Validation of the Maternal Expectations and Attitudes Scale

MISSING IMAGE

Material Information

Title:
Validation of the Maternal Expectations and Attitudes Scale
Physical Description:
vi, 124 leaves : ; 29 cm.
Language:
English
Creator:
Seibring, Angel Renee, 1963-
Publication Date:

Subjects

Subjects / Keywords:
Research   ( mesh )
Reproducibility of Results   ( mesh )
Data Collection   ( mesh )
Questionnaires   ( mesh )
Attitude   ( mesh )
Attitude -- Pregnancy   ( mesh )
Mothers -- psychology   ( mesh )
Pregnancy -- psychology   ( mesh )
Time Factors   ( mesh )
Abnormalities -- Pregnancy   ( mesh )
Psychological Tests   ( mesh )
Department of Clinical and Health Psychology thesis Ph.D   ( mesh )
Dissertations, Academic -- College of Health Related Professions -- Department of Clinical and Health Psychology -- UF   ( mesh )
Genre:
bibliography   ( marcgt )
non-fiction   ( marcgt )

Notes

Thesis:
Thesis (Ph.D.)--University of Florida, 1993.
Bibliography:
Bibliography: leaves 117-123.
Statement of Responsibility:
by Angel R. Seibring.
General Note:
Typescript.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 002295857
oclc - 49807446
notis - ALP9084
System ID:
AA00011842:00001


This item is only available as the following downloads:


Full Text














VALIDATION OF THE MATERNAL EXPECTATIONS AND ATTITUDES SCALE


By

ANGEL R. SEIBRING
























A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA

1993















TABLE OF CONTENTS


PAGE

LIST OF TABLES . . iv
ABSTRACT . . .. v

CHAPTERS
I INTRODUCTION . . 1
II REVIEW OF THE LITERATURE . 5
Expectations . . 5
The Relationship Between Expectations and
Physical Abuse . .. 11
Prenatal Expectation and Postnatal Experience 15
Development of the Maternal Expectations and
Attitudes Scale . .. 19
Statistical Properties of the MEAS 22
Preliminary Validity . .. 23

III STATEMENT OF THE PROBLEM . .. 25

Phase I: Replication of the Factor Structure and
Examination of Test-Retest Reliability 26
Phase II: Construct Validity . 27
Phase III: Construct Validity
of the Child Expectation Subscale 27

IV METHOD . .. 32

Subjects . . 32
Measures . . .32
Procedure . . 33

V RESULTS . . 35
Phase I . .. 35
Analyses . . 36
Factor Analysis of the Maternal Expectations
and Attitudes Scale . 40
Comparing the Factor Structures from the
Two Studies . .54
Test-Retest Reliability and Internal Consistency
for Factor-Derived Subscales 60
Phase II: Construct Validity of the MEAS 62
Subjects . 62
Measures . .. 62

ii









Procedure . ... .66
Analyses . . 66
Social Desirability . 70
Previous Experience and the MEAS 71
Phase III: Construct Validity of the Child
Expectation Subscale . ... .73
Subjects . ... .73
Procedure . ... .75
Measures ... . .. .76
Analyses . .. .76
VI DISCUSSION . . 81
Integration of Findings with other Research 90
Future Use of the MEAS: Rational versus
Factor-Derived . 94
Limitations . . 95
APPENDICES

A MATERNAL EXPECTATION AND ATTITUDES SCALE 98

B MATERNAL SELF-REPORT INVENTORY (MSI) 106

C CHILD ABUSE POTENTIAL INVENTORY (CAP-I) 110

D CATALOG OF PREVIOUS EXPERIENCE WITH INFANTS
(COPE) . . ... 114

REFERENCES . ... 117

BIOGRAPHICAL SKETCH . . .. .124














LIST OF TABLES


TABLE PAGE

5-1 Summary Descriptive Statistics (n=200) 38

5-2 Psychometric Properties of the
Rationally-Derived MEAS Subscales .. .39

5-3 Core MEAS Items: Factor 1
Factor Loadings, Percents of Variance ... .43

5-4 Core MEAS Items: Factor 2
Factor Loadings, Percents of Variance ... .45

5-5 Core MEAS Items: Factor 3
Factor Loadings, Percents of Variance ... .46

5-6 Disappointment Items: Factor 1
Factor Loadings, Percents of Variance ... .50

5-7 Disappointment Items: Factor 2
Factor Loadings, Percents of Variance ... .52

5-8 Correlations between Factors from Seibring
(1991) and Seibring (1993); Cattell's
Salient Similarity Index (s) ... 59

5-9 Test-Retest Reliability and Internal
Consistency for Factor-Derived and
Rationally-Derived MEAS Subscales ... .61

5-10 Pearson Correlations of MEAS Subscale
Scores with MSI Subscales . .. 69

5-11 Summary Descriptive Statistics (n=39)
IUGR Group . ... .74

5-12 Mean Scores on MEAS Subscales ... .80














Abstract of Dissertation Presented
to the Graduate School of the University of Florida
in Partial Fulfillment of the Requirements for the Degree
Doctor of Philosophy

VALIDATION OF THE MATERNAL EXPECTATIONS AND
ATTITUDES SCALE

By
Angel R. Seibring

December, 1993

Chairman: James H. Johnson, Ph.D.
Major Department: Clinical and Health Psychology

The purpose of this study was to generate psychometric

data on the Maternal Expectations and Attitudes Scale, a

self-report instrument for first-time mothers. A primary

focus of this study was to replicate the factor structure of

the MEAS with a large, representative sample. Further,

information was gathered on the construct validity and

reliability of the measure.

The MEAS, validation measures, and a measure of

previous experience with infants were administered to 200

women recruited from an outpatient women's clinic,

childbirth preparation classes, and private obstetric

practices. Construct validity of one subscale of the MEAS

was assessed in a separate phase. In this phase, responses

of two groups of primiparous women were compared. The first

group was comprised of 39 women pregnant with a child








diagnosed as Intrauterine Growth Retarded (IUGR). A

comparison group consisted of women matched for age of the

mother, race, social status, education level, and pregnancy

trimester who did not display known pregnancy complications.

Results of this study provide evidence for five

conclusions: 1) the factor structure of the MEAS generated

in a previous study appears stable and was successfully

replicated in this study; 2) psychometric data support the

reliability and internal consistency of the measure; 3)

there are reliable individual differences in maternal

expectations about motherhood that can be measured with this

self-report instrument; 4) these individual differences in

expectations can discriminate between normal and abnormal

pregnancy status; and 5) support was found for the

prediction that first-time mothers with higher levels of

previous infant experience would report more realistic

expectations on the MEAS.
















CHAPTER I

INTRODUCTION

Over the past two decades, interest in the

transition to parenthood has increased dramatically

(Belsky, Ward, & Rovine, 1986). It is widely

recognized that the first pregnancy and the birth of a

first child represent a major life transition (Fawcett,

1988; Palkovitz & Copes, 1988). The first-time mother

is faced with new responsibilities, new roles,

physiological changes, changes in the family structure,

and in the dynamics of relationships. For many

mothers, the arrival of a first child seems to have a

negative impact (Hobbs & Cole, 1976). Sociologists

interested in the transition to parenthood have focused

primarily upon the degree to which a crisis is

experienced following the birth of the first child.

LeMasters (1957) postulated that crises in families

emerge when the family must reorganize as a social

system. Adding a new member to the family requires

such a change. Eighty-three percent of the parents

interviewed by LeMasters (1957) felt, in retrospect,

that the adjustments to parenthood involved extensive








2

or severe "crisis" transitions. Hobbs and Cole (1976)

conducted a prospective study dealing only with

primiparae and reported that new parents considered the

arrival of a child disruptive, but not a crisis of

severe proportions. They report, however, that the

transition was more difficult for the mothers than for

the fathers. Miller and Sollie (1980) propose that the

changes that result from new parenthood are on the

order of a crisis, but most parents welcome these

changes and are not extremely bothered by them. In

their view, parenthood is a "natural stressor." Yet

they report that during the first year after birth new

mothers reported decreased well-being and significantly

higher levels of personal stress than before the birth.

Concurrent with Hobbs and Cole (1976), they found that

the mothers reported more difficulty during this time

than the fathers. This heightened susceptibility for

women to experience negative change has been a

persistent finding in the transition to parenthood

literature for decades (e.g., Belsky & Pensky, 1988;

Hobbs, 1965; Hobbs & Wimbish, 1977; Russell, 1974).

The consensus seems to be that parenthood can test the

family's coping strategies, particularly those of the

mother.

It is obvious that all mothers do not experience

the transition to motherhood in the same manner. Yet,










to date, little systematic attention has been given to

studying individual differences in the transition to

motherhood. Even less attention has been devoted to

the determinants of these differences. A major purpose

of the present research is to outline one such factor

believed to mediate the adjustment to motherhood. The

prenatal expectations and attitudes held by the first-

time mother are believed to be related to subsequent

adjustment and functioning. These include expectations

regarding the future child and themselves as a new

mother. A previous study conducted by this author

marked the first attempt to develop a measure to

assess the range of maternal prenatal expectations

(Seibring, 1991). The Maternal Expectations and

Attitudes Scale was the product of this research

endeavor. While preliminary data support the

reliability and validity of this measure, further

research is needed to establish its utility and

psychometric qualities. The purpose of the current

study was to test the construct validity of the

measure, and to replicate the factor structure reported

in Seibring (1991) on a larger, more representative

sample of women.








4

The following literature review will outline the

importance of prenatal expectations, the consequences

of discrepant postnatal experiences, and highlight the

need for a psychometrically sound measure of maternal

expectation.
















CHAPTER II

REVIEW OF THE LITERATURE

Expectations

Social psychologists have demonstrated that

expectations are important in human interactions, in

part because they affect behavior toward others (Bond,

1972; Snyder, Tanke, & Berscheid, 1977), and in part

because they affect how others' behavior is perceived

(Schlenker, 1980). Regarding the expectations of

mothers, little is known about what they expect before

a baby is born, and there are few studies which

investigate how expectations discrepant from actual

experience impact on the adjustment to motherhood

(Belsky, Ward, & Rovine, 1986).

Many writers have commented on the fact that

expectant parents frequently romanticize the way in

which becoming parents will affect individual and

family life (Belsky et al., 1986; LeMasters, 1957). It

seems that, as a group, first-time parents hold

inaccurate and unrealistic expectations (Entwisle &

Doering, 1981). Entwisle and Doering (1981) followed

120 couples in a longitudinal study beginning at six








6

months of pregnancy and continuing through six months

after birth. They write that the "women in this sample

had unrealistic expectations in many areas" (p. 35).

For example, these first-time mothers were overly

optimistic about how soon the child would be weaned and

sleep through the night (e.g., over half expected the

infant to sleep through the night before eight weeks).

The tendency of first-time mothers to overestimate

infant ability has been noted by Pharis and Manosevitz

(1980). When asked to estimate the age at which the

average child could perform certain tasks, these

expectant parents only slightly overestimated age.

However, when asked to rate expected ages for their

future child, the tendency to overestimate was greater.

Miller (1988) notes that the tendency to overestimate

ability may result from the fact that parents have a

vested interest in the child's development, perhaps

feel a sense of responsibility, and therefore emphasize

such things as early achievement.

Entwisle and Doering (1981) note that first-time

mothers' expectations for what would transpire during

labor and delivery were also frequently incorrect,

despite the high level of prenatal preparation in the

sample. The errors again were in the optimistic

direction. One possible ramification for unrealistic

expectations for labor and delivery is suggested by










Zeanah et al. (1986). They found that those mothers

who described their experience of labor as very

different from their prenatal expectation were more

likely to change their perception of their infant's

personality pre- to postnatally. This discrepancy in

labor experience distinguished the "changers" of infant

personality ratings from the "nonchangers." Entwisle

and Doering (1981) state that more realistic

expectations would be helpful for new mothers in their

transition to motherhood.

