Parental Attitudes Toward Child Rearing: Instruments, Issues, and Implications

Embed Size (px)

Citation preview

  • 8/4/2019 Parental Attitudes Toward Child Rearing: Instruments, Issues, and Implications

    1/30

    Psychological Bulletin1989. Vol. 106, No. 1,29-58

    Copyright 1989 by the American Psychological Association, Inc.0033-2909/89/S00.75

    Parental Attitudes Toward Child Rearing:Instruments, Issues, and Implications

    George W. Holden and Lee A. EdwardsUniversity of Texas

    Describes historical use of surveys to assess parents' global child-rearing attitudes and reviews thestructure and content of the 83 parent attitude questionnaires published from 1899 through 1986

    designed to quantify variations in parental attitudes and, presumably, parental behavior. Inspection

    of the surveys' psychometric properties reveals marginally acceptable levels of reliability and ques-tionable validity. One suspected source of problems with the instruments, the use of vague andambiguous items, was confirmed in a study of mothers' reactions to one survey. In addition to

    instrument errors, conceptual problems associated with assumptions about the structure of parentalattitudes and how attitudes relate to parental behavior are discussed. Alternative methods for assess-ing parental social cognitions and individual differences in parents are advocated.

    One of the oldest and most important questions in psychol-ogy concerns the role the environment plays in the developmentof an individual. At least in the opening scenes of ontogeny, par-ents are generally recognized to be the protagonists and the fam-ily to be the "primary arena" for socialization (Maccoby, 1984).Parents have frequently been implicated as principal causalagents in their children's behavioral, emotional, personality,and cognitive development. This influence is achieved through avariety of active and passive, reactive and nonreactive processes(Baumrind, 1980; Radke-Yarrow & Zahn-Waxler, 1986; Scarr& McCartney, 1983; Sears, Maccoby, & Levin, 1957; B. B.Whiting, 1980). Consequently, a basic research strategy has

    been to identify variations in parents or parental behavior andcorrelate these differences with child variables.

    The approach used to identify parental variation has oftenbeen to quantify parental attitudes toward child rearing. Reli-ance on parental attitudes as indicative of parent behavior orthe home environment developed rapidly into a major para-digm for the investigation of parent-child relations. The prolif-eration of work in this area is illustrated by the date of the firstreview on the topicover 50 years ago (Stogdill, 1936a). Thefocus on parental attitudes attracted considerable attention be-cause it was intuitively appealing, parsimonious, and expedi-tious. Individual differences in parents' global child-rearing at-titudes were assumed to reflect differences in aggregate parental

    behavior and, consequently, to result in differential child out-

    We thank J. Amis, L. Boyden, J. Leitner, S. Satterwhite, A. Seth, P.Underwood, and D. Zamora for their assistance in the library researchand with the interview study. This work was assisted by grants from theUniversity of Texas Research Institute and IBM's Project Quest. M. A.Easterbrooks, W. A. Goldberg, J. H. Langlois, J. T. Spence, W. B.Swann, R. H. Woodson, J. \bung, and the anonymous reviewers pro-

    vided constructive comments on earlier drafts. Lee Edwards is now atthe University of Maryland.

    Correspondence concerning this article should be addressed toGeorge Holden, Department of Psychology, University of Te\as, Austin.Texas 78712.

    come. Thus, when Frank (1965) could not find that maternalattitudes or other indices of the family were related to child psy-chopathology, he concluded that family variables may not havea direct etiological role (cf. Zuckerman, 1966).

    In fact, evidence has been accumulating about the many waysin which the family environment does indeed affect children'sdevelopment, both in clinical and nonclinical populations (e.g.,Alexander & Malouf, 1983; Clarke-Stewart & Apfel, 1978;Maccoby & Martin, 1983). But the study of parental attitudes

    has contributed little to those findings. As will be argued, paren-tal attitudes do not provide an adequate assessment of the fam-ily environment, for both conceptual and methodological rea-

    sons. For parental attitudes to influence children as hypothe-sized, the attitudes must be stable, be coherent, and reflectbehavior. Furthermore, accurate quantification of parental atti-tudes depends on reliable and valid methods. Though parentalchild-rearing attitudes have sometimes been determinedthrough interviews (e.g., Sears et al., 1957), they have most fre-quently been assessed through paper and pencil surveys; thisreview will be limited to those questionnaires.

    A reappraisal of parental attitudes toward child rearing, asassessed on questionnaires, is needed for at least four reasons.

    As one of the first methods developed for the study of parent-child relations, the surveys continue to be used widely today.Moreover, their use is increasing. Twenty-one new instruments

    appeared in the first half of the 1980s, the most ever publishedover a 5-year period. The question of how the questionnaireshave changed in the face of new knowledge and conceptualiza-tions of parent-child relations will be addressed. A second rea-son to reassess surveys of parental attitudes is that they havereceived insufficient critical review as a methodology. No com-prehensive appraisal of their psychometric properties or limita-tions exists. Third, given the recent surge in attention to thedomain of parental thoughts and beliefs (e.g., Goodnow, 1984;Sigel, 1985a), the placement of parental attitudes on the land-scape of parental social cognition is needed. Finally, the tacitassumptions underlying the use of parental child-rearing atti-tudes raise several important conceptual and theoretical issues

    29

  • 8/4/2019 Parental Attitudes Toward Child Rearing: Instruments, Issues, and Implications

    2/30

    30 GEORGE W. HOLDEN AND LEE A. EDWARDS

    regarding parents and parent-child relations that have gone un-addressed. These issues include the structure of the attitudes,how the attitudes relate to behavior, and how parental attitudesfit into models of parent-child relations.

    For the purpose of organizing and clarifying this research ap-

    proach, parent child-rearing attitude questionnairessubse-quently referred to as PCRAshave been loosely denned. Aswill be shown, the instruments are directed at not only parentalattitudes, but also practices, beliefs, and values. Nevertheless, acoherent set of parental self-report instruments dealing withchild rearing has been gathered together. These PCRAs possessa number of appealing attributes; their popularity as researchinstruments is no surprise. In comparison with the study of be-havior, the benefits of documenting global attitudes are thoughtto lie in the ease of assessing the target attitudes, their resistanceto being affected by situational factors, and their convenienceand utility in predicting behavior (e.g., McGuire, 1985). As pa-per and pencil instruments, they are easy to develop and distrib-ute, inexpensive for both the researchers' and subjects' time,and usually yield data in a form convenient for analysis. Fur-thermore, parental attitudes have been viewed as a useful con-

    cept because they provide a shorthand term that presumablycaptures or summarizes aggregate behavior. Their appeal is en-hanced by the paucity of economical data collection methods;

    observations and interviews offer relatively costly alternatives.

    Although developmental psychologists have been the greatest

    producers and consumers of PCRAs, psychologists from diverseareas have used the method in a variety of research endeavors.Questions in clinical, community, counseling, cross-cultural,educational, personality, school, and social psychology havebeen studied with the surveys. The utility of PCRAs for a widerange of purposes can be illustrated in their having been used to

    quantify the social environment (Breitmayer & Ramey, 1986),study interindividual family differences (Byrne, 1965), reflectthe impact of paternal employment and the social network(Cotterell, 1986), identify differences between ethnic or culturalgroups (Durrett, O'Bryant, & Pennebaker, 1975), examine pa-rental correlates of child pathology (R. V. Freeman & Grayson,1955) or cognitive functioning (Camp, Swift, & Swift, 1982),compare foster mothers with biological mothers (Paulson,

    Grossman, & Shapiro, 1974), assess the effectiveness of parenteducation programs (Radin, 1972), and screen parents at riskfor abuse (Avison, Turner, & Noh, 1986).

    Given the continuing popularity of this methodology, it is cu-rious that no critical reviews have appeared in over 20 years.

    That is not to say PCRAs have gone unchallenged. Almost sincethe inception of the method, multiple concerns have been raised(e.g., Watson, 1933). The article preceding the first appearanceof Schaefer and Bell's (1958) popular Parent Attitude ResearchInstrument (PARI) was a methodological warning on the use of

    questionnairesby the coauthor of the PARI (Bell, 1958). Themost frequent criticisms of the surveys dealt with the ambiguityof the items, susceptibility to response sets such as acquies-cence, the influence of education on responses, and the uncer-

    tain attitude-behavior link. Two reviewers were so adamantabout their censure of the PARI that they concluded their ab-stract with the words: "It has served as an important stepping-stone, but difficulties inherent in its design and structure suggest

    that it would be profitable to work toward new approaches"(Becker & Krug, 1965, p. 329).

    The goal of this article is to catalog the extant PCRAs and toassess their current state as research instruments. This ap-praisal will begin with a brief historical report on the origins ofinterest in parental attitudes. A description of the format andcontent of the questionnaires follows. Next, the reliability andvalidity of the instruments will be reviewed. The results froman interview study, designed to elicit mothers' reactions to itemsfrom a representative PCRA, are then reported. Conceptual is-sues associated with how well PCRAs fit with our knowledge ofparents and parent-child relations are then discussed, followedby proposals for future work in this area.

    Origins of Interest in Parental Attitudes

    In the early part of this century, a confluence of at least threesources implicated the role of parents in general and parentalattitudes in particular for determining child outcome. Thesesources included psychoanalysts and psychologists workingwith clinical populations, psychologists studying parents andchildren by using social psychology's new construct of attitudes,and psychologists and others initiating the parent educationmovement.

