 5.941 vs.6.439; mood stability, F(1, 49) = 17.414, p < .001, with means of 5.883 vs. 6.384;lack of anger, F(1, 48) = 9.072, p < .004, with means of 7.604 vs. 6.873; and lack ofhurt feelings, F(1, 48) = 9.231, .2 < .004, with means of 8.045 vs. 7.358. There wasno difference for patience. On average, participants recalled more positive bodyexperiences, more pleasantness, more stable moods, more anger, and more hurtfeelings than they had reported prospectively.Univariate analyses revealed two significant report by group interactions.First, participants' patterns of prospective and retrospective reports of pleasantnessvaried significantly by group, F(4, 49) = 3.667, p < .011 (see Figure 29). Simple46main effect analyses revealed that IibPMS, F(1, 49) = 18.115, p = .01, and nodxNsaywomen, F(1, 49) = 13.967, p < .01, recalled more pleasantness than they hadreported prospectively. The pairs of means for prospective versus retrospectivereports were 6.223 vs. 7.411 and 5.890 vs. 6.779 for libPMS and nodxNsay women,respectively.Second, participants' patterns of prospective and retrospective reports ofmood stability also varied significantly by group, F(4, 49) = 4.225, p < .005 (seeFigure 30). Simple main effect analyses revealed that IibPMS, F(1, 49) = 21.602, p< .01, and nodxNsay women, F(1, 49) = 12.234, p < .01, recalled more moodstability than they had reported prospectively. The means for prospective andretrospective reports were: 6.080 vs. 7.554 and 5.963 vs. 6.909 for IibPMS andnodxNsay women, respectively.Univariate analyses also revealed that four of the six significant report byweekday interactions were significant. First, the pattern of participants' prospectiveand retrospective reports of body experiences varied across days of the week, F(6,294) = 2.131, p < .050 (see Figure 31). Simple main effect analyses revealed thatreports differed significantly for all days except Monday. Participants recalled morepositive body experiences than they had reported prospectively on Sundays, F(1,294) = 18.622, p < .01, with means of 5.727 vs. 6.491; Tuesdays, F(1, 294) =10.090, p < .01, with means of 5.562 vs. 6.130; Wednesdays, F(1, 294) = 8.990, p <.01, with means of 5.520 vs. 6.056; Thursdays, F(1, 294) = 9.172, p < .01, withmeans of 5.607 vs. 6.148; Fridays, F(1, 294) = 22.463, p < .01, with means of 5.579vs. 6.426; and Saturdays, F(1, 294) = 26.532, p < .01, with means of 5.783 vs.6.704.Second, the pattern of participants' prospective and retrospective reports ofpatience varied across days of the week, F(6, 294) = 3.792, p < .001 (see Figure32). Simple main effects analyses revealed that participants recalled less patience47on Monday than they had reported prospectively, F(1, 294) = 5.789, p < .05, withprospective vs. retrospective means of 6.295 vs. 5.907. In addition, they recalledmore patience on Saturday than they had reported prospectively, F(1, 294) = 4.627,< .05, with prospective versus retrospective means of 6.246 vs. 6.593.Third, the pattern of participants' prospective and retrospective reports ofpleasantness varied across weekdays, F(6, 294) = 5.111, p < .001 (see Figure 33).Simple main effect analyses revealed that participants recalled more pleasantnessthan they had reported prospectively on Sundays, F(1, 294) = 18.582, p < .01, withmeans of 6.022 vs. 6.709; Tuesdays, F(1, 294) = 4.275, p < .05, with means of 5.908vs. 6.241; Thursdays, F(1, 294) = 7.788, p < .01, with means of 5.866 vs. 6.315;Fridays, F(1.294) = 31.268, p < .01, with means of 5.953 vs. 6.852; and Saturdays,F(1, 294) = 33.335, p < .01, with means of 6.072 vs. 7.000.Finally, the pattern of participants' prospective and retrospective reports oflack of anger varied across weekdays, F(6, 288) = 2.677, p < .015 (see Figure 34).Simple main effect analyses revealed that they recalled more anger than they hadreported prospectively on Sundays, F(1, 288) = 37.962, p < .01, with means of 7.640vs. 6.655; Mondays, F(1, 288) = 44.274, p < .01, with means of 7.649 vs. 6.574;Tuesdays, F(1, 288) = 15.536, p < .01, with means of 7.655 vs. 7.019; Wednesdays,F(1, 288) = 16.453, p < .01, with means of 7.618 vs. 6.963; Thursdays, F(1, 288) =18.403, p < .01, with means of 7.567 vs. 6.868; Fridays, F(1, 288) = 11.038, p < .01,with means of 7.481 vs. 6.944; and Saturdays, F(1, 288) = 6.717, p < .01, withmeans of 7.622 vs. 7.204.Prospective versus retrospective weekday summary.  On average,participants recalled more positive mean body experiences, pleasantness, stablemoods, anger and hurt feelings than they had reported prospectively. Thesefindings indicate that participants may have a positive bias when recalling someexperiences (e.g., pleasantness) but a negative bias when recalling others (e.g.,48anger). The groups did not differ in this regard for body experiences, anger, and hurtfeelings, but libPMS and nodxNsay women recalled more pleasantness and moodstability than they had experienced prospectively, whereas the other three groupsdid not. Perhaps these two groups of women have a greater positive bias inrecollections for these variables than the other groups. Finally, retrospectiveexceeded prospective reports of positive body experiences on Sunday and Tuesdaythrough Saturday; patience on Saturday; pleasantness on Sundays, Tuesdays andThursday through Saturday; and anger on all days of the week. In addition,participants recalled less patience on Monday than they had reported prospectively.Participants retrospectively had more positive reports for many of their day of weekexperiences, with the exception of a more negative report of their Monday patience,and more negative (retrospective than prospective) reports of anger on all days.Note that in recalling lack of anger over days of the week, the men in this study didnot differ from the women. Moreover, just as women recalled more anger in severalphases of their menstrual cycle than they had prospectively reported, both men andwomen recalled more anger on most days of the week than they had prospectivelyreported.Review of all weekday results. The prospective weekday cycle resultssupported the hypothesis of a Saturday high but only partially supported thehypothesis of a Friday high and Monday low. When participants recalled theirweekday experiences, libPMS women recalled relatively more positive experiencesthan conPMS women, suggesting that the former group may have a positive biasand the latter a negative bias in recalling day of week data. These biases were notseen in recall of menstrual cycles, however. As expected, participants' recollectionsof the weekday cycle were consistent with the Monday blues and weekend highsstereotypes. Finally, participants retrospectively provided more positive reports ofmany of their day of week experiences, with the exception of more negative reports49of their Monday patience and more negative reports of their anger on all days.Prospective lunar results. The MANOVA analysis of the prospective lunarcycle data revealed that the group main effect was not significant, F(24, 182) =1.495, p < .073, but the main effect for lunar phase, F(18, 449) = 40.425, p < .0001,and the group by phase interaction, F(72, 896) = 1.500, p < .006, were significant.Univariate analyses revealed that, overall, participants' prospective reports of theirmean body experiences varied marginally by lunar phase, F(3, 156) = 2.649, p <.051 (see Figure 35). The means for body experiences were 5.589, 5.679, 5.555,and 5.404 for the new moon, first quarter, full moon, and last quarter phases,respectively. Tukey analyses revealed that participants reported more positive bodyexperiences in the first quarter than the last quarter (2 < .05) of the lunar cycle.Univariate analyses also revealed a significant group by lunar phaseinteraction for prospective reports of body experiences, F(12, 156) = 3.029, p < .001(see Figure 36). When simple main effects were conducted so variations over thelunar phases were examined within each group only IibPMS women showed asignificant lunar pattern, F(3, 158) = 12.254, p < .01. Their lunar means were 6.260,6.520, 5.701, and 5.268 for the new moon, first quarter, full moon, and last quarterphases, respectively. Tukeys revealed that IibPMS women reported more positivemean body experiences in the first quarter and new moon phases than in the lastquarter and full moon phases (2 < .01 for all four comparisons).When simple main effect analyses of the group by lunar phase interactionwere conducted so that group differences were examined for each phase, therewere significant differences among the groups in all phases. First, they varied in thenew moon phase, F(4, 158) = 11.030, p < .01, with means of 4.638, 6.260, 5.766,5.450, and 5.605 for conPMS, IibPMS, nodxsay and nodxNsay women and men,respectively. Tukeys revealed that IibPMS women prospectively reported morepositive mean body experiences during the new moon than did conPMS and50nodxNsay women (2 < .01 for both) and men (p < .05). In addition, conPMS womenreported more negative mean body experiences during the new moon than didnodxsay and nodxNsay women and men (p < .01 for all three).Second, the groups varied at the first quarter, F(4, 158) = 9.546, p < .01, withmeans of 5.260, 6.520, 5.754, 5.334, and 5.677 for conPMS, libPMS, nodxsay andnodxNsay women and men, respectively. Tukeys revealed that IibPMS womenreported more positive mean body experiences than all other groups (2 < .01 for allfour) during the first quarter.Third, the groups varied at the full moon, F(4, 158) = 5.472, p < .01, withmeans of 5.052, 5.701, 6.014, 5.593, and 5.414 for conPMS, libPMS, nodxsay andnodxNsay women and men, respectively. Tukeys revealed that conPMS womenreported less positive mean body experiences during the full moon than did libPMS(2, < .05) and nodxsay (2 < .01) women. In addition, nodxsay women reported morepositive mean body experiences during the full moon than did men (2 < .05).Finally, groups varied in the last quarter phase, F(4, 158) = 5.232, p < .01,with means of 4.924, 5.2678, 5.869, 5.435, and 5.523 for conPMS, libPMS, nodxsayand nodxNsay women and men. Tukeys revealed that conPMS women reportedmore negative mean body experiences during the last quarter phase than didnodxsay women (2 < .01) and men (p < .05). In addition, nodxsay women reportedmore positive mean body experiences during the last quarter phase than did libPMSwomen (p < .05).Prospective lunar summary. Significant prospective lunar patterns werefound only for body experiences. Specifically, participants on average reported morepositive body experiences in the first than in the last quarter. A significant group bylunar phase interaction revealed that IibPMS women reported more positive meanbody experiences in the new moon and first quarter than in the full moon and lastquarter. Looking at the effect from a different perspective, the groups varied51significantly in every lunar phase. ConPMS women experienced relatively morenegative mean body experiences in the new moon, full moon, and last quarter.Nodxsay women experienced relatively more positive mean body experiences in thefull moon and last quarter. These results are not consistent with the hypothesis thatthe groups would not differ over the lunar phases. Instead it appears that libPMSwomen, followed by nodxsay women, reported more positive mean bodyexperiences, whereas conPMS women reported more negative body experiencesduring the various phases.Retrospective lunar results and summary. Fewer than half the participants (N= 25) were willing to provide the recollections of their moods and experiencesaccording to phases of the moon. This reduced N compromised the power of theanalyses to detect a group main effect and group by phase interaction. Neither theoverall MANOVA for group, F(18, 44) = .923, p < .557, nor the group by phaseinteraction, F(54, 338) = .596, p < .989, were significant. The main effect for lunarphase was significant, F(18, 170) = 1.727, p = .039, but none of the univariate follow-up analyses was statistically significant.Prospective versus retrospective lunar results and summary. The overallMANOVAs revealed a significant main effect for report, F(6, 15) . = 4.586, p < .008,but the report by group, F(18, 41) = 1.553, p < .121, report by phase, F(18, 161) =1.282, p < .206, and the report by group by phase, F(54, 320) = .940, p < .597,interactions were not significant. Univariate analyses revealed that, overall,participants recalled more positive body experiences, F(1, 20) = 14.857, p < .001,with means of 5.721 vs. 6.438 and more pleasantness, F(1, 20) = 8.883, p < .007,with means of 6.098 vs. 6.500 than they experienced prospectively.As so few participants provided retrospective reports for the lunar cycle,comparisons with prospective reports must be interpreted with caution. In this case,however, power to detect differences in the MANOVA was not as compromised52because of the repeated measures nature of the design. Only the report main effectwas significant. Univariate follow-up analyses revealed that a difference in meanprospective and retrospective reports occurred specifically for body balance,reflecting a tendency for participants to recall more positive lunar body experiencesthan they experienced prospectively and for pleasantness, reflecting a tendency forparticipants to recall more pleasant moods than they experienced prospectively.Thus, participants recalled their lunar pleasantness and body experiences morepositively than they reported prospectively.Review of all lunar results. The prospective lunar cycle results were notconsistent with the hypothesis that the groups would not differ over lunar phases.Instead, IibPMS women followed by nodxsay women reported more positive meanbody experiences, whereas conPMS women reported more negative bodyexperiences during the various lunar phases. In addition, the participants who werewilling to provide lunar retrospective data tended overall to have a positive bias intheir recall of both body balance and pleasantness. Note that these are the twovariables identified in the forgotten versus remembered analyses as being subject topositive recall bias. It is possible that rather than reflecting a bias specific to lunarstereotypes, this positive bias is more general.Overall Summary of Menstrual, Weekday, and Lunar Parametric Results.Prospectively, IibPMS women appeared to have more positive mean reports forsome variables (i.e., body experience and patience). Unexpectedly, conPMSwomen showed only one difference from the other groups, reporting a more negativemean body experience in the post mid cycle phase than the other groups. The lackof more significant differences is especially surprising given that to be diagnosed asconPMS, each woman had to demonstrate a clear premenstrual down in three ormore variables in their prospective data. But when the prospective data of allconPMS women were combined into one group, there were no demonstrable53premenstrual differences between them and the other groups. The finding ofmenstrual patterns in men's prospective data (who were assigned to pseudo-cycles)also was surprising. The lack of more distinctive patterns in the prospective data ofconPMS women and the finding of significant "menstrual" patterns for men suggeststhat the group parametric analyses may obscure actual patterns for individualparticipants.Whereas only conPMS, and to some extent libPMS women, prospectivelyreported premenstrual downs, all groups of women recalled that they hadexperienced premenstrual downs as well as menstrual downs and mid cycle highs.That nodxNsay women also reported premenstrual downs suggests that menstrualrecall has more to do with menstrual stereotypes than with day-to-day experience.These results are consistent with those of McFarland et al. (1989), which revealedthat the more women held negative menstrual beliefs or theories, the more theybiased their recollection of menstrual experience.The only clear weekday pattern in participants' prospective data was aSaturday high but they recalled Monday lows and Friday and Saturday highs. Agroup difference was detected in weekday recall. LibPMS women appeared to havemore positive reports and conPMS women appeared to have more negative reportswhen recalling their experiences over days of the week (but not over the menstrualcycle). In addition, all groups recalled more anger across weekdays than theyreported prospectively.The prospective lunar data did not reveal any distinct patterns for the fullmoon phase but did reveal more positive body experiences in the first over the lastquarter. In addition, LibPMS and nodxsay women reported more positive meanbody experiences than did conPMS women. This was the only prospective findingsuggesting that conPMS women may have had more negative life experiences thanthe other groups. Analyses of recollections of lunar patterns was hindered because54many participants opted not to provide these data. No patterns were discernible inthe retrospective lunar data. That so few participants were willing to provideretrospective lunar data suggests that the concept of lunar cyclicity was foreign tomany participants in this study. Perhaps if there were a more detailed stereotypeabout lunar cycling in Western culture, there would have been a better response rateand clearer patterns in lunar recollections.What conclusions can be drawn from the parametric group analyses of themenstrual, weekday and lunar prospective and retrospective data?First, stereotypes about the menstrual and weekday cycles may be reliedupon when recalling one's experences. The day-of-week stereotypes seem toprovide a better fit, although still not a good match, to prospective reports than doesthe classic menstrual pattern stereotype. Western culture provides a rather vaguestereotype about a full moon effect and thus may be used less in recalling lunarmoods. Given the apparent influence of menstrual and day of week stereotypes onrecall, one might hypothesize that had there been a stronger lunar stereotype in theculture, more participants might have been willing to go out on a limb and provideretrospective lunar data.Second, there was some evidence that, overall, IibPMS women may have hadsome more positive mean experiences prospectively (e.g., menstrual prospectivepatience) and some more positive reports in recalling their moods (e.g., weekday).In addition, conPMS women reported one negative prospective weekday bodyexperience and showed evidence of more negative reports in recall of someweekday patterns. It is difficult to understand why conPMS and libPMS womenwould stand in apparently stronger contrast than either or both of these groupswould stand by comparison with women not diagnosed with PMS or even men.Third, group parametric (normative) analyses appear to obscure somedifferences that are observable at the individual (idiographic) level (e.g., conPMS55women had clear premenstrual downs) and may even provide spurious results (e.g.,men's "menstrual" patterns). Given this possibility that group parametric results maynot be the optimal way to approach the data, in the next section we will examine theindividual menstrual, weekday, and lunar cycles as well as group differences in thenonparametric data used initially to diagnose PMS and to form the groups.Discriminant Analyses and Final Interview Results. Discriminant analyses aredescribed in detail in Appendix P. Six questionnaires were chosen a priori from theset of final questionnaires (Hassles, Uplifts, Constructive Anger, Body Wisdom,Attitudes toward Feminism, and Memories of Abuse) to be included in multivariateanalyses and subsequent discriminant analyses. A MANOVA revealed that the fivegroups did not differ in their responses to these 6 scales, F(24, 182) = .6147; p =.920. Further discriminant analyses were, therefore, not appropriate.For exploratory purposes, all items in the final questionnaire were examinedfor group differences; the results are presented in Appendix Q. No attempt wasmade to control Type I error since there were 98 separate analyses and Bonferronicorrections would have been prohibitive. In an attempt to slightly reduce the accruedType I error, only coherent patterns of results were interpreted. The theme that bestdescribed the greatest number of items on which group differences were observedwas emotional and sexual abuse. Specifically, conPMS and IibPMS womenreported the highest incidence and men reported the lowest incidence of abuse.This finding is consistent with the theory of PMS as a heightened awareness ofreality or premenstrual sanity. It is possible that the women with PMS or symptomsin this investigation had experienced a higher incidence of abuse than the otherparticipants, and that their PMS symptoms were, in a sense, their bodies' way ofcommunicating the psychological damage done by the abuse. In addition, conPMSwomen reported the highest incidence of emotional abuse with their current partnerand they had been with their (more) abusive partners an average of 11.9 years.56That these women reported the highest levels of PMS symptoms and the highestlevel of emotional abuse over a longer period of time is consistent with theoriesproposing that PMS is related to marital dissatisfaction. Further research with aspecific focus on abuse and PMS is needed, however, before conclusions could bedrawn with confidence.Nonparametric Results As a way to look for patterns in the prediction analysis data, it was decided toexamine patterns in larger clusters of items rather than for individual items. (SeeAppendix K for more detailed description of the prediction analysis data. Thisdiscussion of nonparametric results is also provided in Appendix R and T.)For example, it could be argued that one significant item such as increasedpatience could have been transitory or subtle for a participant, but that a cluster of 3or more positive items would be psychologically more salient to her or him. A clusterof 3 or more positive items was accordingly identified as a "terrific" episode (for thatphase or day), and a cluster of 3 or more negative items was identified as a"miserable" episode. Some participants had large clusters of mixed items, e.g., 3positive and 1 negative item for a particular phase or day. An additional criterion for"terrific" and "miserable" episodes, therefore, was that at least 75% of all the items ina day or phase had to be either positive for terrific or negative for miserableepisodes. For example, 3 positive and 1 negative item would qualify as a terrificepisode but 3 positive and 2 negative items would not. The underlying assumptionin this set of analyses was that the terrific and miserable episodes were moreorganized and reliable units of analysis than would be a procedure involving singleitems. Each episode for each participant was categorized according to theparticipant's group membership and phase or day of the cycle and finally tabulated.Calculations and tables. To find the cycle averages for the menstrual phases,all terrific items within a phase were tallied, and the same was done for all the57miserable items. To illustrate, the frequencies for each group for terrific items in themenstrual phase (shown in Table T1) were 1 for all groups. The total of thesefrequencies, 5 was the number entered as the phase total for the terrific items in themenstrual phase (see Table T2). All other phase averages were calculatedaccordingly. Next, to determine which of these frequencies was relatively high orlow, the mean and standard deviation (SD) for the phase frequencies wascalculated. For example, 5, 7, 0, 2, 5, 3, 0, 6, 1, and 6 were the phase frequenciesfor the menstrual cycle. Their mean was 3.5 and their SD was 2.6. Finally, aconfidence interval was constructed by adding the SD to the mean for the upperbound and by subtracting the SD from the mean for the lower bound. Theconfidence interval in the above example is CI = .9, 6.1. All that remained was tocompare each phase frequency to the confidence interval to see whether it wasexceeded. Phase averages beyond the upper bound were labelled in the table as"high" and beyond the lower bound were labelled as "low". These labelled phasefrequencies were then reported in the results below.To better present and assess the group averages, a third table was made,based on the means from values in the Table T1 described above. For example,Table T3 presents the group averages obtained from values presented in Table Ti.First, means were obtained for terrific episodes for each group by averaging acrossall the terrific episodes in all phases. For example, conPMS women had thefollowing values for terrific episodes: 1, 0, 1, 0, and 0 for the menstrual throughpremenstrual phases respectively. The mean of these numbers is .33 and that valuewas entered in Table T3 under conPMS, terrific episodes. All other cells in the tablewere similarly completed. Next, the means and SDs of all terrific and miserablevalues for all groups was calculated. For example, Table T3 had the followingvalues for terrific and miserable episodes for each group: .33, 1.0, .20, .20, .15, .50,:.08, .33, .27 and .18. The mean was .324. The SD was .265. The confidence58interval became CI = .059, .589. Tabled values that exceeded the confidenceinterval were labelled and reported in the results. The rows of this table wereaveraged to obtain the overall mean frequency for each group and the columns wereaveraged to obtain the overall mean frequency for terrific and for miserable items.These row and column means were also evaluated against the confidence intervalfor this table.It is important for readers to note that these nonparametric or predictionanalyses were conducted using the same prospective data used in the parametricanalyses.Cycle Patterns in "Terrific" and "Miserable" Episodes Menstrual cycle. Remarkably, all but one of the 10 possible combinations(mid cycle, miserable episodes) was either a high or low frequency outlyer (seeTable T1-3; for comparison below the overall mean frequency = 3.5; SD =2.6; andCI = .9, 6.1). Averaging across the groups, the menstrual phase was associatedwith a high frequency of "miserable" (freq = 7) episodes. The post menstrual phasewas associated with a low frequency of terrific (freq = 0) episodes. The mid cyclephase was associated with neither a high nor low frequency of terrific or miserableepisodes. The post mid cycle phase was associated with a low frequency of terrific(freq = 0) episodes. This pattern of miserable menstrual, lack of terrific postmenstrual neutral mid cycle, lack of terrific post mid cycle, and neutral premenstrualphases does not match the menstrual stereotypes very well, with the exception ofthe menstrual down.Weekday cycle. There were five high frequency and five low frequencypatterns according to day of the week (see Table T4-6; for comparison below theoverall mean frequency = 1.0; SD = .877; and CI = .123, 1.877). Sunday wasassociated with a high frequency of terrific (freq = 2) and miserable (freq = 2)episodes. Monday was associated with a low frequency of terrific (freq = 0) and high59frequency of miserable episodes (freq = 2). Tuesday was associated with a highfrequency (freq = 2) of miserable episodes. Wednesday was associated with a lowfrequency