The first few months after childbirth are

considered important for forming bonds between the

mother and child (Cutrona, 1983; Cutrona & Troutman,

1986; Goldsmith, & Alansky, 1987). Attachment between

the mother and child is considered necessary for normal

development (Bowlby, 1969), and there is empirical

support for the view that the quality of the mother-

child interaction directly impacts the attachment

between the two (Bowlby, 1969). Research suggests that

attachment begins as soon as the baby is born or even

during pregnancy (Klaus & Kennell, 1976). Zeanah,

Keener and Anders (1986) found that mothers actively

engage in fantasizing about their unborn infants and

establish "working models" of them long before they are

born. First-time mothers were found to have vivid

prenatal perceptions of their infant's looks,










personality and temperament, and these perceptions

evidenced some stability from pre- to postnatal

assessment. The authors interpret this as providing

evidence for the existence of a maternal bias. That

is, prenatal fantasies lead to expectations about the

child and subsequently bias mother's assessments of

their infants. They state that the pregnant woman is

faced with the task of relinquishing the "imagined

baby" and accepting the actual one.

Stainton (1985) demonstrated that both prospective

mothers and fathers describe individual characteristics

of the fetus (communication, temperament, sleep-wake

cycle), and they interpret the in utero behavior as

having meaning. Interview data (Seibring, 1991)

suggest that some mothers form impressions about the

personality of the infant based on their own personal

characteristics (e.g., "He's going to be mean because I

am mean") and make attributions for fetal movement

("she moves so much I think she will be hyperactive").

The foregoing data suggest that for some women

impressions of the child-to-be are formed by maternal

projections.

This research raises questions about the potential

effects of prenatal expectations on mothers' postnatal

perceptions of, and reactions to, the actual infant.

What consequences might arise from a discrepancy








9

between prenatal expectation and postnatal experience?

The life events literature suggests that situations

that do not reflect what was anticipated invariably

require more coping than ones that turn out as expected

(Belsky et al., 1986; Johnson & Sarason, 1979).

Borrowing from the social psychological literature,

cognitive dissonance may result from the

expectation/experience discrepancy and might lead to

attempts to reduce the dissonance (Wicklund & Brehm,

1976). Dissonance reduction may take the form of

modified maternal perceptions of the child or of

herself as a mother. It is possible then, that as a

means to reduce cognitive dissonance, mothers may bias

their postnatal perceptions of the infant or

themselves. The hypothesis of dissonance reduction has

not been examined empirically; however, there is

evidence to suggest that a maternal bias in infant

perception exists. In some areas, maternal perceptions

may be more influential in infant ratings than actual

characteristics of the infant (Crockenberg & Acredolo,

1983; Sameroff et al., 1982). A review of this

research follows.

Sameroff, Seifer, and Elias (1982) found that

mothers high in self-reported anxiety rated their

children as temperamentally more difficult. These

authors state that individual differences in mothers








10

may be more influential in temperament ratings than

actual differences in the infants being rated (see also

Wolkind & Desalis, 1982). Distorted maternal

perceptions of the infant (e.g., maternal perceptions

which differ from objective infant assessment) have

been associated with maternal behavior. Mothers

exhibiting distorted infant temperament perceptions

have been found to be less responsive to their infants

(Campbell, 1979), less affectionate, and have expressed

less pleasure (Nover et al., 1984).

The relationship between previous infant

experience and maternal behavior has been noted in

several studies. Bates, Freeland, and Lounsbury (1979)

found that multiparous mothers rated their infants as

temperamentally easier than primiparous mothers. It is

possible that these first-time mothers rated their

infants as more temperamentally difficult because the

actual child was different from what they expected.

This hypothesis was not tested, however. An

alternative, though not entirely separate explanation,

is that experience with children is the differentiating

element. Along this line, Ventura (1982) found that

parents of firstborns were more prone to depression

than were experienced parents and proposed that lack of

experience may be contributing to depression.

Experienced parents are perhaps more realistic in their








11

expectations and do not experience the level of

disappointment encountered by the novice parent.

Mothers with more direct experience with babies find

infant cries less aversive than naive raters (Zeskind,

1980), and an important relationship has been noted

between responsiveness to infants and amount of

previous infant experience (Blakemore, 1981).

Based on these findings, it seems that maternal

expectations and attitudes could be significantly

influenced by the level of previous experience with

infants.

The Relationship Between Expectations and

Physical Abuse

The most extreme reaction to violation of

expectations may be physical abuse. General deficits

in understanding normal child development and child-

rearing knowledge have been hypothesized to be the

precipitants of unrealistic expectations (Azar et al.

1984; Kravitz & Driscoll, 1983; Twentyman & Plotkin,

1982; Wolfe, 1985). Typically, researchers in this

area have attempted to test this hypothesis by

examining the results of questionnaires that tap

parents' knowledge of developmental milestones.

Abusive parents have been shown to hold

unrealistic expectations for child development

(Larrance & Twentyman, 1983; Steele & Pollock, 1968;








12

Twentyman & Plotkin, 1982). Most often, parents who

abuse and neglect their children expect too much;

however, in some areas, they expect too little

(Twentyman & Plotkin, 1982). Twentyman and Plotkin

(1982) assessed abusive parents' expectations for when

their child and the "average" child would achieve

developmental milestones. Abusive parents expected

their own child to attain milestones later than the

average child. This appears to contradict the more

common finding that abusive parents' expectations are

unrealistically high. The authors point out, however,

that abuse may result from unrealistically high

expectations in salient areas. Research by Kravitz and

Driscoll (1983) provides some support for this

hypothesis. These authors found that more often

abusive parents gave later expectations for development

than a carefully matched comparison group of non-

abusive parents. However, the abusive parents erred in

the premature direction on items that appear to be

particularly salient for abuse. In particular, they

expect their children to "stop crying when told" and to

"sleep through the night" at premature ages. As a

group, the abusive parents were significantly less

accurate overall in their expectations for development

than the nonabusive parents.








13

Azar, Robinson, Hekimian, and Twentyman (1984)

propose that knowledge of developmental milestones may

be an inadequate test of the relationship between

unrealistic expectations and abuse. These authors

found no significant differences between maltreating

and non-maltreating mothers on a milestone

questionnaire; however, on a measure which required

mothers to generate alternative solutions to ten

typical child-rearing problems, maltreating mothers

reported a significantly greater number of unrealistic

expectations than did the comparison mothers. The

maltreating mothers expected more of young children in

the areas of family responsibility and care of

siblings.

Another research finding demonstrates that abusive

parents exhibit biased perceptions of child behavior.

Reid, Kavanagh, and Baldwin (1987) found that abusive

parents rated their children as more hyperactive and

aggressive even though their children's behavior did

not differ from that of a control group of children

when assessed by independent professional observation

(observations occurred over 6 home visits, 45-60

minutes in length and parent and child behaviors were

coded using the Family Interaction Coding System). The

authors note that the abusive parents' tendency to








14

perceive their children's behavior as more negative

represents a "parental bias."

Early research demonstrated another form of bias

in abusive mothers as evidenced by their tendency to

assign malevolent intent to their children's behavior

(Steele & Pollack, 1968). Larrance and Twentyman

(1983) presented further support for this hypothesis.

In a unique study, abusive and matched nonabusive

mothers were shown sets of pictures of their own child

and another child in six ambiguous situations. The

mothers were asked to tell a story about their child's

behavior including reasons why their child would behave

in such a way. Abusive and neglectful mothers'

expectations were significantly more negative than non-

abusive mothers', indicating that the abusive mothers

may be biased in judging their children's behavior. In

addition, it was found that abusive parents are more

likely to believe their children act intentionally to

annoy them. These mothers clearly evidenced negative

expectations for their children.

The early idea that abusive parents expect too

much of their children too soon has received equivocal

support from the research literature. Current research

suggests that the relationship is more complex--in some

areas abusive parents expect too much, but in other

areas they expect too little. It has been hypothesized








15

that abusive behavior may result from failure to meet

premature expectations in salient areas or "trigger"

situations. Further, in addition to the expectation

research, attempts have been made to identify parental

attitudes which may contribute to abuse (Wolfe, 1985).

There has been little prospective research aimed at

identifying prenatal attitudes and expectations that

postnatally relate to abusive behavior.

Prenatal Expectation and Postnatal Experience

Whether or not difficulty in adjusting to

motherhood arises in part from uninformed or

unrealistic expectations remains an empirical question.

It is plausible that overestimating the gratification

of motherhood and underestimating the difficulties

exacerbates adjustment to the new maternal role (Belsky

et al., 1986). In the few studies that have attempted

to address this question, measures of prenatal

expectation which were used are questionable in their

scope or validity. In fact, only two studies could be

found that longitudinally investigated the relationship

between prenatal expectation and postnatal experience.

Kach and McGhee (1982) assessed expectant parents'

prenatal expectations and postnatal experiences by

having them respond to 29 statements referring to

common problems and gratifications related to

parenthood. A small sample (N=12 couples) was








16

recruited solely from childbirth preparation classes.

Parents attitudes were assessed at 1 month prior to

birth and 2 months postpartum. Across subjects, no

significant differences emerged between prenatal

expectations and postnatal experiences. However, a

discrepancy between maternal prenatal expectations and

postnatal report correlated highly (r=.74, p<.01) with

a measure of problems adjusting to parenthood. This

relationship was true for mothers only; the correlation

for fathers was insignificant. This study provides

evidence to suggest that the accuracy of prenatal

expectations is related to postnatal problems for

mothers. Specifically, mothers with less accurate

expectations about parenthood were most likely to have

problems adjusting to parenthood. While this study

represents a first attempt to assess prenatal

expectations, several limitations are apparent. No

information is provided about the development of the

expectation measure, except to state that questions

were based on "common sources of conflict" identified

from the literature. The scope of the prebirth-

postbirth questionnaire was limited to caretaking

tasks, issues regarding money, time, and social

outings. No items assessed the prebirth-postbirth

expectations the mother held for the child or for

herself. The measure mainly assesses logistical








17

problems; its correlation with postnatal reports of

problems provides support for its use in this manner--

as a problem measure. Further, the use of a small,

nonrepresentative sample limits the generalizability of

this study.

Belsky, Ward, and Rovine (1986) conducted a second

longitudinal study examining the relationship between

prenatal expectation and postnatal experience. These

authors hypothesized that inaccurate or violated

expectations would undermine the adjustment to

parenthood. They suggest that expectations may be

violated in the negative or positive direction--i.e.,

that parenthood may turn out to be more positive or

less negative than expected. The primary focus of

their study was to determine if marital quality is

affected by violated expectations, whatever their

direction. Therefore, the prebirth expectation-

postbirth experience measure primarily assessed domains

pertaining to the marital relationship. Also assessed

were relationships with extended family and friends.

Results suggest that the average parent neither

anticipated nor experienced great positive or negative

consequences of having a baby. Examining individual

differences, however, revealed that when expectations

were violated (events were more negative than

expected), marriages change for the worst. The








18

opposite occurred when events turned out more positive

than expected. These results were strongest for the

mothers. In the case of mothers, up to 25% of the

variance in change in marital satisfaction is accounted

for by violated expectations. There was little

relationship between violated expectations of any kind

and change in the fathers' marital evaluations.

Consistent with previous research, women seem to be the

most adversely affected by negative postnatal events.

These two studies represent the only attempts to

examine the relationship between prenatal expectation

and postnatal experience. The results reveal that

expectations, variously defined, are related to

measures of postnatal adjustment in important ways.

Specifically, violated prenatal expectations in women

are related to increased report of problems in the

postnatal period, and decreased reports of marital

satisfaction. A common problem with the research in

this area is the reliance on data collected from

predominantly middle-class, two-parent families

attending childbirth preparation classes. It is likely

that samples derived primarily (or solely) from

childbirth classes are better prepared and have more

realistic expectations for childbirth and parenting;

the expectations, concerns and problems of this

population are likely to differ from less prepared,








19

lower socioeconomic samples. Belsky et al. (1986)

offer such an explanation for the unanticipated minimal

discrepancy between the prenatal and postnatal reports

found in their sample. The parents in their sample

were volunteers with intact marriages, were middle or

working class, and were likely well-prepared for having

a baby. The discrepancy in pre- to postnatal scores

might have been greater with a more diverse sample. A

second problem with the research in this area is the

lack of a systematic attempt to identify the scope of

maternal expectations, defined by expectant mothers

themselves. A previous study by Seibring (1991)

addressed this issue by developing a measure that

assessed maternal expectations and attitudes prior to

the birth of the baby. This study represented the

first attempt to identify areas of expectation from the

population of interest. The development of this

measure is described in the following section.