    Psychoanalytic theory provided a major impetus for the studyof parental attitudes as a key determinant of child personality.Interestingly, Freud devoted few words to parents' attitudes andtheir effect on child outcome (e.g., Freud, 1936). As one psy-chologist noted,

    the references to the significance of parent-child relationships oc-cur relatively infrequently in Freud's writings.... On the wholeFreud is not highly sensitive to variations in parental attitudes andtheir effect on a child's behavior and personality. Freud's contribu-

    tion consists in outlining some of the main dynamic factors whichoperate in the human economy, and he has left it to his successorsto discover how these apply in the varieties of human relationships.(Symonds, 1949, p. 174)

    The evolvement of Freud's ideas about the parent-child rela-tionship was left to individuals such as E. Jones (1923), Homey(1933), andRibble(!943).

    The major legacy of the psychoanalytic approach to parentalattitudes has been in the study of parental acceptance and rejec-tion, a topic addressed in over 300 studies (Rohner, 1986). Thebasic premise is that the "normal" attitude of the parent is oneof affection (Bakwin & Bakwin, 1940). If the parent's emotionalneeds have not been met at some point, however, the parent

    will then carry these personality needs into his or her parentingbehavior. These needs may then result in overprotecting or re-jecting the child, as theorized by David Levy (1931. 1943).Overprotection, or prolonging infantile care through excessivecontrol, was thought to be produced by a variety of precipita-tors such as current anxiety or previously experienced parentalrejection. In contrast to overprotection, lying at the oppositeend of the bipolar scale, is rejection. Levy's hypotheses aboutthe causes and effects of those salient styles of maternal interac-tion provoked a series of studies into overprotection and rejec-tion in the 1930s by social workers as well as clinicians (Hough,1932; Newell, 1934; Witmer, 1932). Work has continued in thisarea (e.g., Zemlick & Watson, 1953), with the most recent pro-

  • 8/4/2019 Parental Attitudes Toward Child Rearing: Instruments, Issues, and Implications

    3/30

    PARENT CHILD-REARING ATTITUDES 31

    ponent being Rohner (1975, 1986), who has developed the Pa-

    rental Acceptance-Rejection Theory.

    In addition to overprotection and rejection, attention was di-

    rected to several other "attitudinal excesses" in parents, includ-

    ing anxiety, authority, indulgence, perfectionism, permissive-

    ness, responsibility, solicitude, and strictness (Bakwin & Bak-

    win, 1940; Jackson, Klatskin, & Wilkin, 1952; Sewall, 1930;

    Symonds, 1949). One result of this work was a therapeutic focus

    on maternal attitudes (Moore, 1933; Rogers, 1939) and the de-

    velopment of "attitude therapy" designed specifically to modify

    parental attitudes (Bronner, 1936; Garrett, 1936). The child

    guidance clinics begun in the 1920s also focused treatment on

    the parent-child relationship to deal with child maladjustment

    in the home and school (Watson, 1953).

    The psychoanalytic approach had a substantial impact in

    conceptualizing parental attitudes as pervasive regulators of be-

    havior that reflected the "atmospheric conditions of the home"

    (Richards, 1926, p. 241). Parental attitudes emerged as a major

    etiology of children's behavioral maladjustments.

    We recognize more and more that we are dealing with problemenvironments and problem parents rather than problem children.. . . The parents' experiences, their attitudes and behavior, influ-ence the character and behavior of the children, who in turn carryover these attitudes into their later lives, their marital adjustmentsand in relation to their own families. Thus a vicious circle is cre-ated. (M. Field, 1940, p. 293)

    The second major influence on the study of parental attitudes

    was the advent of attitudes as a dominant concept and a quanti-

    fiable entity in social psychology by the 1930s (Allport, 1935;

    Likert, 1932;Thurstone&Chave, 1929). The construct was ap-

    propriated rapidly for the study of parents (Stogdill, 1931). Al-

    though many psychologists recognized that parental attitudes

    did not necessarily reflect behavior, attitudes were thought of

    as filters that indirectly affected parental behavior and thereby

    reflected the child's environment (Francis & Fillmore, 1934;

    Updegraff, 1939). As within the psychoanalytic framework, at-

    titudes rapidly earned a reputation for playing the prominent

    role in a child's development: "Parental attitudes must be of

    paramount importance because the very young child is exposed

    to them continually, and the attitudes themselves are relatively

    fixed and constant" (Pearson, 1931, p. 290). The study of atti-

    tudes expanded rapidly in the 1930s and 1940s with investiga-

    tions into maternal attitudes toward breastfeeding (M. Free-

    man, 1932), hyperactivity in children (Ginsburg, 1934), and sex

    education (Ackerley, 1935), for example. When researchers be-

    gan to realize that particular child-rearing events, such as wean-

    ing, were not associated with child personality development

    (e.g., Orlansky, 1949), the impact of long-term exposure to pa-

    rental attitudes gained added credence.

    Those individuals promoting parent education or research

    to aid parents formed the third source that directed attention

    toward parental attitudes (Stoddard, 1931). A program at the

    University of Iowa was begun in the early 1900s to collect and

    disseminate knowledge about child-rearing: "If research could

    improve corn and hogs it could also improve children" (R. R.

    Sears, 1975, p. 19). Thus, the parent education movement took

    form. Within this approach, parental attitudes were construed

    as beliefs about how to rear children (Fitz-Simons, 1935; Oje-

    mann, 1935a). Along these lines, Wolfenstein (1953) described

    how advances in knowledge about child rearing contributed to

    changes in parents' attitudes. Modifying parental attitudes

    through the provision of new knowledge continues to be a cen-

    tral focus of parent education today (Florin & Dokecki, 1983;

    Wulf & Bartenstein, 1980).

    Separately, the psychoanalytic, social-psychological, and par-

    ent education perspectives highlighted the importance of paren-

    tal attitudes, an ambiguous and ill-defined construct. Some-

    times two of these perspectives coalesced, as illustrated in a text-

    book for physicians: "Defects in parental attitudes may be

    dependent, primarily, on inexperience or ignorance of proper

    methods of child rearing or they may be the expression of dis-

    tortions in the parental personalities" (Bakwin & Bakwin,

    1942, p. 11). Together, these three views promoted parental atti-

    tudes as a major topic for study. One consequence of the promi-

    nence bestowed on parental attitudes was the proliferation of

    questionnaires designed to assess a variety of parental child-

    rearing attitudes. The next section reviews the structure and

    content of those instruments.

    Instruments

    Because of the large number of questionnaires developed for

    collecting data from parents, some limits have been imposed on

    the instruments included here. Although this review is focused

    on questionnaires directed at self-reports of parental attitudes

    toward their children, most surveys tap other elements of social

    cognition as well, such as beliefs and thoughts about child-rear-

    ing practices. In an effort to be comprehensive and inclusive,

    any questionnaire designed for parents with a focus on child-

    rearing attitudes or beliefs was included in this review; thus, a

    few instruments that are not ostensibly "attitude" surveys are

    included. Questionnaires were excluded if they contained fewer

    than 10 items, if the majority of items was not aimed at parental

    attitudes or beliefs, or if the survey was not directed at child

    rearing. For example, instruments designed to assess only pa-

    rental practices (e.g., Crase, Clark, & Pease, 1979; Jackson,

    1956), parental attitudes toward pregnancy (e.g., Despres,

    1937; Kumar, Robson, & Smith, 1984), children's perceptions

    of parental attitudes (Spence & Helmreich, 1978), or parental

    feelings of satisfaction (Ragozin, Basham, Crnic, Greenberg, &

    Robinson, 1982) or stress (Loyd & Abidin, 1985) were omitted.

    Similarly, instruments developed for parents of handicapped or

    exceptional children were omitted, although for a number of

    years this has been an active area of research (e.g., Love, 1970).

    With those restrictions, efforts were made to locate all of the

    parental attitude instruments that were developed through

    1986, using a computer search and handbooks of psychological

    measurement instruments (Goldman & Saunders, 1982; John-

    son, 1976; Shaw & Wright, 1967; Straus, 1969; Touliatos,

    Perlmutter, & Straus, in press). In spite of these multiple re-

    sources, a number of instruments were found that were not in-

    dexed. It is safe to assume that most of the published PCRAs

    were located, but the listing is not exhaustive.

    As those familiar with the origins of developmental psychol-

    ogy in this country may have suspected (see Cairns, 1983), the

    first PCRA was developed in collaboration with G. Stanley Hall

    (C. H. Sears, 1899). The "syllabus" consisted of a series of ques-

    (lext continues on page 35)

  • 8/4/2019 Parental Attitudes Toward Child Rearing: Instruments, Issues, and Implications

    4/30

    32 GEORGE W. HOLDEN AND LEE A. EDWARDS

    Table 1

    Chronological Listing and Description of Parent Child-Rearing Attitude Questionnaires Published from 1899 to 1986

    Instrument

    Untitled

    Unfilled

    Untitled

    Attitude TowardFreedom of Children(ATFC-I and II)

    Untitled

    Untitled

    Untitled

    Attitude Toward SelfReliance

    Attitudes TowardParental Control ofChildren

    Attitudes Toward ChildBehavior

    Parental Control AttitudeScale

    Parents' Inventory

    Parent Attitude Survey(PAS)

    Untitled

    Untitled

    Infra-Family AttitudesSurvey

    Parental OpinionInventory

    Untitled

    ZAR Pregnancy AttitudeScale

    Parental AcceptanceScale

    Child Rearing AttitudeScale

    Traditional FamilyIdeology Scale (TFI)

    Untitled

    Study

    C. H. Sears,1899

    Laws, 1927

    Stogdill, 1931

    Kochetal., 1934

    Stogdill, 1934

    Ackerley, 1935

    Ojemann, 1935a

    Ojemann, 1935b

    Stogdill, 1936b

    Stogdill, 1936b

    Stott, 1940

    Radke, 1946

    Shoben, 1949

    Harris et al..1950

    Roy, 1950

    Itkin, 1952

    Shapiro, 1952

    Mark, 1953

    Zemlick &Watson, 1953

    Porter, 1954

    J. Block, 1955

    Levinson &Huffman,1955

    Wiley, 1955

    Content/subscales

    Views on child punishment

    Physical, Social, andEmotional health

    Undesirable child behavior

    Amount of independenceand autonomy in child

    Effect of parental behavioron child

    24 subscales (e.g., Use offear for child control;Providing sexinformation)