of terrific and miserable episodes (freq = 0, for both). Thursday wasassociated with a low frequency (freq = 0) of terrific episodes. Friday wasassociated with neither high nor low frequencies of terrific or miserable episodes.Finally, Saturday was associated with a low frequency of terrific (freq = 0) and highfrequency of miserable (freq = 2) episodes. This weekday pattern matches theMonday blues stereotype, but contradicts the weekend high stereotype. Theremaining days were either mixed (e.g., Sunday), miserable (e.g., Tuesday) orrelatively neutral with respect to terrific and miserable episodes.Lunar cycle. Two lunar patterns were observed (see Table 17-9; forcomparison below the overall mean frequency = 2.38; SD = 2.20; and CI = .18,4.58). The first quarter was associated with a high frequency of miserable episodes-(freq = 7) and the full moon was associated with a low frequency of terrific episodes(freq = 0, note this value narrowly exceeds the lower bound of the confidenceinterval). These lunar results do not confirm the stereotype of increased activityduring the full moon.Terrific versus miserable episodes. When overall average frequencies forterrific and miserable episodes were compared across all cycles, a clear patternemerged (see Table T10). Miserable episodes were roughly twice as frequent asterrific episodes in each cycle. Perhaps miserable experiences are more salient inparticipants' minds when they think back over the day and so are more likely to bereported. It also is possible that participants simply had more miserable than terrificcyclic episodes in their lives.In addition, the greater frequency of miserable episodes could be an artifact ofthe participants' positive response bias. They tended to have mean responses at themiddle or above the middle of the nine point scales. Participants with high averages60could technically have the upper bound of their confidence interval extending to thetop of the 9-point scale. The effect of the upper bound "hitting the ceiling" of thescale is to make it impossible for that participant to have above averageexperiences. That is, after the confidence interval is constructed around a highpositive mean the only response that can be nonaverage is one that falls below thelower bound, in a negative or "miserable" direction. Thus, the lack of "terrific"episodes may indicate that positive experiences are the norm for that participant andonly "miserable" episodes are nonaverage.Conclusions about cycle results. The menstrual cycle was associated withpatterns of terrific and miserable episodes that do not strongly match the stereotypesfor the menstrual cycle: miserable menstrual, lack of terrific post menstrual, neutralmid cycle, lack of terrific post mid cycle, and neutral premenstrual phases. It was notsurprising to see the stereotype disconfirmed at this level of analysis since only 6 ofthe 58 participants had individual patterns that strongly matched the stereotype.These results are partially consistent with the prospective reports in the McFarlaneet al. (1988) study which failed to show any menstrual pattern. In the current study,there were also several weekday patterns: a miserable Monday, Tuesday, andSaturday, and a mixed Sunday. The former is consistent with the Monday Bluesstereotype and the latter contradicts the Saturday high stereotypes. The significantresults for Sunday and Tuesday also contradicts the stereotype. Finally, the lunarcycle results revealed two lunar patterns: a miserable first quarter and low frequencyof terrific episodes during the full moon (though this effect was marginal). Neither ofthese patterns matches Western lunar stereotypes. Overall, the terrific andmiserable episodes cycled most in weekday patterns, followed by menstrual andlunar patterns.Group Results for Terrific and Miserable Episodes Menstrual group results. ConPMS women had a high average frequency of61miserable episodes (mean = 1.0) (see Table T3; for comparison the overall averagefrequency = .324; SD = .265; and CI = .059, .589). Note that this is a confirmatoryresult of the group selection process and thus it is a good diagnostic check. Noother groups showed significant terrific or miserable patterns. Overall, when terrificand miserable frequencies were averaged, only conPMS women had a high averagefrequency of atypical (terrific or miserable) episodes (mean = .67), but, again it ispossible that this finding is a confirmatory result of the group selection process.Discussion of the patterns in the group results will be reserved until the weekday andlunar results have been presented.Weekday group results. Some groups were characterized by a high or lowfrequency of terrific and/or miserable episodes (see Table T6; for comparison theoverall average frequency = .22; SD = .123; and CI = -.001, .245). LibPMS womenhad a high average frequency of miserable episodes (mean = .40). Only IibPMSwomen had a relatively high overall group mean when terrific and miserableepisodes were averaged.Lunar group results. Some group differences over the lunar cycle wereobserved (see Table T8; for comparison the overall average frequency = .157; SD =.153; and CI = .004, .31). ConPMS women had a low frequency of terrific and a highfrequency of miserable episodes (means = 0 and .50, respectively). An additionalfinding was that libPMS women had a low frequency of miserable episodes (mean =0, note this value narrowly exceeds the lower bound of the confidence interval).When frequency of terrific and miserable episodes was averaged for each group, nogroup had a relatively high or low overall average.Conclusions about group results. Two procedures were used to try to findoverall group patterns. First, the high and low outlying groups for both terrific andmiserable episodes were summarized across all cycles (see Table T11). The lack ofany significant high or low frequency for terrific or miserable episodes in nodxsay62and nodxNsay women and men who was consistent across all three cycles. Thepattern of a high frequency of miserable episodes in the menstrual and lunar cyclesfor conPMS women is possibly a confirmatory effect of the group selection process.The absence of a high frequency of miserable episodes for conPMS women in theweekday cycle supports this confirmatory hypothesis and argues against thepossibility that conPMS women are more miserable in general. No other pattern isvery salient in this table.Second, the results from the overall group averages across all cycles aresummarized in Table T12. Only two groups in one cycle had a high overallfrequency of episodes. First, ConPMS women had a high frequency of episodes,and this finding is likely a confirmatory effect of the group selection process.Second, IibPMS women had a high frequency of episodes in their weekday cycles.This result only narrowly exceeded the confidence interval, however, and should beconsidered marginally significant. No other groups showed a pattern of either low orhigh frequencies of episodes showing a cyclic pattern.In sum, the confirmatory premenstrual pattern in conPMS women aside, clearpatterns in group frequencies of terrific and miserable episodes were not seen. Themost conservative conclusion is that there were no significant differences amongstthe groups with respect to patterns of terrific and miserable episodes.Proportion of participants with any diagnosed cyclicity. To determine thenumber of individuals for whom cyclical episodes were diagnosed across theirmenstrual, weekday, and lunar cycles, Table T13 was constructed. The data fromthis table were first summarized in Table T14. The mean (.710), standard deviation(.213), and confidence interval (.497, .923) were calculated for the proportion ofparticipants with greater than zero diagnosed cyclical episodes. A high proportion(100% or p = 1.0) of episodes of conPMS women had one or more diagnosed cyclicepisodes, as would be expected from the group selection process. A relatively low63proportion (p = .42) of nodxNsay women had one or more diagnosed cyclicepisodes. This proportion was even lower than that for men (g = .63), the groupleast likely to cycle according to cultural stereotypes. The proportion for IibPMS andnodxsay women were p = .70 and p = .80, respectively.Next, the proportion (averaged across groups) of participants who had noneto four diagnosed cyclical episodes was .288, .322, .290, .090 and .010,respectively. Clearly, the majority of participants, roughly 69%, had diagnosablecyclical patterns. It was not rare to have no diagnosable cyclical pattern (30%), but itcertainly was not the norm.Because of the dilemma of synchronicity between the menstrual and lunarcycles, the proportions of cyclic episodes could have been overestimated. If themenstrual and lunar cycles were highly synchronized it would be possible to havethe same cyclic episodes show up in both the menstrual and lunar summary andthus be counted twice. To correct for this possible overestimate, when an episodewas diagnosed for both the menstrual and lunar cycle only one of these was enteredinto the final tabulation. This corrective procedure is likely to underestimate the trueproportions, however, because not all menstrual and lunar cycles weresynchronized. The true proportions lie somewhere between those shown in Tables14 and 15.With the conservative correction for possible synchronistic cycles, theproportions of participants with none to three diagnosable cyclic episodes were:.288, .410, .258 and .04, respectively. These proportions reveal that half theparticipants, i.e., 41%, had a single diagnosable cyclical episode. Participants wereroughly just as likely to have two as no diagnosable cyclical episodes and a verysmall minority had three diagnosable cyclical episodes. As was the case beforecorrection for synchronized menstrual and lunar cycles, roughly 71% of this samplehad one or more diagnosed episodes.64It is possible that participants who had more cyclicity than normal volunteeredfor this study and are thus overrepresented in this sample. Anecdotally, someparticipants who found that their responses did not vary from day to day were"bored" or sometimes distressed by the lack of variation and dropped from the study.It also is possible that people who were aware of the same invariant characteristicsof their daily lives might not find a daily journal study appealing and thus would notvolunteer. It would be difficult to determine whether 69-71 % is representative ofpeople with cyclical episodes. Even if this number is exaggerated because of anoverrepresentation of cycling individuals in this sample, it is clear that there are moreindividuals who cycle than cultural stereotypes would predict. It was concluded,therefore, that behavioural cyclicity appears to be the norm and not the exception.An illness model of cyclicity, as in the case of menstrual cyclicity and PMS,therefore, seems inappropriate.Non-Parametric vs. Parametric Results Contrary to the parametric group results, the nonparametric analysesrevealed that conPMS women were the only participants who had premenstrualmiserable episodes. This significant premenstrual pattern is confirmatory evidenceof the group selection process. It is interesting that parametric analyses did notreveal this confirmation.The nonparametric analyses revealed several weekday patterns: a miserableMonday, Tuesday, and Saturday and a mixed Sunday whereas parametric analysesonly revealed a Saturday high. In addition, nonparametric analyses revealed"miserable" first quarter and lack of "terrific" full moon patterns whereas parametricanalyses revealed only higher mean positive body experiences in the first quarter.The parametric and nonparametric analyses contradict each other for both the lunarand day of week cycles. Unfortunately, there is no equivalent to the obviousmenstrual artifact to help decide which set of weekday results may be more65accurate.What the individual nonparametric analyses provided was an assessment ofthe nature of a particular terrific or miserable cluster of items which varied fromindividual to individual and, within individuals, from phase to phase and cycle tocycle. That is, the specific items constituting a terrific cluster could, and did, varywithin and amongst individuals. This advantage of the prediction analyses was notpossible in the parametric analyses. Beyond the initial multivariate stage, which didconsider all individuals' items (i.e., dependent variables), there was no way toconduct the analyses on individually varying clusters of three or more items. Weresuch an analysis available maybe the results of group parametric analyses would bedifferent.It is important to note that multivariate ANOVAs are designed to examinegroup and not individual differences. To the extent that members of a group do notbehave as a group in response to the phenomena being studied, analyses such asMANOVAs designed to examine group difference become less appropriate. It ispossible that individuals in this study experienced cycles idiosyncratically and thatany grouping scheme would obscure their individual patterns.The most basic hypotheses that could be made about group differences inthis study are that conPMS women (who were a priori selected individually forpremenstrual "miserable" episodes) would show premenstrual downs in theirprospective data and that men (assigned to pseudo-menstrual cycles that randomlystart on different days and are highly unlikely to coincide in a meaningful way acrossall men's cycles) would not have menstrual cycles in their prospective data. Thenonparametric results were consistent with both of these hypotheses and theparametric results contradicted both of them. It was concluded that thenonparametric analyses yielded results that were more accurate than the parametricanalyses because the latter obscured individual cyclic patterns whereas the formerdid not.General ConclusionsThis investigation was designed to better understand a number of issuesconcerning women's PMS experiences. One purpose was to shed light on the largediscrepancy between the majority of women who say they experience premenstrualproblems and the small minority who meet diagnostic criteria for PMS. Addressingthese issues required an appropriate and conservative procedure for diagnosingPMS. The technique developed for this study, using each woman's own mean andstandard deviation to identify unusual experiences for her, and then using predictionanalyses to determine whether there were any patterns in these "outlyers," wassuccessful. The procedure was easy to learn and the only equipment necessarywas paper, pencil, and calculator. The diagnosis of PMS and the calculationsleading up to it took about 45-60 minutes per participant.Comparison of this new diagnostic technique with the DSM-IIIR and NIMHdiagnostic techniques was beyond the scope of this thesis, but is planned for futureresearch.The procedure developed for diagnosing PMS revealed patterns in all phasesof the menstrual cycle. The diagnostic results revealed that nodxsay women had thegreatest number of "misdiagnosable" patterns. That is, nodxsay women had downsin phases other than the premenstrual phase which, if not carefully noted, could bemistaken for PMS, or they had highs that occurred during another phase of theirmenstrual cycle that might make their premenstrual phase appear to be "down" or"miserable" by contrast. Perhaps some nodxsay women, who were the majority ofwomen in this study and indeed may be like the majority of women seen byprofessionals, have cyclical patterns in phases other than the premenstrual phasethat can be just as conservatively diagnosed as can "strict" PMS. On the otherhand, only 65% of the nodxsay women in this study had a "misdiagnosable" pattern.6667Clearly there is more happening in their lives, leading them to believe that they havePMS, than just being out of phase with popular diagnostic procedures.It is important to underscore the finding that only 6 or 12.5% of the women inthis study were conservatively diagnosed with PMS, but two (one-third) of these 6women did not identify themselves as having PMS, leaving only 4 or 8.3% of thewomen with self-identified and conservatively diagnosed PMS. Future qualitativeanalyses are planned to see how the two conPMS women who did not identifythemselves as having PMS experienced their menstrual mood changes. Did theyhave a vague awareness that something was wrong in their lives that they could notidentify or were they aware of the menstrual cyclicity but did not consider it to be aproblem even though their PMS met diagnostic criteria? Perhaps qualitativeanalyses will show that these two women had unique patterns in their daily lives, butclearly more than two women are needed to confirm these results. Women whocould be diagnosed with PMS but who do not identify their PMS as a problem clearlyhave much to teach us about menstrual moods and women's sense of well-being.Another issue arising from the diagnostic results is that only a smallproportion, i.e., 8.8% of all diagnosable episodes or 13% of all miserable episodes,discovered in this sample occurred during the premenstrual phase. Premenstrualmood cyclicity does not appear to be a unique phenomenon or "syndrome" at all.There were 40 other "miserable" menstrual, weekday, and lunar episodes (7 ofwhich occurred for men) that did not differ diagnostically from the premenstrualepisodes. The vast amount of research and media attention narrowly focused onPMS does a disservice to both women and men who experience nonPMS cyclicity.If the proportions found in this sample correspond to those of the population, it willbe important to acknowledge the reality of cyclicity as well as to recognize that tolabel one specific small type of cyclicity, such as PMS, as a "syndrome" or medicaldisease, cannot serve to promote the understanding of women's (and men's)68experiences or their well-being.Our criteria for PMS were strict in this investigation and hence a smallminority of women, 8.3%, said they had PMS and were diagnosed with PMS. Butwomen are more often exposed to far more liberal definitions of PMS. These vagueor looser definitions may be an important reason for so many women saying theyhave PMS. For example, some researchers and popular media authors include themenstrual phase with the premenstrual phase, and/or 2 weeks instead of 1 weekprior to onset of menstruation. By the time these 2 premenstrual weeks and themenstrual phase are combined most of the cycle is accounted for, so most negativemoods would be "premenstrual". In addition, these vague definitions neitherdistinguish between a single negative item and a cluster of negative items nor dothey consider that positive experiences could disconfirm PMS. One might speculatethat if only vague definitions were used, more women would be "diagnosed" withPMS than not. It is not surprising given the very vague criteria in the media andsome research that so many women say they have PMS.The prospective parametric analyses revealed a few patterns of groupdifferences in menstrual cycle experiences, but the occurrence of menstrual phasefluctuations for men suggests that these differences may be spurious. It wassurprising that these analyses did not show the PMS "down" of conPMS women,which had been expected as a confirmatory effect of the group selection process.Weekday parametric analyses revealed a Saturday high but other results wereinconclusive. Prospective analyses on the lunar cycle revealed more positive meanbody experiences in the first than last quarter.Results of the retrospective analyses are consistent with those of McFarlandet al. (1989) and suggested that participants used cultural stereotypes to help themrecall their cycles. All groups of women, even nodxNsay women, recalled that theyhad experienced a classic menstrual mood pattern, and men as well as women69recalled having experienced weekend highs and Monday blues. These results wereconsistent with those obtained for younger adults by McFarlane et al. (1988). Theseresults are also consistent with schema theory. Even though no individual womanreported the "classic" menstrual pattern in her day-to-day experiences, in recallingher pattern she may have searched for data consistent with this cultural menstrualstereotype. For example, she may have tried to recall a time when she was in agreat mood at mid cycle, and may have ignored data inconsistent with thestereotype, such as times she was in just as good a mood in non-mid-cycle phases(McFarlane et al., 1988). The analyses of retrospective lunar data were limitedbecause fewer than half the participants were willing or able to provide them.Perhaps if Western culture had more specific and/or better established lunarstereotypes participants would have been more willing to provide retrospectivereports oft-heir lunarmoods, assisted by the lunar stereotypes.Might the retrospective reports have been more accurate than the prospectiveones? For example, perhaps a woman was not aware that she was angry or sadwhen she completed her daily report but later on, when thinking back, she realizedshe actually was angry and so reflected this new awareness in her retrospectivereports. Ruble (1977) found that when she was able to convince women they were"premenstrual" based on incorrect information provided by the experimenter, theyreported more symptoms than did women who believed (incorrectly) that they were"intermenstrual". Her results support the contention that rather than providingreports that more closely match daily reports, retrospective reports are likely to beinfluenced by beliefs and stereotypes. Throughout the social science disciplinesretrospective reports are not considered the best reports because they are removedin time from the daily experience and more likely to be influenced by stereotypes andintervening life events. To argue for : a different paradigm for menstrual experienceswould be difficult to justify. Moreover, in this particular study, one would have to70justify a paradigm switch for both day of week and menstrual cycle reports, orexplain why the switch should be specific to the menstrual cycle.It is important to recall that the purpose of this study, and much research andpractice, was to diagnose individual differences in cyclicity. Had retrospectivereports been used then the majority of women, including women who did not believethey have PMS (nodxNsay), would have been diagnosed with PMS. The finding thatall groups of women recalled precisely the same menstrual cycle pattern for all 14variables also suggests that retrospective reports would be useless for diagnosis ofindividual differences.The parametric analyses seem to have been an excellent tool for revealingthe role of participants' (similar) stereotypes in their retrospective data. Theanalyses' failure to reveal the conPMS confirmation of premenstrual downs, coupledwith the finding of menstruaLpattems for men as a group in the prospective data,does suggest, however, that these parametric group analyses were not optimal forthe prospective daily data.When the nonparametric or diagnostic data were averaged for each groupand across cycles, the conPMS confirmation was clearly seen, as expected.Moreover, although two individual men did have a "menstrual" or monthly pattern,men as a group were not characterized by a menstrual pattern, also as expected.As a means of analyzing and understanding patterns of cyclicity, therefore, thenonparametric approach used in this study seems to have greater validity than theparticular parametric approach we used.The nonparametric analyses revealed that the women's prospective data,when averaged together overall, did not show a pattern similar to the classicmenstrual pattern, aside from a menstrual down.The nonparametric prospective analyses revealed a Monday, Tuesday andSaturday "down" as well as a mixed Sunday. These results again suggest that71participants' recollections of Monday blues may be magnifications of the Mondaydowns revealed in daily reports by some participants. The Saturday down wasunexpected and difficult to explain. Anecdotally, when asked about these Saturday"miserable" episodes, several participants said that Saturdays were days when theywere most likely to plan all chores and shopping and also to be poorly rested fromstaying up late on Friday. It was not until Sunday that they reported slowing down torest. Perhaps the mixed Sunday was a combination of reaping the rewards for ahard worked Saturday and some trepidation about starting a new week. It is difficultto explain the down Tuesday in terms of either the weekday stereotype or anecdotalevidence.Finally, the nonparametric analyses revealed first quarter downs and a lack ofterrific episodes in the full moon phase. Unfortunately, it is difficult to interpret themeaning of these findings within the context of Western culture, in which lunarbeliefs outside a "Transylvania" effect tend not to be held.Traditionally, in psychology and other social sciences, when data permitparametric analyses they are considered to be the best choice because they arethought to be more rigorous and to allow some control over Type I error.Nonparametric analyses typically are reserved for categorical data. Interval data, ifavailable, must be reduced to categorical data for nonparametric techniques, losingsome potential information, and there are fewer options (e.g., no multivariatetechniques) for controlling Type I error. The study of cyclical patterns may call for anexception to this convention. The results of this study indicate that cyclical patternsmay be very individual and relatively common, but when individual participants' dataare combined for group analyses the patterns are obscured. There are parametricauto-correlational and time series techniques that can reveal cyclical patterns, butthey require considerable statistical expertise, computer equipment, and software,making them unavailable to most PMS researchers, to physicians, and certainly to72the women who say they have PMS. The diagnostic procedure developed for thisstudy is sensitive to cycles, and could be used by anyone who can use a basicstatistical calculator.If in the PMS field what we want to do is understand individual women'scycles, then the only reason to consider group data would be to understand whatproportion of a group had a specific pattern. But group summarization of the datastill would have the technical problem of possibly obscuring individual patterns.There was tentative post hoc evidence to suggest that conPMS and libPMSwomen may have experienced a higher incidence than other women and men ofemotional and sexual abuse. Unfortunately, analyses could be conducted only onan exploratory basis so no conclusions about possible links between PMS andabuse can be made. Other researchers are independently beginning to mentionsimilar hints in their data, so it is important that more research on possiblerelationships between PMS and abuse be done.To rephrase these results in terms of the hypotheses, some were confirmed,others not. It was expected that more nodxsay women would have "down" phasesthat were not premenstrual than would be true of the other groups. In keeping withthis expectation, nodxsay