Development of the Maternal Expectations and
Attitudes Scale

The Maternal Expectations and Attitudes Scale

(MEAS) is a self-report instrument for first-time

mothers. Item generation was guided primarily by

extensive interviews with primiparous women. Forty

women (ages 15 to 35) participated in open-ended

interviews to determine the range of expectations they

have for their infants and themselves. The women were








20

sampled across the three trimesters. Ten women were

interviewed a second time postnatally to assess how

their experiences differed from their prenatal

expectations. Two obstetric nurses and a developmental

psychologist were also interviewed to assess common

misexpectations for birth, delivery, and the neonatal

period. Items were further derived from the research

literature on physical abuse to tap prenatal

pathological attitudes toward parenting. These

procedures resulted in the construction of thirty

preliminary items which were assessed for

appropriateness, content validity, clarity and

readability. The resulting item pool was heterogeneous

in nature and dealt with a range of issues. In an

attempt to develop meaningful subscales, five clinical

child psychologists assisted in grouping the items.

They sorted items into as many categories as they

deemed appropriate and provided a rationale for their

grouping. Four clear categories emerged which

reflected the following domains: maternal expectations

for self, maternal expectations for the child,

expectations for what makes a good or competent mother,

and attitudes thought to relate to pathological

parenting styles. These four rationally derived

subscales were labeled as follows: Maternal Expectation

(ME), Child Expectation (CE), Maternal Competence (MC),








21

and Pathological Parenting (PP). The judges sorted the

items again, this time into the four rationally derived

subscales. Items were assigned to a subscale by

majority agreement (4 of 5). Three items failed to be

categorized consistently and were subsequently omitted

from the measure. The remaining 27 items were

consistently assigned to one of the four subscales.

An unanticipated aspect of scale development arose

from the interviews. Many women described an affective

component to expectation, namely "disappointment,"

relating to both maternal and child expectations.

These women projected what their potential affective

reactions might be to unmet expectations. For example,

one mother reported that she would be very disappointed

if her baby did not smile at her right from birth. A

second mother reported that while she expected her baby

to smile at her right from birth, she would not be too

disappointed if this did not happen. It became clear

that women varied in the strength with which they held

certain expectations, and unexpectedly, they varied in

the extent to which they would be disappointed if the

expectation was not confirmed. Questionnaire items

were created to assess "disappointment" by rephrasing

the content of the original question. For example, "I

will be disappointed if does not happen." The

disappointment items were created relative to each core








22

item for the Child Expectation, Maternal Expectation,

and Maternal Competence subscales. This resulted in

three disappointment subscales: Maternal Expectation

Disappointment, Child Expectation Disappointment, and

Maternal Competence Disappointment. The resulting 47-

item, seven subscale measure is presented in Appendix

A. Items are rated on a 5-point continuum from

strongly agree to strongly disagree. A higher score on

an item (e.g., 5) is indicative of a more unrealistic

expectation.

Statistical Properties of the MEAS

Factor analyses were performed on the responses of

100 primiparous pregnant women to determine if

empirical support existed for the rationally derived

subscales. The empirically derived factors provide a

close approximation of three of the four rationally

derived subscales, and two of the disappointment

subscales. Maternal Competence failed to emerge as an

independent factor. A general maternal expectation

factor emerged comprised of maternal competence and

maternal expectation items. Test-retest reliability

information was obtained for a period of one to three

weeks. Results suggest that the MEAS exhibits adequate

test-retest reliability. Coefficients for the rational

subscales range in magnitude from .63 to .89, with a

reliability coefficient for the Total MEAS score of








23

.90. Cronbach's coefficient alpha was computed as a

measure of internal consistency. The coefficients were

sufficiently large to indicate an adequate degree of

internal consistency (coefficients for subscales range

in magnitude from .52 to .74 with a coefficient .86 for

the total score). Internal consistency and test-retest

reliability data for the factor-derived subscales were

equivalent to the rationally derived subscales.

Overall, the psychometric data suggest that the

rationally derived and factor-derived methods of

subscale derivation are equivalent.

Preliminary Validity

Measures of pregnancy anxiety and previous

experience with infants were administered to all women

in the sample. Support was found for the prediction

that first-time mothers with higher levels of previous

infant experience would report more realistic

expectations on the MEAS. Women with more formal

experiences with infants (e.g., day-care work) endorsed

items indicative of more realistic infant and maternal

expectations. In addition, higher levels of 1) overall

previous infant experience, 2) estimates of parenting

knowledge, and 3) formal experiences with infants were

negatively related to attitudes thought to reflect

pathological parenting. That is, women who report

higher levels of previous experience with infants are








24

less likely to endorse attitudes thought to reflect

pathological parenting attitudes. These results

provide some preliminary support for the construct

validity of the measure. Pregnancy anxiety was found

to be only marginally related to maternal expectations

and attitudes.
















CHAPTER III

STATEMENT OF THE PROBLEM

The development of the MEAS represents the first

attempt to assess maternal expectations and attitudes

in a range of areas, determined by the population of

interest. The introduction of this measure represents

a significant addition to the prenatal expectation

literature. Two studies were previously reviewed that

used measures of prenatal expectation. The measure

used by Kach and McGhee (1982) was limited in scope and

excluded domains identified by Seibring (1991) as

salient to primiparous women. The instrument used in

Belsky et al. (1986) was similarly developed from the

research literature; yet it focused on expectations for

the marital relationship. While this area of

expectation is undoubtedly related to postnatal

adjustment, it represents a domain that did not emerge

as salient in the development of the MEAS. The method

of item generation for the MEAS maximizes the

likelihood that the expectations and attitudes assessed

are salient and relevant to first-time mothers.








26

While preliminary data support the reliability and

validity of the MEAS, further research is needed to

establish its utility and validity. The purpose of the

current study was to attempt to replicate the factor

structure obtained in Seibring (1991) with a larger,

more representative sample, to provide information

regarding the construct validity of the measure, and to

provide more adequate data regarding test-retest

reliability. A limitation of the previous study was

the failure to assess the degree of social

defensiveness associated with responses on the MEAS.

Maternal responses on the MEAS may reflect a tendency

to present expectations in the best possible light.

This possibility was examined in the current study.

Finally, the relative psychometric properties of the

rational versus factor subscales were compared as a

secondary aspect of this study.

Phase I: Replication of the Factor Structure and
Examination of Test-Retest Reliability

The purpose of this phase was to examine the

replicability of the MEAS factor structure with a

larger, more representative sample of women. While the

sample size in the original study was adequate for the

analyses performed (e.g., Tabachnick & Fidell, 1989),

some authors have suggested that larger samples are

more appropriate for factor analyses (e.g., 200

subjects; Comrey, 1978). Therefore, 200 subjects were








27

recruited for the current study. A second aspect of

this phase was to assess the test-retest reliability of

the MEAS for a two-week interval. While Seibring

(1991) found adequate support for the test-retest

reliability, one limitation of the study was the

variability in the test-retest interval (second

administration ranged from 7-21 days). The current

study attempted to establish the test-retest

reliability of the MEAS within a more restricted,

consistent time period.

Phase II: Construct Validity

Construct validity of the MEAS was assessed by

examining the correlations between MEAS subscales and

other measures believed to be theoretically related to

constructs assessed by these subscales. Comparison

measures were found for the Maternal Expectation,

Maternal Competence, and Pathological Parenting

subscales. No appropriate comparison measure was found

for the Child Expectation subscale.

Phase III: Construct Validity

of the Child Expectation Subscale

Given that no appropriate comparison measure was

found to investigate the construct validity of the

Child Expectation subscale, this phase was designed to

provide information regarding its ability to

discriminate between two groups of expectant women.








28

The first group was comprised of women referred

for testing for a condition known as Intrauterine

Growth Retardation (IUGR). Intrauterine Growth

Retardation denotes a condition that results in the

failure of the fetus to attain its inherent growth

potential (Pollack & Divon, 1992). There are many

factors associated with the development of IUGR.

Etiological factors include genetic disorders,

congenital malformations, placental abnormalities (and

concomitant inferior fetal nutrition), and maternal

environmental causes (such as nutritional deprivation,

alcohol consumption, and cigarette smoking) (Pollack &

Divon, 1992). The clinical significance of IUGR is

based on the fact that birth weight is the most

important indicator of perinatal morbidity and

mortality (Watt, 1989). Possible consequences for IUGR

include developmental lags including increased

incidence of speech and language problems, cognitive

delays, minor attentional deficits, and minor

neurologic findings (Watt, 1989). The incidence of

neurodevelopmental problems increases significantly

with IUGR. In her review, Watt (1989) notes evidence

for behavioral problems as well. Maternal postpartum

reactions to IUGR infants commonly include worry and

concern over the health of their infants due to their

apparent frailty and small size (Watt, 1989).








29

Prenatal maternal reactions to deviations from a

normal pregnancy often include significant stress, due

in part to the violations of prior expectations (Parke

& Beitel, 1988). That is, a change in pregnancy status

to "abnormal" often causes anxiety and emotional

distress for the mother. Fearn, Hibbard, Laurence,

Roberts and Robinson (1982) assessed women with

elevated alpha fetoprotein (AFP) levels--a condition

that can be associated with fetal neural tube defects

and possible other congenital malformations--and found

all of the women in their sample to be extremely

anxious at the time of reassessment of the AFP

concentration. Even women who did not require further

evaluations (e.g., amniocentesis) after reassessment

continued to have some residual anxiety. Berne-

Fromell, Uddenberg, and Kjessler (1983) found that

there was a range of reactions from panic to denial of

fear after disclosure of abnormal serum AFP levels;

however, the majority of women experienced considerable

anxiety. All women shared their concern with their

spouses and 86% discussed this situation with others,

especially the women's own mothers.

No research has examined the potential impact an

abnormal pregnancy has on the mothers' expectations for

the child. However, Sjogren and Uddenberg (1988)

studied the emotional engagement with the fetus among








30

women undergoing prenatal diagnosis by amniocentesis or

chorionic villi biopsy. They found that 73% of the

women reported that they withdrew their feelings for

the child-to-be to some extent, perhaps preparing

themselves for the possible loss of the fetus (Sjogren

& Uddenberg, 1988). These results suggest that mothers

undergoing prenatal diagnosis may experience altered

feelings toward the fetus. Further, there is evidence

to suggest that the women's feelings about themselves

as mothers are also altered (Kemp & Page, 1987). Kemp

and Page (1987) report that women experiencing a high-

risk pregnancy report lowered maternal self-esteem

compared to women experiencing a normal pregnancy.

Based on the above, it seems reasonable to

hypothesize that maternal expectations for the child

might be altered as a result of an abnormal pregnancy.

To test this hypothesis in the current study, the

responses of women in the IUGR group were compared to

those of a matched comparison sample. Women in the

IUGR group were expected to report child expectations

that were significantly different from the "normal"

group based on their awareness of possible fetal

abnormality. In addition, we hypothesized that their

disappointment scores would differ from women's scores

in the comparison group. Confirmation of these








31

hypotheses would support the construct validity of the

Child Expectation subscale.














CHAPTER IV
METHOD

The purpose of this phase was to examine the

replicability of the factor structure of the Maternal

Expectations and Attitudes Scale with a larger, more

representative sample of women than that used in the initial

development of the measure. Further, test-retest

reliability of the MEAS was measured within a consistent

time frame using a subset of this sample.

Subjects

Subjects were 200 primiparous pregnant women from the

Central Florida area recruited from several locations: 1)

the prenatal clinic (N=133) or prenatal childbirth

preparation classes at Shands Hospital, Gainesville, FL

(N=33); 2) prenatal childbirth preparation class at North

Florida Regional Medical Center, Gainesville, FL (N=13); 3)

private obstetric practices in Gainesville, FL (N=21).

Measures

Background Information Questionnaire (Appendix B).

This measure assesses relevant background information

including maternal and paternal age, marital status, level

of education, race, occupation, whether or not the pregnancy

was planned, number of pregnancies not resulting in birth,

number of weeks pregnant, and whether the parents were

32










enrolled in any pregnancy preparation classes. Further,

women were asked to indicate whether or not they had

received any test results suggesting fetal abnormality.