    Developmental importanceof parent or child actions

    Self-reliance in children

    Approval of control and/orfreedom

    Various child behaviors

    Control of children'sactivities

    Authority and Discipline

    Dominating, Ignoring,Possessive,Miscellaneous

    Authoritarian, Permissive,Rigidity, and 3 othersubscales

    Authoritarian vs.Permissive

    Acceptance, Dominance,and View of own child

    Restrictive, Mixed,Analytical, Neutral

    Control, Warmth,Objectivity

    Rejection of pregnancy andmotherhood

    Acceptance

    Restrictiveness vs.Permissiveness

    Democratic vs. autocraticorientation

    8 subscales (e.g., HomePractices, Hostility, Oral

    No. of items

    Unspecified

    346 in 4parts

    70

    Two 33-itemforms

    60

    367

    319

    5 land 52(preschoolform); 37(elemen-taryform); 25(highschoolform);

    70

    99

    30

    127

    85

    (a) 36(b)35

    (c) 10

    Unspecified

    1 3 1

    40

    139

    40

    20

    20

    40

    160

    Response scale

    Unspecified

    Multiple responseformats

    10-pt. seriousnessorundesirability

    3-pt. agree/disagree andomit

    10-pt. seriousness

    Multiple responseformats

    3-pt. importance

    Fill inappropriate age

    7-pt. agree/disagree

    7-pt. agree/disagree

    4-pt. agree/disagree

    3-pt. frequency orintensity and 2open-endedquestions

    4-pt. agree/disagree

    (a) True/false(b) 6-pt.

    frequency(c) 5 or 6

    alternativemultiple choice

    1 1-pt. control/freedom

    5-pt. Likert scale

    5-pt. agree/disagree

    4-pt. agree/disagree

    6-pt. frequency

    5 alternativemultiple choice

    4-pt. agree/disagree

    7-pt. agree/disagree

    5-pt. agree/disagree

    Reliabilityinformation"

    None

    None

    None

    2

    None

    2

    1,2

    2

    1,2

    2

    2

    None

    2

    None

    None

    2

    None

    None

    2

    2

    None

    1,2

    2

    Activities)

  • 8/4/2019 Parental Attitudes Toward Child Rearing: Instruments, Issues, and Implications

    5/30

    PARENT CHILD-REARING ATTITUDES 33

    Table 1 (continued)

    Instrument

    Parent AttitudeInventory (PAI)

    Parent Attitude ResearchInstrument (PARI-IV)

    Maternal Attitudes Scales

    Stanford Parent AttitudeQuestionnaire

    Parent Attitude Survey

    Maternal AttitudeInventory

    Child Rearing Practices

    Report (CRPR)

    Manifest Rejection Index(MR)

    Untitled

    Maryland ParentAttitude Survey(MPAS)

    Untitled

    Untitled

    Sex-Role Attitude Test(SRAT)

    Parental RoleQuestionnaire (PRQ)

    Family Problems ScaleIII

    Maternal Attitude Scale(MAS-Form DD)

    Family AttitudesMeasure (FAM)

    Use of Parental Authority

    Untitled

    Parent Attitude Inquiry(PAD

    Child Rearing PracticesQuestionnaire (CRPQ)

    Child Behavior Inventory

    Study

    Leton, 1958

    Schaefer & Bell,1958

    Gildea et al.,1961

    Winder & Rau,1962

    Hereford, 1963

    Pitfield &Oppenheim,1964

    J. H. Block,

    1965

    Hurley, 1965

    R. R. Sears,1965

    Pumroy, 1966

    Rothbart &Maccoby,1966

    Anders &Dayan, 1967

    Lansky, 1967

    Emmerich, 1969

    Ernhart &Loevinger,1969

    Cohler et al.,1970

    Delhees et al.,1970

    T. Gordon, 1970

    Gutelius, 1970

    Baumrind, 1971

    Dielman et al.,1971

    Hurley & Hohn,1971

    Content/subscales

    Unspecified

    23 subscales (e.g..Strictness, MaritalConflict)

    5 subscales (e.g.,Responsibility,Rejection)

    27 scales (e.g., Affection,Restrictiveness,Reasoning, Self-Esteem)

    Confidence, Causation,Acceptance, Trust,Understanding

    10 subscales (e.g.,Overprotection,Rejection)

    2 1 scales (e.g., Worry,

    Control, Express Affect)

    Acceptance vs. Rejection

    Permissiveness andPunitiveness

    Disciplinarian, Rejecting,Protective, Indulgent

    Attitudes about existingand desired sexdifferences

    Permissive vs. Restrictive

    Same and cross-sexbehavior in children

    Values, Means-EndsBeliefs, Means-EndsCapacities, Goals

    Authoritarianism (AFI)and 4 other clusters

    Assesses adaptive attitudeswith 1 3 scales (e.g.,Reciprocal Exchange)

    24 attitudes assessed (e.g..Assertiveness,Protectiveness)

    Control, Punishments, andReward

    Orientation towarddiscipline

    Authoritarianism, earlymaturity demands, and 7other clusters

    16 factors (e.g., Strictdiscipline, lack of self-confidence)

    Overprotection, ManifestRejection, AchievementPressure

    No. of items

    1 1 8

    1 1 5

    17

    491

    75

    50

    91

    30

    79

    90 paireditems

    (a)40,(b)Unspecified

    45

    50

    140

    209 paireditems

    233

    96

    40

    12

    59

    101

    179

    Response scale

    Unspecified

    4-pt. agree/disagree

    4-pt. agree/disagree

    4-pt. agree/disagree

    5-pt. agree/disagree

    5-pt agree/disagree

    Q sort into 7-pt.

    most/leastdescriptive

    5-pt. agree/disagree

    5-pt agree/disagree

    Forced choice and5 miscellaneousitems

    (a) Multiplechoice, (b) 5-pt.importance

    Unspecified

    Pairedcomparisons

    and multiplechoice

    3 different Likertscales

    Forced choice

    6-pt. agree/disagree

    in 4 formats (e.g.,paired wordchoice, learningassociationtask)

    3-pt. likely/unlikely/uncertain

    3-5 alternativemultiplechoice and 1open-ended

    Unspecified

    Unspecified

    5-pt. agree/disagree

    Reliabilityinformation'

    None

    1,2

    1

    2

    2

    1

    1

    None

    2

    1,2

    None

    None

    None

    1

    2

    2

    None

    1

    None

    2

    None

    1

    (table continues)

  • 8/4/2019 Parental Attitudes Toward Child Rearing: Instruments, Issues, and Implications

    6/30

    34

    Table 1 (continued)

    Instrument

    Parent Attitude Scale(PAT)

    Parent as a TeacherInventory (PAAT)

    Parent-Attitude Survey

    Parent Trait Survey

    Untitled

    Untitled

    Attitudes TowardFathering Scale

    Downstate ChildbearingQuestionnaire

    Michigan ScreeningProfile of Parenting(MSPP)

    Untitled

    Parental Acceptance-RejectionQuestionnaire (PARQ)

    Untitled

    Untitled

    Maternal DevelopmentalExpections andChildbearing AttitudesSurvey (MDECAS)

    Child-Rearing Sex-RoleAttitude Scale(CRSRAS)

    Untitled

    Prenatal MaternalQuestionnaire

    Modified Child RearingPractices Report

    NC-l58Q-SortInventory of Parenting

    Behaviors

    Ideal Child Scale b

    Parenting DiscrepancyScale" (PDS)

    Parental AttitudesToward Childrearing(PACR)

    Untitled

    GEORGE W. HOLDEN AND LEE A. EDWARDS

    Study

    Li, 1973

    Strom &Johnson, 1974

    Holtzmanetal.,1975

    Holtzman et al.,1975

    Atkinson &Endsley, 1976

    Wagaw, 1976

    Bigner, 1977

    Gaines et al.,1978

    Heifer etal.,1978

    Hitchcock, 1978

    Rohner et al.,1978

    Endsley et al.,1979

    Falender&Mehrabian,1980

    T. M. Field etal.,1980

    Burge, 1981

    Sheintuch &Lewin, 1981

    Crockenberg &Smith, 1982

    Rickel &Biasatti, 1982

    Lawtonetal.,1983

    Paguio, 1983

    DeSalvo &Zurcher, 1984

    Easterbrooks &Goldberg,1984

    Hock etal., 1984

    Content/subscales

    Dominance, Obedience,Communication

    Creativity, Control,Frustration, Teaching-Learning, Play

    Determinism, Acceptance,Confidence, andDiscipline

    Values for child

    Attitudes toward sex-typedactions

    19 subscales (e.g., Weaning,Sibling Relations)

    Traditional vs.developmental inorientation

    Child abuse proneness

    Coping, Emotional Needs;Expectations; and 2other subscales

    Achievement, Obedience,Responsibility Training

    Warmth, Rejection,Aggression, Neglect

    2 components ofauthoritarianism

    Pleasure, Arousal,Dominance

    (a) Knowledge ofmilestones, (b) Attitudestoward feeding, teaching.and discipline

    Traditional vs.Nontraditional

    Permissive vs. Restrictive,Directive

    Responsiveness andFlexibility

    Restrictiveness,Nurturance

    Physical, Intellectual,Social, and Emotional

    Development

    Factors of Adjustment,Conforming, Creative,Negativistic

    Actual and ideal parentingattitudes, skills, andpractices

    Encourage Independence,Strictness, Warmth, andAggravation

    Exclusive maternal care,employment orientation

    No. of items

    20

    50

    68

    15

    14

    (a)174,(b)84

    36

    42

    50

    30

    60

    16

    46

    (a) 8, (b) 13

    28

    40

    34

    40

    72

    66

    20

    51

    10

    Response scale

    4-pt. agree/disagree

    4-pt agree/disagree

    5-pt. agree/disagree

    Ranking

    7-pt. like/dislikeand 7-pt.importance

    (a) 5-pt. Likert-type, (b) Forcedchoice

    5-pt. agree/disagree

    6-pt. Likert-type

    7-pt. Likert-type

    4-pt. agreement

    4-pt. "almostnever" to

    "almost alwaystrue of me"