women had relatively more misdiagnosable patterns thandid the other groups. In addition, it was expected that nodxNsay women would havethe fewest number of diagnosable "downs". Contrary to this hypothesis, men hadthe lowest number of downs, followed by libPMS women. It also was expected thatconPMS would have reported more negative daily experiences overall than wouldthe other groups. Other than the expected premenstrual downs (a confirmatoryeffect of the group selection procedure), conPMS women did not report morenegative daily patterns. It also was hypothesized that conPMS, IibPMS, andnodxsay women would recall more negative premenstrual moods than wouldnodxNsay women, but all groups of women recalled a PMS pattern. Finally, it was73hypothesized that participants' prospective day of the week cycles would reveal aweekend high and possibly a Monday low and that weekday recollections would beconsistent with this pattern. Participants did show the weekend high and Mondaylow in their weekday recollections, but their prospective reports revealed Monday,Tuesday, and Saturday lows and a mixed Sunday in the nonparametric analyses.Overall, the results of this investigation are consistent with several theories,especially some of the recent feminist theories. The theory of PMS as a heightenedsense of reality is consistent with tentative evidence of a link between the conPMSand IibPMS women's negative premenstrual reports of their moods and theirresponses to the questions regarding abuse. This tentative result also suggests thatrecent attempts to link PMS to the after effects of abuse may be important.The results of this study also address the PMS-as-"abnormal" theory. Clearly,with 69-71% of all participants, including men, experiencing diagnosable cyclicity,cyclicity is more normal than abnormal. In addition, "downs" or "miserable" episodeswere diagnosed for every phase of the menstrual cycle as well as for other cycles.The premenstrual phase acounted for only 6 of these downs as opposed to 40 other"miserable" menstrual, weekday, and lunar episodes. PMS, rather than being"abnormal", is just one of several possible and normal ways women can cycle.Some exploratory analyses regarding marital dissatisfaction and PMS wereconducted but a link between the two was not confirmed. It is possible thatmeasures with better reliability and validity data would have shown a link and thatfuture qualitative analyses may reveal some links.The stress-and-PMS theory was tested both in an exploratory fashion as wellas in discriminant analyses with the analyses of the Hassles and Uplifts scales. Thegroups did not vary on these scales. To the extent that physical, sexual, andemotional abuse adds stress to participants' lives, however, exploratory analyses didreveal that conPMS and IibPMS women may have experienced a higher incidence of74abuse. In short, there was no support for a general link between hassles and upliftsand PMS but the specific stress of abuse may be associated with PMS.Testing the theory that PMS was an "out-of-character-for-me" experience wasbeyond the limits of this investigation, but relevant data were collected. Futurequalitative analyses of the Out-of-Character-for-Me Questionnaire will address thisquestion. Finally, the theory of PMS as a Culture Bound Syndrome and as anindicator of role strain was tested by several demographic analyses, as well as thescales measuring anger and sadness in the home and workplace. No groupdifferences were found for any of these variables.The results of this study have implications for a biological model of PMS. Asmall minority of women with conservatively diagnosed PMS was included in thesample, even though they were not specifically recruited and the menstrual purposeof the study was disguised. So too, however, were women (and men) withdiagnosable downs in every other menstrual phase. Moreover, these downs takentogether were more common than the premenstrual downs. It would be very difficultto construct a biological model based on hormonal changes that could explainpotential downs and highs in every phase. In addition, it is not clear how a biologicalmodel would explain the finding that one-third (2 of 6) of the women for whom PMSwas conservatively diagnosed did not themselves believe they had PMS.Unfortunately, no hormonal or other physiological data were collected. To do sowould likely have revealed the menstrual purpose of the study, and would have beenprohibitively expensive. Any discussion of possible relationships between hormonaland behavioural patterns is, therefore, not possible. Nevertheless, the resultssuggest that a biomedical approach that focuses narrowly on a PMS pattern to theexclusion of all other patterns may be misleading.This investigation, although extensive in several ways, was neverthelesslimited in others. First, the group sizes were relatively small (N = 6 to 20) and75although the parametric results showed acceptable power (mean group power =.845; range .53 to 1.00 for all prospective and retrospective analyses; see AppendixN), the nonparametric group and cycle results might have been different if there hadbeen more participants in each group. Thus, the results presented here may not berepresentative of women and men, in general. In addition, this was a predominantlywhite, middle class, heterosexual, and healthy sample. It is possible that peoplefrom other demographic backgrounds would have significantly different cycles.Participants who volunteer for a 4-month journal study in which they agree toprovide introspective data about their daily lives are not a representative sample.Indeed, several participants dropped from the study because they found itsintrospective nature too painful, stressful, or time consuming. There are somecharacteristics of the sample which are nevertheless noteworthy with respect torepresentativeness. First, participants did not know the menstrual interests of theresearchers and thus results were likely not to be biased by the demandcharacteristic of many menstrual studies (e.g., Ruble, 1977; Ruble & Brooks-Gunn,1979). Second, participants were not full-time students and were older than mostuniversity samples (mean age = 34), thus the sample was more representative of thegeneral community. Third, this sample had a good age range as well as a goodrange on marital status, education, and spirituality. Fourth, analyses of those whocompleted, partially completed, or dropped from the study revealed no demographicdifferences except that more participants of colour dropped. It would be preferableto have a sample more characteristic of Canada's multicultural society, but theseparticipants are hard to recruit because of different languages, etc. It will beimportant to get a multicultural sample, however, in future research, especially to testthe PMS-as-a-culture-bound-syndrome hypothesis. Fifth, daily data were collectedfor 4 months, making cycle results more representative of the individual participants'usual patterns than only 2 or 3 months of data would have provided. Sixth, it is76important to note that the goal was not to conduct an epidemiological study but tobetter understand women who say they have PMS. A large proportion of thesample, i.e., 62.5%, were indeed women who said they had PMS.Finally, in the current study, because it was important to keep participantsblind to the menstrual purpose, it was not possible to collect blood or urine samplesor to measure temperature to assess hormone levels and other indicators ofovulation. Thus, while many diagnosable menstrual patterns were observed, noconclusions can be made about how these experiences might map onto thesewomen's hormonal fluctuations. In order to appropriately test for the correlationbetween hormonal changes and diagnosable menstrual episodes a different sampleand methodology would be needed. The results from the current study, therefore,cannot shed light on the issue of hormone-behaviour connections in PMS.What of the future of PMS research? Clearly, more investigation using thediagnostic procedure developed in this study are warranted, to determine if the"terrific" and "miserable" menstrual, weekday and lunar cyclical episodes can bereplicated. Indeed, the McFarlane et al. (1988) data are now being reanalyzed forjust this purpose. If these results are replicated by other researchers with a varietyof samples then PMS researchers, and especially PMS businesses (e.g., clinics,treatments, etc.), are faced with a tough decision.We are not saying that there do not exist women who experience severe PMSwith a biological basis, but that clinical studies in which biological data were collectedand strict criteria used to diagnose PMS would be necessary to establish a biologicalbasis to PMS. Clearly, however, most women who say they have PMS would notmeet these strict criteria. Years of momentum have gathered around PMS, andcareers and businesses have been founded with PMS as the focus. If premenstrualfluctuations are just one of many possible and common patterns then it is notappropriate for researchers and businesses to focus on PMS as a special syndrome77or disease. PMS clinic business professionals would have to be willing to go againstthat business interest. The considerable media pressure and the culturalstereotypes that have developed would make re-education about normal menstrual,weekday, and lunar cycles difficult.It will be important, although time-consuming, to identify more women whohave conservatively diagnosed PMS, but do not say they have PMS. These womenwould be able to teach us what is different in their lives that they do not use the PMSlabel, a label popular in the media and used by the majority of women (62.5% in thesample) to describe themselves, even when they experience and prospectivelyreport a pattern that could accurately be diagnosed as PMS. Do these women havemore or less awareness or acceptance of their menstrual cyclicity? Do they havedifferent levels of stress or abuse in their lives? More research is needed to answerthese questions.The results of this study call into question the proposed inclusion of LateLuteal Phase Dysphoric Disorder (LLPDD) in the Diagnostic and Statistical Manual.To date, including the results of this study, premenstrual miserable episodes or othernegative experiences have not yet been found to be qualitatively different thandowns during nonpremenstrual phases. Until a qualitative difference is found for thepremenstrual down, the existence of a premenstrual down cannot be "diagnostic" ofany disorder. LLPDD or premenstrual syndrome, therefore, does not belong in aDiagnostic Manual because it has not yet been proven to be uniquely or qualitatively"diagnostic" of any disorder.It is important to distinguish amongst the various types of premenstrual andother menstrually-related changes. Whereas some changes may indicate illness,others reflect healthy physical and emotional changes that normally occur over themenstrual cycle. Some researchers refer to these healthy changes as molimina.Healthy changes should not be confused with symptoms, which by definition indicate78an illness or disorder (e.g., dysmennhorea or painful menstruation). The situation isfurther complicated by the fact that some women report changes that would appearto qualify as (illness) symptoms and say that they are troublesome whereas othersreport the same degree and type of changes but do not find them troublesome. Thephrase premenstrual syndrome is restricted to the co-occurrence of a cluster ofsymptoms in the premenstrual phase that do not occur in other phases of themenstrual cycle. In this study, few women met this criterion, and some of those whodid, did not consider themselves to have troublesome premenstrual symptoms.Moreover, a number of women met the "syndrome" criterion of a cluster of"symptoms" in one phase that remitted in other phases, but their symptomatic phasewas not premenstrual (e.g., post mid-cycle).The findings of this study underscore the importance of recognizing thatcyclical changes are not necessarily abnormal or unhealthy. The common beliefsseem to be that women's moods vary considerably over the menstrual cycle, that allwomen's moods do so in the same pattern, and that this is undesirable. Implicitlylinked to these beliefs are others, namely, that men's moods do not vary and thatwomen's moods vary more over the menstrual cycle than other cycles (such as dayof week, lunar, school year, seasonal, etc.). Instead, in this study the researchevidence indicated that most women experienced healthy changes over theirmenstrual cycle, some women experienced some changes as positive, someexperienced other changes as troublesome symptoms, and very few womenexperienced a specific cyclic pattern of changes that meet diagnostic criteria forPMS. Moreover, both women and men experienced mood changes over days of theweek that were either negative or positive or both. Whether or not this turns out tobe a representative sample, these results indicate that the common beliefs regardingcyclical changes are incorrect for at least some, and perhaps most, adult women andmen.79In conclusion, the best service PMS researchers can provide for women whosay they have PMS is to place it in context. This includes broadening the scope oftheir work to encompass more cycles for comparison, and using diagnosticprocedures that allow for the possibility of positive and negative cycles in allmenstrual phases, days of the week, phases of the moon, and other cycles (e.g.,related to paid and domestic work, seasons, holidays, etc.). Perhaps a portion ofwomen who say they have PMS can be diagnosed with PMS, perhaps anotherportion have other cyclic diagnosable nonpremenstrual patterns that are mistaken forPMS, and perhaps still another portion of women who say they have PMS areactually finding that the label PMS is a good way to cope with and talk about maritaldissatisfaction, the after effects of physical, sexual, or emotional abuse in theirchildhood or adulthood, etc. Perhaps some women are confused because they arenot aware of the distinction between menstrual cycUcity as a healthy sign and as anillness symptom or syndrome. Doubtless, there are still more possibilities andcombinations to account for other women who say they have PMS, but do not meetdiagnostic criteria. Much work investigating these possibilities remains to be done.References(see Appendix A)80Appendix AA Literature review of Premenstrual Syndrome (PMS)81We gratefully acknowledge Leanne Hammond's and Sharon Eng's help with proof-reading and Elizabeth McCririck's help in word processing.pms3.doc/emPremenstrual Syndrome82Overview of Problems in Previous Research Prior to the 1970s researchers found support for what has come to be known asthe classic menstrual mood pattern (Altman, Knowles, & Bull, 1941; Benedek &Rubenstein, 1939a, 1939b; Golub, 1976; Ivey & Bardwick, 1968; Janowsky, Berens, &Davis, 1973; Luschen & Pierce, 1972; Moos, Kopell, Melges, Yalom, Lunde, Clayton, &Hamburg, 1969; Patkai, Johnannson, & Post, 1974; Rossi & Rossi, 1977). This patternconsists of pleasant affect in mid-cycle (when ovarian hormones typically are high),followed by negative affect in the premenstrual (when ovarian hormone concentrationsare dropping) and menstrual (when hormones are at their lowest) phases. Claims thatmost women (up to 90%) experience PMS are common in the popular literature. Bycontrast, in the scientific literature researchers increasingly report no evidence of moodfluctuations over the menstrual cycle (Ainscough, 1990; Golub & Harrington, 1981;Lahmeyer, Miller, & DeLeon-Jones, 1982; Little & Zahn, 1974; O'Neil, Lancee, &Freeman, 1984; Sommer, 1973; Swandby, 1981; Wilcoxon, Schrader, & Sherif, 1976;Zimmerman & Parlee, 1973). Several methodological problems in most early and somerecent studies have been reviewed (e.g., Parlee, 1973, 1974; Ruble, 1977; Ruble &Brooks-Gunn, 1979).The first problem is that when women know they are in a study of the menstrualcycle they tend to report mood variations, but menstrually related changes are seldomreported when the purpose is concealed (Parlee, 1974). Participants' expectationsapparently influence their reports. Ruble (1977) found that women who believed theywere "premenstrual" (on the basis of incorrect information provided by the experimenter)reported more symptoms than women who believed they were "intermenstrual."Feldman (1989; DA) 1 conducted a study in which women were either "aware" or"blind" to the menstrual purpose of the study and believed or did not believe they hadPMS. Feldman found that awareness of purpose more than belief in PMS wasassociated with significant symptom severity.Premenstrual Syndrome83The second set of problems concerns mood checklists, most of which consistmainly or entirely of negative moods or "symptoms", e.g., Moos Menstrual DistressQuestionnaire or MDQ (Moos, 1968). People may respond more negatively than theywould if the measures included positive options. Moreover, using only sociallyundesirable moods makes it impossible to detect fluctuations in positive moods (seeRossi & Rossi, 1977). There is some evidence that all premenstrual changes are notnegative. Logue and Moos (1988) reported that positive changes are quite prevalent.Brush (1938) found that 30% of 100 college women reported well-being and bursts ofenergy. Moos (1968) found that 20% of 800 women reported moderate ratings on anArousal factor which included affection, orderliness, excitement, feelings of well-being,and bursts of energy or activity. Moss and Leiderman (1978) reported that 8% ofwomen reported increased arousal and virtually no negative experiences. Tocomplicate matters further, some women who report premenstrual depression andanxiety also report premenstrual energy and increased sexual arousal. Halbreich,Endicott, Schachts, and Nee (1982) stated that 5-15% of the women in their studyreported increased energy and well-being. Koeske (1980) reported that thepremenstrual phase may be a time of enhanced sensitivity to strong stimuli and womenmay show much variability in their responses to this heightened sensitivity. Morris andUdry (1972) found increased activity in the premenstrual phase for some women.Endicott and Halbreich (1982) found that some women reported decreased energy butincreased activity in the premenstrual phase. Lazarov (1982) found that premenstrualwomen performed better than women in other menstrual phases on a facial affectrecognition task. Finally, Chaturvedi and Chandra (1990) found that 46 of 48 womennursing students restrospectively reported they had positive experiences during thepremenstrual phase. These included: feeling of relief (45% of women), craving specialfoods (42%), increased work performance (35%), feeling of well-being (31%), feeling ofexcitement (31 %), increased sexual desire (29%), being affectionate (29%), feelingPremenstrual Syndrome84happiness and joy (27%), and increased energy and vigour (17%).A third methodological problem with PMS research is the growing evidence thatretrospective reports differ from concurrent daily reports obtained in prospective studies(e.g., Ascher-Svanum, 1984; AuBuchon & Calhoun, 1985; Englander-Golden, Chang,Whitmore, & Dienstbier, 1980; May, 1976; McCance, Luff, & Widdowson, 1937;McFarlane, Martin, & Williams, 1988; Olasov & Jackson, 1987; Parlee, 1974; Ruble,1977; Ruble & Brooks-Gunn, 1979; Slade, 1984). Several of these researchers contendthat retrospective reports are more likely to reflect participants' attitudes or stereotypesthan their recall of actual experiences. McFarland et al. (1989) found that the morewomen believed that menstruation was distressing the greater they biased theirrecollections of their menstrual experiences in the direction of their negative menstrualtheory. This, and other perplexing findings, for example, the ineffectiveness ofhormonal treatments, have led some researchers to propose psychological and/orsociological models (discussed later as alternatives to the medical model) to explainsome women's belief that they have PMS.A fourth research problem is failure to examine mood fluctuations in context.Relatively little scientific evidence is available about fluctuations in the emotional andphysical feelings of healthy people, so consideration of comparison groups is important.In McFarlane et al. (1988), normally cycling women were compared with women takingoral contraceptives and with men. Prospectively, there were very few differencesamong the groups, and none related to the premenstrual phase. Menstrual cyclefluctuations also were compared with day of week and lunar cycle fluctuations as asecond means of considering context. Larger prospective fluctuations were found forday of week than for the menstrual or lunar cycles.A fifth problem is that many researchers treat the menstrual cycle as only anindependent variable that can influence women's moods and/or behaviour. Thesuccess of using PMS as a legal defense relies on this one-way interpretation ofPremenstrual Syndrome85influence. Data indicating that the menstrual cycle can be influenced by psychosocialfactors suggest that it should also be conceptualized as a potential dependent variable.For example, the menstrual cycle length can be influenced by stress (Osofsky & Fisher,1967; Russell, 1972; Parlee, 1976; Koeske, 1980). The effect of stress on themenstrual cycle provides an important reinterpretation of Dalton's (1964) claim that themenstrual cycle, particularly the premenstrual and menstrual phases, plays an importantrole in the timing of accidents, suicides, admissions to psychiatric hospitals, and crimesby women.McClintock (1971) found that the menstrual cycles of women living in dormitoriessynchronized over time. One interpretation of this finding is that the menstrual cycle canbe influenced by the psychosocial environment. In a replication of the McClintock(1971) investigation, Jarett (1984) studied 86 pairs of roommates at a women's collegeand found that these women were more likely to have synchronized cycles if: they usedsanitary napkins (rather than tampons); had a longer menstrual flow; obtained a lowscore on an affiliation personality scale and a high score on a social recognition (desireto be held in high esteem by others) scale; and did not mention stress in guessing thepurpose of the study. Overall, roommates showed only a trend toward synchrony. Theanalyses of the data were problematic, but despite this shortcoming, there was modestsupport for the hypothesis that the menstrual cycle can be influenced by psychosocial,as well as by biological, factors.Koeske (1977) reanalyzed data obtained prospectively by Wilcoxon, Schrader, andSherif (1976) to test their assumption that the menstrual cycle is an independentvariable. Wilcoxon et al. (1976) studied young women on the pill, women not on the pill,and men for 35 days. They measured pleasant events, stressful events, mood, andgave the MDQ. Women showed premenstrual and menstrual increases in negativeaffect, impaired concentration, stressful events, pain, and water retention. Stressfulevents accounted for more of the variance in the mood and concentration measuresPremenstrual Syndrome86than did the menstrual cycle, but the menstrual cycle "explained" more of the variance inpain and water retention. When Koeske (1977) ascertained the calendar dates ofpremenstrual days in the study, she discovered that, for women not on oralcontraceptives, the premenstrual days clustered around the midterm and final examtimes, when stress would likely peak. The stress pattern for women on the pill and menwas different. Having established a plausible link between stressful events and themenstrual cycle, Koeske (1977) contended that using analysis of covariance to seewhether events or cycle phases have more influence or better "explain" the variance indifferent measures may be misleading because of confounds or interactions betweenevents and cycles. She argued that Wilcoxon et al.'s (1976) interpretation of theirresults should be restated that, "under conditions of moderate to high stress and eventvariability there is little independent influence of cycle phase on moods and symptomsother than pain and water retention" (Koeske, 1977). Nothing about the relativecontribution of cycle phase and events could be inferred, however, because thestressful events were confounded with the premenstrual phase. To determine therelative contributions of events and menstrual cycle the timing of study would have to bevaried to include different patterns of positive and negative events so that thedistribution of events was not completely confounded with cycle phases (Koeske, 1980).In sum, investigations of the psychosocial factors that may influence the menstrualcycle are relatively rare compared with those in which the menstrual cycle is treated asan independent variable that affects women's physical and psychosocial well-being.The danger in the predominance of research on the effects of the menstrual cycle onwomen is that it may be assumed that this direction of influence is the only one possible.When interpreting their data, researchers may forget to ask how the events thatoccurred during their study could have affected women's menstrual cycles. Sensitivityto the two-way direction of influence between the menstrual cycle and the events inwomen's lives is important.Premenstrual Syndrome87The methodological problems discussed above apply to studies conducted fromseveral theoretical perspectives. These methodological problems also are present inmany diagnostic procedures and treatment outcome studies. In the next sections of thispaper the various theoretical perspectives will be discussed. Then preliminary work byMcFarlane, Martin, and Williams (1988) designed to address many of themethodological problems will be presented. Finally, diagnostic procedures andtreatments for PMS will be reviewed.Biological Theories of PMS There are several excellent reviews of the various theories of pathophysiologies ofwomen with PMS (e.g., Halbreich & Endicott, 1985; Halbreich, Alt, & Paul, 1988; Parry& Rausch, 1988; Reid, 1985; Reid & Yen, 1981; Rubinow & Roy-Byrne, 1984).Biological theories about the causes of PMS include: progesterone deficiency; excessestrogen relative to progesterone; excess aldosterone; a possible role of vasopressin;excess prolactin; ingestion of sugary and salty food; deficiency of vitamin B6; changesin insulin metabolism leading to premenstrual hypoglycemia; allergies to endogenousprogesterone; endogenous opiate withdrawal; excess prostaglandins; possiblemediating roles of serotonin, dopamine, norepinephrine, acetylcholine, beta-endorphinand 5-hydroxytryptamine; and elevated cortisol levels. Many of these theories aretested by administration of the deficient substance or of an antagonist of the excesssubstance to women who say they have PMS. Evidence of improvement aftertreatment is viewed as evidence of the underlying cause of PMS. Bignami's (1982)criticism of this form of inference is presented below in the section on treatments ofPMS. Detailed discussion of each theory will not be presented here. All reviewers ofthe various biological theories have concluded, however, that to