This information was used to determine the demographic

characteristics of the sample, and provided necessary

information to estimate social status (Hollingshead, 1975).

Maternal Expectations and Attitudes Scale (MEAS)

(Appendix A). The development of this measure and its

statistical properties were previously described in the

literature review. The MEAS was administered to all

subjects.

Procedure

Each woman entering the prenatal clinic or childbirth

preparation class was asked if she was pregnant with her

first child, and the number of weeks pregnant was

determined. The women were given a standardized explanation

of the study stating an interest in the nature of first-time

mothers' expectations for their child and themselves as new

mothers. The questionnaires were administered while the

women waited for their appointment or class. Completion of

the questionnaires took approximately 35 minutes. As an

incentive to fully complete the study, subjects were told

that if all questionnaires were completed they could enter a

raffle contest for a $50.00 gift certificate redeemable at a

local toy store.








34

Fifty women completed the Maternal Expectations and

Attitudes Scale a second time to determine test-retest

reliability. The MEAS was administered the second time

approximately 2 weeks after the first administration (range

12-15 days).

Criteria for inclusion in the study were (1) first

pregnancy within the third trimester, (2) no children living

in the home, (3) no knowledge of fetal abnormality. Women

were not excluded on the basis of prior miscarriage or

abortion.














CHAPTER V

RESULTS

Phase I

Of the 200 women who agreed to participate, 76.3% were

Caucasian, 18.3% were African-American, 3.6% were Asian, and

1.8% were Hispanic. The women ranged in age from 16 to 44

years, with a mean age of 23 years. Fifty-eight percent of

the women were married, 38% were single, and 3.5% were

divorced. Sixty-nine percent lived with the baby's father,

while 30.6% did not. Twenty-nine percent of the subjects

reported that they had experienced at least one pregnancy

not resulting in birth (either by miscarriage or abortion);

all women who participated in the study were considered

primiparae. Fifty-five percent of the subjects reported

that the current pregnancy was not planned. They reported a

desire to have two children, on the average, with a range

reported from zero to four children. Nine of the women in

the sample indicated that they did not desire to have any

children at all. Approximately one-third (34%) were

planning to take, or had already taken, prepared childbirth

classes. The mothers' education levels ranged from 7 to 20

years with an average 12.8 years of education. The fathers'

education levels ranged from 7.5 to 21.5 years, with an










average of 13.2 years of education. Paternal age ranged

from 16-50 years old, with an average age of 26.

The Hollingshead Four-Factor Index of Social Status was

computed for each subject using education, occupation, sex,

and marital status to estimate level of social status.

Descriptive statistics for this sample are presented in

Table 5-1.

Analyses

Correlations with demographic data. To determine if

the MEAS might be biased when used with certain subjects,

correlations with a variety of subjects characteristics were

computed. Total scores on the MEAS were not found to be

related to race, marital status, number of weeks pregnant,

or residence. They were, however, related to maternal

education level and maternal age. Older women and those who

are more educated receive lower total scores on the MEAS

than younger, less educated women (r=-.30, E=.001; r=-.34

p=.001, respectively).

Test-retest reliability. Pearson product-moment

correlation coefficients were computed to estimate test-

retest correlations on the rationally derived subscale and

total scores of the MEAS. The correlation coefficients show

that the reliability of the Maternal Expectations and

Attitudes Scale measured within a period of 12-15 days is

adequate (Anastasi, 1988). The coefficients for the

subscales range from .69 to .86, with a reliability










coefficient for the Total MEAS score of .85. Test-retest

reliability results are presented in Table 5-2.

Internal consistency. Cronbach's Coefficient Alpha was

computed as a measure of internal consistency. Coefficient

alpha reflects the degree of homogeneity within a test and

is in effect a measure of the test's construct validity

(Anastasi, 1988). Alpha reliability coefficients were

determined for each subscale and the total score; the

coefficients are presented in Table 5-2. The rationally

derived version yielded coefficients that range in magnitude

from .57 to .88. The alpha reliability coefficient for the

total MEAS score is .88. These coefficients are

sufficiently large to indicate an adequate degree of

internal consistency for this measure. Indeed, the alpha

level for the total scale suggests a high degree of internal

consistency. While the alpha levels for subscales are

lower, the smaller number of items on each subscale limits

the coefficient alpha level that can be obtained

(Kleinmuntz, 1982). These data are presented in Table 5-2.

Test-retest reliability and internal consistency data

are presented for the factor-derived subscales following the

results of the factor analyses.









38

Table 5-1
Summary Descriptive Statistics (n=200)

Variable

Age Mean/Range
Maternal Age 23 16-44
Paternal Age 26 16-50

Race Percent
White 76.3%
Black 18.3%
Asian 3.6%
Hispanic 1.8%

Mothers' Education Level Percent
8th Grade or less 4.1%
9th to 12th Grade 46.2%
Community College 22.2%
College Graduate 16.4%
Graduate School 11.1%

Fathers' Education Level Percent
8th Grade or less 2.5%
9th to 12th Grade 51.6%
Community College 21.4%
College Graduate 8.8%
Graduate School 15.7%

Marital Status Percent
Single 38.2%
Married 58.2%
Divorced 3.5%

Hollingshead Index of Social Status
Unskilled 32.2%
Semiskilled 17.5%
Skilled, Clerical 11.7%
Minor Professional 17.5%
Major Professional 21.1%

Residence Percent
Home 34.5%
Parents' Home 22.8%
Apartment 21.1%
Trailer 18.1%
Campus 2.9%
Rooming House .6%











Table 5-2
Psychometric Properties of the
Rationally Derived MEAS Subscales



Subscale Internal
Number Consistency Test-Retest
of (Coefficient Reliability
Items Alpha) r


Child Expectation 8 .57 .86

Child Expectation
Disappointment 8 .72 .69

Maternal Expectation 7 .73 .80

Maternal Expectation
Disappointment 7 .70 .83

Maternal Competence 5 .60 .73

Maternal Competence
Disappointment 5 .70 .70

Pathological Parenting 7 .63 .81

Total Score 47 .88 .85











Factor Analysis of the Maternal Expectations and

Attitudes Scale

Replication of the Factor Structure. Preliminary

analyses were conducted to determine if factor analysis was

warranted, and to assess for potential problems of

multicollinearity or singularity (Tabachnick & Fidell,

1989). Kaiser's measure of sampling adequacy (MSA) was

further examined to ensure the appropriateness of factor

analysis.

Correlation patterns among the 47 items demonstrated

that variables appeared sufficiently correlated to warrant

factor analysis. Correlations, however, were not so large

as to indicate problems with multicollinearity or

singularity (Tabachnick & Fidell, 1989).

Principal factors extraction with an orthogonal varimax

rotation was performed (SAS, 1985) on the MEAS items. Here,

two separate factor analyses were performed: the first

analyzed the 27 expectation and attitude items on the MEAS;

the second analyzed the 20 disappointment items related to

the Maternal Expectation, Child Expectation, and Maternal

Competence subscales. Extraction and rotation techniques

conducted were identical to those used in Seibring (1991) to

allow for comparison of the factor structure.

Factor analysis of core items. An initial exploratory

factor analysis was performed without specification of the

number of factors to be retained. No eigenvalue cut-off was








41

specified for initial examination (Cliff, 1988). For the

core questions, three factors emerged with eigenvalues

greater than one. Examination of the scree plot (Cattell,

1966) provided evidence for the extraction of three to four

factors.

In separate analyses, both three and four factors were

rotated, and adequacy of the rotation was assessed by

examining the simple structure of the factors (Tabachnick &

Fidell, 1989). These findings revealed that a 3-factor

solution was most interpretable. As indicated by the

squared multiple correlations, all factors were internally

consistent and well-defined by the variables; the lowest of

the squared multiple correlations for factors was .27. In

general, items received high loadings on only one factor,

and loadings within factors appropriately reflected several

high and many low values (Tabachnick & Fidell, 1989). These

findings reveal an adequate rotation. With .40 specified as

the loading cutoff value (16% of the variance overlap

between variable and factor) no items loaded on more than

one factor; however, 2 of the 27 questions failed to load on

any factor with the specified cutoff (questions 1 and 2).

Tables 5-3 through 5-5 list the items loading on each

factor, with the rational scale assignment from Seibring

(1991) listed parenthetically. Factor loadings and the

percent of variance accounted for by the factors are also










shown in these tables. Variables are ordered and grouped by

size of loading.

The three factors together accounted for 83.04% of the

variance. Factor 1, accounting for 34.02% of the total

variance, was labeled Pathological Parenting. The items

loading on this factor represent an overlap of items that

were rationally labeled as pathological parenting and child

expectation items.

Four items loaded on this factor that were labeled by

the judges in Seibring (1991) as child expectation items;

however, three of these items were generated from the

literature pertaining to physical abuse. That is, the

research literature suggests that premature expectations in

certain salient, or "trigger" situations have been related

to parental abusive behavior. Ideas for the following three

items evolved from the physical abuse literature: "By six

months of age a baby should be able to stop crying when told

to" (item 24); "A newborn should be able to sleep through

the night before 8 weeks of age" (item 27); "During the

first six months, my child will understand what 'no' means"

(item 20). These items loaded on the first factor in this

study with strong relationships to other pathological

parenting items. The addition of these questions to the








43

Table 5-3
Core MEAS Items: Factor 1
Factor Loadings, Percents of Variance


Item


Fl: Pathological Parenting Factor
24 By 6 months of age a baby
should stop crying when
told to (CE)

22 Some babies are bad right
from the start and need to be
put in their place early on
so they can get a good start
in life (PP)

27 A newborn should be able to
sleep through the night before
8 weeks of age. (CE)

20 During the first 6 months, my
child will understand what
"NO" means (CE)

10 Sometimes it will be necessary
to spank my new baby to get
him/her to stop crying. (PP)

13 I believe that sometimes infants
act up just to embarrass their
mothers or make them mad. (PP)

21 I expect that after my baby is
born, my free time will not
change much (ME)


F1 F2 F3




.65 .06 .08





.59 .06 .02



.50 .07 .06



.50 .16 .09



.49 .06 .28


.48


.02 .20


.46 .21 .04


6 It is alright to shake an
infant hard to get them to
stop crying

4 If a baby has a difficult
personality, it's probably
the mother's fault


23 I expect that my baby will
smile at me right from birth (CE)


(PP) .44. .21 .07


(MC) .44 .02


.03


.41 .01 .36








44

Table 5-3 (continued)

Item F1 F2 F3

16 Mothers who pick their babies
up when they cry are spoiling
them (PP) .40 .27 .02

Percent variance explained by this factor: 34.02


Note: Designation in parentheses refers to the judges'
rational subscale assignment for the item.








45

Table 5-4
Core MEAS Items: Factor 2
Factor Loadings, Percents of Variance


Item


Fl F2 F3


F2: Maternal Expectation Factor

25 As a new mother, there will
be times I'd rather do other
things than take care of
my baby (ME) .16 .76 .20

18 Sometimes I will become angry
with my newborn (ME) .01 .65 .08

7 I don't think my new baby can
do anything to "get on my
nerves" (ME) .06 .55 .23

14 I believe there will be days
when I feel frustrated because
I can't do things I used to do
before the baby was born (ME) .17 .54 -.26

11 Even good mothers have days
when they have mixed feelings
about being a mother (MC) .29 .51 -.09

3 I imagine that it will be
tiring to get up at night
to feed the baby (ME) .06 .51 .08


19 Even if a baby is crying
for no reason, a good mother
should be able to calm him/her(MC) .30

8 A mother who gets frustrated
because she has to get up at
night isn't a very good mother(MC) .29


.41 .36



.40 .30


26 I expect that any mother would
lose her temper if her baby
won't stop crying (PP) .30 -.39 .05

Percent variance explained by this factor: 31.26


Note: Designation in parentheses refers to the judges'
rational subscale assignment for the item.