    5-pt. agree/disagree

    9-pt. agree/disagree

    (a) Fill in theblank (b) 2alternativeforced choice

    5-pt. agree/disagree

    4-pt. agree/disagree

    4-pt. Likert&2alternativeforced choice

    6-pt. "descriptiveof me"

    Q sort into 9categories of

    perceptions ofactual practicesand idealpractices

    6-pt. encourage/discourage

    Unspecified

    6-pt. agree/disagree

    5-pt. agree/disagree

    Reliabilityinformation'

    None

    2

    None

    None

    1,2

    1

    1

    2

    1

    None

    2

    1

    2

    None

    2

    None

    None

    2

    None

    None

    2

    2

    2

  • 8/4/2019 Parental Attitudes Toward Child Rearing: Instruments, Issues, and Implications

    7/30

    PARENT CHILD-REARING ATTITUDES 35

    Table 1 (continued)

    Instrument

    Untitled

    Role of the Father(ROFQ)

    Maternal Expectations,Altitudes, and Beliefs(MEABI)

    Attitudes TowardTraditional Parentaland Values

    Ideas about parenting

    Parent Attitude TowardChild EmotionalExpressiveness Scale(PACES)

    Concepts ofDevelopmentQuestionnaire(CODQ)

    Parental ModernityInventory (PM)

    Untitled

    Role Disposition

    QuestionnaireBlack Parental Attitudes

    Scale"

    Parental Locus ofControl Scale (PLOC)

    Parental BeliefsQuestionnaire

    Little Parenting StylesScale

    Family Beliefs Inventory(FBI)

    Study

    Keller etal.,1984

    Palkovitz, 1984

    Rickard et al..1984

    Sacks &Donnenfeld,1984

    Cowan etal.,1985

    Saarni, 1985

    Sameroff& Feil,1985

    Schaefer&Edgerton,1985

    Segal, 1985

    Segal, 1985

    Branch &Newcombe,1986

    Campisetal.,1986

    Elias & Ubriaco,1986

    Little, 1986

    Roehling&Robin, 1986

    Content/subscales No. of items

    (a) Attitudes, (b) Beliefs, (c) (a) 7, (b) 10,Expectations, (d) Sources (c) 1 2, (d)of Information 9

    Importance of father for 1 5child

    9 scales (e.g.. Need for 67Approval, Reaction toDeviant Behavior)

    (a) Traditional parenting, (a) 1 7, (b) 7(b) values for child andself

    Authoritarian Control, 56Child-Cemeredness,

    Permissive-Protective

    Degree of control toward 20child's emotionalexpressive behavior

    Categorical and 20perspectivistic thinking

    Progressive and tradit ional 30factors

    Values in 6 areas (e.g., 30obedience, school)

    Parent-as-teacher 24

    Racial-Social Awareness in 75Child Rearing and 5other subscales

    Efficacy, Child Control, 47Parent Control, Belief inFate, Responsibility

    Importance of various Unspecifiedinfluences on child'ssocial competence

    6 parenting styles (e.g., 53Rejecting,Overprotective,Overinduigent)

    6 "distorted" beliefs: 60Ruination,

    Perfectionism, Approval,Obedience, Self-Blame,and Malicious Intent

    Response scale

    (a) Yes/no forcedchoice, (b)Open-ended,

    (c) Open-ended, (d) Yes/no forcedchoice

    5-pt. agree/disagree

    7-pt. agree/disagree

    (a)Likerttype, (b)4-pt. "believein and wantchild to believein"

    9-pt. agree/disagree own

    and partner'sview

    4 alternativemultiple choice

    4-pt. agree/disagree (alsoin Q-sortformat)

    5-pt. agree/disagree

    6-category Q sort

    5-pt. agree/

    disagreeLikert-type

    5-pt. agree/disagree

    5-pt. influence

    6-pt. agree/disagree

    7-pt. agree/disagree

    Reliabilityinformation*

    None

    None

    1,2

    None

    2

    None

    2

    1,2

    None

    None

    1

    2

    None

    1,2

    2

    Note- The maternal form of the instrument is described if it differs from the paternal form.a I - Test-retest data are available; 2 = internal consistency data are available.b The author of the scale was someone other than those cited.

    tions designed to disclose the opinions of adults concerning the ob-

    ject of punishment, the types of child behaviors that should be pun-

    ished, and the types of punishment to be used. Since then and

    through 1986, 82 other parental attitude questionnaires have ap-

    peared in English-language journals or books. As will be seen, a

    number of questionnaires sample from distinct domains and

    thereby justify their existence. However, many instruments dupli-

    cate earlier efforts. A summary of the characteristics of the instru-

    ments and their content follows. Each instrument, along with a de-

    scription of its format and content, can be found in Table 1.

  • 8/4/2019 Parental Attitudes Toward Child Rearing: Instruments, Issues, and Implications

    8/30

    36 GEORGE W. HOLDEN AND LEE A. EDWARDS

    As determined by the stated or inferred rationale, the major-

    ity of the instruments (57%) were developed for descriptive

    purposes, such as to identify attitudinal differences across cul-

    tures (Holtzman, Diaz-Guerrero, & Swartz, 1975; Keller, Mi-

    randa, & Gauda, 1984) or to discover fathers' attitudes about

    child rearing (Signer, 1977; Palkovitz, 1984). The next most

    common rationale involved theory testing (24% of the PCRAs).

    Instruments have been developed within the perspective of psy-

    choanalytic theory and its derivatives (e.g., Cohler, Weiss, &

    Grunebaum, 1970), the authoritarian personality (Levinson &

    Huffman, 1955), and family systems (Roehling & Robin, 1986).

    Two other rationales have been provided for the development of

    instruments: the need for a clinical or applied tool (11%), such

    as a device to identify individuals at risk for child abuse (Gaines,

    Sandgrund, Green, & Power, 1978), and the desire for method-

    ological improvements on previous surveys (9%). For example,

    one instrument (Pumroy, 1966) was designed explicitly to avoid

    contamination from social desirability effects, or the tendency

    to select a response solely because it gave a favorable picture of

    oneself.

    The questionnaires also differed greatly in how they were de-

    veloped. Most instrument developers provide little information

    about the origins of the items, though some items were based

    on empirical observations or behavioral reports (J. H. Block,

    1965; Porter, 1954), developed from the Thurstone approach

    (Ackerley, 1935; Koch, Dentler, Dysart, & Streit, 1934), derived

    from a series of studies involving item analyses (Schaefer &

    Bell, 1958), created on the basis of previously developed inter-

    views or interview data (Dielman, Cattell, Lepper, & Rhoades,

    1971; Keller et al., 1984; R. R. Sears, 1965; Winder & Rau,

    1962), or borrowed from previously developed instruments

    (Cowan et al., 1985; Easterbrooks & Goldberg, 1984; Harris,

    Gough, & Martin, 1950; Holtzman et al., 1975; Hurley &

    Hohn, 1971;Leton, 1958; Pumroy, 1966;Stott, 1940). Despite

    their varied origins, the instruments often manifest a similar

    appearance.

    Structure

    Four basic ingredients of a questionnaire include the format

    of the question, the number and phrasing of the items, and the

    type of response elicited. Although a variety of question formats

    have been tried, including word associations (Laws, 1927),

    paired comparisons (Pumroy, 1966), fill-in the blanks (Oje-

    mann, I935b), a memory test (Delhees, Cattell, & Sweney,

    1970), and short vignettes (Ackerley, 1935), these efforts repre-

    sent exceptions. The great majority of surveys use single sen-

    tence phrases depicting something about a parent or child to

    which the parent then makes a judgment.

    There is considerable variability concerning the number of

    items on the surveys, with a range of 10 to 491 items (median =

    50 items; see Table 1). Less variance is found on the important

    design decision concerning how the items should be worded.

    The surveys developed in the 1930s and 1940s typically re-

    flected either the influence of Thurstone's (Thurstone & Chave,

    1929) or Likert's (1932) views on assessing attitudes. However,

    in the late 1940s, Edward Shoben (1949) became the authorita-

    tive reference. After considering various formats, Shoben opted

    for short statements in the third-person format for his Parent

    Attitude Survey (PAS). The third-person format was selected

    because it avoided potential bias due to "confessions" of par-

    ents (which, he feared, could affect negatively the clinical rela-

    tionship). Other instrument developers have concurred about

    the need to "avoid arousing self-conscious and defensive ten-

    dencies" (Loevinger & Sweet, 1961, p. 110). About three-quar-

    ters of the instruments have been written in the third person,

    although in recent years more instruments are being worded in

    the first person, following the lead of Radke (1946). As will be

    seen later, the decision to word items in the third person has

    contributed greatly to interpretation problems.

    Shoben made a second influential decision about the phras-

    ing of items. Rather than presenting specific situations, which

    he feared would only result in the data becoming "stimulus

    bound," he worded his questionnaire in the form of cliches ("A

    child should be seen and not heard"), rationalizations ("A child

    should have strict discipline in order to develop a fine, strong

    character"), and truisms ("Children should have lots of paren-

    tal supervision"). The majority of PCRAs reflect that level of

    generality. In stark contrast are a few surveys that clearly specify

    the behavior in question and the situation in which the behavior

    occurred. Such context-specific items usually appear in the

    form of a series of sentences organized into a short vignette.