date, clear evidencesupporting any of these theories has not been found.Some biological theorists have recently proposed models in which women'sbiological constitution interacts with their environment. For example, Halbreich, Alt, andPremenstrual Syndrome88Paul (1988) propose a theory of PMS in which hormonal homeostasis (the rate and typeof change of the endogenous hormones co-occurring in harmony) is upset, leading toincreased "vulnerability" to negative events. Another example of an interactionist modelis Rubinow and Schmidt's (1989) suggestion that "the menstrual cycle physiology... maybiologically facilitate or choreograph state changes rather than produce specificsymptoms. That is, the menstrual cycle may mark a transition from one experientialstate to another and thus women's perceptions more so than behaviours changethroughout the cycle." These models of biology-environment interaction have yet to beempirically tested.Attribution Theories of PMS Some researchers (e.g., Bains & Slade, 1988; Koeske, 1975; Koeske, 1977;Koeske & Koeske, 1975; Parlee, 1982; Rodin, 1976) contend that situational factors andsubsequent attributions, rather than physiology, determine how moods are experiencedand labelled. For example, Koeske and Koeske (1975) presented vignettes whichvaried according to menstrual cycle phase (i.e., pre vs. intermenstrual), environment(pleasant or unpleasant), and mood (positive vs. negative). Forty-nine men and 53women were asked to identify the cause of the mood experienced by the character inthe vignette. Negative moods (depression, irritability) were linked to the premenstruum.In addition, participants made more internal attributions (e.g., personality) when moodwas inappropriate and made more situational attributions when mood was appropriateto the situation.In a second study, Koeske (1975) found that biology was used to explain negativebut not positive emotions. Specifically, biology was used to explain only negativepremenstrual moods. Evidence was found to support the hypothesis that hostility anddepression were seen as out-of-role for women and that it was assumed that abiological imbalance in the woman must be present. When negative behaviour wasattributed to the premenstrual phase, the social factors simultaneously influencing itPremenstrual Syndrome89were discounted in importance and behaviour was rated as more extreme.Bains and Slade (1988) conducted an investigation modelled after Koeske andKoeske (1975) in which they asked nonstudent participants to make causal attributionsfor the moods of characters in vignettes. Each participant viewed four vignettes whichvaried according to whether the characters were premenstrual or intermenstrual andalso whether they had positive or negative moods. Participants could choose from fivecategories of causal explanations: the characters' background, health, workresponsibilities, events in the day, and personality. Women attributed negative moodsoccurring premenstrually to health factors but negative moods occurringintermenstrually to work or personality factors. Positive moods were explained bybackground, events of the day, and personality factors regardless of phase of themenstrual cycle.Koeske (1977) modelled her predictions on Schacter and Singer's (1962) arousaltheory of emotion. They suggested that emotionality results when an aroused personinteracts with emotion labels provided by the situation. The existence of the label linkingpremenstruum with negative mood might lead to discounting situational factors. Koeske(1977) proposed that women may experience an increase in arousal in the premenstrualphase but that situational cues present then will determine whether the womanexperiences positive or negative moods. Koeske (1977) measured variousphysiological and psychological variables in women who did not know the researchersinterest in the menstrual cycle and did find some evidence to support a general increasein arousal in the premenstrual phase. There also were wide individual differences in thisincreased arousal tendency.Bains and Slade (1988) replicated Koeske and Koeske's (1975) findings. Bainsand Slade (1988) studied nonstudents of various ages as well as students and foundthe same results as in the original study, which was conducted only with students.Women attributed negative emotions occurring premenstrually to health factors but thePremenstrual Syndrome90same emotions occurring intermenstrually to work or personality factors. Positiveemotions were attributed to background variables, events of the day, and personality.The implications for PMS are that women will be unlikely to notice disconfirming data forPMS, since positive emotions occurring during the premenstrual phase and negativeemotions occurring intermenstrually will be attributed to causes unrelated to themenstrual cycle.In another study by Koeske (1980a) college women and men were asked to reporttheir mood, recent positive and negative events, and their menstrual history (imbeddedin a larger questionnaire) at low, moderate, or high stress times of the year, asdetermined by proximity to final exams. Positive premenstrual moods were morestrongly related to the current situation than were moods at other phases. That is,positive moods which occurred in all but the premenstrual phases were attributed to theself whereas positive moods experienced during the premenstrual phase were attributedto the current situation rather than the self.Attributing negative experiences to the self rather than to the situation can havemixed results. Rodin (1976) reported that women who complained of premenstrualsymptomatology were less likely to commit suicide than were women who did not havethese complaints. In contrast to other women, women who had premenstrualcomplaints did not ascribe "increases in depression and irritability to sources such aspersonal instability, hostility from others, or overpowering situational demands" (Rodin,1976). Rodin (1976) found that women performed better on several tasks if they couldreattribute their anxiety to premenstrual symptoms rather than to situational factors.Some active assertive women apparently disregarded cyclical changes and were morepersistent when performing in difficult circumstances. Thus, rather than showingdiminished performance, women who knew they were in their premenstrual phase hadimproved performance. In contrast to Rodin's (1976) findirigs, Valins and Nisbett (1971)have argued that self-esteem suffers if behaviour patterns are internally attributed andPremerictrual Syndrome91social factors ignored. Women feel guilty and, as a result, less likely to make changes.Parlee (1982) suggested further implications of women's menstrual attributionsafter studying 7 women for 90 days. They reported increased activation and vigour anddecreased fatigue, deactivation, and confusion in the premenstrual phase bycomparison with their periovulatory phases. She suggested that maybe women who arenot satisfied with their lives experience nonspecific bodily arousal associated with thepremenstrual phase and react to this physical experience with depression, whereaswomen who are content with their lives label their experience as increased vigour.PMS as a Heightened Sense of Reality"Suppose that society is a lie, and the period is a moment of truth which willnot sustain lies." (Shuttle & Redgrove, 1978; p. 58)"If we are to respect ourselves as women, we have to own all our states of^-being as parts of ourselves, even, and perhaps especially, the painful ones.If we are angry or sad before our periods, there is anger or sadness in us,and there are reasons for it. The menstrual cycle does not imposeextraneous problems on a woman--it is part of her." (Laws, 1985; p. 57-58)Another alternative to the usual description and interpretation of PMS is that awoman's perceptions may be more accurate then, albeit sometimes unpleasant to thosearound her. Hamilton, Parry, Alagna, Blumenthal, and Herz (1984) contend that thequestion should be how and why these women keep the lid on their anger for 3 out of 4weeks rather than why they express their anger premenstrually. Because women havebeen socialized not to express anger they may "split-off" this affect, conceptualizing it asoccurring at a time when they are "not themselves", or may even see the PMSbehaviour as the actions of an alternate self. Hamilton et al. (1984) discuss thehypothesis of a colleague that women>"misperceive" coworkers' statements as personalattacks in the premenstrual phase when at other times they "intellectually know thesePremenstrual Syndrome92comments are innocuous". Hamilton et al. (1984) contend that this hypothesisoverlooks the possibility that some women's perceptions are more accurate in thepremenstrual phase despite their tendency during the rest of their cycle to use denial orintellectualization.Perhaps heightened sensitivity and more accurate perceptions in the premenstrualphase motivate a woman to take issue with her family or friends over matters that on thesurface appear minor or trivial but actually are the "tip of the iceberg" she finally has theenergy to confront.Slade (1989) worked with one woman who had prospectively diagnosed PMS andhad "difficulty in decision-making premenstrually exemplified by her problem in decidingon meals for the family when she was faced with conflict between financial andnutritional requirements." Clearly, any mother faced with the possibility of loweringnutritional standards for her family because of limited financial resources has a right tobe angry about her circumstances. This woman's experience that the nutrition vs.money dilemma became more frustrating premenstrually may have reflected heightenedawareness premenstrually to the injustice of her poverty.Van den Akker and Steptoe (1989) showed women with and without PMS anemotionally provocative film (edited scenes from "The Day After", a film about theeffects of nuclear holocaust), and had them complete mental arithmetic and video gametasks both pre and post menstrually. The only significant finding was that whether ornot a woman had PMS, greater upset was experienced after seeing the filmpremenstrually than postmenstrually. This finding is consistent with the hypothesis thatduring the premenstrual phase some women may be more open to experience, withfewer defenses to external stimuli, and thus be more upset by upsetting material thanthey would be in the postmenstrual phase, when they are less open to experience andmore defended. 'Van der Molen, Merckelbach, and Van den Hout (1988) found that healthy womenPremenstrual Syndrome93in the premenstrual phase were quicker to acquire a "fear" (as measured by skinconductance) response to electric shock and took longer to extinguish the responsethan did healthy women not in the premenstrual phase. The authors interpreted theseresults as evidence that women's physiology creates "heightened susceptibility tophobic fears that return each month, (and) may be a reason for the high incidence offemale phobic responses." They did not, however, collect psychological data on thewomen's subjective reports of fear or phobia. An alternative explanation for their resultsis that women in their premenstrual phase have heightened awareness of and lesstolerance for continuation of obnoxious physical stimuli.Some theorists who regard changes that occur premenstrually as a heightenedawareness of reality contend that before her period a woman may become aware of themany sources of her oppression and that is why she becomes angry, tearful, and/ordepressed (Shuttle & Redgrove, 1978). The premenstrual experience may be that allthe bad feelings of the month are re-experienced and felt more intensely in thepremenstrual phase (Laws, 1985). Some clinicians, believing that women in theirpremenstrual phase are closer to their true feelings, encourage women to accept thosefeelings and use the enhanced energy at this time to make change (Laws, 1985;Vergare, 1987).Premenstrual Behaviour as "Abnormal" Goodman (1986) challenged the view that women's behaviour in theirpremenstrual phase is abnormal, contending that women's behaviour in thepremenstrual phase would not be considered abnormal if it were compared with men'sbehaviour, since men commit 90% of the crimes and have waged nearly all the wars.Alagna and Hamilton (1986) compared the reports of women who were menstruating,intermenstrual, or premenstrual with the reports of men. Participants did not know theresearchers' interest in the menstrual cycle. Women who were premenstrual ratedthemselves as feeling more dominant, energetic, indifferent, negative, and somewhatPremenstrual Syndrome94more tense than did women who were menstruating or intermenstrual. Aside frommarginal feelings of tension, reports of women in the premenstrual phase did not differfrom the reports of men, indicating that these women in their premenstrual phase werenot abnormal by comparison with men.Lever (1980) stated that PMS is a cause of criminal offences by women, babybattering and accidents to children, attacks on husbands, suicide attempts, heavydrinking, accidents in the home and on the road, broken marriages--even wife battering."His wife provokes him by her own violent behavior when suffering from PM(S)." (p.69).A woman's anger is regarded as a problem when its expression disrupts familyharmony or affects family members. Martin (1987) criticized Lever (1981) for sayingthat "[the woman's] own violent feelings and actions while suffering from PM(S) couldsupply the spark that causes him to blow up." Lever (1981) described a woman whowas screaming and pounding on her husband's chest and then ripped the collar andbuttons off his shirt. The woman was then knocked against the wall by the husband, hitacross the face, suffered two cracked teeth, a dislocated jaw, and was bleeding fromhaving bitten her tongue. The woman required emergency medical care. Lever (1981)concluded her description of this incident by saying the husband moved out because"he was afraid he might hit her again because she was so uncontrollable when she wasin a rage" (emphasis added by Martin, 1987). Martin (1987) pointed out that althoughthe woman's behaviour cannot be condoned, her fists hitting her husband's chest andthe ripping of his clothes did not produce bodily injuries requiring emergency treatment.The woman chose a more symbolic means of expressing her anger (she could havekicked him in the groin or stabbed him if she wanted to injure him) but he did not choosea symbolic, noninjurious means of expression. Lever (1981) remarked that the womanwas uncontrollable but did not question the uncontrollability of the husband (Martin,1987).Premenstrual Syndrome95Only men's anger is sanctioned by society. Bardwick (1979) claimed that "womenare not entitled to anger. Anger, except in some girlish tantrum, is unfeminine. Direct,bold, eyeball-to-eyeball, confronting, dominating, resisting, insisting anger has beentraditionally forbidden to women" (p. 48). Women who break society's sanction againstanger pay heavy costs in terms of their own psychological well-being.Miller (1983) contends that women are afraid of disruptions to their relationships atthe first stirrings of their own anger. They have learned that even a small amount ofanger is dangerous so they do not express it. Miller (1983) described a spiral of anger.Because of social pressure not to express anger, women experience repeatedinstances of supressed anger. This leads to increased frustration and inaction, which inturn leads to feelings of weakness and lowered self-esteem, and then to an increase inher sense of unworthiness. Finally, the woman's unworthy feelings make her even moreangry and the spiral of anger begins all over again. The spiral makes anger seemmagnified so women may feel out of control and "hysterical." Miller further explains thatbecause of the social prohibitions against women's expression of anger, it is usuallyonly expressed through symptoms such as depression. It is a damaging social myththat the person who is the always-present caretaker of relationships cannot be theperson who expresses anger. This myth places an unfair burden on women who do themajority of the caretaking. Because of this social myth, women are put in the position ofprotecting their identities as nurturing and squelching their anger. Men, on the otherhand, can express their anger and not be expected to simultaneously caretake theirrelationships with others. In addition, since there are fewer social pressures againstmen's anger, the feelings of unworthiness associated with having angry feelings arereduced. Miller (1983) concludes that social sanctions aside, women's "anger providesa powerful (and useful) recognition of discomfort and motivation for action to bring abouta change in immediate conditions."Social sanctions against anger are not all that women must overcome. TaylorPremenstrual Syndrome96(1988) contends there is a double standard. Men's moods change but they do not needto explain their temper tantrums. Male violence is accepted as part of men's nature.Indeed, the traditional role of a wife was to placate men's moods. But when women'smoods change or they express their anger, this is labelled PMS and the source of theirfrustration is ignored. Women are then given medication to soothe them and makethem less disruptive (Laws, Hey, & Eagen, 1985).PMS as a Construct for Social Control of Women Strong proponents of the PMS-as-disease theories have been criticized for usingthe PMS label as a way to exert social control over women (Eagan, 1983; Lander, 1988;Laws, 1985; Martin, 1987; Rome, 1983). The following quotes illustrate how a woman'sfailure to maintain domestic harmony is interpreted as a symptom of the disease PMS.Proponents of the disease model of PMS look for confirmatory data in the testimonies ofhusbands and coworkers but not in the word of the women themselves. Treating thewoman's disease is seen as a way to restore harmonious domestic relationships. Thefirst nine statements that follow were made by Katherine Dalton, a leading proponent ofprogesterone therapy for PMS, and an expert witness in the legal defenses of somewomen in England who have alleged that their violent behaviour was caused by PMS."...sudden mood changes, emotional behaviour and bursting into tears forno apparent reason are bewildering (to a husband), while suddenaggression and violence are deeply disturbing when, with little warning andno justification, his darling little lovebird suddenly becomes an angry,argumentative, shouting, abusive bitch" (Dalton, 1978, pp. 79-80)."Once-a-month, with monotonous regularity, chaos is inflicted on Americanhomes" (Dalton, 1979, p. xi)."Her irritability and tiredness were hindering her husband" (Dalton, 1978, p.84).Premenstrual Syndrome97"How many wives batter their husbands during their paramenstruum isunknown, nor do we know how often the husband is provoked beyondendurance and batters her" (Dalton, 1978, p. 84)."...the mother's tears caused the fraught husband to beat his daughter"(Dalton, 1978, p. 86)."I think women have a duty if they know they are going to break somethingand going to be irritable to be treated and to look after themselves. Theyowe it to themselves and women in general. Otherwise they will get whatthey deserve from men" (Dalton interviewed by Sarah Crompton, CoventryEvening Telegraph, 15 July 1982, cited in Laws, 1985, p. 31)."Many a husband has commented after the first course of injections that hiswife is now more like the woman he knew at their marriage" (Dalton, 1969,p. 73)."Then suddenly her irritability ends. She is once more her usual sweettempered and placid self, or she may be filled with guilt and remorse at theproblems her actions have caused" (Dalton, 1969, p. 62)."Among PMS women, increased libido is occasionally noticed in thepremenstruum, a fact recorded by Israel back in 1938. All too often it is thisnymphomaniac urge in adolescents which is responsible for young girlsrunning away from home, or custody, only to be found wandering in thepark or following the boys. These girls can be helped, and their criminalcareer abruptly ended with hormone therapy" (Dalton, 1982a; cited in Laws,1985, p. 22).Premenstrual Syndrome98Dalton is not the only physician to link women's expressions of protest, familycasualties and the need for treatment to restore happy family relations."It is estimated that from 25 to 100% of women suffer from some form ofpre-menstrual emotional disturbance... Eicher makes the discerning pointthat the few women who do not admit to premenstrual tension are basicallyunaware of it, but one needs only talk to their husbands, or co-workers, toconfirm its existence (O'Connor, Shelly, & Stern, 1973, p. 7; cited inWeideger, 1975, p. 47).Reid (1985) stated"Episodes of depression or crying, irritability leading to emotional outbursts,and agitated or irrational behavior occurring without forewarning and withapparently little reason create a sense of frustration and tension in even themost understanding of husbands. Frequent harsh reprimands of thechildren over matters of minor importance at times produce bewildermentand fear in the offspring, who are quick to perceive changes in theirmother's temperament. The concerned father who comes to a child'sdefense may only aggravate the situation by inciting further recriminations.Friends or relatives who fall victim to an acrimonious tirade may beunderstandably reluctant to resume their relationship with the same degreeof enthusiasm in the future. Afflicted women... are quick to perceive the tollthat their behavior is taking in terms of alienation of loved ones." (pp. 16-17).Reid (1985) suggests that not just family but society as a whole is affected byPMS."Not infrequently such (PMS) women will offer vivid acounts of the mannerin which their symptoms alter their interactions with others or disrupt theirPremenstrual Syndrome99ability to work, lending at least some credibility to published reports aboutthe adverse effects of PMS on society." (p. 18)He provides some anecdotes from his practice to substantiate this statement. Healso says that more research is needed, but does not make clear that the publishedreports about the negative effects of PMS on society (e.g., diminished workperformance on the job; Langley, 1988) are weak in methodology and are not credible inand of themselves (e.g., Laws, 1985; Martin, 1987).Ronald Taylor, professor of obstetrics and gynaecology at St. Thomas's Hospital,London writes "I have known of no other condition in which so many husbands take timeoff to accompany their wives to the first clinic visit (Women's Health Care undated, citedin Laws, 1985, p. 36). The following quote by Reid (1985) suggests this tendency ofhusbands accompanying wives to PMS clinic visits may be common. "(T)he overridingconcerns about deterioration of family life expressed by many couples attending PMSclincs attest to the importance of this problem" (p. 17, emphasis added). It is suspiciousnot that husbands accompany their wives but that they may do so in greater numbersfor PMS clinic visits than visits for other reasons. If couples complain to a clinician ofmarital or family discord, triggered by PMS or anything else, treating or giving therapy toboth partners would seem appropriate. When a husband is interviewed, however, hetypically is asked about his wife's behaviour problem, and his contributing role is ignoredor minimized.Harrison (1984) described one husband who said his wife was"fine for two weeks out of the month. She's friendly and a good wife. Thehouse is clean. Then she ovulates and suddenly she's not happy about herlife. She wants a job. She wants to go back to school. Then her periodcomes and she is all right again."This husband wanted Harrison to medicate his wife so that she would be a "good wife"throughout her cycle (p. 50). The connection between marital dissatisfaction and PMSPremenstrual Syndrome100will be further explored in the next section of this review.Dalton (1982b) said that when using PMS as a legal defense, diagnosis of PMSshould be made by the physician with substantiating evidence from the woman'semployers, friends or family, or by searching through diaries, medical files, policerecords, or prison documents. Unfortunately, all of these methods are retrospective.Dalton (1982b) believes that a true PMS sufferer can be identified by herwillingness to enter treatment and by an improvement in her condition with treatment.According to Dalton, women who say they have PMS but are not willing to haveprogesterone therapy, are not "true" PMS sufferers. Finding that a woman has improvedafter treatment with progesterone is not, however, sufficient to prove the woman hadPMS. Given the high placebo response rate reported in the treatment of PMS(discussed later under treatments), Dalton's reliance on improvement as proof of PMS ismisdirected (Bignami, 1982).Martin (1987) observed that when Frank (1931) first published his paper describinghow women were affected by their ovarian hormones he was also interested in thedeleterious effects PMS would have on women as employees. Frank (1931) describeda premenstrual feeling of "undescribable tension," irritability, and "a desire to find reliefby foolish and ill-considered actions." Frank (1931) discussed the necessary provisionsand leaves employers would need to make for women. Martin (1987) noted that Frank's(1931) and others' (e.g., Billings, 1934; Brush, 1938; McCance, Luff, & Widdowson,1937; Seward, 1934) concern about women's diminished capabilities as workerscoincided with the Depression after World War I and the pressure on women to give uptheir jobs to men.Martin (1987) pointed out that at the start of World War II, when women wereneeded to rejoin the paid workforce, studies were published demonstrating thatwomen's ability to work was not adversely affected by their menstrual cycles (e.g.,Altmann, Knowles, & Bull, 1941; Anderson, 1941; Brinton, 1943; Novak, 1941; Percival,Premenstrual Syndrome1011943). After World War II, when jobs held by women were needed by men, researchersfound that women were disabled by their hormones (e.g., Dalton began her work onPMS in the 1940s and published in Dalton & Greene, 1953). The strongest interest inPMS by researchers coincided with the second wave of the women's movement aswomen have re-entered the paid workforce in large numbers (Laws, 1985; Martin,1987).A recent example of concern over women's productivity is Langley's (1988)position that, "(a)s greater numbers of women continue to increase in economic value totheir work organizations, PMS can be expected to increase as a hidden, productivity-sapping issue. As women continue to increase in economic value to their workorganizations, the economic costs of this syndrome also will continue to rise."In summary, the clinicians and researchers who view PMS as a disease havebeen criticized by others who consider this perspective to be a mechanism of socialcontrol. "Calling these times of rage, symptoms of disease, is a handy way of notlooking at what women are upset about and why" (Rome, 1983, p. 1).PMS and Marital Dissatisfaction Cohen (1987; DA) tested the hypothesis that premenstrual