46

Table 5-5
Core MEAS Items: Factor 3
Factor Loadings, Percents of Variance


F1 F2 F3


F3: Expected Relationship Factor


12 My baby will recognize me
right from birth

5 I expect that my baby will
want to be with me more than
anyone else

15 I believe that a good mother
will always have the energy
she needs

9 I believe that my newborn will
enjoy playing with me right
from birth

17 I will be able to calm my
baby better than anyone
else can


(CE) -.10 .07


.55


(CE) .03 .03 .47



(MC) .34 .42 .43



(CE) .14 .28 .42



(ME) .17 .02 .40


Percent variance explained by this factor:


17.76


Note: Designation in parentheses refers to the judges'
rational subscale assignment for the item.


Item








47

pathological parenting factor shifts the nature of the

factor more in line with the literature.

Factor 2 accounted for 31.26% of the variance and was

comprised of a combination of items labeled by the judges as

Maternal Expectation and Maternal Competence. As in

Seibring (1991), the Maternal Competence items did not

emerge as a separate factor, providing consistent evidence

for one maternal factor reflecting expectations the mother

holds for herself. This factor was labeled Maternal

Expectation. Another possible interpretation of this factor

is that the items reflect expectations that are more

moderate and truthfully ambivalent in nature. Where factor

1 represents more extreme, pathological parenting notions,

factor 2 seems to capture more reasonable attitudes.

The third and final factor accounts for 17.76% of the

variance and corresponds most closely with the Child

Expectation factor found in Seibring (1991). The emphasis

of this factor in the current study seems to reflect

attitudes about the expected mother-child relationship.

That is, while continuing to reflect expectations for the

child, a more precise interpretation for this factor would

include the relational element of the items. This factor

was labeled Expected Relationship.

A direct comparison between the factor structure

obtained in Seibring (1991) and the current study will be








48

discussed following an examination of the disappointment

factors.

Factor analysis of disappointment items. A second,

separate factor analysis was performed on the 20

disappointment items related to the Maternal Expectation,

Child Expectation, and Maternal Competence subscales to

determine if dimensions of disappointment, corresponding to

maternal and child expectations, could be defined. Of

particular interest was whether the disappointment

dimensions approximate those defined by the factor analysis

of core items. To this end, principal factors extraction

with varimax rotation was performed (SAS, 1985) without

specification for number of factors to be retained.

Two factors emerged with eigenvalues greater than 1,

and examination of the scree plot suggested the retention of

two or three factors. The two-factor and three-factor

solutions were rotated and compared for interpretability.

The two-factor solution yielded the most interpretable

dimensions which together accounted for 66.15% of the

variance. No items loaded on more than one factor, and with

a specified factor loading cutoff of .40, two disappointment

items failed to load on either factor (items 5a and 12a).

Tables 5-6 and 5-7 list the items loading on each

factor with the assigned factor labels. Factor loadings and

percent of variance are also presented in these tables.

Variables are ordered and grouped by size of loading to aid








49

interpretation. Variables appear adequately defined by the

factor solution with final communality estimates ranging

from .17 (question 5a which failed to load) to .50.

Factor 1, accounting for 35.15% of the variance was

labeled Maternal Expectation Disappointment. Ten questions

loaded on this factor, all paired with either a Maternal

Expectation or Maternal Competence question. This reflects

the finding in the core questions that only one "maternal"

expectation factor emerged. The second factor accounted for

30.99% of the variance and was comprised of disappointment

items paired with 6 Child Expectation items, 1 Maternal

Competence item, and 1 Maternal Expectation item. The

Disappointment item for question 1 was deleted as the core

question failed to load on any factor. The preponderance of

questions loading on this factor pertain to child

expectation, and it was thus labeled Child Expectation

Disappointment factor.








50

Table 5-6
Disappointment Items: Factor 1
Factor Loadings, Percents of Variance


Core Item (presented for reference)
Disappointment Item Fl F2

Factor 1: Maternal Disappointment Factor

18 Sometimes I will become angry with
my newborn

18a I will be disappointed in myself
as a mother if this happens (ME) .67 .02

7 I don't think my new baby can do
anything to "get on my nerves"

7a I will be disappointed if I let my
baby "get on my nerves" (ME) .56 .29

11 Even good mothers have days when
they have mixed feelings about
being a mother

lla I will be disappointed in myself if
at times I have mixed feelings
about being a mother (ME) .55 .07

25 As a new mother, there will be
times that I'd rather do other
things than take care of my baby

25a I will be disappointed in myself
as a mother if this happens (ME) .55 .05

17 I will be able to calm my baby
better than anyone else can

17a I will be disappointed in myself
as a mother if I can't calm my
baby better than anyone else (ME) .54 .33

14 I believe there will be days when
I feel frustrated because I can't
do things I used to do before the
baby was born.










Table 5-6 (continued)

Core Item (presented for reference)
Disappointment Item F1 F2

14a I will be disappointed in myself
as a mother if this happens (ME) .54 .07

15 I believe that a good mother will
always have the energy she needs

15a I will be disappointed in myself
if I don't always have the energy
I need (MC) .54 .27

19 Even if a baby is crying for
no reason, a good mother should
be able to calm him/her

19a I will be disappointed in myself
if I am not able to calm my baby
when he/she is crying for no reason (MC).52 .36

8 A mother who gets frustrated because
she has to get up at night isn't a
very good mother

8a I will be disappointed in myself if
I get frustrated about getting
up at night (MC) .50 .12

3 I imagine that it will be tiring
to get up at night to feed the baby

3a I will be disappointed in myself if
I end up feeling tired (ME) .41 .38

Note: Designation in parentheses refers to the judges'
rational subscale assignment for the items.








52

Table 5-7
Disappointment Items: Factor 2
Factor Loadings, Percents of Variance


Core Item (presented for reference)
Disappointment Item Fl F2

Factor 2: Child Disappointment Factor

24 By 6 months of age a baby should be
able to stop crying when told to

24a I will be disappointed if this does
not happen (CE) .10 .70

23 I expect that my baby will smile
at me right from birth

23a I will be disappointed if this
does not happen (CE) .25 .66

27 A newborn should be able to sleep
through the night before 8 weeks

27a I will be disappointed if this
does not happen (CE) .03 .56

4 If a baby has a difficult
personality, it's probably
the mother's fault

4a I will be disappointed in myself
if my baby has a difficult
personality (MC) .34 .48

21 I expect that after my baby is
born my free time will not
change much

21a I will be disappointed if my
free time changes a lot (ME) .06 .47

20 During the first 6 months, my
child will understand what
"NO" means

20a I will be disappointed if this
does not happen (CE) .17 .47








53

Table 5-7 (continued)

Core Item (presented for reference)
Disappointment Item F1 F2

9 I believe that my newborn will
enjoy playing with me right from
birth

9a I will be disappointed if this does
not happen (CE) .35 .44

Note: Designation in parentheses refers to the judges'
rational subscale assignment for the item.










Comparing the Factor Structures from the Two Studies

The factor structures derived from the core and

disappointment items were compared to those obtained in

Seibring (1991) to determine whether the basic factor

structure of the MEAS had been replicated. The factor

structures were compared to determine if 1) both samples

generated the same number of factors; 2) the majority of

variables loaded highly on the same factors as in Seibring

(1991); 3) the same factor labels appear applicable to the

results of both studies.

Results revealed that the same number of core-item and

disappointment-item factors emerged. In both Seibring

(1991) and the current study, three strong factors for the

core items, and two factors for the disappointment items

were found.

The first factor in the current study matches most

closely with the Pathological Parenting factor in Seibring

(1991). Six of the seven salient items from Seibring (1991)

loaded on the first factor in this study. This factor is

further strengthened by the addition of four questions

believed to relate to pathological parenting expectations,

thus providing more support for the existence of a

pathological parenting factor. This factor was labeled

Pathological Parenting.

The second factor in the current study was labeled

Maternal Expectation. This factor most closely matches the








55

Maternal Expectation Factor from Seibring (1991). Six of

eight salient items from the Seibring (1991) Maternal

Expectation factor loaded, and two additional items were

added to the factor in the present analyses.

The third factor from the current study matches most

closely with the Child Expectation factor from Seibring

(1991) with three of the five salient items again loading on

this factor. These items, combined with two additional new

items, seem to reflect expectations that pertain to the

mother-child relationship. Thus, while encompassing some of

the core child expectation items, the present third factor

may more accurately be described as a relational factor, and

was labeled Expected Relationship for the purposes of

discussion.

Two disappointment factors were found in the current

study. The first directly corresponds to the Maternal

Expectation Disappointment factor in Seibring (1991) and was

similarly labeled in the current study. The same six

salient items loaded on this factor in both studies, and,

the current Maternal Expectation Disappointment factor was

strengthened by the addition of four other maternal

disappointment items. The second Disappointment factor

corresponds closely to the Child Expectation Disappointment

Factor found previously. Eight of 12 salient Child

Expectation items again loaded on this factor. Three of the

items that failed to load on this factor were "maternal"








56

items that loaded on the Maternal Expectation Disappointment

factor, thus making this a cleaner Child Expectation

Disappointment factor.

Overall, the structural criteria strongly suggests that

the factor composition for the core items has been

successfully replicated. To provide further, more

objective, support for this conclusion, Cattell's salient

similarity index, s, was computed (Cattell & Baggaley, 1960;

Cattell et al., 1969). This technique compared the patterns

of loadings from the two separate factor analyses.

The similarity index, s, was calculated from the full

set of factor loadings for the Pathological Parenting factor

from Seibring (1991) and the Pathological Parenting Factor

in the current study (factor 1). The s value was .87 which

exceeded the value expected by chance at E<.001. By this

test, the factors are highly similar. Next, the Maternal

Expectation factors were compared for the two studies

(Factor 2 in the current study). The s value was .73 which

exceeded the value expected by chance at p<.001, again

reflecting a strong similarity between these two factors.

Finally, s was calculated for the Child Expectation factor

from Seibring (1991) and the third factor in the current

study, labeled Expected Relationship factor. The s value

was .71, again exceeding the valued expected by chance at

p<.001, highlighting the similarity between these two

factors.










The similarity index was calculated from the full set

of factor loadings for the Maternal Expectation

Disappointment factors in both studies. The s value was

.85, exceeding the chance value. Similarly, the Child

Expectation Disappointment factors from the two studies were

highly similar (s=.89, g<.001).

A final objective step in evaluating the degree to

which the factor structure found in Seibring (1991) was

replicated was to compute Pearson's r for the full set of

loadings between pairs of factors obtained from the two

studies (Tabachnick & Fidell, 1988). For the core-item

factors, the three factor dyad comparisons were highly

significant, providing further evidence for the successful

replication of the factor structure. The Pathological

Parenting Factor (factor 1 in the current study) and the

Pathological Parenting factor from Seibring, 1991 were

correlated (r=.75, E=.001). The Maternal Expectation

factors from both studies correlated r=.88, D=.0001.

Finally, the correlation between the Child Expectation

factor (Seibring, 1991) and the Child Expectation or

Expected Relationship factor was r=.62, p=.0006.

Correlations between the Child Expectation

Disappointment and Maternal Expectation Disappointment

factors from both studies were highly significant, again

suggesting a successful factor replication for the

disappointment items (Maternal Expectation Disappointment








58
factor in both studies: =.90, E=.0001; Child Expectation

Disappointment r.77, R=.001).

Overall, these findings suggest that the factor pattern

found in Seibring (1991) has been successfully replicated.

The same number of factors was obtained in both studies, and

the structural criteria strongly suggest that the factor

composition for the core and disappointment items was

successfully replicated. Two more objective indices were

also used to measure the success of replication. Both

Cattell's index of similarity and Pearson's correlation

between pairs of factors from the two studies provide strong

evidence to suggest that the factor pattern in the current

study successfully replicates the results of Seibring

(1991).

The correlations and similarity indices are presented

in Table 5-8.











Table 5-8
Correlations between factors from Seibring (1991) and
Seibring (1993) and Cattell's Salient Similarity Index(s)


Pearson's r Similarity Index
Pathological Parenting
Factor (1991), (1993) r=.75 s=.87
p=.001 P<.001

Maternal Expectation Factor
(1991), (1993) r=.88 s=.73
p=.0001 P<.001

Child Expectation Factor(1991)
Expected Relationship r=.62 s=.71
Factor (1993) R=.0006 P<.001

Maternal Expectation
Disappointment Factors
(1991), (1993) r=.90 s=.85
E=.0001 p=.001

Child Expectation
Disappointment Factors
(1991), (1993) r=.77 s=.89
P=.001 p=.001








60

Test-Retest Reliability and Internal Consistency

for Factor-Derived Subscales

Test-retest reliability and internal consistency data

for the rationally derived subscales (Seibring, 1991) were

previously presented (Table 5-2). Data for the factor-

derived subscales are presented here for comparison.