    Ackerley (1935) was the first to do this; since then others have

    followed suit (e.g., Atkinson & Endsley, 1976; Gutelius, 1970;

    Harris etal., 1950).

    The last major instrument design decision concerns the type

    of response format to use. Preference for Likert-type response

    scales is strong; 68% of the surveys use that format exclusively,

    probably because of the ease of administration and analysis.

    Of those PCRAs using the Likert format, all but 20% of the

    instruments have agreement/disagreement response scales. In

    contrast to the popularity of the agreement ratings, no accord

    exists on the number of response points to use. Response scales

    range from 3 to 11 points (median and mode = 5). Likert-type

    scales continue to be the most popular response format for in-

    clusion in new surveys or as a way to modify older instruments

    with other formats (e.g., Rickel & Biasatti, 1982).

    A few other response formats were used, including Q sorts,

    multiple choice, ranking, fill-in the blanks, forced choice, or a

    mixture of formats. The Q-sort approach, as a method to con-

    trol for social desirability effects, has been appearing more fre-

    quently over the past 5 years (Lawton et al., 1983; Sameroff &

    Feil, 1985; Segal, 1985).

    Content

    The content of the instruments will be reviewed in two ways.

    First, the types of subscales found in PCRAs, as labeled by the

    developers of the instruments, will be summarized. Then, the

    particular items that compose the subscales will be examined.

    Subscales

    The PCRAs tap a vast domain of parental attitudes; over 140

    topics are sampled by the 83 instruments. The number of sub-

    scales per instrument varied widely (median = 3, range = 1-

    27), though almost half of the instruments assessed just one or

    two attitudes. The subscales can be divided into three categories

  • 8/4/2019 Parental Attitudes Toward Child Rearing: Instruments, Issues, and Implications

    9/30

    PARENT CHILD-REARING ATTITUDES 37

    Table 2

    Definitions of Domains in Parent Social Cognition Commonly Assessed by Parent

    Child-Rearing Attitude Questionnaires With Examples of Items

    Domain Definition Examples

    Attitudes

    Behavioral intentions

    Beliefs

    Self-perceptions

    Values

    An individual's predisposition, reaction to, or affectiveevaluation of the supposed facts about an objector situation are labeled attitudes. Thus, attitudesare a function of beliefs.

    Expressing general orientations toward behavioralpractices, or contingencies, or the willingness toact in harmony with one's beliefs are labeled asbehavioral intentions.

    Ideas about facts or expectancies that could be, inprinciple, proved or disproved are classified asbeliefs. These are synonymous with knowledge,judgments, conceptions, and sometimesperceptions. Two classes of parental beliefs aredifferentiated: descriptive and instrumental.

    Descriptive: Items that concern expectations aboutchildren, characteristics of boys and girls,developmental timetables, or other ideas aboutchildren or parents in general are descriptivebeliefs. Stereotypes would fall under this category.

    Instrumental: Items that assess the repertoire ofideas about how to achieve given goals areinstrumental beliefs. These beliefs are basedprincipally on experience with children and oftenconcern one's own parenting.

    Items assessing parents' own reactions or feelingsabout parenting or their children are labeled self-perceptions.

    Items that assess abstract goals or a coherent set ofattitudes to a class of stimuli are labeled values.These include long-term goals.

    A good mother should shelter her child from life'slittle difficulties."

    I put the wishes of my mate before the wishes ofmychild. b

    Taking care of baby is much more work thanpleasure.'

    When I am angry with my child, I let him knowit. b

    Most children are toilet trained by 15 months ofage."

    When a child is born he (she) already has apersonality of his (her) own."

    You must always keep tight hold of baby duringhis bath for in a careless moment he mightslip."

    I believe physical punishment to be the best wayto discipline."

    Holding and caressing a baby when he (she) criesis good for him (her).'

    I find some of my greatest satisfactions in mychild. 6

    It is a terribly frustrating task to care for anewborn infant, because he (she) can't let youknow what he (she) needs.'

    Loyalty to parents comes before anything else.*I encourage my child to be independent of me. b

    Children should be raised so that everyone in theneighborhood feels they are good children.'

    jVo/f. Superscripts identify the parent child-rearing attitude questionnaire from which the items were taken: a Parent Attitude Research Instrument(Schaefer & Bell, 1958); b Child Rearing Practices Report (J. H. Block, 1965); c Maternal Attitude Survey (Cohler et al., 1970).

    concerning parents and their children. About 80% of the sub-

    scales focus predominantly on the orientations toward,

    thoughts about, or consequences of child-rearing practices. The

    subscales in this category range from acceptance and aggrava-

    tion to warmth and weaning. The second category, comprising

    about 17% of the subscales, concerns parents' views about chil-

    dren in genera] or their own children. Examples of subscales

    concerning child characteristics or behavior include depen-

    dency, emotional health, fearfulness, and trustfulness. The

    third category is encountered infrequently on the surveys, but

    it deals with marital relations and includes subscales such as

    marital conflict and husbands' inconsiderateness (see Table 1).

    Despite the range of topics, certain subscales reappeared fre-

    quently. The most common subscales concern concepts related

    to how to control a young child. Subscales on about 20 instru-

    ments assess parental orientation to control, discipline, or pun-

    ishment (e.g., "The most important thing children should learn

    is obedience to their parents"; Gildea, Glidewell, & Kantor,

    1961); another 20 instruments contain similar subscales but are

    labeled as identifying the controlling style of the parent in terms

    of authoritarian or permissive parenting ("Parents should pun-

    ish small children when they use naughty words"; Ernhart &

    Loevinger, 1969). More than 10 different instruments contain

    subscales designed to reveal parents' orientation toward accep-

    tance or rejection of children; subscales assessing attitudes to-

    ward sex roles, sex differences, or sexuality in children havebeen equally popular ("Only male children should be spanked";

    Burge, 1981). Other subscales that have appeared in at least six

    instruments include parental overinvolvement or overprotec-

    tion and parental attitudes toward the independence of children

    ("Children are being allowed too much freedom"; Koch et al.,

    1934).

    Given that the same construct is often measured by different

    instruments, it should be noted that the content of the subscales

    can differ dramatically. In the case of acceptance and rejection,

    there is little agreement about the nature of the construct or

    how it should be assessed. Some instruments attempt to tap pa-

    rental acceptance or rejection by focusing on tangential paren-

  • 8/4/2019 Parental Attitudes Toward Child Rearing: Instruments, Issues, and Implications

    10/30

    38 GEORGE W. HOLDEN AND LEE A. EDWARDS

    tal beliefs about children in general, such as Hereford's (1963)

    item, "Children should be toilet trained at the earliest possible

    time." Other instruments try to assess the same domain more

    directly, by presenting items concerning parental self-percep-

    tions ("Looking after children really demands too much of me";

    Pitfield & Oppenheim, 1964) or having parents report on their

    behavior ("I am harsh with my child"; Rohner, Saavedra, &

    Granum, 1978).

    As is evident, the preponderance of subscales concern social

    development or social interactions. Though some subscales deal

    with emotional development or issues related to affect, very few

    address parental attitudes or beliefs concerning cognitive devel-

    opment (e.g., parental views on teaching or creativity in chil-

    dren). It is true that PCRAs have addressed an impressive num-

    ber of topics in the domain of social development; unfortu-

    nately, the focus of many subscales has tended to be on

    problems or negative topics such as parental discipline or ag-

    gressiveness in children. PCRAs have not achieved a balanced

    sampling of the domains of parental attitudes.

    Items

    When the content of the items is examined, it is clear that

    the label "child-rearing attitude survey" is more nominal than

    accurate; attitude questions constitute only a portion of the

    items. Many items are directed toward beliefs, behavioral inten-

    tions, or self-perceptions. Though most instrument developers

    did not intentionally mingle the items assessing different cogni-

    tive domains, some researchers did. Strom and Slaughter (1978)

    included an amalgam of items because parents' "emotional and

    behavioral responses to their own parent-child relationship are

    a combination of present parenting experiences, value-laden

    expectations, and beliefs regarding child behavior" (p. 45).

    In an effort to clarify those cognitive domains that have been

    assessed, definitions of "attitude" and related concepts will fol-

    low, with the goal of establishing a standardized lexicon in this

    area. Although there is considerable dispute about what consti-

    tutes an attitude and how to define it, the reader is referred else-

    where for a discussion of those issues (e.g., Allport, 1935;

    Oskamp, 1977; McGuire, 1985;Symonds, 1927). The first dis-

    tinction to be made, a common one in social psychology, is that

    attitudes can be subdivided into three areas: a cognitive or belief

    component, an affective or evaluative part, and a behavioral in-

    tention element (e.g., Ajzen & Fishbein, 1980; Oskamp, 1977).

    Two additional distinctions will be made involving values and

    self-perceptions. To clarify meanings and avoid the terminolog-

    ical confusion that is rampant in the area of parental social cog-

    nition, definitions of the different social cognitive elements are

    provided in Table 2, along with examples from three of the most

    frequently used PCRAs.

    To summarize Table 2, the term "attitude," under the tripar-

    tite definition, is generally recognized to refer to the affective

    evaluation of the favorableness or unfavorableness of an atti-

    tude object. Although unresolved issues remain with this divi-

    sion (e.g., Dawes & Smith, 1985), it will be adopted for heuristic

    purposes. "Beliefs" refers to the cognitive component of an atti-

    tude consisting of knowledge or ideas that are accepted as truth

    (Antill, 1987;Sigel, 1985b). Beliefs can be subdivided in various

    ways, but the differentiation by Stolz (1967) into descriptive and

    instrumental beliefs is especially appropriate for parents. De-

    scriptive beliefs refer to those about children or parents in gen-

    eral, whereas instrumental beliefs concern how one achieves

    particular goals. Parents hold a large number of beliefs concern-

    ing their children: Stolz (1967) identified nearly 5,000 beliefs

    based on her interviews of 78 parents. The behavioral compo-

    nent consists of "behavioral intentions," or the intention to re-

    spond in a particular way. Although some instruments are sup-

    posedly directed at "practices" (e.g., J. H. Block, 1965; Diel-

    man et al., 1971), those items are typically too vague to be

    assessments of practices. To collect self-reports of parental

    practices, an instrument would have to include items directed

    at identifying the frequency of which a particular behavior oc-

    curred over a particular duration; none do so. Consequently,

    items directed at so-called parental practices are actually assess-

    ing generalized behavioral intentions.