distress might be asymptom of marital dissatisfaction. Multiple regression analyses of the data from 98healthy married women revealed that some marital factors did predict premenstrualdistress. Marital dissatisfaction and poorer communciation predicted concentrationdifficulties and water retention during the premenstrual phase. Marital dissatisfactionalso predicted intermenstrual pain, concentration difficulty, autonomic reactions,negative affect and water retention. Conglin (1987; DA) also investigated therelationship between marital dissatisfaction and premenstrual symptoms. A significantrelationship between marital dissatisfaction and premenstrual symptoms was found forthe 150 women in the sample, but life event stressors were better predictors ofpremenstrual symptom intensity. More women with careers than women who were notPremenstrual Syndrome102in the labour force had problematic premenstrual symptoms .Coppen and Kessel (1963) and Tonks, Rack, and Rose (1968) found premenstrualsyndrome was more prevalent in 30 to 45 year old women living with a man than inyounger women or women aged 30 to 45 who were not living with a man.Hamilton (1988; DA) studied 30 couples in which the wife reported low levels ofpremenstrual distress and 30 couples in which the wife reported high levels ofpremenstrual distress. The former group had significantly better marital adjustment andcommunication, and higher levels of intimacy and trust, than did the latter group. Bothgroups experienced significantly better marital adjustment, communication, intimacy,and trust in the wife's intermenstrual phase than in her premenstrual phase.In a more qualitative investigation of the relationship between PMS and maritaldissatisfaction, Sara (1987; DA) studied six women with self-defined PMS and theirhusbands over one month. Decreases in the husband's marital satisfaction were foundto be associated with increases in his perceptions of his wife's negative affect in five ofthe six cases. In four of the six cases, the husband perceived an increase indispleasing behaviours by his wife when he perceived an increase in her symptoms.Since PMS was determined by self-report it is likely that a significant proportion of thesewomen did not have clinically demonstratable PMS. Nevertheless, this study isinstructive in showing the link husbands make between their perceptions of wife'sbehaviour and PMS.Finally, Rattray (1987; DA) attempted to study the effects of premenstrualsymptoms (as measured retrospectively) on the marital satisfaction of husbands andwives. Significant correlations were found between severity of premenstrual symptomsand marital dissatisfaction. Unfortunately, the author causally interprets the correlationsas demonstrating "the destructive effects which PMS may have upon the marriage andthe family." Since causation cannot be inferred from correlation, a hypothesis thatmarital dissatisfaction is a contributing factor to PMS is equally plausible.Premenstrual Syndrome103In sum, in a number of unpublished doctoral dissertations, PMS has been found tobe associated with reported marital dissatisfaction. Further research is needed toascertain the processes and direction(s) of influence involved in this association.PMS as a Culture Bound Syndrome An alternative to a strictly medical model has been raised by Johnson (1987), whoreviewed evidence that premenstrual symptoms have been reported since Aristotle butthat premenstrual syndrome is specific to this historical period in Western culture. Hedescribes PMS as a culture bound syndrome. He proposes that PMS is a "negotiatedreality" between women and society, and points out that women have been the majorforce in pressuring the medical community to recognize the existence of PMS. Hesuggests that PMS be thought of as a "barometer" of role strain in women's lives, whichhas increased as they have entered the labour force in large numbers while continuingto carry primary responsibility for their households and children. He hypothesized thatas the issues around women's multiple roles are resolved then reports of PMS shoulddecline.Prince and Tcheng-Laroche (1987) define culture bound syndromes as "acollection of signs and symptoms (excluding notions of cause) which is restricted to alimited number of cultures, primarily by reason of certain of their psychosocial features"(Prince, 1985). Use of the concept of culture-bound syndrome varies greatly and it haseven been proposed that all illnesses are culture bound (Cassidy, 1982); the abovedefinition is a compromise.Allen (1984) illustrates the social construction that goes into making an illness bysuggesting that we might"pose the category of 'pre-breakfast syndrome' in which to lump together allthe various complaints which could ever, in any individual, be shown toappear regularly in the first hours after waking and then to subside. Thesecould include such diverse problems as habitual hangover, morningPremenstrual Syndrome104sickness, smoker's cough, lethargy, or excitability, reduced or increasedlibido, irritability, intellectual impairment and numerous others." (cited inLaws, 1985, p. 35)These experiences can be unpleasant, but they are not known to be part of a disease.With media attention to "pre-breakfast syndrome", however, some individuals mightcome to view these experiences as part of an illness (Laws, 1985).At any given time 90% of the population has a symptom which if reported to aphysician would be considered "clinically serious" (Pearse & Crocker, 1949; Siegel,1963). Laws (1985) points out that most of the time we do not see ourselves as ill. Thequestion Laws (1985) asks is not why people do not report symptoms but rather, giventhis 90% symptom rate, why we choose to report the symptoms when we do so. Forexample, if women are exposed to much media coverage on PMS, they may selectivelylook for symptoms they have learned should occur but not notice the same symptoms atother times. Zola (1966, 1973) investigated the reasons for which patients presentedthemselves to physicians. Rather than the seriousness of the illness, the constraints oftheir social situations, e.g., the fit or lack of fit of their symptoms with culturalexpectations, determined when patients reported to their physicians.One way to test for PMS as a culture-bound syndrome is to do cross-culturalresearch. For example, Chaturvedi, Santach, Chandra, Prabha and Issac (1990) havereported that women residing in Canada as compared to women in India experienced agreater number of negative premenstrual changes such as irritability, sadness, fatigue,sore breasts, tension, depression, and decreased sexual desire, whereas Indian womenreported more positive premenstrual changes such as increased energy, increasedwork performance, calmness, well-being, and increased socializing. Other researchersare also studying cross-cultural aspects of PMS (e.g., Buckley & Gottlieb, 1988; Dan &Al-Gasseer, 1991; Dye, 1991; Monagle, 1991a; Monagle, 1991b).Premenstrual Syndrome105PMS and Stress Several researchers have established a link between PMS and stress. Forexample, Chuong, Colligan, Coulam and Bergstralh (1988) confirmed PMS in 20 womenwho kept daily information for one menstrual cycle. Women with PMS had less effectivepsychological functioning and increased stress (as measured by the MMPI) in the lutealas compared to the follicular phase.Beck (1989; DA) also studied the relationships among mood, physical symptoms,stress, and levels of estrogen and progesterone in 25 women with severe PMS. Shefound no relationship between physical symptoms or mood and levels of estrogen orprogesterone. In addition, the amount of stress a woman experienced early in themenstrual cycle was not significantly related to premenstrual symptom severity. Bruto(1988; DA) reported that 18 of 33 women who said they had PMS had mood changesnot restricted to the premenstrual phase. The remaining 15 women who had confirmedPMS showed significantly different patterns of stress, coping, and endocrine functionsthan did women who were not confirmed with PMS.Further evidence of a link between stress and PMS is provided by Hanson (1987;DA), who reported that women with severe PMS viewed menstruation as moredebilitating, reported greater severity of hassles, and were more external in health locusof control orientation. Both women with and without PMS reported a greater number ofhassles premenstrually. Kerstner (1987; DA) investigated 121 career women and theirretrospective reports of premenstrual symptoms. Using multiple regression techniques,life stress and masculine gender-role orientation significantly predicted recalledpremenstrual symptoms; specifically, life stress predicted positively and masculinegender role orientation predicted negatively the recalled incidence of premenstrualsymptoms. The relationship between PMS and stress also was explored by Meyer(1989), who studied 19 women with PMS and 16 cohtrol women to test the hypothesisthat women with PMS are more physiologically reactive to stress during thePremenstrual Syndrome106premenstrual phase. She measured heart rate, pulse transit time, and electrodermalresponses. She found that women with PMS were physiologically different than thecontrol women not only in the premenstrual phase but in other phases as well. Meyer(1989) suggested that women with PMS are more sensitive than are women who do nothave PMS to both biological fluctuations associated with the menstrual cycle andexternal changes. This increased sensitivity may mean these women are more reactiveto stress.A final example of a link between PMS and stress is Morgan's (1988; DA) studythat compared 40 women with PMS to 20 women without PMS and 20 men. Womenwith PMS reported experiencing significantly more symptoms, more stressors, and ahigher level of stress than did the two comparison groups. The women with PMS didnot report a history of more sexually traumatic events than did women without PMS.The men reported no sexually traumatic events.Some researchers theorize that stress interacts with other variables to exacerbatepremenstrual symptoms. For example, Shapiro (1990; DA) proposed that PMS resultsfrom an interaction among biological, psychological, and stress factors. Heilbrun andFrank (1989) hypothesized that self-preoccupaton and general stress may act togetherto sensitize a woman to premenstrual symptoms.A few researchers have failed to confirm the link between stress and menstrually-related symptoms. For example, Schechter (1988; DA) found that within subjectanalyses of women with PMS revealed no relationship between degree of life stress andPMS symptoms. Between subject analyses, however, revealed chronically higherstress levels in women with the greatest degree of premenstrual change. Clarvit (1988)was unable to establish a link between menstrual dysfunction and stress in femalemedical students. Clarvit focused on oligomenorrhea, i.e., irregular cycles; amenorrhea,i.e., lack of menstruation; and dygmenorrhea, i.e., painful menstruation, not PMS.The following researchers have suggested variables that might form a protectivePremenstrual Syndrome107mechanism against the effects of stress. Morse, Dennerstein, Varnavides and Burrows(1988) found that a group of healthy women had higher self-esteem and an absence ofdepression by comparison with women with prospectively diagnosed PMS. Theysuggested that high self-esteem and absence of depression act together as a"protection" factor and these two resources, along with social support, buffer healthywomen against stress and illness (Cobb, 1976).PMS and Abuse The theme of an association between a history of abuse and premenstrualsyndrome was discussed several times (though no papers were presented on thistheme) at the 9th National Conference of the Society for Menstrual Cycle Research.After a very thorough literature review, only one reference was found linking sexualabuse to self-reported premenstrual syndrome. Dekker (1986; DA) found a significantpositive association between women's self-reported premenstrual symptomatology andthe degree of distress experienced after unwanted sexual experiences.Dekker (1986) had expected a strong link between post-traumatic stress disorder(as diagnosed by the DSM-III criteria) and unwanted sexual experiences, but thisassociation was not significant. In other words, PMS was more strongly linked withunwanted sexual experiences than was post traumatic stress disorder.Given the many different approaches in the literature to understanding PMS, it issurprising that researchers have not systematically explored possible links betweenhistory of physical, sexual, and emotional abuses and PMS.Other Theories of PMS Harrison (1984) observed that some women first notice premenstrual symptomsafter: starting or stopping using birth control pills, a hysterectomy, tubal ligation, or thebirth of a baby. She contended that some of the premenstrual symptoms may beiatrogenic, or caused by medical intervention.Lander (1988) concluded that feeling out of control is one of the most disturbingPremenstrual Syndrome108aspects of the premenstrual experience for women. She suggested that perhaps PMSis a "metaphor for women's common position of not being in control of their situation" (p.97).Blechman and Clay (1987) proposed a danger-signal hypothesis of PMS. Womenwho have painful periods may search for any evidence of an oncoming period, and theirhypervigilance may make them irritable and hence manifest PMS. This hypothesis wasbased on Seligman's (1968) safety-signal theory of escape avoidance learning.Participants in his studies preferred to listen for a cue signalling shock rather than tomusic, even when the shock was unavoidable. Vigilance was described by participantsin his experiment as anxious anticipation, tension, or stress. Blechman and Clay (1987)noted some parallels between this experimental experience of vigilance and PMS. Oneway to test the danger signal hypothesis of PMS would be to take a thorough menstrualhistory of women who do and do not have EMS and determine whether they differaccording to dysmenorrhea or vigilance. Steege, Stout, and Rupp (1985) have foundthat some subtypes of premenstrual complaints (water retention and atypicaldepression), but not others, are correlated with dysmenorrhea.Another psychological construct was considered by Marabotto-Sloan (1988; DA),who theorized that when a woman has a "firm sense of her body boundary," thisstructure acts as a protective wall against the premenstrual and menstrual physicalsensations. Unfortunately, a clear definition of what was meant by body boundary wasnot supplied in the brief abstract. This study is included here to illustrate anotherapproach to the study of PMS.Another theory regarding PMS was proposed by Morse and Dennerstein (1988;cited in Gallant, 1991) who suggested three causes of PMS: (1) family history and pastpsychiatric disorders predispose women to PMS; (2) biochemical changes (such asgoing on or off oral contraceptive pills) or external factors such as stress may precipitate PMS; and (3) life style, diet, and (lack of) exercise habits can sustain PMS. The causePremenstrual Syndrome109of PMS can also be any combination of the above.Schmidt, Grover, Hoban and Rubinow (1990) suggested that women withconfirmed PMS have state-related alterations in mood and perception or an interactionbetween their affective state and their environment. They asked women with confirmedPMS and control women to complete an inventory of pleasant and unpleasant events ontwo occasions, once during the follicular phase, and once during the luteal phase. Theinstructions were to recall all events listed in the inventory that had occurred over thepreceding 5 days. Women with confirmed PMS recalled significantly more unpleasantlife events in the luteal phase than did control women. In addition, significantly morewomen with PMS than control women reported the same events as being moreunpleasant in the luteal phase than in the follicular phase. A limitation of this study isthat the life events schedule was retrospective and the purpose of the investigation wasknown. This combination of methodological flaws could have biased the women'sreports.The relationship of family attitudes to PMS was explored by Williams (1990; DA)who conducted a cross-sectional survey of the menstrual attitudes and behaviour ofmothers and daughters. She found that mothers and daughters held similar attitudestoward menstruation. There was a positive association between the daughtersperception of her mother's encouragement of sick role behaviours and the daughtersown premenstrual distress.Role of Stereotypes in PMS Reports Our previous research (McFarlane et al., 1988) was designed to address themethodological problems discussed earlier. Participants were studied daily over 70days to ensure the inclusion of at least two menstrual cycles (Parlee, 1982). To providecomparisons, three groups were studied: 12 women who were normally cycling(womenNC), 15 women taking oral contraceptives (womenOC), and 15 men. To avoidbiasing reports, interest in the menstrual cycle was camouflaged (Parlee, 1974). MoodPremenstrual Syndrome110data were obtained both prospectively (concurrently), to assess actual mood changes,and retrospectively. The prospective data consisted of daily reports of arousal,pleasantness, and stability. The retrospective data were the participants' recollectionsat the end of the study of their average mood for each phase of the menstrual cycle(women only) and for each day of the week. To avoid mood labelling problems,decrease the likelihood of participant attrition, and to provide positive as well asnegative mood reports, a simple bipolar mood scale (Russell, Weiss, & Mehrabian,1987) was used. Mood fluctuations were studied in the context of other cycles, includingthe weekday cycle.For analyses of daily moods over the menstrual cycle, men were randomlyassigned to cycles. The menstrual cycle was divided into five phases (Rossi & Rossi,1977): Menstrual = Day 1-5, with Daylthe first day of menstruation; Follicular = Day 6-12;Ovulatory = Day 13-17; Luteal = Day 18-24; and Premenstrual = Day 25-28. Hormonalassays were not conducted, so the phase labels do not necessarily reflect the actualhormonal fluctuations experienced by individual women (Blechman & Clay, 1987). Thepremenstrual and menstrual phases, which were of primary interest, would neverthelessbe accurate. Rossi and Rossi's (1977) procedure was used to adjust longer and shortercycles to 28 days.The prospective daily reports did not reveal the classic menstrual mood pattern.There was no evidence that women had more negative moods in the premenstrual ormenstrual phase than during other phases of their cycles, and women's moods were noton average more negative than men's moods. Indeed, the only differences contradictedthe stereotype. WomenNC (not on the pill) reported more positive moods during boththe follicular phase (the seven days between menstruation and ovulation or mid-cycle)and the menstrual phase than did men and womenOC. Women's moods fluctuatedmore over days of the week than over phases of the menstrual cycle and bothwomenOC and womenNC prospectively reported exactly the same weekdayPremenstrual Syndrome111fluctuations as men's moods, that is, Friday - Saturday highs. Just as menstrual andpremenstrual lows were not found, Monday Blues were not found in the prospectivedata. Finally, there was no evidence in the mood stability ratings that women's moodswere more changeable than men's, either within a day or from day to day.The women's reports of how they remembered feeling during the premenstrual andother phases of their cycles were stereotypical. They remembered experiencingsignificantly more negative moods during the premenstrual and menstrual phases.These women would have been diagnosed as having PMS if only their retrospectivereports were used, as has been the case in some research and clinical practice.Why would there be a difference between what women reported on a day-to-daybasis and what they remembered? The most likely explanation is that they wereinfluenced by stereotypes. When they did not know the researchers were interested inthe menstrual cycle they did not feel as if they were experiencing any PMS symptoms,but these same women believed the stereotype that most women, including them, havemenstrual mood changes like PMS. When asked to remember their menstrual moodsthey reported what they believed was their typical menstrual pattern, i.e., the stereotype,even though it was inaccurate for them. This interpretation gains credence from twoother findings. First, the recollections for the follicular phase of both WomenNC andWomenOC were significantly more positive than they had reported concurrently.Moreover, whereas womenOC had not concurrently reported experiencing the increasein positive mood during the follicular phase that womenNC concurrently reported, theyapparently were aware that it happened to some women and incorrectly rememberedthat it had occurred for them in their most recent cycles. The menstrual stereotypesapparently operated in both directions.How might stereotypes about PMS arise? The finding of more positive moods inthe follicular phase and exaggerated recall of these positive moods suggests onehypothesis (McFarlane et al., 1988). Perhaps women (and their partners) believe thesePremenstrual Syndrome112especially positive moods are "normal" and the lower (but still positive) moods thatfollow them are "abnormal". The prospective or concurrent data indicate instead thatpremenstrual moods are as positive as moods in all other phases except the follicularphase, when the moods of normally cycling women are even more positive.The second finding that lends credence to our interpretation of the menstrual datais that a similar pattern occurred for weekly cycles. When men and women were askedto remember their average moods for each day of the week, over the last few months,they exaggerated their day-to-day moods in keeping with the stereotype of "Mondayblues" and "Thank-God-Its-Friday" feelings. They did not report Monday Blues in theirdaily records, but they recalled Mondays as bluer than other days, just as Stone,Hedges, Neale, and Satin (1985) found. They did prospectively report Friday/Saturdayhighs in their daily records but they recalled the weekends as even more positive thanthey had reported concurrently. Thus the contrast effect found for thepremenstrual/menstrual versus follicular phases also seems to apply to days of theweek. The consistency of the findings for two different kinds of cycles, menstrual andday of week, lends strength to this contrast hypothesis.The difficulty with all stereotypes is their resistance to change (Martin & Halverson,1981, 1983). People tend to notice things that are stereotype-consistent and ignoreinconsistent evidence. Moreover, even if inconsistent evidence is noticed, it is likely tobe forgotten, or distorted in memory. In the case of PMS and stereotypes about moodfluctuations over the menstrual cycle, a similar process may occur. If a woman feels"down" on a particular day she will wonder why and look for a reason. She usuallyknows if her period is due, and does know if she is menstruating, so if either is true (andthey will be true 25-35% of the time), she may decide, "I'm down because of my period."Most people are more likely to search for biological explanations than for environmentalones for their negative feelings, e.g., work-related or interpersonal events (e.g., Bains &Slade, 1988; Koeske, 1975; Koeske & Koeske, 1975; Slade, 1984). Moreover, thePremenstrual Syndrome113woman's partner or friends may reinforce this process. If they are having an argument,her partner or friend may say, "What is the matter with you today? Are you getting yourperiod?" Neither the woman nor her partner is likely to say on other occasions, "Youare in a terrific mood today," and attribute that to her menstrual cycle. Evidence thatcontradicts the stereotype is likely to be ignored, so all that is required to maintain thestereotype is an occasional piece of consistent evidence. The woman may only feel"down" once or twice a year during her premenstrual phase but she is more likely-tonotice that and to ignore the more frequent occasions when she feels fine.It may well be that some or even most women do on occasion experience an "out-of-character-for-me" mood or reaction during the premenstrual phase (McFarlane et al.,1988). For example, if they are over-tired, and in a very frustrating situation over whichthey have no control (and maybe x and y other things co-occur), and they arepremenstrual, they may be more likely to cry, or express anger, or act irritable, etc. Theimportant question is how often all these events co-occur to produce this reaction andwhether it is debilitating for the woman. If it only happens once every year or so, howmuch attention does the premenstrual experience deserve? Maybe the same womanhas that or some other "out-of-character-for me" reaction when another set of events notinvolving the menstrual cycle co-occur, and for that matter, maybe this is true for manymen.One limitation of our initial study is the sample. It is possible that the younguniversity women studied were anovulatory for many or most of their cycles (Prior,personal communication, 1988). For this and other physiological, sociological, andpsychological reasons, older women may be more likely to experience PMS.Premenstrual Symptoms versus Syndrome One of the central problems for both research and diagnosis is confusion aboutwhat constitutes PMS. In the first 30 years of research over 150 different symptomswere said to vary with the menstrual cycle (Rubinow & Roy-Byrne, 1984). SeveralPremenstrual Syndrome114researchers now emphasize the importance of distinguishing between premenstrualsymptoms and syndrome. Diagnoses have often been made on the timing of symptomsrather than on the basis of specific symptom clusters (Hamilton, Parry, Alagna,Blumenthal, & Herz, 1984). If a woman observes a cyclical change in one or moresymptoms, she may wonder if she has premenstrual syndrome. This distinctionbetween symptoms and syndrome continues to pose problems for researchers,clinicians, and women who think they may have PMS. It is possible that a woman mayexperience some changes, i.e., symptoms, but not have PMS according to diagnosticcriteria. It is also possible that a woman may be diagnosed as having premenstrualsyndrome but because of the many PMS criteria may not experience consistent changein any one symptom (e.g., breast pain may alternate with headache and irritability duringthe premenstrual phase but taken as a whole the PMS criteria still may be met).Endicott Nee, Cohen, and Halbreich (1986) found evidence for 5 differentdimensions of premenstrual change in women's daily reports. One of their analysesincluded a comparison of 5 premenstrual with 5 menstrual days, and they argued thatpremenstrual changes is a more accurate term than premenstrual syndrome. They didnot use the daily data for the 18 intermenstrual days and covered only one cycle,however, so the five dimensions might not be replicated if more data were used.Brooks-Gunn (1986) summarized major problems with much of the PMS research:(1) the rigorous criteria of Steiner, Haskett and Carroll (1980) are not used; (2)recruitment only of women who identify themselves as PMS sufferers; and (3) errorsmade in diagnosis are more often errors of inclusion rather than exclusion. Brooks-Gunn (1986) distinguished premenstrual symptom from syndrome research andsuggested that premenstrual syndrome may be biochemically based whereassymptoms may have a psychological etiology.There is a diversity cross-culturally of patterns of reports of premenstrualsymptoms. In one cross-cultural study of premenstrual symptoms (not syndrome),Premenstrual Syndrome115Janiger, Riffenburgh and Kersh (1972) compared 50 Greek, 100 Japanese, 51 Turkish,35 Nigerian, and 28 Aboriginal United States women with 135 United States studentsand housewives. The most distressing symptom was reported to be lower abdominalpain. Every respondent reported at least one symptom. The data were retrospective,with symptoms rated for the week prior to most recent menstruation. Symptoms oftension in the premenstrual phase were reported in all cultures, with some specificdifferences across cultures. There was a paucity of breast complaints by Japanesewomen (who also have a low incidence of breast cancer). There was a high frequencyof headache in Nigerian women. Japanese women reported least premenstrualdistress, the U.S. women were intermediate, and Turkish and Nigerian women reportedthe most. These researchers concluded that premenstrual distress is a universalphenomenon but with no consistent cross cultural pattern. The World HealthOrganization conducted a study of premenstrual symptoms in 10 countries; 50°/0 offemales retrospectively reported physical discomfort before menstruation and 38%reported mood changes (Snowden & Christian, 1983). Boyle, Berkowitz, and Kelsey(1987) conducted an epidemiological study of various demographic and health variablesas they related to premenstrual symptoms. They found that premenstrual irritability ormoodiness was moderately associated with the white race and less so with a history ofvaricose veins and high blood pressure. Woods, Dery, and Most (1982) reported thatmore white than black women said they had PMS. Several additional studies showevidence of cultural differences in premenstrual symptom reporting. Ferguson andVermillion (1957) found white and black U.S. women reported the same incidence of PMsymptoms except that blacks reported a much higher frequency of headaches. Weisand Janiger (cited in Janiger et al., 1972) found that homosexual women in acorrectional institute reported fewer PM symptoms than did women in a heterosexualcontrol group.In addition to varying cross-culturally, premenstrual symptoms may vary accordingPremenstrual Syndrome116'to age, marital status, and type of employment. There is some evidence to suggest thatpremenstrual symptoms are more likely to occur for women in their 30s and 40s thanwomen in their 20s (Moos, 1968; Dalton, 1964; Golub, 1976, 1988; Kramp, 1968;Abplanalp, Haskett, & Rose, 1980; Rees, 1953a; Rees, 1953b; Kashiwagi, McClure, &Wetzel, 1976). Golub (1976) studied healthy women over the age of 35 and found thata significant premenstrual mood change did occur but that the change was small inmagnitude. Golub and Harrington (1981), using the same measures as Golub (1976),found no premenstrual mood change in high school students aged 15-16.Endicott and Halbreich (1988) conducted studies of 14 different groups of women.Women executives from New York banks and law firms who had "made it" reported thelowest rates of premenstrual depression.Predominance of Poor Research on PMS Evidence (cited earlier) that restrospective reports of PMS reflect attitudes andstereotypes more than they reflect actual premenstrual experience was first presentedin the late 1930s (McCance, Luff, & Widdowson, 1937). This point has beenemphasized consistently since the mid 1970s (e.g., Parlee, 1974; Ruble, 1977). It is adisappointing statement on progress in the PMS research field that more than 10 yearslater, publications abound in which the researchers continue to define PMSretrospectively (e.g., Brayshaw & Brayshaw, 1987; Cumming, 1988, DA; Hallman,Oreland, Edman, & Schalling, 1987; Hallman & Georgiev, 1987; Heilbrun & Renert,1988; latrakis, Kourounis, Sakellaropoulous, & Gallos, 1988; Larkin, 1987, DA; Mauri,Reid, & MacLean, 1988; Metcalf, Livesey, Hudson, & Wells, 1988; Van der Ploeg, 1989.It is noteworthy that the majority of these citations are from psychiatric rather thanpsychological journals.). Continuing to publish retrospective research on PMS runs therisk of further clouding rather than clarifying the PMS picture. Given that PMS haspsychological, health, and legal consequences for women, it should be incumbent onPMS researchers to maintain a reasonable standard of quality. In sum, the time hasPremenstrual Syndrome117long since passed for researchers to conduct retrospective PMS research and forretrospective PMS work to be accepted for publication (unless the retrospectivephenomenon is of specific interest) .Even when prospective techniques are used, there are problems withinterpretation of data and results. The following paper is one example ofmisinterpretation of the data, but many others can be found. Dennerstein, Morse andVarnavides (1988) reported that more women with prospectively confirmed PMS thanwomen for whom it was not confirmed recalled having postpartum depression. Theproportion of PMS women as compared to controls recalling postpartum depression wasindeed higher in absolute terms, but the authors conducted Chi-square analyses whichdemonstrated that the difference was not statistically significant. They neverthelessregarded their data as being "largely confirmatory" of an association between PMS andpost partum depression. They concluded that their data provide "... further support forthe aetiological role of biological factors" in PMS as well as being evidence of thebiological vulnerability of women to mood disorders (p. 51). These conclusions areinconsistent with the results of the statistical analyses.Problems in Diagnosing PMS The following study was chosen to illustrate the necessity for improvement andeducation about diagnostic techniques for PMS. Pepper-Kozita (1990; DA) sent a casevignette of a woman with PMS to a randomly selected group of psychotherapists inCalifornia. Fewer than half accurately diagnosed PMS and slightly more than halfsuggested appropriate treatment. To improve the diagnoses and treatments womenreceive, it will be important to clearly define the diagnostic criteria for PMS as well as tohave readily available and easy to understand diagnostic tools.The first important issue in the diagnosis of premenstrual syndrome, whether forresearch or clinical purposes, is the use of prospective or day-to-day reports versusretrospective reports. There is growing evidence (cited earlier) that prospective reportsPremenstrual Syndrome118are the only acceptable way to demonstrate relationships among mood, behaviouralchanges, and the menstrual cycle (Rubinow & Roy-Byrne, 1984). It should be noted,however, that even prospective reports can be biased if a woman knows the purpose ofthe study (Olasov & Jackson, 1987). Women recruited because they say they havePMS will always know the menstrual interest of the clinician, so attempts should bemade whenever possible to diminish the effect of this knowledge and/or to considerpossible bias when evaluating daily reports.The second important problem in PMS diagnosis and research is the widevariation in definition. Kramp (1968) defined the premenstrual phase as the last sixpremenstrual days and the first two days of menstruation. Dalton (1964) defined a"paramenstruum" as four days before and four days after menstruation begins. Taylor(1988) also used a "perimenstrual" phase. By contrast, Sutherland and Stewart (1965)contended that PMS symptoms must vanish- at the onset of menstruation for a validdiagnosis of PMS. May (1976) reported that half of the 30 women studied reportedmost symptoms in the premenstrual phase whereas the other half reported most in themenstrual phase. Researchers have not agreed on a clear distinction between thepremenstrual and menstrual phases. There is some evidence that some premenstrualcomplaints (e.g., water retention and atypical depression) are correlated withdysmenorrhea (painful menstruation) but that other premenstrual complaints are not(Steege, Stout, & Rupp, 1985). Graham and Sherwin (1987) found a strong relationshipbetween severity of dysmenorrhea and overall severity of premenstrual changes. Thisconfusion raises the additional question of whether the women participating in the .menstrual research are clear about the distinction. It is possible that some women mayhave only premenstrual symptoms (e.g., PMS), some only menstrual symptoms (e.g.,dysmennorhea), and some both, but because the distinction is not made completelyclear, they all say they have PMS when asked.A third problem is that an ongoing psychiatric illness can exacerbate premenstrualPremenstrual Syndrome119symptoms and these changes may or may not be distinct from PMS. For example,Siegel, Meyers, and Dineen (1986) grouped 156 women into low, medium, and highscorers on the Beck Depression Inventory. They also measured hypochondriasis. PMsymptoms were measured using the MDQ daily for each of the premenstrual days of thenext cycle. Nondepressed women listed food cravings, swelling, fatigue, and weightgain as their most severe physical symptoms, and irritability, mood swings, tension,anxiety, and lowered concentration as their greatest premenstrual emotional problems.The severely depressed group showed the same pattern. The moderately depressedgroup scored each of these items more severely than did the other groups butpresented the same ranking of items. Haskett, Steiner, Osmun, and Carroll (1980)considered that women with "pure" PMS show a normal psychological profile throughoutthe menstrual cycle. They have a significant incidence of emotional and behaviouralsymptoms premenstrually, but not necessarily depression. Women with •psychopathology, on the other hand, may show particularly high scores in thepremenstrual phase that reflect premenstrual exacerbation of psychopathology presentthroughout their cycle. A more detailed discussion of the link between PMS andaffective disorders is presented in the next section.A fourth problem in diagnosing PMS is the question of what women find mosttroublesome about PMS. At least 150 premenstrual symptoms have been listed(Rubinow & Roy-Byrne, 1984), but Rubinow, Roy-Byrne, Hoban, Gold and Post (1984)found that 96% of the 220 women who completed their retrospective questionnairereported that their menstrually related mood disturbance was their most upsetting andincapacitating problem. Women's greater difficulty with mood changes than withsomatic changes is consistent with a sociological or psychological interpretation ofpremenstrual syndrome. Woods (1985) entered several variables of interest to PMSresearchers into a causal model and found that of all symptoms, negative affect was themost important cause of perimenstrual disability. Woods, Most, and LongeneckerPremenstrual Syndrome120(1985) studied 74 women daily for two months. The women reported stressors,completed the MDQ, and recorded major life events. Daily stressors were morestrongly related to perimenstrual experience than were major life events.A fifth diagnostic problem is that a woman may be classified as having PMS byone set of criteria but not by another. Only 20-50% of women who retrospectively saidthey had PMS actually showed a PMS pattern in prospective reports collected over onecycle (e.g., Kessler, DeLongis, Haskett, & Tal, 1988; Hamilton, Parry, Alagna,Blumenthal, & Herz, 1984) and when two consecutive cycles were used the proportionwas even lower (approximately 5%; Kessler et al., 1988). Retrospective reports ofsevere premenstrual changes are more likely to be confirmed by prospective reportsthan are retrospective reports of mild changes (Endicott, & Halbreich, 1982).When prospective reports do not support a PMS diagnosis, there are severalpossible explanations. First, these women may not have PMS. Second, the womenmay have PMS but the measures may not tap the domains in which these women wouldshow PMS patterns. Third, the particular month(s) studied may have been atypical, soother cycles might reveal PMS patterns. For diagnosis it would seem prudent tocontinue with prospective assessment, and for research purposes, to collect data forthree cycles (e.g., O'Boyle, Severino, & Hurt, 1988; Woods, 1991).A sixth diagnostic problem is the bias inherent in determining what behaviour isunhealthy and warrants being called a symptom. Taylor and James (1979) found thatwomen in a PMS clinic reported "loss of libido" as the second most commonly citedsymptom. Laws (1985) pointed out that cyclic change in sexual energies is defined as asymptom of sickness only in cultures in which women are expected to be responsive tomen at all times. Laws went on to say that where health is equated with constant sexualavailability, appealing to illness may be a woman's only way out of unwanted sex. Careshould be taken by the diagnostician to avoid labelling an experience reported by awoman as a symptom, unless the woman sees it as a symptom herself.Premenstrual Syndrome121PMS and Affective Disorders Some authors suggest there is a link between PMS and affective disorders. Forexample, Ascher-Svanum and Miller (1990) found that 21% of inpatients diagnosed withmood disorders had significant premenstrual changes, as determined by retrospectiveprocedures. In addition, 10 of 11 patients with significant premenstrual changes (again,retrospectively determined) were diagnosed with mood disorder (the one exception wasdiagnosed as having atypical psychosis). Malikian (1987; DA) found evidence tosuggest that the premenstrual phase intensified symptoms in women diagnosed withmajor or minor depression. Using retrospective techniques, Chisholm, Jung, Cumming,Fox and Cumming (1990) discovered two subgroups of women: those showing "purePMS" and those with a premenstrual exacerbation of existing anxiety and depression.Graze, Nee, and Endicott (1990) found that retrospective reports of PMS weresignificant predictors of major depressive episodes that occurred in the subsequent 2-4years. Indeed, retrospective reports of PMS were stronger predictors of majordepressive episodes than the two known risk factors: family history of depression andprior personal history of depression. Evidence of a link between PMS and affectivedisorder also was obtained by Harrison, Sandberg, Gorman, Fyer, Nee, Uly, andEndicott (1989). They found that 24 women diagnosed with Late Luteal PhaseDysphoric Disorder (LLPDD; DSMIII-R) were more susceptible to panic attacks aftercarbon dioxide inhalation than were controls. They hypothesized that susceptibility tolab induced panic attacks suggested a susceptibility to panic in natural settings. Stein,Schmidt, Rubinow, and Uhde (1989) found no association, however, between timing ofpanic attacks and the phase of the menstrual cycle.O'Boyle, Severino, and Hurt (1988) found no relationship between PMS, asdiagnosed prospectively with 3 months data, and psychiatric diagnoses. The proportionof women with none, past, or current psychiatric diagnosis was virtually identical forwomen with and without PMS. These authors do not present their data as disconfirmingPremenstrual Syndrome122the hypothesis that PMS is associated with affective disorders. Rather, they say theissue is "unsettled" and cite others who have reported a link between mood disordersand PMS. Note, however, that many of the others relied on retrospective reportswhereas their data were prospective.Severino, Hurt, and Shindledecker (1989) reported that prospectively diagnosedPMS was three times more prevalent among women with a past or current episode of apsychiatric disorder than among women without a past or current psychiatric disorder.These authors did not distinguish, however, between premenstrual exacerbation of acurrent psychiatric disorder and PMS in which at least one phase is free of affectivesymptoms. If this distinction were made, the association between "pure" PMS andaffective disorder may have been weaker.Other researchers who have used prospective methodologies have found no linkbetween PMS and affective disorders. Christensen, Oei, and Calan (1989) studied 13women with prospectively confirmed PMS and 13 control women. None of the womenwith confirmed PMS had a current mood disorder. The control women, whoseprospective data did not confirm their self-reported PMS, were not statistically differenton any measure (aside from retrospectively reporting PMS) than control women.Endicott, Halbreich, Schacht et al. (1985; cited in Endicott & Halbreich, 1988)found that approximately 65% of women with a history of major depression disorderhave PMS or premenstrual exacerbation of symptoms. Approximately 80% of womenwith current PMS have had a prior period of major depressive disorder. These authorsalso found an association between PMS and subsequent episodes of major depressivedisorder. In later work, however, a marked difference was found if retrospective ratherthan prospective diagnoses of PMS were used. Retrospective reports of PMS weresignificant predictors of major depressive disorder which occurred in the subsequent 2-4years (Graze, Nee, & Endicott, 1990). By contrast, women with prospectivelydiagnosed PMS showed no significant differences in psychiatric history than controlsPremenstrual Syndrome123aside from the premenstrual changes (Harrison, Endicott, Nee, Glick, & Rabkin, 1989).Harrison et al. also found that women who sought treatment for PMS had a relativelyhigh incidence of lifetime history of depression, panic disorder, suicide attempts, andsubstance abuse, by comparison with controls. Moreover, women who met stringentDSM-III-R criteria for LLPDD showed few significant differences from controls, asidefrom premenstrual symptoms (Harrison et al., 1989). Lesonsky (1989; DA) found thatwomen with a psychiatric disorder reported more symptoms throughout the menstrualcycle than did women with no psychiatric disorder. The women with psychiatricdisorders also reported symptoms not typically associated with PMS. She argued thatthese response biases may have been an artifact in previous research in which it wasargued that there is a link between psychiatric disorder and PMS symptom severity.It is possible that there may be no causal link between PMS and affectivedisorders, but the menstrual cycle may influence a woman's experience of an affectivedisorder. Blumenthal and Nadelson (1988) reviewed the literature and summarizedseveral ways the menstrual cycle may modulate psychiatric symptomatology: (1)premenstrual exacerbation of the course of psychiatric disorders; (2) episodic clusteringof psychiatric symptoms (e.g., bulimia and panic attacks) during the premenstrualphase; (3) the cycle may function as a "zeitgeber," or synchronizer of periodicpsychiatric disorders so that the expression of these illnesses depends on the phase ofthe menstrual cycle; (4) mood changes related to the menstrual cycle phase may serveas sensitizing stimuli which exert a "kindling" effect on the development and course ofpsychiatric disorders. That is, just as repetitive administration of a electrical stimulus tothe brain may produce increasing effects and profound long-term alterations in brainactivity and behaviour, a dysphoric state that is repetitively experienced in associationwith the premenstrual phase might facilitate the gradual development or expression ofan affective illness in vulnerable individuals; and,(5) there may be an artifact ofresponse bias to report more frequent and more severe symptoms all through thePremenstrual Syndrome124menstrual cycle in women with psychiatric disorders.Hamilton, Gallant, and Lloyd (1989) estimate that 5% of menstruating women inepidemiological studies of depression may be "false positive" because they are in theirpremenstrual phase. Thus, some of the greater incidence of depression in women thanin men may be an artifact of transient premenstrual moods.Diagnostic MeasuresIn one of the first attempts to develop a diagnostic measure of PMS, Moos (1968)produced the MDQ, a 47-item symptom checklist containing eight symptom subscales.The focus is primarily on negative somatic changes; very few items are related topsychological, emotional, or behavioural changes. Moos did not supply inclusion andexclusion criteria for determining who has PMS. The MDQ has most often been usedretrospectively.Steiner, Hackett, & Carroll (1980) used a modified MDQ and several psychiatricmeasures to develop diagnostic criteria. They excluded somatic symptoms and womenwith a psychiatric history, and compiled a list of eight mood and behavioural symptoms,five of which were necessary for a diagnosis of PMS. Using these criteria the womenclassified as having PMS were more homogeneous. Rubinow & Roy-Byrne (1984)criticized this procedure on the grounds that it does not account for many women'scomplaints of PMS and is too conservative.Abraham and Hargrove (1980) devised a classification scheme for women whosay they have PMS on the basis of their retrospective reports to a 19-itemquestionnaire. The classifications differ conceptually on the basis of physiologicaletiology: (a) anxiety and irritability, (b) water retention, (c) increased appetite and foodcravings, and (d) depressed affect with cognitive impairment. This test was notconstructed using standard psychometric procedures and its sensitivity and specificitywere questioned by Rubinow : and Roy-Byrne (1984).Halbreich, Endicott, Schachts, and Nee (1982) developed a retrospective 95-itemPremenstrual Syndrome125Premenstrual Assessment Form (PAF) which allowed them to establish inclusion andexclusion criteria for different sub-syndromes of PMS. Although the PAF hassatisfactory usefulness as an a priori screening tool for obtaining homogeneous groupsof women (Gerstein, Reznikoff, Severino, & Hurt, 1984; Rubinow & Roy-Byrne, 1984), itis a retrospective measure and not appropriate for diagnosing PMS. For that purpose aPAF-Time chart was developed; it was designed to record prospective daily ratings witha focus on "target changes of interest." No criteria for PMS diagnosis based on dailyPAF-T ratings were provided (Halbreich & Endicott, 1982). In later research, Endicott,Nee, Cohen, and Halbreich (1986) abandoned the PAF-T and used a much simplifiedvisual analogue scale in their prospective data collection. It is not clear how Halbreichand Endicott (1982) defined the premenstrual phase, i.e., which days of the menstrualcycle. They appear to have asked women to give ratings for their "premenstrualperiod." This is vague and the use of "period" could have been confusing in thiscontext.There are two major prospective diagnostic procedures for PMS. First is the set ofcriteria proposed by the U.S. National Institute for Mental Health (NIMH) in April, 1983(Hamilton, Parry, Alagna, Blumenthal, & Herz, 1984): 1) a change of 30% in theintensity of symptoms measured intermenstrually (days 5 to 10 of the menstrual cycle)as compared to premenstrually (6 days before menstruation) and 2) documentation ofthese changes for at least two consecutive cycles. Rubinow et al. (1984) used thesecriteria in their prospective study. There are four potential concerns about the NIMHcriteria. First, prospective reporting is not explicitly stated so retrospective reports thatmeet the criteria could be used. Second, if data are collected prospectively, then only11 of at least 28 days of the cycle are used for diagnosis. With over half the availableinformation overlooked, accuracy in diagnosis is likely diminished. If changes in thepremenstrual phase were compared to changes for all other parts of the cycle,researchers and clinicians would have a more complete context within which to placePremenstrual Syndrome126PMS changes. Third, if normally cycling women experience more pleasant mood in thefollicular phase, i.e., following menstruation, than in all other phases, which do not differ,as we found (McFarlane et al., 1988), then the NIMH criterion of comparing theintermenstrual and premenstrual phases could be misleading. A woman whosepremenstrual phase was no different from all other phases except the follicular phasecould meet one of the NIMH criteria because of her follicular positive experience.Findings obtained by Morse, Dennerstein, Varnavides and Burrows (1988) also argueagainst contrasting the premenstral and intermenstrual phases. They used appropriateprospective techniques for assessing PMS, and conducted a factor analysis on datafrom 75 women with confirmed PMS. Surprisingly, the largest factor by far waspsychological distress in the follicular phase. (This factor had an eigenvalue of 6.70; thesecond highest eigenvalue was 1.72.) This finding of distress in the follicular phase isimportant given that most researchers use this phase as a symptom free control phasewith which to compare the premenstrual phase. The use of only the follicular phase asa comparison for the premenstrual phase would be misleading if either follicular phasedistress (Morse et al.) or increased pleasantness of mood (McFarlane et al.) is commonto many women. The use of data from the whole menstrual cycle seems moreappropriate when diagnosing PMS.In a Lancet (1981) editorial, the issue of how to interpret change over themenstrual cycle was addressed. "With some exceptions, the data seem equallyconsistent with the hypothesis of a mid-cycle syndrome of lowered crime, fewer epilepticseizures, increased self-esteem and elation, and increased sexual desire and activity. Itwould be incomplete to say only that women perform worse at certain times in the cyclethan others; their performance may at times be better than average performance ofmales on the task in question" (cited in Laws, 1985, p. 57). In a related line of criticism,Martin (1987) questioned whether "symptoms" are true losses or whether they could bemanifestations of other skills or perceptions. Martin speculated that perhaps thePremenstrual Syndrome127symptoms or "losses" in ability are accompanied by gains in complementary areas. Forexample, a woman may experience a loss of ability to concentrate but a gain in theability to free associate; a loss of muscle control but a gain in the ability to relax; anddecreased efficiency in performing a large number of tasks, but increased attention to asmaller number of tasks.A fourth problem with the NIMH diagnostic criteria is the use of 30% change fromthe intermenstrual to premenstrual phase. It is intended to isolate "marked" changefrom "normal" or "slight" change. Perhaps, however, a 30% change from phase to phase(or day to day) is considered normal or unremarkable by some women but remarkableand bothersome to other women. The only way to know what is characteristic oruncharacteristic change for each individual is to use a measure of change based on heror his own mood fluctuations. For