Test-retest correlation coefficients for the factor-

analytically derived subscales range in magnitude from .69

to .82, with a reliability coefficient for the total MEAS

score of .84. These coefficients are not markedly different

from those obtained for the rationally derived subscales.

Cronbach's alpha coefficients were determined for each

subscale. The coefficients ranged in magnitude from .60 to

.82, with estimates consistently higher than those obtained

for the rationally derived version. The alpha level for the

total factor-derived scale was .88, equal to that for the

rationally derived total.

A one-to-one direct comparison between the two methods

is not possible as each method of scale construction

involves different items and numbers of items on each

subscale. However, a general comparison to characterize

these data does not provide strong support for the

statistical superiority of one method of subscale derivation

over the other, though the factor-derived version is more

internally consistent. These data are presented in Table 5-

9 for comparison.








61

Table 5-9
Test-Retest Reliability and Internal Consistency for
Factor-Derived and Rationally Derived MEAS Subscales


Number Internal Test-Retest
of Items Consistency Reliability (r)

Rational/Factor Rational/Factor Rational/Factor

PP/Factor 1


7 11


.63 .76


ME/Factor 2


7 9



7 10


CE/Factor 3

8 5


8 7


Total Score

47 42


.73 .75 .80 .80



.70 .82 .83 .80



.57 .60 .86 .82


.72 .76 .69 .69



.88 .88 .85 .84


Subscale Names
Factor 1
Factor 2
Factor 3


Pathological Parenting Factor
Maternal Expectation Factor
Expected Relationship Factor


CE Child Expectation
CED Child Expectation Disappointment
ME Maternal Expectation
MED Maternal Expectation Disappointment
PP Pathological Parenting
** Maternal Competence and Maternal Competence
Disappointment Subscales excluded as no comparable
factor emerged


:








62

Phase II: Construct Validity of the MEAS

Construct validity of the MEAS was assessed by

examining the correlations between MEAS subscales and other

measures believed to assess theoretically related

constructs. The comparison scales are described below.

Subjects

Subjects were the same 200 women described in phase

one. In addition to the measures described in that phase,

subjects also completed the following measures.

Measures

Maternal Self-Report Inventory (Appendix B). The MSI

(Shea & Tronick, 1988) is a 100-item self report inventory

which is described as a measure of maternal self-esteem

(Appendix D). The measure assesses five dimensions

identified from the literature which are believed to affect

adaptation to motherhood. These scales include measures of:

1) Caretaking ability (e.g., "I feel confident at being able

to know what my baby wants"); 2) General Ability and

Preparedness for Mothering Role (e.g., "I don't have much

confidence in my ability to help my baby learn new things");

3) Acceptance of Baby (e.g., "I have real doubts about

whether my baby will develop normally") ; 4) Expected

Relationship with Baby (e.g., "I worry about whether my baby

will like me"); 5) Parental Influence (e.g., "I expect that

I will be at least as good a mother as my mother was").

Subjects rate how true a statement is about them on a 5-








63

point scale ranging from completely false to completely

true, yielding five subscale scores. Maternal scores on

the MSI have been shown to be positively related to maternal

health, family support, maternal perception of the infant,

and mother-infant interaction (Shea & Tronick, 1988).

Specifically, mothers in good health with higher levels of

family support, who perceived their infants as above

average, and who displayed "competent" interactions with

their infants had higher levels of self-esteem as measured

by the MSI. The measure demonstrated adequate test-retest

reliability over a 4-week period (r=.85), and internal

consistency (Cronbach's alpha for the subscales ranged from

.66 to .89). For the current study, items phrased in the

present or past tense (e.g., "Feeding my baby is fun") were

rephrased to refer to future expectations for the child.

This measure was included to provide information

regarding the construct validity of the Maternal Expectation

and Maternal Competence subscales of the MEAS. These

subscales assess expectations the mother holds for herself,

and expectations for what a good or competent mother should

represent, respectively.

Child Abuse Potential Inventory (CAP-I) (Appendix C)

(Milner & Wimberly, 1979). This measure was included to

provide information regarding the construct validity of the

Pathological Parenting subscale of the MEAS. The CAP-I was

designed to identify individuals who abuse and neglect their








64

children. Subjects indicate whether they agree or disagree

with 334 items. Factor analyses of the CAP-I produced the

following four subscales: 1) Loneliness; 2) Rigidity; 3)

Problems; and 4) Control. The Loneliness factor contains

items that express feelings of being alone in the world;

items reflecting rigid beliefs about one's home, children,

and self comprise the Rigidity factor; the Problems factor

contains items describing problems with self, family,

friends, and things in general; the final factor, Control,

is comprised of items assessing a lack of social and self

control. The validity of the CAP-I has been established in

a series of studies (Milner & Wimberly, 1980; Milner, Gold &

Wimberly, 1986; Milner, Gold, Ayoub, & Jacewitz, 1984).

Scores on the CAP-I have clearly discriminated abusive

parents from nonabusive parents, with a reported overall

correct classification rate of 96% for abusive and

nonabusive individuals (Milner, 1979). Most errors are in

the direction of failing to label known abusers as abusive

(Milner, 1982).

A Lie Scale is incorporated within the CAP-I to detect

"fake-good" response sets (Milner, 1982). The statements

included on this subscale represent socially acceptable

values and behaviors not characteristic of many individuals

(e.g., "I always tell the truth"). The Lie subscale has

demonstrated moderate internal consistency (KR-20

coefficient=.72) and construct validity. The Lie Scale was








65

successful in identifying response sets of subjects

consciously attempting to deceive and present themselves in

a socially desirable manner (Robertson & Milner, 1985). The

Lie subscale was used in the current study to assess whether

subjects exhibited a "fake good" response bias.

Catalog of Previous Experience with Infants (COPE)

(Appendix D). This measure is comprised of 17 items which

assess the extent of the respondent's experience with

babies, ages birth to three years (MacPhee, 1981). Two

items were deleted for the purposes of this study because

they inquired about experience with the subject's own child.

Factors derived from the COPE relate to 1) formal and

informal experiences with infants (e.g., college classes and

direct observation of infants, respectively), 2) confidence

in one's knowledge about infants, and 3) knowledge obtained

from daycare experiences and spouse. MacPhee (1981)

established the validity of the COPE; however, the

reliability of the instrument remains untested.

Procedure

The procedure was the same as that described in Phase

I. Subjects were recruited at the time of their regularly

scheduled appointment. They received a standardized

explanation of the study, and completed the questionnaires

while waiting for their appointment.










Analyses

Construct validity. Pearson product-moment

correlations were computed between MEAS subscale scores and

scores on each of the validation measures. Correlations were

computed between the MEAS subscales as they were rationally

defined in Seibring, 1991 and the comparison measures.

Correlations between the three core factors generated in the

current study and the comparison measures are noted as well.

Examination of Table 5-10 reveals that two of the seven

predictions for the rationally derived subscales were

confirmed.

The maternal scales. Regarding specific subscales

of the MSI, we predicted that the Caretaking Ability

subscale would relate most strongly to the Maternal

Expectation subscale of the MEAS. Items on both the

Maternal Expectation subscale and the Caretaking Ability

subscale (MSI) reflect expectations the mother holds for

herself. Some items tap maternal feelings of competence

and were therefore predicted to relate secondarily to the

Maternal Competence subscale of the MEAS. Results

confirmed the hypothesis that the caretaking ability

subscale of the MSI is related to the Maternal

Expectations subscale of the MEAS (r=.32, E=.0005). The

MEAS Maternal Competence subscale, however, did not

significantly correlate with the caretaking scale.








67

The General Ability/Preparedness for Mothering Role

subscale of the MSI was hypothesized to relate to both

the Maternal Competence and Maternal Expectation

subscales of the MEAS. Items on the General Ability MSI

subscale predominantly reflect maternal feelings of

competence (e.g., "I feel like I will be a failure as a

mother"; "I feel confident about being able to teach my

baby new things"). Results confirmed that the General

Ability/Preparedness for Mothering subscale was

positively correlated with the Maternal Expectation

subscale (r=.29, p=.03), however, it was unrelated to the

Maternal Competence subscale.

The MSI Expected Relationship with Baby subscale

contains items regarding expectations for the mother-

child relationship, and was expected to relate most

strongly to the Maternal Competence subscale of the MEAS.

Some items tap maternal expectations not related to

competence, and the prediction was made that this

subscale would be moderately correlated with the Maternal

Expectation subscale of the MEAS as well. The MSI

Expected Relationship with Baby subscale was uncorrelated

with either the Maternal Expectation or Maternal

Competence subscales of the MEAS. In light of the

factor-analytic results which revealed that the Maternal

Competence subscale did not emerge as a separate factor

in either Seibring, 1991 or the current study, we








68

correlated the Maternal Expectation factor (a combination

of MC and ME items) with the MSI subscales and found that

it was significantly correlated with the MSI Caretaking

subscale (r=.26, E=.004), and while not significantly

related to the General Ability and Preparedness for

Mothering subscale, revealed a trend in that direction

(r=.16, p=.07). The Maternal Expectation factor was not

related to the Expected Relationship with Baby subscale,

however. Of note, is the relationship between the MEAS

Expected Relationship factor and the MSI Expected

Relationship with Baby subscale (r=.19, E=.02).

The Child Expectation subscale of the MEAS was

unrelated to the MSI Acceptance of Baby Scale as had been

proposed.

No specific predictions were made regarding how the

Parental Influence subscale of the MSI would relate to

any MEAS subscales. Results revealed a moderate negative

correlation between this scale and the Pathological

Parenting subscale of the MEAS (r=-.27, E=.001).









69

Table 5-10
Pearson Correlations of MEAS subscale scores with MSI
subscales


MEAS: ME MC CE PP
MSI Subscale
Caretaking Ability .32** NS

General Ability &
Preparedness for
Mothering .29* NS

Expected Relationship
with Baby NS NS

Acceptance of Baby -- -- NS

Parental Influence -- -- -.26**


MEAS FACTOR: PPfact MEfact ERfact
NSI Subscale
Caretaking Ability -.23** .25** .26**

General Ability &
Preparedness for
Mothering -.24** NS .19*

Expected Relationship
with Baby -.26** NS .20*

Acceptance of Baby NS NS NS

Parental Influence -.19* NS .20*



*E<.05. **P<.01.

ME=Maternal Expectation
MC=Maternal Competence
CE=Child Expectation
PP=Pathological Parenting
PPfact=Pathological Parenting Factor
MEfact=Maternal Expectation Factor
ERfact=Expected Relationship/Child Expectation Factor
--=no prediction
NS=not significant










Pathological parenting. The Child Abuse Potential

Inventory (CAP-I) was used in the current study to test

the construct validity of the Pathological Parenting

subscale of the MEAS. High scores on the Pathological

Parenting subscale are believed to reflect attitudes

consistent with a pathological parenting style. We

predicted a high positive correlation between scores on

the CAP-I and the Pathological Parenting subscale of the

MEAS. Results revealed a moderate correlation between

the two scales (r=.28, p=.03). The correlation between

the CAP-I and the Pathological Parenting Factor was

significant as well r=.20, R=.02.

Social Desirability

The Lie subscale of the CAP-I was used in the

current study as an index of social desirability. A

higher score on the Lie subscale would suggest a tendency

to endorse socially acceptable items in a way not

characteristic of many people. Likewise, higher scores

on the MEAS are indicative of more unrealistic

expectations, and higher scores may reflect a tendency to

answer in a socially desirable manner. Significant

positive correlations were obtained between the Lie scale

and the Maternal Expectation and Maternal Competence

subscales of the MEAS (r=.41, p=.0001; r=.32, p=.0001,

respectively).