    In addition to beliefs, affective evaluations (which will be la-

    beled attitudes for simplicity's sake throughout this review),

    and behavioral intentions, two other elements of social cogni-

    tion are commonly found in PCRAs. Values, sometimes used

    synonymously with attitudes (e.g., J. W. M. Whiting & Child,

    1953), provide a useful distinction as a superordinate category

    (Kohn, 1977; Rheingold, 1973; Rokeach, 1979; Stolz, 1967).

    Unlike attitudes, values transcend the specifics of objects or sit-

    uations (e.g., independence in a child), represent abstract goals

    (e.g., to lead a fulfilling life), or reflect a coherent set of attitudes

    to a class of stimuli (e.g., respect one's parents). The final term

    concerns parental "self-perceptions." These perceptions consist

    of parents' beliefs about their own parenting ability, reactions

    to the parenting role, or views about their relationships with

    their children.

    The preponderance of items in most PCRAs concern descrip-

    tive and instrumental beliefs. Next most often appear queries

    of attitudes and self-perceptions. Items directed at values and

    behavioral intentions are included less frequently. Depending

    on the instrument, many surveys as well as subscales include an

    undefined mixture of types of items. Relatively few question-

    naires sample exclusively from one domain, such as beliefs (e.g.,

    Elias & Ubriaco, 1986; Roehling & Robin, 1986), behavioral

    intentions (Lawton et al., 1983), or values (Holtzman et al.,

    1975; Segal, 1985).

    Problems With Content

    The result of this medley of items is a confounding of the

    elements of parental social cognition. Consequently, various

    problems are created for clarifying parental social cognition

    and its relation to child-rearing practices. First, there is reason

    to suspect that each cognitive element serves different functions

    and may relate to behavior somewhat differently. Beliefs, for ex-

    ample, form the foundation of attitudes and therefore may

    affect behavior less directly than attitudes (Ajzen & Fishbein,

    1980). A second problem resulting from the failure to disentan-

    gle related yet distinct constructs is that some elements of pa-

    rental social cognition are more relevant to particular domains

    of parenting than others. Attitudes about methods of discipline

    do not function in parallel with beliefs about how children learn

    or how they acquire gender identification. Because of the

    differences in how the elements interact and in turn relate to

  • 8/4/2019 Parental Attitudes Toward Child Rearing: Instruments, Issues, and Implications

    11/30

    PARENT CHILD-REARING ATTITUDES 39

    behavior, a differentiation of the components of parental socialcognition is essential. Factor-analytic techniques provide oneway to separate some of these elements, as can be found withdata from the Michigan Screening Profile of Parenting (MSPP;Heifer, Hoffmeister, & Schneider, 1978).

    Another principal problem concerns the ambiguity of items.This issue has long been recognized as a possible pitfall of child-

    rearing attitude questionnaires (Anderson, 1931; Koch et al.,1934; C. H. Sears, 1899; Watson, 1933). Likert (1932), as afounding father of attitude questionnaires, was well aware ofthis potential problem. He admonished developers of surveysto (a) word items as clear, concise, and straightforward state-ments; (b) omit "double-barreled" or dual questions; and (c)remember that "above a l l . . . each statement must avoid everykind of ambiguity" (Likert, 1932, pp. 45-46). Likert's recom-mendations, though often neglected, have withstood the test oftime (J. M. Converse & Presser, 1986). Ambiguity is enhancedby the common practice of phrasing items in the third-personformat (see examples in Table 2). As a result, parents couldrespond in terms of "cultural norms, professional opinion, em-pirical facts, or beliefs about what is best for others, none ofwhich may have anything to do with what the parent actuallydoes with his own child" (Becker & Krug, 1965, p. 361).

    Confusion has also resulted from other variables not ade-quately considered, most notably the age of the child describedby the questionnaire items. The majority of questionnaires donot specify the age of the child but present issues concerningtoddlers or preschoolers. Some surveys, however, include itemsthat span development issues from the transition to parenthoodto beyond the preschooler period (e.g., Cohler et al., 1970;Schaefer & Bell, 1958)! The consequence of including a rangeof age-specific items is that parents are forced to retrospectabout what they did and believed maybe years earlier or to pro-ject about what they will think or do in the future.

    A second age-related problem concerns the error that resultsfrom collecting data from samples of parents whose children'sages are not homogeneous. Two parents may have responded toa scale identically when their children were at the same age, butbecause they were sampled when their children were differentages, their responses were quite different. Behavioral intentionsconcerning punishment, independence, and other constructsthat are mediated by the developmental level of the child areparticularly at risk for this problem. Many investigators havebeen careful in controlling for the child's age (e.g., Lawton etal., 1983); others have not (e.g., Little, 1986). Remarkably, theonly investigator who explicitly incorporated the age of the

    child into the design of his questionnaire was Ojemann (1935b),who developed three different forms describing children ofdifferent ages.

    In a similar fashion, a variety of other important contextualinformation has been omitted from the items that is likely tohave a substantial effect on parental response. Prime variablesinclude the setting (e.g., home vs. public), the parent's goals orintentions (e.g., to discipline the child or to maintain familyharmony), the presence of others (father or siblings), and thequality of "ownness" (whether the child is their own or not).Each of these variables has been neglected in the pursuit of atenuous objective: to identify the pervasive, global parenting at-titude that transcends all variables.

    Maybe due to the baldness of the items, the content of certainitems stands out as antiquated. Many PCRAs still used duringthe 1980s include sexist assumptions and language ("Too manywomen forget that a mother's place is in the home"), outmodedconcepts ("The 'Puritan' method of bringing up children is thebest method"), or inapplicable questions ("What you read inbooks about a boy's being afraid of having his penis cut off isjust nonsense"). A more subtle problem lies in the word selec-tion. Words or phrases such as "breaking the will," "ruin," or"stunting a personality" are out of place today in most parents'phraseologies.

    Continuity and Change in PCRAs

    PCRAs, as a class of research stimuli, are characterized by alack of change. Some of the oldest instruments (e.g., Koch et al.,1934) continue to be used today (Campis, Lyman, & Prentice-Dunn, 1986). Similarly, many of the instruments developedover the past decade are indistinguishable in form or contentfrom their 50-year-old predecessors. However, three subtletrends can be gleaned from examining recent instruments.First, context-specific items are appearing more frequently in-stead of generalizations. Of the 11 instruments developed in thelast 2 years covered by this review, three instruments containsituation-specific items (Elias & Ubriaco, 1986; Roehling &Robin, 1986; Saarni, 1985). A second trend concerns the in-creasing differentiation of the parental social cognition do-mains. Although some early efforts at separating elements ofparental social cognition can be found in Ackerley's (1935) dis-tinction between attitudes and information and Emmerich's(1969) identification of both instrumental beliefs and values,most investigators made no such efforts. A few recent instru-ments have been developed to focus exclusively on one or two

    domains but provide only a limited or superficial assessment ofthe topic (e.g.. Keller et al., 1984; Rickard, Graziano, & Fore-hand, 1984; Sacks & Donnenfeld, 1984). The third trend is theconsideration of new types of attitudes or beliefs. Examples in-clude parental attitudes toward pleasure in children (Falender& Mehrabian, 1980), emotional expression in children (Saarni,1985), atypical or "distorted" beliefs (Roehling & Robin,1986), rearing children with racial awareness (Branch & New-combe, 1986), attitudes about parental control over their chil-dren's development (Campis et al., 1986), and views on mater-nal employment (Hock, Gnezda, & McBride, 1984). Note thatmany of these new topics lie in the domain of parental self-per-ception. One new and distinctive instrument attempts to assess

    the level of Piagetian-type thinking parents use when reflectingon their children (Sameroff & Feil, 1985). Whether this novelapproach will be successful for categorizing parents' thinkingabout their children awaits further tests.

    Though PCRAs have experienced relatively little modifica-tion over time, some investigators have taken great liberties inaltering the instruments. The popular PARI has seen at leasteight modifications, including a reduction in length (Camp &Morgan, 1984; Chorost, 1962; Cross &Kawash, 1968; Emmer-ich, 1969), a translation into a foreign language (Seth & Sak-sena, 1978), and modifications in unspecified ways (e.g., Fu etal., 1984). Mercifully for respondents, the 491-item survey de-veloped by Winder and Rau (1962) has been shortened almost

  • 8/4/2019 Parental Attitudes Toward Child Rearing: Instruments, Issues, and Implications

    12/30

    40 GEORGE W. HOLDEN AND LEE A. EDWARDS

    by half in a recent study (Chandler, Wolf, Cook, & Dugovics,

    1980). However, modifications of a survey raise the question

    of whether the psychometric properties (if known) have been

    maintained.