example, one standard deviation above or below theindividual's mean response in other phases could be considered an uncharacteristicchange. The degree of change would be similar from woman to woman becauseindividual variations in "usual" fluctuations would be controlled with the standarddeviation criterion. Larsen (1987) contended that the standard deviation of daily moodscores is a measure of mood extremity, not mood change frequency. He suggestedthat a spectral analysis of daily mood scores would provide information on mood changefrequency. Researchers need not settle for an arbitrary amount of change which couldbe either trivial or devastating when change can be measured so as to control forindividual differences.The second major prospective set of criteria used in the diagnosis of PMS is theU.S. Psychiatric DSM-IIIR: A) in most menstrual cycles during the past year, symptomsin B below occurred during the week before and remitted 2 days after menstruation; B)at least five of the following symptoms have been present most of the time during eachweek before menstruation, and at least one symptom is 1, 2, 3, or 4: 1) marked moodlability, 2) anger or irritability, 3) marked anxiety, 4) markedly depressed mood orPremenstrual Syndrome128thoughts, 5) decreased interest in usual activities, work, friends, hobbies, 6) lethargy, 7)difficulty concentrating, 8) marked change in appetite, 9) hypersomnia, or insomnia, and10) other physical symptoms (breast tenderness, bloating, headache, etc.); C)disturbance which seriously interferes with work or social activities; D) not justpremenstrual exacerbation of other psychiatric problems; E) criteria A, B, C, and D areconfirmed by prospective daily self-ratings during at least two symptomatic cycles.Severino, Hurt, and Shindledecker (1989) conducted spectral density analyses ofdata from two consecutive menstrual cycles of 58 women self-referred for PMS. Theyfound evidence to corroborate the rationale behind DSM-III-R criteria for PMS. Forexample, women with confirmed PMS had at least 30% increase in symptom severity;this symptom severity increase is the one recommended by the DSM-III-R. In addition,five items were sensitive to the effect of premenstrual severity group (enjoyment, moodswings, depression, anxiety, and irritability or anger). This item sensitivity correspondswell to the DSM-III-R criterion that one of four specific symptoms (i.e., affect lability,anger, anxiety, and depression) be present.The DSM-IIIR criteria do explicitly state the use of prospective reports and in thatsense are an improvement over the NIMH criteria. The DSM-IIIR criteria do not,however, explicitly state whether all or some of the "non-premenstrual" days are to beused for comparison with premenstrual days. A second concern is the use of the term"marked" or "markedly" in describing changes. It raises again the issue of whatconstitutes significant change for each woman. The DSM-IIIR leaves decisions as towhat constitutes "marked" change up to the individual researcher or clinician, andvariability among these individuals leaves open the possibility of non-comparable groupsof women. This ambiguity could be settled by using a criterion for change based on awoman's own levels of fluctuation, e.g., the standard deviation discussed earlier.There is a contradiction in the diagnostic criteria of the DSM-IIIR. It states incriterion A that symptoms must occur in "most" menstrual cycles over the past year.Premenstrual Syndrome129The only way a researcher would know about the woman's year before coming fordiagnosis would be to ask her to recall her experiences, but such retrospective reportsare known to be inaccurate. Perhaps the writers of the DSM-IIIR meant that for a PMSdiagnosis to be accurate, the PMS must occur in "most" cycles. This presents anothercontradiction, however, because a woman might meet all other criteria for PMS in twoprospectively measured cycles but no information would be available to help adiagnostician know whether these PMS changes occur in "most" of the woman's cycles,unless she were prospectively studied for more than half her menstrual cycles in a year,a daunting prospect. In short, the DSM-IIIR does not make it clear whether diagnosisshould be based on inaccurate retrospective reports of the past year, two cycles ofprospective data, and/or prospective data collected during the majority of a woman'scycles.The following paper has been chosen to demonstrate that problems with diagnosispersist even in recent literature. Endicott and Halbreich (1988) discussed their use ofretrospective measures for diagnosing PMS, accompanied by two interviews. They didnot acknowledge that retrospective measures tend to provide information that is biasedby cultural stereotypes. They did describe daily ratings as "almost indispensable", butadded the disclaimer that changes in daily reports apply only to that particular monthand may not capture the severity of change occurring in other cycles. Readers arelikely to infer from their review article that retrospective measures provide a betterpicture of the full scope of the woman's PMS, whereas prospective daily measuresprovide only a partial glimpse. Endicott and Halbreich (1988) also contended thatreports from spouses, siblings, and roommates are"useful because a woman who is not particularly introspective may not beaware of some of the specific changes she displays or experiences. Suchreports also help judge the severity of changes and the degree to whichthey impair social functioning, work, getting along with others, and otherPremenstrual Syndrome130expected role behaviours." (p. 2)Endicott and Halbreich (1988) state that a problem that frequently causes women toseek treatment is hostility, "because it causes psychosocial difficulties with spouses,boyfriends, children and co-workers" (p. 3). Others' reports, although potentially asource of additional information, are likely to be biased by stereotypes because theytypically will be based on recall, i.e., be retrospective. Endicott and Halbreich also implythat the most objectionable "symptom" of a woman's "illness" is social conflict, and thather hostility is not warranted. Whether or not they intended to do so, they have set updiagnostic criteria that include techniques known to more accurately reflect stereotypesthan actual behaviour. These criteria may be used for the social control of womenperceived to cause trouble for their family members or others, without considering oraddressing the source of the conflict. There is one additional source of bias in howthese researchers treated their data. They reported that they typically "eyeball" the dailyratings of women, mentioning that this process can be difficult. They did not, however,consider that their own stereotypes might affect their "eyeballing" and thus theirdiagnoses. Endicott and Halbreich's (1988) chapter has been singled out not as anaberrant example of poor diagnostic procedures but rather as an example of the kindsof PMS diagnostic techniques used by many physicians and clinicians today.In a well-conducted study, Metcalf, Livesey, and Wells (1989) compared sevenmethods of diagnosing PMS: (1) the mean symptom score in the last 10 days of themenstrual cycle is at least twice that reported during the first 10 days (Muse, Cetel,Futterman, & Yen, 1984); (2) the premenstrual score (7 days prior to menstruation ishigher than the follicular score (7 days after flow stopped) (O'Brien, Craven, Selby, &Symonds, 1979); (3) the mean score in the last week of the menstrual cycle is 30% ormore above that in the week following the cessation of bleeding (Rubinow, Roy-Byrne,Hoban, Gold, & Post, 1984); (4) the mean score in the last week of the menstrual cycleis 30% or more above that in the 10 days following flow (Metcalf & Hudson, 1985); notePremenstrual Syndrome131that Metcalf et al. (1989) criticized these first four techniques because no statisticalrationale was given for the methods. The remaining techniques they evaluated were:(5) fitting menstrual data to a sine wave model and using the amplitude of the peak as ameasure of PMS. PMS can be determined statistically with this method, but the sinewave is a poor model for the asymmetric PMS patterns. In addition, this technique doesnot compensate for autocorrelation after removal of trend (Sampson & Jenner, 1977);(6) applying trend analyses to the daily reports, with a positive followed by a negativetrend as evidence of PMS. This technique allows for autocorrelation (the moodsexperienced on one day contribute, after removal of trend, to the moods experienced onsubsequent days) in the data and gives a smoothed curve that unfortunately lagsseveral days behind the actual observation of events (Magos & Studd, 1986); and (7) amodified version of Sampson and Jenner's (1977) technique in which a model of twosine waves of different lengths is fitted to the data (Metcalf et al., 1989).The third (Rubinow et al., 1989) and fourth (Metcalf & Hudson, 1985) proceduresdescribed in the preceding paragraph provide a more satisfactory compromise betweensensitivity and specificity than do the first (Muse et al., 1984) and second (O'Brien et al.,1979) techniques. Metcalf et al. (1989) concluded that all but the Muse et al. (1984)procedure produced comparable diagnoses of PMS. They also concluded that thefollowing 5 conditions should be met a PMS diagnostic technique to be satisfactory:(1) (The method) should be applicable to probabilistic cycles with a highlevel of background noise; menstrual cycles are not of constant durationand both moods and physical symptoms fluctuate from day to day. (2) (Themethod) should take into account the characteristic pattern of symptoms inwomen who suffer from PM(S); symptoms are minimal 5-14 days aftermenstrual onset and maximal in the last 5 days of the menstrual cycle. (3)PM(S) severity should be measured as a difference or increment, ratherthan as a ratio. Ratios cannot be used to analyze unipolar symptomPremenstrual Syndrome132ratings. (4) Statistically valid criteria for the recognition of PM(S) should bestated. (5) If thresholds are used to distinguish PM(S) present from PM(S)absent cycles, they should be optimized (i.e., specificity balanced withsensitivity).In summary, there likely are variations among researchers and clinicians on what is used to measure each symptom or experience but the DSM-IIIR seems to be themost recent articulation of how to diagnose PMS (APA, 1987). Improvements couldnevertheless be made to the DSM-IIIR criteria to clear up some of its ambiguities.First, the criteria could specify that diagnoses should be based on all theprospective data collected, not just a certain portion of the cycle, thus providing the mostreliable information obtainable and maximizing the strengths of longitudinal designs.Second, whenever possible, references to "marked" or "problematic" changescould be avoided and replaced with criteria based on the woman's own experience. Forexample, the use of standard deviation as a measure of change would control forindividual differences in mood fluctuations from phase to phase or day to day. Tworesearchers have used standard deviation as a measure of significant change. Watkins,Williamson, and Falkowski (1989) used the standard deviation of the control group toevaluate the clinical significance of changes observed for each woman with LLPDD.Unfortunately, this means each LLPDD woman is evaluated against a standard derivedcompletely from others' rather than her own experiences. Watkins et al. found that 5 of16 (31%) women with provisional LLPDD had significant prospectively confirmedpremenstrual increases in negative affect (by comparison with 2 of 30 (7%) controlwomen). Only 3 LLPDD women (18% of the original 16) by comparison with 1 of 30(3.3%) control women with confirmed premenstrual increases in negative affectexperienced a decrease in their symptoms following the premenstrual week. Symptomsmust remit after menstruation to be considered true LLPDD symptoms. Thus it wasinappropriate for Watkins et al. to emphasize both in the discussion and abstract of theirPremenstrual Syndrome133article the value of 31% over the 18% as the observed rate of confirmed LLPDD. Thisemphasis misrepresents the true incidence of LLPDD in this study.Schnurr (1988) used a standard deviation as the effect size necessary todistinguish significant changes in symptoms. The standard deviation was calculated oneach woman's entire cycle. Then the difference between her post menstrual andpremenstrual phase was calculated. If that difference exceeded one standard deviation,the premenstrual change was deemed significant.Schnurr (1988) listed the advantages and disadvantages of using a standarddeviation or effect size as a measure of significant mood change: (1) it provides ameasure of change that is related to the woman's own day to day experiences; (2) it isdimensionless, like a Z-score, and easily permits comparisons amongst individuals; and(3) it suggests a conceptually meaningful cut off point for determining significant moodsor changes. The disadvantage is that the standard deviation calculations would berelatively time consuming for a clinician with a patient with possible PMS.A third proposed improvement to the DSM-IIIR criteria is that rather than assumingthat a physical "symptom" is troublesome, the woman could be asked how troublesomeshe considers the symptom to be. Finally, some decision will have to be made as towhat constitutes evidence that PMS occurs in "most" cycles. Clearly the diagnosticianwill have to make a trade-off between what would be ideal, e.g., prospective collectionof data for one year, and what is practical, e.g., prospective data collection for a shorterperiod of time. Perhaps demonstration of PMS in two of three contiguous cycles wouldbe a good compromise.Treatment of PMSOne of the difficulties in evaluating PMS treatments is that temporary improvementin symptoms has been observed when women begin to complete daily diaries withoutany other treatment. It has been suggested, therefore, that treatment begin only afterthis temporary benefit has disappeared (Rubinow & Roy-Byrne, 1984). Second, there isPremenstrual Syndrome134an unusually high placebo rate, i.e., 19-88%, with a 60% mean response rate in PMSstudies (Hamilton et al., 1984), that makes demonstrating the efficacy of any PMStreatment difficult. To date, no treatment has been found to be more effective thanplacebo in double-blind placebo-controlled cross-over studies (Norris, 1987; Rubinow &Roy-Byrne, 1984). Third, to the extent that diagnostic procedures are flawed,treatments are unlikely to be shown to be effective. For example, if treatments areevaluated on women diagnosed retrospectively, the group will include some womenwho would be diagnosed prospectively as having PMS, but more who would not(Kessler et al., 1988), and inclusion of the latter could mask evidence of treatmenteffectiveness for the former group. An additional problem is that for some treatments itis difficult to conduct a truly blind trial because of obvious side effects, for example,micronized progesterone (obvious sedative side effects) and spironolactone (obviousincrease in urination) (Robinson & Garfinkel, 1990).Some investigators have interpreted apparent improvement while on treatment asevidence of the underlying cause of premenstrual syndrome. Reasoning backward fromwhat seems to work is not, however, a good way to determine cause and effect.Physicians may end up reifying a treatment and an illness without true evidence(Bignami, 1982). Speculation about why a treatment may be effective can be useful indiscovering important areas for future research into the cause of PMS, but directinferences of causation from an apparently effective treatment is misguided.Over 327 treatments for PMS have been suggested (Hamilton, Parry, Alagna,Blumenthal, & Herz, 1984). Treating women who say they have PMS with progesteroneis the most widely publicized pharmacological therapy. For example, Norris (1987)stated that Dalton's treatment of women with progesterone for the last 30 yearsencompasses "the largest uncontrolled trial in history of a single medication by onephysician for one disorder!" Dalton has never completed double-blind placebocontrolled studies of progesterone or long term epidemiological follow-up studies of sidePremenstrual Syndrome135effects (Norris, 1987). Dalton (1984) did, however, monitor 120 women for 5 years andfound no side effects. Norris (1987) studied 50 women for three years and also foundno side effects of progesterone therapy. Day (1979) and Dalton (1982a) are reluctant tocarry out controlled trials because of the desperate state of their patients. Daltonargues that it would be ethically wrong to include severe high-risk PMT (PMS) sufferersin a placebo study since she believes progesterone is an effective treatment.Norris (1987) pointed out that some researchers have inaccurately referred to"controlled" studies of progesterone therapy. For example, Glick (1985) cited Smith(1976) as providing evidence of a controlled progesterone therapy study but Smith(1976) is a review article with no data. Smith (1975) is similarly inaccurately cited.Norris (1987) described the conditions necessary for an adequate double-blind,placebo-controlled, cross-over study (of progesterone): (1) acceptable operationaldefinition of PMS, (2) homogeneous study population, (3) prospective daily assessmentthrough several cycles, (4) the same daily instrument used through placebo cycles, and(5) measurement of serum progesterone to control for different levels of absorption.There is growing evidence that progesterone is not an effective treatment. Smith,Cleghorn, Streiner, and Younglai (1975) cited negative results of a "double-blindcontrolled study" of progesterone therapy. Unfortunately, they did not report how theydiagnosed PMS, how women were selected, or in what doses the drug was given.Sampson (1979) conducted a double-blind controlled trial of progesterone, but thisstudy was flawed by the retrospective use of the MDQ to diagnose PMS. In addition,the case descriptions of the women in this study indicate that some might have beensuffering from affective disorders rather than, or as well as, PMS. A third problem is thatSampson (1979) used 200 mg and 500 mg of progesterone two times daily but Norris(1983) showed that only 62% of PMS sufferers would respond to these doses.Sampson (1981) found progesterone not to be more effective than placebo in asubsequently improved double-blind placebo controlled cross-over study.Premenstrual Syndrome136Maddocks, Hahn, Moller, and Reid (1986) studied 20 women in a double-blindplacebo-controlled cross-over study. Prospective measures of affect andsymptomatology were collected every 3 days during an 8 month period. The first andlast month were baseline data. Both progesterone and placebo trials resulted inmarginally improved PMS, but the progesterone was not significantly better thanplacebo. In another stringently controlled double-blind placebo-controlled study ofprogesterone, Maddocks (1987; DA) found that progesterone was not significantly betterthan placebo for 20 women with confirmed PMS.Finally, Sampson, Heathcote, Wordsworth, Prescott, and Hodgson (1988)conducted a double-blind cross-over study of the effect of dydrogesterone (a form ofprogesterone) versus placebo on women diagnosed has having PMS on the basis ofone month of prospective data. No difference between the hormone and placebo wereseen except that the hormone was associated with an increase in breast tendernessand decrease in pain with menstrual flow.In the United States, Norris (1983) applied to the Food and Drug Administration(FDA), with confirmed financial support of L.D. Collins, Ltd., a British manufacturer ofprogesterone suppositories, to test the efficacy of progesterone in treating PMS.Attempts by Norris (1987) to receive funding for well-designed double-blind cross-overstudies was denied by the FDA because of the failure to obtain support from aprominent physician in the PMS field.The issue of safe doses of progesterone was briefly discussed at the FDA Fertilityand Maternal Health Drugs Advisory Committee in November 1981. Withoutpresentation of data or expert opinion, the FDA committee decided that 200 mg per daywas safe even though previous research had shown this dose to be ineffective (Dalton,1977; Sampson, 1979; Norris, 1983). At a subsequent meeting (February, 1983), theFDA confirmed their previous recommendations.Other treatments also have been advocated for PMS. These include exercise,Premenstrual Syndrome137vitamin B6, and prolactin suppressant. Their efficacy has not yet been adequatelydemonstrated. For example, Baumbatt and Winston (1970) is often cited as findingevidence that vitamin B6 is an effective treatment for PMS, but, as Rubinow and Roy-Byrne (1984) pointed out, Baumbatt and Winston (1970) studied depressed womentaking oral contraceptives, not women diagnosed as having PMS. Stokes and Mendels(1972) found that vitamin B6 was not more effective than placebo in a double-blindplacebo-controlled cross-over study. Abraham and Hargrove (1980) used a similardesign; they found that luteal phase symptoms were significantly improved by vitaminB6 but the statistical analyses were flawed. At least 50 t-tests were conducted on datafrom 25 women without precautions against accruing type I error, however, so theirresults are difficult to interpret. Williams, Harris, and Dean (1985) studied 434 womenwith PMS; 82% improved in the vitamin B6 group, 70% improved in the placebo group,and this difference was significant. The patients' reports of symptoms also did not differ.Such a high placebo rate makes it difficult to determine the efficacy of vitamin B6.Malmgren, Collins, and Nilsson (1987) gave women with prospectively confirmed PMSvitamin B6 in a double-blind cross-over design. The vitamin B6 was not significantlybetter than placebo in relieving premenstrual symptoms. In sum, vitamin B6 or othertreatments may yet be shown to be effective for some women, but to date this has notbeen the case.Robinson and Garfinkel (1990) reviewed various treatments of PMS and theirefficacy. They found that no evidence exists to show that progesterone, progestins,Vitamin B complex, spironolactone (a diuretic), bromocriptine (a dopamine agonist andprolactin inhibitor), clonidine (an alpha-2 agonist which decreases amount ofnorepinephrine), and immunotherapy (injection of candida antigen) are effective inreducing PMS symptoms. Two treatments were found to be significant in reducing PMSsymptoms in preliminary studies. Mefenamic acid (prostaglandin synthetase inhibitor)was used as a PMS treatment in 15 women but found to be effective only in women withPremenstrual Syndrome138PMS that was associated with dysmenorrhea or menorrhagia (Mira, McNeil, Fraser etal., 1986; cited in Robinson & Garfinkel, 1990). More research on mefenamic acid isneeded. Some researchers (Puglakk, Makaramen, Viinikka et al., 1985; cited inRobinson & Garfinkel, 1990) have found that evening oil of primrose (a source ofgamma-linoleic acid and precursor of prostaglandin E) is significantly more effectivethan placebo in reducing PMS symptoms, but more controlled studies are needed toconfirm these results. Robinson and Garfinkel (1990) concluded that although notreatments for PMS have been effectively proven, as a last resort physicians may trysafer treatments such as mefenamic acid or oil of evening primose which have not yetbeen disproven.Some researchers have tried to relieve PMS by focusing on symptoms. No strongevidence exists demonstrating a specific benefit of well balanced diet or other dietaryapproaches for PMS symptoms. Exercise has been found to be effective (Prior &Vigna, 1987). An improvement was found in women with PMS following partial sleepdeprivation but further study is needed to confirm these results (Parry & Wehr, 1987).The findings for psychoactive drugs are mixed. Lithium carbonate was found notto be effective in double-blind studies (Steiner, Haskett, Osmun, et al., 1980, cited inRobinson & Garfinkel, 1990; Singer, Cheng, & Schori, 1974). Harrison, Endicott,Rabkin et al. (1987, cited in Robinson & Garfinkel, 1990) found alprazolam to beeffective for 34 women with PMS in a double-blind placebo conrolled study. Smith,Rinehart, Ruddock, et al. (1987, cited in Robinson & Garfinkel, 1990) also foundalprazolam to be effective in a double-blind placebo controlled, randomized multiplecross-over study.The apparent efficacy of alprazolam, or any other tranquilizers, in the treatment ofPMS, deserves thoughtful consideration. The tranquilizers act to mask, deaden, ornumb the affect of the drug taker. Emotions that are experienced as distressingtypically occur in response to an event that is distressing. This event could be eitherPremenstrual Syndrome139endogenous or external; the emotional responses are merely the messengers thatindicate how the event is experienced by the person. Halting or attenuating theseemotional reactions to an unpleasant event is the equivalent of "shooting themessenger." Not only does tranquilizer therapy wrongly focus on the message ratherthan the unpleasant event, the numbing characteristic of the drug cause the additionalproblem of obscuring the drug taker's real feelings and knowledge of their real self.Being out of touch with these real feelings may mean that a woman will stay longer in aproblem situation because she is out of touch with her anger, sadness, and her will tomake change.As an illustration, it is technically possible to treat a woman with a broken arm withtranquilizers alone. At sufficient doses the drug would help the woman detach from herpain. In fact, this woman taking tranquilizers and detached from the pain may try to useher arm or not protect it appropriately and in doing so compound the original injury. Noethical physician would choose this drug treatment, however, because it is clear that thebroken arm is the cause of the distress. The physician would set the injured arm (a verypainful procedure) and then as the woman's natural healing processes begin the painwill lessen. Similarly, if there is emotional or some other kind of injury, every attemptshould be made to locate and deal with the injurious event. Emotions never occur in avacuum, all by themselves. Emotions communicate to the self about experience. Thelogic of tranquilizer therapy is flawed, therefore (except in extreme cases when aperson's emotional responses are so extreme they may be motivated to harmthemselves or others) because the tranquilizers do not address the root problem.Tranquilizers can, however, make a woman appear more stable and easier to manage,a quality some physicians and husbands may find desirable.Finally, not only does tranquilizer therapy not treat the "dis-ease" and only thesymptom, but there also is a significant risk of drug dependency. Women who havePMS do not need the added burden of a significant risk of drug dependency. In ourPremenstrual Syndrome140view, tranquilizers such as alprazolam are not an acceptable treatment for women withPMS.A final category of treatment involves attempts to suppress physiological