71

To summarize, the predicted relationships between

the MEAS Maternal Expectation subscale and several

subscales of the MSI were confirmed, providing some

support for the construct validity of this subscale. The

Maternal Expectation Factor generated in the current

study was also significantly correlated with one subscale

of the MSI. The MEAS Maternal Competence scale was not

related in predicted ways to the MSI. Construct validity

of the MEAS Pathological Parenting subscale and the

Pathological Parenting factor was supported by the

moderate correlations with the CAP-I. Finally, responses

on the Maternal Expectation and Maternal Competence

subscales appear to be influenced by social desirability,

complicating the interpretation of scores on these

subscales.

Previous Experience and the MEAS

Previous experience with infants (assessed by the

Catolog of Previous Experience) was found to be related

to scores on the MEAS in the first study. The Catalog of

Previous Experience with infants (COPE) was again used to

further assess the relationship between scores on the

MEAS and previous experience with infants. Results were

similar to those in Seibring (1991). Specifically, the

MEAS total score was found to be significantly related to

a COPE factor which assesses various formal experiences

with infants (e.g., working in a professional childcare








72

setting, infant classes) (r=-.34, p=.0001). As would be

expected, increased levels of professional work

experience with infants and higher education classes are

associated with more accurate expectations (e.g., lower

scores on the MEAS).

The total score on the COPE was found to be

negatively related to the Pathological Parenting subscale

of the MEAS (r=-.24, E=.005). Higher scores on the COPE

reflect increasing levels of experience with, and

information about, infants. Thus, women with higher

levels of information and experience attain lower scores

on a subscale believed to tap pathological parenting

attitudes.








73

PHASE III: CONSTRUCT VALIDITY OF THE

CHILD EXPECTATION SUBSCALE

Given that no appropriate comparison measure was

found to assess the validity of the Child Expectation

(CE) subscale, this phase was designed to assess

construct validity by examining response differences in

two groups of expectant women. The groups are described

below.

Subjects

Subjects in Group A were 35 women recruited from the

Fetal Non-stress Test Clinic of Shands Hospital. Women

are referred to this service from the Shands Hospital

Women's Clinic and local obstetricians following

determination that they are carrying a child who is

growth retarded. All women had been given the diagnosis

for their child of Intrauterine Growth Retardation

(IUGR).

A second group was formed for comparison (Group B).

Group B was comprised of 35 women matched for age of the

mother, race, social status (Hollingshead score),

education level, and pregnancy trimester. In two cases,

the race of the subjects was not matched; however, these

subjects were matched on all other variables. Subjects

in Group B were recruited from the outpatient Women's

Clinic at Shands Hospital and local private obstetric










practices. Demographic information for the IUGR group is

presented in Table 5-11.

Table 5-11
Summary Descriptive Statistics (n=35) IUGR Group


Variable


Race Percent
White 72.0%
Black 24.0%
Hispanic 4.0%


Mothers' Education Level Percent
8th Grade or less 8.0%
9th to 12th Grade 64.0%
Community College 20.0%
College Graduate 8.0%
Graduate School 0%

Fathers' Education Level Percent
8th Grade or less 4.0%
9th to 12th Grade 84.0%
Community College 8.0%
College Graduate 4.0%
Graduate School 0%

Marital Status Percent
Single 40.0%
Married 56.0%
Divorced 4.0%

Hollingshead Index of Social Status
Unskilled 12.0%
Semiskilled 44.0%
Skilled, Clerical 32.0%
Minor Professional 12.0%
Major Professional 0%

Residence Percent
Home 32.0%
Parents' Home 20.0%
Apartment 20.0%
Trailer 28.0%












Procedure

The subjects in Group A were recruited from the

Nonstress Test clinic. The nonstress test involves

external fetal monitoring using a Dopler ultrasound

machine for approximately 30-60 minutes. The mother is

positioned on her back and two belts are strapped around

her abdomen: one contains a transducer that measures the

baby's heart rate, the other measures uterine

contractions. The mother is asked to keep a record of

when she feels a movement. The purpose of the test is to

verify an increase in the infants' heartrate when

movement occurs. The same obstetric nurse provided

education about the procedure and information about the

purpose of the test. The mothers were told that they

would receive feedback about their babies' status

following the procedure. Subjects completed the

questionnaires while they were in the midst of the fetal

nonstress test. A "manipulation" check was used to

ascertain whether or not subjects recognized the purpose

of the nonstress test. They were asked why they were

receiving the nonstress test, and what the nature of the

feedback was they were to receive. The purpose of this

inquiry was to assess whether the subjects understood

that they were receiving the sonogram because their

infant was growth-retarded, and to get a sense of their








76

understanding of the implications of this test.

Sufficient understanding was assumed if the subject

demonstrated an expressed knowledge that the current

status of their pregnancy was abnormal.

The subjects in the comparison group were recruited

at the time of their regularly scheduled prenatal

appointment. It was determined that none of these women

had any reason to believe that their pregnancy was

abnormal.

Measures

The Background Information Questionnaire and

Maternal Expectations and Attitudes Scale were completed

by all women. These measures were described in the

preceding phase.

Analyses

While the primary focus of this phase was to assess

group differences on the Child Expectation subscale,

differences on other subscales of the MEAS were also of

interest. Analyses were performed separately for the

rational and factor-derived versions of the measure.

Rationally Derived MEAS. Multivariate analysis of

variance was performed on the seven subscales as

dependent variables: Child Expectation, Child Expectation

Disappointment, Maternal Expectation, Maternal

Expectation Disappointment, Maternal Competence, Maternal

Competence Disappointment, and Pathological Parenting










(Seibring, 1991). Group served as the independent

variable with two levels (IUGR and Normal). The primary

purpose of this analysis was to test whether the two

groups differed significantly on their responses to the

Child Expectation subscale of the MEAS, therefore, this

dependent variable was given priority in the model.

The overall MANOVA was significant, using Wilks' Lambda

as criterion, F (7,64) = 6.05, R<.0001, indicating that

the pattern of responses on the MEAS scales was different

for the two groups.

The following MEAS subscales were found to

discriminate between groups: Child Expectation F (1,70) =

20.57, R<.0001; Maternal Competence Disappointment F

(1,70) = 3.79, p<.05, and Pathological Parenting F (1,70)

= 5.10, R<.05. As presented in Table 5-12, significant

differences on the MEAS subscale scores were in the

predicted direction, with IUGR mothers reporting child

expectations that were lowered in comparison to normal

mothers. No prediction was made about the Pathological

Parenting scale which discriminated between the two

groups. Normal mothers scored higher on the PP subscale

than IUGR subjects. The prediction that IUGR mothers

would report significantly higher child disappointment

scores was not confirmed. However, an unexpected finding

was that they reported significantly higher scores on the

Maternal Competence Disappointment scale, suggesting a










higher level of anticipated disappointment in themselves

if their expectations are not met.

Factor-Derived MEAS. Multivariate analysis of

variance was performed on the five factor-derived

subscales as dependent variables: Expected

Relationship/Child Expectation Factor, Maternal

Expectation Factor, Pathological Parenting Factor, Child

Expectation Disappointment Factor, and Maternal

Expectation Disappointment Factor.

The overall MANOVA was significant, using Wilks'

Lambda as criterion, F (,) = ., p<.001, indicating that

the responses differed for the two groups on the MEAS

factor-derived subscales. The following factor-derived

subscales were found to discriminate between groups:

Child Expectation Factor F (1,70) = 7.07, p<.01; and

Pathological Parenting Factor F (1,70) = 7.43, p<.01.

The group difference on the Child Expectation Factor was

again in the predicted direction, with IUGR mothers

reporting lowered expectations for their child relative

to normal mothers. There was no significant difference

in their anticipated child disappointment scores as had

been predicted. As with the rationally derived scale,

the Pathological Parenting factor discriminated between

the two groups with normal mothers receiving higher (more

deviant) scores.








79

In summary, the Child Expectation subscale of the

MEAS and the factor-derived version of this subscale

successfully discriminated between groups of subjects

predicted to differ in their expectations. The

hypothesis that the groups would further differ in their

anticipated level of disappointment was not borne out.

An exception to this was that IUGR subjects reported

higher levels of anticipated disappointment in themselves

(Maternal Competence Disappointment subscale). An

unanticipated finding was a significant difference in

pathological parenting scores with normal subjects

attaining higher average scores. These results are

presented in Table 5-12.













Table 5-12 Mean Scores on MEAS Subscales


Group

MEAS SUBSCALE IUGR NORMAL F alpha



Rationally Derived MEAS:
CE 19.13 22.67 F=20.57 ***
CED 18.13 18.61
ME 21.18 19.91
MED 18.56 18.58
PP 11.51 13.10 F=5.10 *
MC 12.38 12.03
MCD 13.67 12.36 F=3.79 *
Total 114.56 117.24
.........................................................
Factor-Derived MEAS:
ER/CE Factor 14.72 16.36 F=7.07 **
Maternal Factor 25.87 24.70
PP Factor 20.41 22.94 F=7.43 **
MED Factor 27.87 26.42
CED Factor 18.13 18.61
Total Factor 103.49 105.70


CE= Child Expectation
CED=Child Expectation Disappointment
ME=Maternal Expectation
MED=Maternal Expectation Disappointment
PP=Pathological Parenting
MC=Maternal Competence
MCD=Maternal Competence Disappointment

Factor-Derived MEAS:
Maternal Factor
ER/CE=Expected Relationship/Child Expectation Factor
PP=Pathological Parenting Factor
MED=Maternal Expectation Disappointment Factor
CED=Child Expectation Disappointment Factor

***E<.0001
** E<.01
* R<.05














CHAPTER VI

DISCUSSION

The purpose of the current study was to generate

psychometric data on the Maternal Expectations and

Attitudes Scale. A primary focus was to replicate the

factor structure of this measure with a larger and more

representative sample than was employed in the original

development of the measure. Further, information was

gathered on the construct validity and reliability of the

measure.

Results of the present study provide evidence for

five conclusions: 1) The factor structure of the MEAS

generated in Seibring (1991) appears stable, and was

successfully replicated in the current study; 2)

psychometric data support the reliability and internal

consistency of the measure; 3) reliable individual

differences in maternal expectations about motherhood can

be measured with the MEAS; 4) these individual

differences in expectations can discriminate between

mothers with normal and abnormal pregnancies; and 5)

support was found for the prediction that first-time

mothers with higher levels of previous infant experience

would report more realistic expectations on the MEAS.








82

In Phase I, several indices were used to assess

whether the factor pattern found in Seibring (1991) was

successfully replicated. As Cattell and Baggaley (1960)

note, "a factor once found remains merely a hypothesis

about a pattern; it is verified only after the pattern

has been found again ." (p. 33). Indices of

comparison for the two factor solutions included

structural comparison, comparisons of factor content,

factor correlation and indices of salient similarity.

All indices confirmed that the factor solution was

successfully replicated in this second study. This

finding minimizes the possibility that the results

obtained in Seibring (1991) reflect the chance

characteristics of one sample (Gorsuch, 1983).

To review, in Seibring (1991) four primary subscales

were derived by a rational/intuitive approach to test

construction. These subscales were labeled Maternal

Expectation, Child Expectation, Maternal Competence, and

Pathological Parenting. Three disappointment scales were

created relative to the Maternal Expectation, Child

Expectation and Maternal Competence subscales to assess

affective reaction to expectations. The factor analysis

of Seibring (1991) produced a three-factor solution for

the core items corroborating the MEAS rational subscales;

however, the Maternal Competence subscale failed to

emerge as a separate factor. A two-factor solution for








83

the disappointment items yielded Child Expectation and

Maternal Expectation Disappointment factors.

In the current study, results from factor analyses

provided empirical support for the same number of core-

item and disappointment-item factors. In both Seibring

(1991) and the current study, three strong factors

derived from core items, and two factors from analyses of

disappointment items were found. This factor analytic

model is strikingly similar to the a priori model

outlining domains of expectation generated from the

initial item-construction interviews (Seibring, 1991).

Again, the Maternal Competence subscale failed to emerge

as a separate factor, but was subsumed within a more

general Maternal Expectation Factor.

The three factors for the core items essentially

retained the essence of those generated in the first

study. However, the redistribution of several items in

the present study provides a possible shift in factor

interpretation. The Pathological Parenting factor was

strengthened by the addition of items that were derived

from the physical abuse literature, but were identified

by the judges in Seibring (1991) as related to

expectations for the child (items noted on page 47).