    Psychometric Issues

    On the continuum of research methodologies from those

    closely related to the primary object of studyin this case, pa-

    rental behaviorto the furthest removed, the PCRA lies at the

    far end (Lytton, 1971). For this reason and others, question-

    naires in general, and parental questionnaires in particular,

    have sometimes been regarded with suspicion. John Anderson,

    in his methodology chapter of the first Handbook of Child Psy-

    chology, wrote,

    there is much distrust of the questionnaire. This is in part due tothe fact that confusion exists in the minds of question makers andthose who fill them out with reference to matters of opinion andmatters of fact ... It is subject to errors of memory, misunder-

    standing of terms, and mental sets imposed by the questions. Ingeneral, although the results may be suggestive, verification byother methods is necessary. (Anderson, 1931, p. 14)

    Over the next two editions of the handbook, spanning 23 years,

    Anderson saw no reason to change substantively those words.

    Cautions about the quality of parental self-report data were also

    prominent in the 1960s (Maccoby & Martin, 1983; Yarrow,

    1963).Given the concern over questionnaire data and particularly

    parental self-reports, it is all the more important for the psycho-

    metric properties of PCRAs to be both documented and sound.

    Such has not been the case. Of the instruments reviewed, reli-

    ability data were found for just over half (59%), and only nine

    instruments have both principal types of reliabilitytest-re-

    test and internal consistency (Table 1). In most cases, the reli-

    ability data were found with the original description of the in-

    strument; however, in some cases where no psychometric data

    were provided (e.g., T. Gordon, 1970), other investigators have

    subsequently supplied such information (Flynn, 1979). Validity

    data are even more scarce; such data were found for only 43%

    of the instruments.'

    Reliability

    The reliability of an instrument is determined usually by a

    test-retest assessment or an appraisal of the internal consis-tency. Ideally, both methods should be used, as each yields a

    different type of information. These two types of reliability pro-

    vide an index of the amount of measurement error associated

    with an instrument. Because variance is increased by measure-

    ment errors, which in turn makes it more difficult to identify

    significant differences or relationships, using reliable instru-

    ments is especially important in correlational studies (Mc-

    Nemar, 1969). A third type of reliability is sometimes deter-

    mined from the comparison of two or more assessments of the

    factor structure of the instrument. Only one study (Koch et al.,

    1934) used a fourth approach to assess an instrument's reliabil-

    ity: Two alternate, parallel forms were correlated.

    Test-Retest Reliability

    Test-retest data reveal how consistently an individual re-

    sponds to the same items over a short period of time. Though

    the retest method of establishing reliability is problematic be-

    cause subjects may remember their earlier responses on the sec-

    ond administration, and thus the correlation between the two

    administrations may be erroneously high, the method is usefulas an initial reliability assessment. As one psychometrician

    pointed out, "If a test does not correlate with itself when admin-

    istered on two occasions, it is hopeless to seek other evidence

    of reliability and hopeless to employ the test in correlational

    studies" (Nunnally, 1978, p. 234). Published test-retest data are

    available for 21 of the instruments (Table 1).

    Interpreting the test-retest results is difficult because there

    has been little standardization of assessment procedures or, in

    two cases, correlations were not computed. For those reliability

    assessments that do report Pearson correlations, there is consid-

    erable variation with sample sizes (median = 63 subjects,

    range = 8-645) and intervals (median and mode = 4 weeks,

    range = 1-156). Given that no prescribed interval between test-ing exists, some researchers have used a 12-month interval to

    assess reliability (J. H. Block, 1965),' and others have conceived

    of the same time period as a suitable duration to examine conti-

    nuity or change in parental attitudes (e.g., Davids & Holden,

    1970; Ramey, Farran, & Campbell, 1979). Averaged across all

    studies and subscales, the mean test-retest correlation was .76.

    There is also little concurrence as to the sample composition

    for test-retest assessments. In five of the assessments, nonpar-

    ents were used as subjects; their retest correlations averaged .79

    (range = .67-.95). However, nonparents form an inappropriate

    sample to test the stability of parental attitudes because one can

    assume that nonparents' (i.e., undergraduate students) attitudes

    and beliefs about child rearing are less developed and less proneto change than are those of parents. Some support is provided

    by the 12 correlations involving only samples of parents; those

    coefficients averaged .74 (range = .61-.89).

    Internal Consistency

    The internal consistency of a scale provides an index of how

    well a set of items are able to assess one construct. Information

    about the internal consistency of the questionnaire was found

    for 45% of the instruments (Table 1). Split-half reliability, usu-

    ally corrected by the Spearman-Brown Prophecy formula, aver-

    aged .87 over the 18 instruments with such information. Addi-

    tional evidence indicating that nonparents may provide inflated

    reliability data was reported by Stogdill (1936b). Across his two

    instruments, the nonparents' split-half reliability averaged .89,

    whereas smaller samples of parents averaged .66.

    Given that the split-half reliability of an instrument may vary

    as a result of how items are divided, a better assessment of inter-

    nal consistency is obtained with Cronbach's coefficient alpha

    (Nunnally, 1978), or the related KR-20 statistic (Kuder & Rich-

    ardson, 1937). Since 1971 internal consistency has been deter-

    mined predominantly with these two statistics. The overall aver-

    1Tables with the reliability and validity data are available from the

    first author.

  • 8/4/2019 Parental Attitudes Toward Child Rearing: Instruments, Issues, and Implications

    13/30

    PARENT CHILD-REARING ATTITUDES 41

    age for the 20 instruments reporting these types of consistency

    was .74 (median = .76, range = .5S-.92); only 6 instruments

    attained internal reliability levels at or above Nunnally's recom-

    mended .80 for basic research.

    factor Structure Reliability

    It has become a relatively common practice to factor analyze

    questionnaire data. Reports on the factor structure can be

    found on more than 11 instruments developed since I960, al-

    though there is only one instrument whose factor structure has

    received repeated scrutiny: the PARI. Because of the relatively

    large number of items, the popularity, and the longevity of the

    instrument, more than 10 investigations have tested its factor

    structure. Schaefer and Bell (1957) initially described five

    oblique factors based on a sample of 100 students. Subsequent

    samples with parents revealed three factors: authoritarian con-

    trol, hostility-rejection, and democratic attitudes (Schaefer,

    1961: Zuckerman, Ribback, Monashkin, & Norton, 1958).

    These three basic factors have been replicated with only minor

    differences in samples of clinically heterogeneous mothers, low

    income mothers, fathers, and college students. Sims and Pao-

    lucci (1975) identified 10 clusters on a modified PARI that

    differed from an earlier cluster analysis (Cross & Kawash,

    1968), but the clusters are comparable to the three original fac-

    tors. In contrast to those replications, some investigators (Beck-

    with, 1971: Wyer, 1965) have reportedly found other factor

    structures; small sample sizes or insufficient information pre-

    clude acceptance of those claims.

    The factor structures of three other instruments have also

    been tested with less consistent results. The structure of the Par-

    ent as a Teacher Inventory (Strom & Johnson, 1974) was origi-

    nally described as having five 10-item factors. In subsequent

    work, the six factors identified bore little resemblance to the

    hypothesized ones (Thornburg, Ispa, Gray, & Ponder, 1983).

    Similarly, Cattell and his colleagues (Barton, Dielman, & Cat-

    tell, 1973) found a more complex factor structure on the Family

    Attitude Measure (FAM) than had been found in an earlier

    study (Delhees et'al., 1970).' With a second PCRA, Dielman,

    Barton, and Cattell (1973) replicated only about half of the fac-

    tors they had identified previously.

    Summary of Reliability Assessments

    Insufficient attention has been devoted to determining the

    reliability of PCRAs; there is inadequate information with

    which to evaluate almost half of the instruments. A mere 11 % ofthe surveys have been subjected to both test-retest and internal

    consistency reliability assessments. Collecting both types of re-

    liability information should not be considered optionaleach

    type provides important yet distinctive information.

    The quality of the reliability data, when available, is tem-

    pered by the frequent use of nonparents to assess the properties

    of instruments designed for use with parents. Test-retest data

    are further limited by the wide range of intervals between tests.

    With those qualifications, the available reliability information

    indicates PCRAs have, in general, marginally acceptable test-

    retest correlations. The internal consistency, as measured by

    split-half reliability, is acceptable on most of the surveys that

    provide such information. When consistency is assessed by the

    preferred calculation of Cronbach's alpha, few instruments at-

    tain an adequate level of internal consistency. Given the some-

    times low reliability coefficients and, in a few cases, variable

    factor structure assessments, caution should be exercised in as-

    suming that an a priori subscale measures what it was designed

    to. The consequence of the tenuous reliability of many of the

    instruments means that relations between parental attitudes

    and whatever else is measured are underestimated. As a result,

    it is more difficult to find reliable relations or to establish the

    validity, a topic turned to next.

    Validity

    Nunnally (1978) has proposed that "the degree to which it

    is necessary and difficult to validate measures of psychological

    variables is proportional to the degree to which the variable is

    concrete or abstract" (p. 95). This maxim holds for the valida-

    tion of instruments to measure parental attitudes. To their

    credit, almost half of the developers of PCRAs have recognized

    the need to validate their surveys. However, the evidence is often

    less than compelling, as will be shown. For example, the evi-

    dence for the validity of an instrument sometimes appears as

    a significant correlation between two sets of measures, in the

    absence of any specific predictions (e.g., Shapiro, 1952). To an-

    other investigator, validity was manifested when three of five

    judges agreed on the ranking of items (Porter, 1954).

    After disregarding certain validity data that provided no such

    evidence, the available information was marshaled and orga-

    nized under Messick's (1980) description of the kinds of valid-

    ity. He differentiated three basic categories: content validity, cri-

    terion validity, and construct validity. Content validity, or the

    adequacy with which a domain is sampled, will not be discussed

    here for two reasons. First, the topic was discussed under the

    review of the content of the PCRAs. Moreover, when evidence

    for content validity is provided, it is essentially useless informa-

    tion: "Five expert judges examined and concurred on the con-

    tent validity of the items" (DeSalvo & Zurcher, 1984, p. 10).

    Evidence is available, however, for both construct and criterion

    validity.

    Construct Validity

    Construct validity indicates that the instrument does, in fact,

    measure the particular psychological phenomena of interest.

    Three types of construct validity have been provided as evi-

    dence for the validity of PCRAs: convergent, discriminant, andpopulation validity.

    Convergent validity. Evidence for convergent validity usually

    comes in the form of how well the measure coheres with other

    measures of the same construct. Given the number of PCRAs

    developed and the frequent use of certain types of subscales

    (e.g., authoritarian control and acceptance), all too few studies

    have determined the convergent validity of subscales from

    different questionnaires. The available evidence provides gener-

    ally weak support for the convergent validity of the instruments.

    For example, G. F. Brody (1965) found four reliable corre-

    lations between the PARI and the Maryland Parent Attitude

    Survey (MPAS), including the Authoritarian and Disciplinar-

  • 8/4/2019 Parental Attitudes Toward Child Rearing: Instruments, Issues, and Implications

    14/30

    42 GEORGE W. HOLDEN AND LEE A. EDWARDS

    ian subscales, r(50) = .30, p < .05, though the Hostility-Rejec-tion and the Rejecting subscales did not reach significance,r(50) = .26, p >*5. Although a few investigators have beenmore successful in finding relatively high associations betweenthe predicted variables (Levinson & Huffman, 1955; Rohner etal., 1978), more often than not the correlations are nonsignifi-cant, and some even fail to be trends (e.g., Campis et al., 198(>;Leton, 1958).

    Only one methodological study was located that comparedthe quality of data collected from questionnaires, interviews,and observations. Using a sample of 40 mothers of 4- and 5-year-old children, R. R. Sears (1965) found that mothers' self-reports from a survey and interview correlated .53 across fivevariables. When two of those variables were correlated with rat-ings from observations of maternal punishment of aggressionin the laboratory, the survey data were found to correlate morehighly than the interview data. The survey's punitiveness scalecorrelated significantly with ratings of observed punishment(.31) and ratings of permissiveness (.40). The results providedmore support for the survey than the interview method for re-flecting behavior, but Sears concluded that a replication studywas required. Such a study is still needed, more than 20 yearslater.

    Discriminant validity. Discriminant validity, indicatingwhether a measure differs from other distinct constructs, hasbeen assessed rarely; this is probably a reflection of the lackof understanding about why one parental attitude should differfrom another attitude or from another psychological construct.The evidence for discriminant validity has frequently consistedof finding that two or more subscales on an instrument do notcorrelate (Falender & Mehrabian, 1980; Rickard et al., 1984;Rohner et al., 1978). Though such analyses do provide an initialstep, more convincing discrimination needs to be providedwith the use of proven instruments differing on theoreticalgrounds.

    Population validity. Population validity refers to how well

    scores on a test generalize to different populations. No evidenceexists on this topic; however, with varying degrees of complete-ness, relatively comprehensive normative data could be foundfor four of the instruments. Schaefer and Bell's (1958) initialnormative population on the PAR! was a dubious one: It con-sisted of a total of 120 wives of military personnel who filledout the survey 1 to 4 days postpartum. Subsequently, othershave collected more thorough normative information and iden-tified how variables such as the age and sex of the child, or theparents' age, education, socioeconomic status, and marital sta-

    tus, can influence attitude scores (e.g., Leeper, Milo, & Collins,1983; Mohan, 1981;Phelps, 1969).

    Other instruments that are accompanied with at least somenormative data include the MPAS (Pumroy, 1966), the Mater-nal Attitude Survey (Cohler et al., 1970), and the Family Prob-lems Scale (Ernhart & Loevinger, 1969). Ernhart and Loevinger(1969) made the most extensive efforts to establish norms fortheir instrument by administering it to 1,584 women of differ-ent backgrounds. Scores from about 400 men and women anda smaller sample of mothers of developmentally delayed chil-dren are available for the MPAS (Slough, Kogan, & Tyler, 1978?/. 1"In genera], however, normative data are scarce; little is knownabout the response ranges for most of the surveys, how family

    characteristics influence those responses, or whether the scoresfrom one sample of subjects generalize to other groups.

    Criterion Validity

    Criterion validity concerns the relations between the instru-ment and something external to the instrument; validity datarelated to two subcategories of concurrent validity and one ofpredictive validity will be reviewed.

    Concurrent validity: Criterion groups. How well attitude sur-veys discriminate two or more groups of individuals of knowndifferences has provided mixed results. Stogdill pioneered thisapproach by comparing the attitudes of parents with mentalhygienists (Stogdill, 1931), child specialists (Stogdill, 1934), andpsychologists and undergraduates (Stogdill, 1936b). Since thattime, support for the validity of at least 19 other instrumentshas relied on differences between two or more groups. Often theselection of the criterion groups is made in the absence of atheoretical rationale for linking particular attitudes with partic-ular groups of individuals. Group differences have provided aneasy source of validity evidence, although the magnitude of thegroup differences is often quite small (e.g., T. M. Field, Wid-mayer, Stringer, & Ignatoff, 1980;Saarni, 1985).

    An early example of the lack of theoretical rationale can befound in the series of conflicting studies involving the assess-ment of the attitudes of mothers of schizophrenics or behavior-disturbed children; as such, they were flawed tests of the hy-pothesis that parental attitudes were causally linked to childoutcome. Mark (1953), and to a lesser extent R. V. Freeman andGrayson (1955), discovered mothers of disturbed children to bemore overpossessive, overprotective, restrictive, or passive thana comparison sample. But at least four subsequent investiga-tions, with larger and better controlled samples, were unable toreplicate those results (e.g., Pitneld & Oppenheim, 1964).

    Since that time, PCRAs have been used to discriminate vari-ous other clinical samples. For instance, particular surveys havebeen able to reveal that depressed or psychotic mothers have lessadaptive child-rearing attitudes than a control group of mothers(Cohler, Grunebaum, Weiss, Hartman, & Gallant, 1976; Sus-man, Trickett, lannotti, Hollenbeck, & Zahn-Waxier, 1985);parents experiencing problems have a more external parentallocus of control than parents not experiencing problems(Campis et al., 1986); and abusive mothers differ from nonabus-ers in their responses (Avison et al., 1986; Brunnquell, Crich-ton, & Egeland, 1981).

    PCRAs have also been effective in differentiating between pa-

    rental attitudes of nonclinical populations. For example, indi-viduals from different religious groups hold different child-rear-ing attitudes (Anders & Dayan, 1967; Levinson & Huffman,1955); mothers who revolunteer for a psychological study tendto have more adaptive attitudes than those who do not (Cohler,Woolsey, Weiss, & Grunebaum, 1968); and fathers who scoredhigh on a restrictive subscale differ in personality from permis-sive fathers (J. Block, 1955). One of the most consistent androbust findings with PCRAs concerns the ability of the instru-ments to discriminate parents from differing social classes andeducational levels (e.g., Fu et al., 1984; Sheintuch & Lewin,1981). In fact, the group differences reported in many studiesusing the PARI could be accounted for by differences in educa-

  • 8/4/2019 Parental Attitudes Toward Child Rearing: Instruments, Issues, and Implications

    15/30

    PARENT CHILD-REARING ATTITUDES 43

    tional level, according to Becker and Krug (1965). Many studies

    have not controlled for educational status (e.g., Moran &O'Brien, 1984; Turner & Harris, 1984). The problem with

    differing education levels in respondents is that one does notknow whether observed differences reflect previously existingparental attitudes or are simply artifacts of reactions to the

    wording of the questions. Becker and Krug (1965) concludedthat both explanations may be accurate, but the necessary ex-

    perimental manipulation has yet to be conducted.Despite the many studies reporting group differences, it is

    also true that results are sometimes the opposite of what was

    predicted. For example, Melnick and Hurley (1969) reportedthat abusive mothers had less rejecting attitudes toward theirchildren than did a sample of control mothers. More com-monly, predicted group differences have not always been found(e.g., Hitchcock, 1978;Staples*Smith, 1954).

    A second approach designed to establish the concurrent va-

    lidity of PCRAs has been to examine the associations betweenparental attitude scores and child behavior. Most studies reportlittle or no relation between maternal attitudes and assessments

    of child behavior or personality (e.g., G. F. Brody, 1969; Zunich,1962) or parent ratings of the child (Leton, 1958; Robinson &Anderson, 1983), although there is some conflicting evidence(cf. Barton, Dielman, & Cattell, 1977; Read, 1945). A numberof studies have compared parental att itudes with the incidence

    of children's behavior problems. Again, most studies find fewor no relations, although there are exceptions (e.g., Rees &

    Wilborn, 1983). In the largest study conducted on this topic,Wi nder and Rau (1962) compared the sociometric status of 710boys with their parents' attitudes. Out of 340 analyses of vari-

    ance (ANOVAS) conducted, only 17% were significant at the .05

    level or beyond. Post hoc explanations were then created to de-

    scribe the pattern of results.

    Concurrent validity: Behavioral assessments. The singlemost important validity question is how well do responses onPCRAs correspond with parental behavior? To date, only a

    dozen studies have tested the links between parental behaviorand attitudes as assessed by PCRAs. The first investigator to

    address this question (J. E. Gordon, 1957) arrived at an inauspi-cious conclusion: No relation was found between rankings of

    mothers based on 12 days of behavioral observations and their

    attitude scores on Shoben's (1949) PAS.Five studies have assessed the behavioral validity of the PARI.

    Zunich (1961) compared maternal attitudes with maternal be-

    havior in the laboratory. Of the 272 comparisons made, only 12correlations attained significance at the .05 levelfewer than

    would be expected by chance 'G. F. Brody (1965) comparedmaternal attitudes on the PARI and the MPAS with observa-

    tions in a preschool. One out of 23 PARI but three out of fourMPAS subscales were related to observed behavioral differences(at the .05 level); when