menstrualchanges with the use of oral contraceptives (to suppress ovulation), danazol (anattenuated androgen which suppresses the menstrual cycle), gonadotropin-releasinghormone analogues (which inhibit gonadotropin release and suppress the menstrualcycle), or estradiol implant and cyclic oral norethisterone, which suppress ovulation.With the exception of oral contraceptives, there is some evidence suggesting a minimalimprovement in PMS symptoms accompanied by unpleasant side effects (e.g.,masculinization with danazol). Lack of knowledge regarding the safety of long term useof gonadotropin-releasing hormone analogues makes the use of some of thesetreatments dubious.Robinson and Garfinkel (1990) concluded their review of PMS treatments byrecommending a conservative symptomatic approach. Foremost is an accuratediagnosis according to DSM-III-R, followed by supportive counselling, sensible diet, andexercise. Exercise and a well balanced diet are also recommended by Osofsky,Keppel, and Kuczmierczyk (1988). If necessary, Robinson and Garfinkel (1990)suggest a woman could try alprazolam daily one week prior to menstruation to relieveanxiety, irritability, and depression. Another possibility would be to try the relatively safebut as yet unproven mefenamic acid or evening oil of primrose. What is lacking inRobinson and Garfinkel's (1990) review, however, is an adequate review of thepsychological or sociological theories of PMS. They did not consider why women mayhave symptoms of anxiety, irritability, or depression. Rather than medicating women torelieve these symptoms temporarily, a more conservative approach would be to try todiscover through psychotherapy, journal keeping, and self-help groups the possiblefactors in women's lives that may be the source of their troubles.In an attempt to test the hypothesis that ovulation suppression would reducePremenstrual Syndrome141symptoms of PMS, Parry and Rausch (1988) administered two synthetic progestogens,norethisterone and medroxyprogesterone acetate (MPA) to women presenting problemsattributed to premenstrual tension. MPA was more significant than placebo in reducingbreast symptoms but neither progestogen improved psychological symptoms. Inanother study, Parry and Rausch (1988) administered a synthetic agonist analogue ofLHRH, goserelin (zoladex), which suppresses the pituitary-ovarian activity andmenstruation. Women who received goserelin experienced complete relief of physicalsymptoms and partial relief of psychological symptoms. The first ovulatory cyclefollowing administration of goseralin was associated with complete return of allsymptoms, suggesting a link between pituitary-ovarian activity and mood.Walker and Bancroft (1990) compared women on oral contraceptives (bothtriphasic and monophasic) with women using nonsteroidal methods of birth control forsymptoms of PMS. They concluded that there were more similarities than differencesamong these groups. The striking exception was that monophasic contraceptive pillusers reported less cyclical change in breast tenderness than did triphasic pill users andnon-pill users. The authors concluded that ovulation status appeared not to influencewomen's reporting. One limitation of the study was that physiological indicatois ofovulation such as presence of an LH surge and subsequent rise in blood concentrationof progesterone were not measured, that is, presence or absence of ovulation wasassumed rather than measured.Greenblatt, Teran, Barfield, and Bohler (1987) reviewed the literature on PMS andconcluded:... suppression of cyclical ovarian function by inhibition of ovulation withestrogens (implants) appears to be the most rational method yet proposedfor the alleviation of the syndrome. Quite evident is the fact that men arenot subjected to cyclic hormonal swings and accompanying psychologicalchanges. It is this very reason that prompted Professor Higgins to ask 'whyPremenstrual Syndrome142can't a woman be more like a man'.In light of the mixed evidence concerning the effectiveness of hormones for treatingPMS, and the fact that estrogens have been shown to increase risk of cervical cancer ifunopposed by progesterone, the recommendation for complete cessation of a woman'snormal reproductive cycle seems inappropriate. One wonders whether chemicalsuppression of male reproductive processes would be recommended in the same glibvein.Other psychopharmacological treatments for PMS have been tried. For example,Harrison, Endicott and Nee (1989) treated 11 women with nortriptyline (tricyclicantidepressant Aventyl, Pamelor). These women had previously shown no benefit fromplacebo or alprazolam (Xanax). Eight of the 11 women showed some improvement withnortriptyline. Unfortunately, however, there were no control or placebo procedures inthis study. Harrison, Endicott, and Nee (1990) also treated 30 women whoseprospective ratings over two menstrual cycles showed clear LLPDD. They were givenalprozolam (Xanax) in a double-blind procedure with some success. Rausch,Janowsky, Golshan, Kuhn, and Risch (1988) studied the affect of atenolol (a beta-blocker which decreases plasma renin activity and inhibits urinary excretion ofaldosterone) on women who met provisional (i.e., retrospective) DSM-III-R diagnosticcriteria for LLPDD. Atenolol was associated with limited but significant improvement inLLPDD symptoms over placebo. Another drug treatment was tried by Deicken (1988)who gave verapamil (a calcium channel blocker which interferes with entry of calciuminto cells and is used in the treatment of mania, depression, and panic disorder) to onewomen with confirmed PMS for 3 months and reported an improvement in symptoms.Eriksson, Lisjo, Sunblad, Andersson, Andersch, and Modigh (1990) gave clomipramineto five women with severe premenstrual irritability (who had previously not responded toplacebo) for five menstrual cycles. The women reported almost total relief ofpremenstrual irritability and sadness from the second cycle onward. This studyPremenstrual Syndrome143unfortunately had no controls or placebo trials.In an attempt to link (3-endorphin withdrawal to PMS, Giannini, Sullivan,Sarachene, and Loiselle (1988) studied 38 women who had "PMS-associated"symptoms. No description was given regarding the method of assessing PMSsymptoms, so it may have been done retrospectively. The women's 3-endorphin levelswere measured in the preluteal phase, and compared with established norms. Theauthors thus identified 24 women with moderate to severe decreases in (3-endorphin.Half were given clonidine (an alpha 2-agonist, which decreases amount ofnorepinephrine; note that norepinephrine rapidly increases after rapid withdrawal from(3-endorphin) for two cycles and then switched to placebo for two cycles. The other 12women received the reversed schedule. Symptoms were significantly improved withclonidine over placebo treatment. The authors concluded that clonidine was significantin relieving PMS symptoms. If PMS was indeed diagnosed retrospectively rather thanprospectively, however, this conclusion is not warranted. What was shown, at most,was that when women were prescreened for moderate to severe 13-endorphinwithdrawal in the preluteal phase, clonidine was more effective than placebo inalleviating symptoms of this withdrawal. Clonidine's effect is not surprising since it isused in alleviating symptoms of opiate withdrawal. Any connection between rate of 13-endorphin-withdrawal and PMS has yet to be shown.Nonpharmacological approaches also have been used to treat of PMS. To test thehypothesis that PMS is related to a phase advance in circadian rhythms, Parry, Berga,Mostofi, Sependa, Kripke, and Gillin (1989) gave six women with confirmed PMS twohours of bright light in the evening for seven days during the luteal phase and found thatdepression ratings were significantly reduced. Parry, Rosenthal, Tamarkin, and Wehr(1987) have suggested a link between seasonal affective disorder and PMS, andsuggested the use of both beta blockers, which inhibit melatonin, and bright lighttherapy to manage PMS. Parry and Wehr (1987) found that one night of total sleepPremenstrual Syndrome144deprivation improved premenstrual symptoms in 8 of 10 women diagnosed with PMS.Further, late-night partial sleep deprivation was more effective than early-nightdeprivation.The very high placebo response in PMS treatments is worth considering. Is itpossible that the placebo actually contains a partially and temporarily effectivetreatment? Rome (1983) and Lander (1988) have suggested that women may feelbetter in placebo trials if they believe they are being taken seriously, perhaps for the firsttime. Lander (1988) suggests that if "not being taken seriously" is one of thecontributing factors to PMS, then one would expect the traditional placebo to improvewomen's PMS. Rome (1983) contended that receiving placebo may lead to improvedself-esteem because the women asserted themselves and found the "right" treatment.An additional quality of placebo that may be a beneficial "treatment" is that the woman'spartner or coworkers may believe that since she is receiving "treatment", she is cured.Behaviour they would otherwise attribute to PMS may now be attributed to othersources (e.g., "she's angry because she was in a bad situation"). This change in others'views of the woman's behaviour may mean that she receives less negative feedbackabout her PMS and thus feels better.Several researchers advocate psychotherapy or educational interventions forPMS. Unfortunately, insufficient detail is provided in most of these studies to determinewhether adequate methodological controls (e.g., prospective diagnosis) were included.Alberts (1990; DA) analyzed the data of 42 women, using a multiple regression model,and found that only coping style (i.e., towards/away from people), expectation oftreatment effectiveness, and number of telephone contacts were significant predictors ofreduced symptoms. Byers (1988; DA) found evidence that cognitive behaviouraltherapy, as compared with education alone, significantly reduced symptoms ofdepression and dysfunctional attitudes in women with premenstrual symptoms. Weiss(1988; DA) found cognitive behavioural therapy to be effective in alleviating somePremenstrual Syndrome145symptoms of PMS. Note that all three authors focused on PM symptoms rather thanPM syndrome.Cook (1989; DA) developed a course consisting of a series of five classes on"Emotional Aspects of Premenstrual Syndrome" and found that 41 women with PMShad significantly greater positive changes than 41 control women who did not attend theclasses on: (1) reduction in depression; (2) ability to live more in the here-and-now in ameaningful manner; (3) becoming less dependent on the approval of others; (4)becoming more self-motivated and directed; (5) reduction of anxiety; (b) living less in thethere-and-then; (7) awareness of their own needs and feelings; and (8) fear whenexpressing their feelings in social situations. Young (1987; DA) found that a PMSeducation program was helpful in alleviating PMS symptoms. These results suggestthat similar educational interventions might be included in any conservative treatmentplan for women experiencing symptoms.Goodale (1990; DA) found progressive relaxation, 30 mins a day, to be moreeffective than both the equivalent time spent leisurely reading and a no-instructioncontrol. Relaxation reduced both physical and emotional PMS symptoms. Reed (1987;DA) compared the effectiveness of three therapy interventions: stress management,group therapy, and support group therapy. All three therapy interventions were useful inhelping PMS women, but no significant differences among the three therapies werefound. Slade (1989) conducted psychotherapy with four women prospectivelydiagnosed as having PMS and found significant improvement. Unfortunately, therewere no control procedures.Some writers urge women to join self-help groups to discuss PMS. Rome (1983)stated that self-help groups can be very useful because women can share informationand set up a telephone buddy system for when they are feeling desperate, depressed,and/or angry. In addition, they can discuss the aspects of their lives they would like tochange. By working together and discussing their premenstrual experiences, RomePremenstrual Syndrome146argued, women may come to understand PMS as something other than a disease.There are many self-help books available for women who believe they have PMS.Laws (1985) has criticized many of these books as being manuals on how to survive ina man's world. They tend to focus on what medications and therapies will help womenadjust to rather than change their patriarchical living conditions (Laws, 1985). Anexample of a self-help oriented book that avoids the theme that women must adjust withthe help of medications is Harrison (1984) .Some of the women interviewed by Martin (1987) spoke of an anger with noimmediate identifiable cause. That is, nothing in the immediate circumstances couldaccount for the intensity of the rage they felt. When this "unnamed anger" occurs, awoman may attribute it to something in herself and feel guilty about it. When anger isexperienced as justifiable, however, it makes people feel good. Martin (1987) suggeststhat if women got together in groups and talked about the unnamed anger, they coulddiscover the causes and convert it into productive anger.Summary and Conclusion A large body of literature on the theories, diagnosis and treatment of PMS wasreviewed. To date, biomedical models have been predominant in the generation ofhypotheses about the causes and treatment of PMS, but even after decades of workneither a biomedical cause nor an acceptable effective treatment for PMS has beenfound. Nevertheless, some researchers and clinicians continue to adhere to abiomedical model. The literature on the methodological difficulties in studying PMS alsowas reviewed and it was concluded that rigorous methodology is crucial. Continuing topublish studies based on retrospective diagnoses hampers progress in understandingPMS. The literature dealing with social, psychological, and feminist perspectives onPMS also was reviewed. These non-biomedical approaches offer a new way to look atPMS and the possibility of understanding the factors in the environment that maycontribute to PMS. These new approaches also offer new avenues of treatment andPremenstrual Syndrome147intervention (e.g., psychotherapy, self-help groups) not previously used by biomedicallyoriented professionals. Unfortunately, these newer theories need further empiricaltesting with rigorous methodology. Since the causes and treatments of PMS are still notestablished, the most responsible approach for helping women who say they have PMSis a conservative one based on treatments with the least harmful side effects. Forexample, eating a balanced diet and exercising regularly is a sensible approach foranyone and unlikely to cause harm, whereas alprazolam and other tranquilizers areaddictive and may merely cause symptoms and issues to go "underground". In additionto physical self-care, prospectively charting her experiences daily will help a womanbecome aware of her mood patterns and the troublesome issues in her life.Psychotherapy and/or self-help groups could provide a safe place to discuss herdiscoveries and resolve issues. Perhaps by taking these conservative steps to listen toand help women experiencing premenstrual syndrome now, while at the same timemaking a commitment to rigorous work in several disciplines on the causes andtreatments of PMS, premenstrual syndrome will finally be understood.It is important to distinguish amongst the various types of premenstrual and othermenstrually-related changes. Whereas some changes may indicate illness, othersreflect healthy physical and emotional changes that normally occur over the menstrualcycle. Some researchers refer to these healthy changes as molimina. Healthy changesshould not be confused with symptoms, which by definition indicate an illness ordisorder (e.g., dysmennhorea or painful menstruation). The situation is furthercomplicated by the fact that some women find changes that would appear to qualify as(illness) symptoms to be troublesome whereas others report the same degree and typeof changes but do not find them troublesome. The phrase premenstrual syndrome isrestricted to the co-occurrence of a cluster of symptoms in the premenstrual phase thatdo not occur in other phases of the menstrual cycle. In this study, few women met thiscriterion, and some of those who did, did not consider themselves to have troublesomePremenstrual Syndrome148premenstrual symptoms. Moreover, a number of women met the "syndrome" criterionof a cluster of "symptoms" in one phase that remitted in other phases, but theirsymptomatic phase was not premenstrual (e.g., post mid-cycle).The findings of this study underscore the importance of recognizing that cyclicalchanges are not necessarily abnormal or unhealthy. The common beliefs seem to bethat women's moods vary considerably over the menstrual cycle, that all women'smoods do so in the same pattern, and that this is undesirable. Implicitly linked to thesebeliefs are others, namely, that men's moods do not vary and that women's moods varymore over the menstrual cycle than other cycles (such as day of week, lunar, schoolyear, seasonal, etc.). Instead, in this study, most women experienced healthy changesover their menstrual cycle, some women experienced some changes as positive, someexperienced other changes as troublesome symptoms, and very few womenexperienced a specific cyclic pattern of changes that meet diagnostic criteria for PMS.Moreover, both women and men experienced mood changes over days of the week thatwere either negative or positive or both. Whether or not this turns out to be arepresentative sample, these results indicate that the common beliefs regarding cyclicalchanges are incorrect for some adult women and men.149ReferencesAbplanalp, J. M., Haskett, R., & Rose, R. (1980). The premenstrual syndrome.Psychiatric Clinic of North America, 3, 327.Abraham, G. E., & Hargrove, J. T. (1980). Effect of vitamin B-6 on premenstrualsymptomatology in women with premenstrual tension syndromes: A double blindcrossover study. Infertility, 3, 155-165.Ainscough, C. E. (1990). Premenstrual emotional changes: A prospective study ofsymptomatology in normal women. Journal of Psychosomatic Research, 34, 35-44.Alagna, S. W., & Hamilton, J. A. (1986). Social stimulus perception and self-evaluation:Effects of menstrual cycle phase. Psychology of Women Quarterly, 10, 327-338.Alberts, M. S. (1990). Predicting treatment success in the multi-modal treatment ofpremenstrual syndrome. Dissertation Abstracts International, 50, 5304-B.Allen, H. (1984). At the mercy of her hormones: Premenstrual tension and the law.M/F, 9, 19-44.Altman, M., Knowles, E., & Bull, H. D. (1941). A psychosomatic study of the sex cyclein women. Psychosomatic Medicine, 3, 199-225.Anderson, M. (1941). Some health aspects of putting women to work in war industries.Industrial Hygiene Foundation 7th Annual Meeting, 165-169.Ascher-Svanum, H. (1984, August). Biased recollections of menstrual discomfort.Paper presented at the meetings of the American Psychological Association,Toronto.Ascher-Svanum, H., & Miller, M. J. (1990). Premenstrual changes andpsychopathology among psychiatric inpatients. Hospital and CommunityPsychiatry, 41, 86-88.150AuBuchon, P. G., & Calhoun, K. S. (1985). Menstrual cycle symptomatology: The roleof social expectancy and experimental demand characteristics. PsychosomaticMedicine, 47, 35-45.Bains, G. K., & Slade, P. (1988). Attributional patterns, moods, and the menstrualcycle. Psychosomatic Medicine, 50(5), 469-476.Baumblatt, M. J., & Winston, F. (1970). Pyridoxine and the pill. Lancet, 1, 832-833.Beck, L. E. (1989). The relationship between hormones, mood, physical symptoms andstress in women who suffer severe premenstrual syndrome. Dissertation Abstracts International, 49, 3993-B.Benedek, T., & Rubenstein, B. B. (1939a). The correlations between ovarian activityand psychodynamic processes: I. The ovulative phase. Psychosomatic Medicine,1, 245-270.Benedek, T., & Rubenstein B. B. (1939b). The correlations between ovarian activityand psychodynamic processes: II. The menstrual phase. Psychosomatic Medicine, 1, 461-485.Bignami, G. (1982). Disease models and reductionist thinking. In S. Rose (Ed.),Against biological determinism. London: Allison and Busby.Billings, E. G. (1933). The occurrence of cyclic variations in motor activity in relation tothe menstrual cycle in the human female. Bulletin of Johns Hopkins Hospital, 54,440-454.Blechman, E. A., & Clay, C. J. (1987). The scientific method and ethical treatment ofpremenstrual complaints. In B. E. Ginsburg & B. F. Carter (Eds.), Premenstrual Syndrome: Ethical and legal implications in a biomedical perspective (pp. 223-236). New York: Plenum Press.151Blumenthal, S. J., & Nadelson, C. C. (1988). Late luteal phase dysphoric disorder(Premenstrual Syndromes): Clinical implications. Journal of Clinical Psychiatry,49(12), 469-474.Bordwick,^(1979). Anger and aggression: An essay on emotion. New York:Springer-Verlag. [Cited in Averill (1982]Boyle, C., Berkowitz, G. S., & Kelsey, J. L. (1987). Epidemiology of premenstrualsymptoms. American Public Health, 77, 349-350.Brayshaw, N. D., & Brayshaw, D. D. (1987). Premenstrual syndrome and thyroiddysfunction. Integrative Psychiatry, 5(3), 179-184.Brinton, H. P. (1943). Women in industry. In the Manual of Industrial Hygiene and Medical Service in War Industries (pp. 395-419). National Institutes of Health,Division of Industrial Hygiene. Philadelphia: W.B. Saunders.Brooks-Gunn, J. (1986). Differentiating premenstrual symptoms and syndromes.Psychosomatic Medicine, 48, 385-387.Brush, A. L. (1938). Attitudes, emotional and physical symptoms commonly associatedwith menstruation in 100 women. American Journal of Orthopsychiatry, 8, 286-301.Bruto, V. C. (1988). Premenstrual syndrome: Psychological and endocrine profiles.Dissertation Abstracts International, 49(6-B), 2371.Buckley, T., & Gottlieb, A. (1988). Blood magic. Berkeley, CA: University of CaliforniaPress.Byers, S. B. (1988). A comparison of the efficacy of education alone or in combinationwith cognitive therapy in the treatment of premenstrual syndrome. Dissertation Abstracts International, 48(12-B, Pt. 1), 3674-3675.Cassidy, C. M. (1982). Protein-energy malnutrition as a culture-bound syndrome.Culture, Medicine and Psychiatry, 6, 325-345.152Chaturvedi, S. K., & Chandra, P. S. (1990). Stress-protective functions of positiveexperiences during the premenstrual period. Stress Medicine, 6, 53-55.Chaturvedi, S. K., Santosh, K., Chandra, M. K., Prabha, S., & Issac, M. K. (1990).Premenstrual changes. Acta Psychiatrica Scandinavica, 81, 206.Chisholm, G., Jung, S. 0. J., Cumming, C. E., & Fox, E. E., & Cumming, D. C. (1990).Premenstrual anxiety and depression: Comparison of objective psychological testswith a retrospective questionnaire. Acta Psychiatricia Scadinavica, 81, 52-57.Christensen, A. P., & Oei, T. P. S. (1989). Correlates of confirmed premenstrualdysphoria. Journal of Psychosomatic Research, 33, 307-313.Christensen, A. P., Oei, T. P. S., & Calan, V. J. (1989). The relationship betweenpremenstrual dysphoria and daily ratings dimensions. Journal of Affective Disorders, 16, 127-132.Chuong, C. J., Colligan, R. C., Coulam, C. B., & Bergstralh, E. J. (1988). The MMPI asan aid in evaluating patients with premenstrual syndrome. Psychosomatics, 29(2),197-202.Clarvit, S. R. (1988). Stress and menstrual dysfunction in medical students.Psychosomatics, 29(4), 404-409.Cobb, S. (1976). Social support as a moderator of life stress. PsychosomaticMedicine, 38, 300-314.Cohen, M. G. (1987). The relationship between premenstrual distress and maritalsatisfaction. Dissertation Abstracts International, 47(11-B), 4645.Cook, G. G. (1989). The effectiveness of short term Psycho Dynamic EducationalClasses in the reduction of emotional symptoms related to premenstrual syndrome(PMS). Dissertation Abstracts International, 49, 2152-A.Coppen, A. J., & Kessel, N. (1963). Menstruation and personality. British Journal ofPsychiatry, 109, 711-721.153Coughlin, P. G. (1987). Marital satisfaction and premenstrual syndrome severity: Aninverse relationship mediated by career choice. Dissertation Abstracts International, 47(11-A), 4190.Cumming, C. E. (1988). Psychological profiles of women with premenstrual complaint.Dissertation Abstracts International, 48(10-A), 2540.Dalton, K. (1964). The Premenstrual Syndrome. Springfield, Illinois: Charles C.Thomas.Dalton, K. (1969). The Menstrual Cycle. Harmondsworth: Penguin.Dalton, K. (1977). Premenstrual syndrome and progesterone therapy. London:Williams Heinemann Medical Books.Dalton, K. (1978). Once a Month. London: Fontana Paperbacks.Dalton, K. (1979). Once a Month. Pomona, CA: Hunter House.Dalton, K. (1982a). What is the premenstrual syndrome? Journal of the Royal College of General Practitioners, 32, 717-723.Dalton, K. (1982b). Legal implications of premenstrual syndrome. World Medicine, 17,93-94.Dalton, K. (1984). Premenstrual syndrome and progesterone therapy (2nd ed.).London: William Heinemann Medical Books.Dalton, K., & Greene, R. (1953). The premenstrual syndrome. British Medical Journal,1016-1017.Dan, A., & Al-Gasseer, N. (1991, June). Perimenstrual symptoms among Bahraini women. Paper presented at the 9th National Conference of the Society ofMenstrual Cycle Research.Day, J. B. (1979). Clinical trials in the premenstrual syndrome. Current Medical Research and Opinion, 6 (Supp. 5), 40-45.Deicken, R. F. (1988). Verapamil treatment of premenstrual syndrome. Biological Psychiatry, 24(6), 689-692.154Dekker, D. L. (1986). The relationship between self-report unwanted sexualexperiences, post-traumatic stress disorder and premenstrual tensionsymptomatology. Dissertation Abstracts International, 47(6-A), 2022-2023.DeLongis, A., Tal, M., Kessler, R. C., & Haskett, R. (1986, May). Marital stress and coping among women who report premenstrual tension. Paper presented at theMid-Western Psychological Association Convention, Chicago.Dennerstein, L., Morse, C. A., & Varnavides, K. (1988). Premenstrual tension anddepression-is there a relationship? Journal of Psychosomatic Obstetrics and Gynaecology, 8(1), 45-52.Dye, L. (1991, June). Attitudes to and experience of menstruation in a West German University sample. Paper presented at the 9th National Conference of the Societyof Menstrual Cycle Research.Eagan, A. (1983, October). The selling-of premenstrual syndrome: Who profits frommaking PMS "the disease of the 80's?" MS.Endicott, J., & Halbreich, U. (1982). Retrospective report of premenstrual depressivechanges: Factors affecting confirmation by daily ratings. Psychopharmacological Bulletin, 18, 109-112.Endicott, J., & Halbreich, U. (1988). Practical problems in evaluation. In L. H. Gise, N.G. Kase, & R. L. Berkowitz (Eds.), The Premenstrual Syndrome (pp. 1-6). NewYork: Churchill Livingstone.Endicott, J., Nee, J., Cohen, J., & Halbreich, U. (1986). Premenstrual changes:Patterns and correlates of daily ratings. Journal of Affective Disord