These items are clearly expectations for the child, but

are extreme in nature and more appropriately classified

as pathological parenting attitudes. The Maternal Factor








84

in the current study represents a mix of both intuitively

labeled Maternal Competence and Maternal Expectation

items. Further, many items on this factor can be viewed

as representing more moderate or ambivalent expectations.

Finally, the third factor essentially replicates the

Child Expectation factor from the first study, but shifts

in tone to reflect a relational component pertaining to

the expected mother-child relationship and was thus

labeled Expected Relationship factor.

Phase II of the current study assessed the construct

validity of three of the rationally derived subscales

(Maternal Expectation, Maternal Competence, and

Pathological Parenting) and the three core-item factors

(Maternal Expectation, Pathological Parenting, and

Expected Relationship). The results of this phase provide

evidence for the construct validity of the overall MEAS

and two of its subscales as measures of expectation.

The construct validity of the Maternal Expectation

and Maternal Competence subscales was assessed by

correlating these measures with designated subscales of

the Maternal Self-Report Inventory (MSI), a measure of

maternal self-esteem. The predicted moderate

correlations between the MEAS Maternal Expectation Scale

and the MSI subscales received some support, indicating

that the Maternal Expectation subscale taps a similar but

not identical, construct to that measured by the MSI








85

Caretaking Ability, and General Ability/Preparedness for

Mothering subscales. Further, the Maternal Expectation

factor generated in the current study was also

significantly related to the MSI Caretaking Ability

subscale, providing some support for the construct

validity of this factor. The Maternal Competence

subscale, however, was unrelated to the MSI. This

finding, in conjunction with the factor analytic results,

supports the conclusion that the Maternal Competence

subscale does not represent a unique construct in the

MEAS, but rather is a source of variance best subsumed

within a more general Maternal Expectation factor.

The MSI, as noted, was designed as a measure of

maternal self-esteem, such that higher scores reflect a

style of holding ones maternal abilities in higher

esteem. Correlations with the MEAS Maternal Expectation

subscale and Maternal Factor suggest that mothers with

higher expectations for themselves possess concomitant

higher levels of maternal self-esteem. An alternative

interpretation is that higher scores on both measures

represent overidealized ideas of motherhood. Future

research must ascertain whether high scores on this MEAS

scale are adaptive or predictive of future difficulty.

For the purposes of the current study, the Maternal

Expectation subscale appears to bear some relationship to








86

the construct of maternal self-esteem as measured by the

MSI.

The Child Abuse Potential Inventory (CAP-I) was used

in the current study to assess the construct validity of

the MEAS Pathological Parenting subscale. Scores on the

CAP-I have been shown to discriminate between physically

abusive and non-abusive parents. Higher scores on the

CAP-I Abuse subscale reflect a tendency to endorse

attitudes associated with physically abusive behavior.

The correlation between the CAP-I Abuse subscale and the

MEAS Pathological Parenting subscale and Pathological

Parenting factor suggests that they are similarly

measuring the construct of abusive parenting attitudes.

Significant correlations between the MEAS subscales,

MEAS factors and the validation measures, provide some

evidence for the construct validity of these scales

(Cronbach & Meehl, 1955). The comparison scales were

selected for their perceived relationship to the

construct tapped on each of the MEAS subscales.

A final step toward assessing the validity of the

MEAS was to compare the scores of two groups of women on

the Child Expectation subscale as no appropriate

comparison measure was found. In this phase, the Child

Expectation subscale and Expected Relationship Factor

were able to discriminate between mothers carrying

infants with the diagnosis of Intrauterine Growth








87

Retardation and those of normal pregnancy status. The

IUGR group mothers scored significantly lower on a

subscale designed to tap expectations for the child.

That is, their expectations tended to be more realistic.

It seems that experiencing an abnormal pregnancy and

receiving information about possible long-term sequelae

may minimize a tendency toward overoptimism or idealism

in forming expectations for the child. This finding is

in line with research that suggests that some women

undergoing prenatal diagnostic tests report a withdrawal

of feelings for the child (Sjogren & Udenberg, 1988).

The results of the current study strengthen the finding

that maternal attitudes toward the child are affected by

diagnostic information. The ability of the MEAS to

discriminate between different diagnostic groups of

mothers provides strong support for its validity and

potential clinical utility.

The prediction that the IUGR group would report

significantly higher Child Expectation Disappointment

scores than mothers in the normal group was not

confirmed. However, an unexpected finding was that they

reported significantly higher scores on the Maternal

Competence Disappointment scale, suggesting a higher

level of anticipated disappointment in themselves if

their expectations are not met. This finding provides

some support for the hypothesis that women enduring high-








88

risk pregnancies may experience a sense of lowered self-

esteem (Kemp & Page, 1987). On the MEAS, this lowering

of maternal self-esteem may be reflected in the tendency

to report anticipated disappointment in themselves.

Future research must determine if this tendency endures

beyond the pregnancy and postpartum period.

Another unexpected difference emerged between the

two groups of expectant women. Both the rational and

factor-derived Pathological Parenting subscale

discriminated between the two groups with normal mothers

scoring higher on the subscale than IUGR subjects. This

finding may further support the hypothesized tendency to

lower child expectations in the midst of experiencing a

high-risk pregnancy. This hypothesized tendency to lower

expectations may extend to those more extreme in nature

as found on the Pathological Parenting subscale.

Overall, significant differences occurred between the

IUGR and Normal pregnancy groups on three of the seven

rationally derived MEAS subscales, and two of the five

MEAS factors. The ability of the MEAS to discriminate

between groups of mothers provides promising evidence not

only for its validity, but for its future utility in

research and clinical application.

One difficulty in establishing the validity of the

MEAS is the correlation between two of the MEAS subscales

(ME and MC) and the Lie scale of the CAP-I. It appears








89

that higher scores on the Maternal Expectation and

Maternal Competence subscales may result from subjects'

tendency to portray themselves favorably; it becomes

difficult to separate this influence from actual

differences in expectations. Future research should

attempt to control for this factor, as it seems that a

portion of the variability in MEAS scores may be

attributable to a social desirability factor.

A second difficulty establishing the validity of the

MEAS is the relationship between the MEAS total score and

demographic characteristics (e.g., the negative

correlation between the MEAS total and maternal age and

level of education). Older, more educated women

produced, overall, lower MEAS total scores. One possible

explanation for this finding may be the effect of

increasing levels of experience on expectations.

Specifically, older mothers may have more experience with

infants and children, and consequently, may hold more

realistic expectations. Indeed, in the current study,

increased levels of previous experience with infants were

related to more accurate expectations (e.g., lower scores

on the MEAS). Research previously reviewed on the

relationship between experience and expectations revealed

that experienced multiparouss) mothers rate their infants

as temperamentally easier (Bates, Freeland & Lounsbury,

1979), are less prone to depression (Ventura, 1982), find








90

infant cries less aversive (Zeskind, 1980), and are more

responsive to infants than inexperienced parents

(Blakemore, 1981). In the current study, older and more

educated women (presumed to have more informal experience

with infants due to their age, and more formal experience

through their education) overall reported expectations

for their child-to-be that were more realistic in nature.

Future research, however, must continue to assess the

problematic relationship between MEAS scores and

demographic characteristics.

A final objective in the current study was to

provide more information regarding the psychometric

properties of the MEAS. One aim was to assess test-

retest reliability within a consistent time frame.

Results suggest that the MEAS is reliable across a

timeframe of 12-15 days. Further, internal consistency

was quite high. In both Seibring (1991) and the current

study, the MEAS has been shown to have both adequate

internal consistency and test-retest reliability. This

finding is true for both the rational and the factor-

derived versions of the scale.

Integration of Findings with other Research

This research project began with a basic question:

why are some women more affected by the transition to

motherhood than others? Review of the literature reveals

that little systematic attention has been devoted to










studying this important transition. The current study is

the second by this author aimed at identifying one area

believed salient in the transition to motherhood. A

major premise in this work is that prenatal maternal

expectations and attitudes mediate the adjustment to

motherhood.

Research suggests that first-time mothers, in

particular, tend to overestimate the gratification of

motherhood and underestimate the difficulties (Entwisle &

Doering, 1981). Some studies have attempted to address

the consequences that might arise from a discrepancy

between prenatal expectations and postnatal experience

(Kach & McGhee, 1982; Belsky, Ward, & Rovine, 1986).

Kach and McGhee (1982) note that the prenatal-postnatal

discrepancy relates to reports of increased adjustment

problems. The higher the discrepancy, the more

difficulty mothers report adjusting. One explanation for

this increased difficulty adjusting to motherhood may be

offered by the present study. In response to imagined

unmet expectations, new mothers in our study were prone

to report more expected disappointment in themselves--an

element which might adversely affect the adjustment phase

and may be reflected in increased reports of problems.

Similarly, Belsky et al. (1986) found that when

expectations were violated in the negative direction

(worse than expected), marriages changed for the worst.








92

Our findings converge with their results in that a change

in pregnancy status (worse than expected) seems to

deflate the buffer of overidealized views of motherhood

(lowered expectations) and increase anticipated

disappointment. To elaborate, deviations from a normal

pregnancy are experienced as a violation of expectations

(Park & Beitel, 1988), causing increased anxiety (Fearn

et al., 1982) and, we hypothesized, an alteration in

attitudes and expectations for the child-to-be.

Sjogren and Uddenberg (1989) noted a change in emotional

engagement with the fetus as a withdrawal of feelings for

the child-to-be. No research, prior to this study, has

directly examined the impact of violated expectations (in

the form of changed pregnancy status) on maternal

expectations for the child. The results suggest that

indeed a change in pregnancy status is associated with

differences in expectations for the child.

There is little known about changes in the mothers'

feelings toward herself as a result of an abnormal

pregnancy. Kemp and Page (1987) document a lowering of

maternal self-esteem in response to a high-risk

pregnancy. The effect such an event has on the mothers'

expectations for herself has not been studied.

Results of the current study support and advance the

findings of this previous research. Consistent with Kemp

and Page (1987), there was a difference in maternal self-








93

perception between mothers experiencing a high-risk

versus normal pregnancy. Mothers in the high-risk group

anticipated experiencing significantly more

disappointment in themselves in the face of unmet

expectations.

The change in IUGR mothers' expectations for the

child toward more realistic ideas leads to some

conjecture. It may be that in the normal course of

events new mothers bolster their maternal self-esteem by

overidealizing or holding unrealistic views of the child-

-perhaps a necessary aid in the early, and often

difficult, phase of transition. Receiving negative

information about the pregnancy, the child-to-be, or the

self, may override this mechanism and produce more

realistic beliefs. However, the impact this has on the

transition to motherhood is unknown. The results of the

current study suggest that violation of expectations,

even prenatally, significantly influence maternal

perceptions. It is our belief that future research in

the postnatal period will demonstrate even more profound

effects of violated expectations.

This research has some implications for applied

concerns. It may be possible to develop interventions

aimed at providing accurate information and realistic

expectations in more extreme cases. It may be feasible,

for example, to begin screening expectations and








94

attitudes as mothers enter clinics or classes for

prenatal preparation, using an inventory such as the

MEAS.

Future Use of the MEAS: Rational versus Factor-Derived

The successful replication of the factor structure

for the MEAS provides some support for its continued use

in future research. Following is a review of the

psychometric comparability of the rational scales versus

the factor analytic scales generated in the current

study. Comparison between the two methods of subscale

construction reveals the following: the factor method of

construction produces consistently higher estimates of

internal consistency while test-retest reliability for

both versions is comparable.

Construct validity findings were presented for both

methods of subscale construction. Predictions about the

relationship between the MEAS and comparison measures

were supported in three instances for the rational

version, and in two instances for the factor-derived

version. The rational version significantly

discriminated between groups in Phase III on three

subscales, versus two significant effects for the factor

subscales.

Overall, it appears that the psychometric properties

for the two methods of subscale construction are fairly

equivalent; however, the rationally derived version was




Full Text
xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID EQ8HEPGZN_VO8INH INGEST_TIME 2012-09-24T14:34:05Z PACKAGE AA00011842_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES