10
Inlanl Menlal Heahli Iouriial. Vol. X, No. 2, Suniiinu 19x7 Temperament and Play Interaction Behavior Across Infancy TIFFANY FIELD, SHERILYN ADLER, NITZA VEGA-LAHR, FRANK SCAFIDI AND SHERI GOLDSTEIN Mailman Center for Child Development University of Miami Medical School ABSTRACT: Temporal stability and convergent validity of infant temperament ratings and play interac- tion behaviors were examined across early to late infancy. Stability was noted for the temperament dimensions activity, rhythmicity, intensity, mood, and persistence. Stability also was noted for the infant play in- teraction behaviors looking, smiling, and vocalizing. In addition, convergent validity was noted for attentive- affective behaviors and comparable temperament ratings; for example, gaze aversion and nondistractability, and distress brow behavior and negative mood. Generally easier temperament infants were noted to vocalize more and to cry less frequently during their play interactions. RESUME: Une etude sur la stabilitk temporelle et la validite convergente du caractere du norrisson a ete menee jusqu’a la fin de la petite enfance. La stabilite a ite etudiee pour I’activite, le rythme, I’intensite, I’hurneur et la persistence, ainsi que pour les comportements ludiques du nourisson au niveau du regard, du sourire et de la vocalisation. En plus, la validite convergente a ete etudiee pour les comportements attention-affection, et des measures de caractere, comme par exemple, le refus de regarder et de se laisser distraire, et les froncements du front marquant le mecontentement et les humeurs negatives. En general, les enfants de nature plus facile vocalisaient davantage et pleuraient moins souvent pendant les jeux. RESUMEN: Se examino la estabilidad temporal y la validez convergente de la clasificacion del temperamento infantil y las conductas de interaccion Iudica, a traves de la infancia. Se noto estabilidad en la actividad de las dirnensiones del temperamento, ritmo, intensidad, humor y persistencia. Tambien se hallo estabilidad en las conductas de la interaccion ludica infantil siguientes: contact0 visual, sonrisa y vocalizacion. Ademas, se noto validez convergente en conductas de atencionfecto y clasificaciones comparables de temperamento, por ejernplo, aversion visual y falta de distraccion, enfurruiiamiento y ma1 humor. Generalmente se noto que 10s niiios con un temperamento mas llevadero vocalizaban mas y lloraban menos frecuentemente durante las interacciones Iudicas. Temperament is one of those few developmental phenomena that imply continuity. Parents commonly claim that their children continue to show the same temperament (Plomin, 1982), and significant stability has been noted for parent ratings of infant This research was supported by NIMH Research Scientist Development Award #MH00331 to the first author. We wish to thank the teachers, parents, and infants who participated in this study and Kerry Col- lins, Marcia Soto, and Wendy Tuttle for their research assistance. Correspondence should be addressed to Tiffany Field, Mailman Center for Child Development, University of Miami Medical School, P.O. Box 016820, Miami, FL 33101. OMichigan Association fnr lnFnni h l ~ n p - 1 Uanlrh

Temperament and play interaction behavior across infancy

Embed Size (px)

Citation preview

In lan l Menlal H e a h l i Iour i ia l . V o l . X, N o . 2 , Suni i inu 19x7

Temperament and Play Interaction Behavior Across Infancy

TIFFANY FIELD, SHERILYN ADLER, NITZA VEGA-LAHR, FRANK SCAFIDI A N D SHERI GOLDSTEIN

Mailman Center fo r Child Development University of Miami Medical School

ABSTRACT: Temporal stability and convergent validity of infant temperament ratings and play interac- tion behaviors were examined across early to late infancy. Stability was noted for the temperament dimensions activity, rhythmicity, intensity, mood, and persistence. Stability also was noted for the infant play in- teraction behaviors looking, smiling, and vocalizing. In addition, convergent validity was noted for attentive- affective behaviors and comparable temperament ratings; for example, gaze aversion and nondistractability, and distress brow behavior and negative mood. Generally easier temperament infants were noted to vocalize more and to cry less frequently during their play interactions.

RESUME: Une etude sur la stabilitk temporelle et la validite convergente du caractere du norrisson a ete menee jusqu’a la fin de la petite enfance. La stabilite a i t e etudiee pour I’activite, le rythme, I’intensite, I’hurneur et la persistence, ainsi que pour les comportements ludiques du nourisson au niveau du regard, du sourire et de la vocalisation. En plus, la validite convergente a ete etudiee pour les comportements attention-affection, et des measures de caractere, comme par exemple, le refus de regarder et de se laisser distraire, et les froncements du front marquant le mecontentement et les humeurs negatives. En general, les enfants de nature plus facile vocalisaient davantage et pleuraient moins souvent pendant les jeux.

RESUMEN: Se examino la estabilidad temporal y la validez convergente de la clasificacion del temperamento infantil y las conductas de interaccion Iudica, a traves de la infancia. Se noto estabilidad en la actividad de las dirnensiones del temperamento, ritmo, intensidad, humor y persistencia. Tambien se hallo estabilidad en las conductas de la interaccion ludica infantil siguientes: contact0 visual, sonrisa y vocalizacion. Ademas, se noto validez convergente en conductas de atencionfecto y clasificaciones comparables de temperamento, por ejernplo, aversion visual y falta de distraccion, enfurruiiamiento y ma1 humor. Generalmente se noto que 10s niiios con un temperamento mas llevadero vocalizaban mas y lloraban menos frecuentemente durante las interacciones Iudicas.

Temperament is one of those few developmental phenomena that imply continuity. Parents commonly claim that their children continue to show the same temperament (Plomin, 1982), and significant stability has been noted for parent ratings of infant

This research was supported by NIMH Research Scientist Development Award #MH00331 to the first author. We wish to thank the teachers, parents, and infants who participated in this study and Kerry Col- lins, Marcia Soto, and Wendy Tuttle for their research assistance.

Correspondence should be addressed to Tiffany Field, Mailman Center for Child Development, University of Miami Medical School, P.O. Box 016820, Miami, FL 33101.

OMichigan Association fnr lnFnni h l ~ n p - 1 U a n l r h

T. Field, S . Adler, N . Vega-Lahr, F. Scafidi and S . Goldstein 157

temperament (Hagekull & Bohlin, 1981; McDevitt & Carey, 1981; Rothbart & Der- ryberry, 198 1). Although stability coefficients are only moderate, these ratings may underestimate the stability of temperament in part because parent reports potentially are biased by their perceptions or expectations (Goldsmith & Campos, 1986; Plomin, 1982; Vaughn, Taraldson, Crichton, & Egeland, 1981).

To investigate this potential parent bias in temperament ratings, researchers have examined the convergent validity of temperament characteristics reported by parents and infant behaviors recorded by observers. Convergence has been measured between parent ratings of infant temperament and infant behavior observed in the context of laboratory tasks (Wilson & Matheny, 1983), test-taking situations (Roth, Eisenberg, & Sell, 1984), and the strange situation-attachment classification paradigm (Goldsmith & Campos, 1986; Sroufe, 1985). Others have observed infant behaviors such as cry- ing and fussiness in the home and have correlated these with temperament dimen- sions (Bates, Freeland, & Lounsbury, 1979). Low convergence between temperament scale ratings and behavior observations confirmed investigators’ interpretations of temperament data as parental perceptions of temperament rather than accurate descriptions of infants’ temperamental characteristics (Bates et al., 1979).

Because mothers’ ratings of infant temperament may be based primarily on the behaviors that they observe during their interactions with their infants, greater con- vergence might be expected between their ratings of temperament and observers’ ratings of infant behavior during mother-infant interactions. Vaughn and his col- leagues (Vaughn, Taraldson, Crichton, & Egeland, 198 1) observed mother-infant feeding and play interactions when the infants were 6 months old and compared com- posite ratings of those interactions for easy vs. difficult temperament infants. Because no differences were noted between the easy/difficult groups, they concluded that difficult temperament was not related to the behaviors of either the infant or the mother during feeding and play interactions. Most of the behavior factors used in their cor- relations, however, were based on mother behavior ratings, and the authors did not indicate what infant behaviors comprised their only infant factor, labeled “Baby’s involvement in play.” The ratings for the infant factor, just as the easy-difficult sum- mary temperament rating, may be too global and, thus, may mask relationships between specific temperamental dimensions and specific infant play behaviors. In ad- dition, ratings per se rather than coded behavior may be too subjective. In a study by Greenberg and Field (1982), for example, temperament ratings based on observed mother-infant interactions were related only modestly to the corresponding tempera- ment ratings by mothers. Specific temperamental characteristics may be related to specific play behaviors, e.g., the activity temperament dimension to motor activity during play, when those behaviors are coded rather than rated.

Rothbart (1986) investigated the relationships among specific infant behaviors that were coded at 3, 6, and 9 months during mother-infant feeding, bathing, dressing, and play interactions in the home and specific temperament dimensions of the same label. The behaviors included activity level, smiling and laughter, fear, distress to limitations, vocal activity, and reactivity, the same dimensions that comprise Rothbart’s Infant Behavior Questionnaire. Given that the same speciflc behaviors were assessed by both mothers and observers and that the observations were coded rather than rated, it is perhaps not surprising that moderate intercorrelations were noted

158 Infant Mental Health Journal

for observed behavior and parent temperament ratings. These relationships probably were facilitated by the specificity of the temperament dimensions on the Infant Behavior Questionnaire. Relations between temperament dimensions and play interac- tion behaviors were perhaps rarely investigated prior to Rothbart’s (1986) study because the traditional temperament scales featured dimensions that were more global and subjective (e.g., mood) and, thus, less quantifiable than interaction behaviors (e.g., smiling and laughter).

Rothbart (1986) also noted significant stability across the first 9 months, not only for temperament ratings, but also for interaction behaviors. While several studies have been conducted on stability of infant temperament ratings, stability of infants’ play interaction behavior rarely has been assessed across infancy, presumably because different behaviors have been observed at different time periods, e.g., face-to-face social behaviors during early infancy and toy play behavior during later infancy. Another possible reason why stability of play interaction behaviors has not been assessed across infancy is that very little stability has been noted across short inter- vals during early infancy. For example, Field et al. (Field, Vega-Lahr, Goldstein, & Scafidi, 1987a) noted stability only in distress brow behavior from 4 to 8 months, and others have noted greater stability in parental social behaviors than in infant behaviors across the first 9 months (Belsky, Gilstrap, & Rovine, 1984) and from 6 to 13 months (Pettit & Bates, 1984). Thus, even less stability might be expected across infancy because it is a longer interval. Nonetheless, the stability of play interaction behavior reported by Rothbart (1986) suggests that when temperament-like behaviors (e.g., activity level and SmilingAaughter) are observed in a play interaction context, stability can be found at least across the first 9 months of infancy.

The present study assessed the stability of infants’ temperament ratings, the stability of their play interaction behaviors, and the relations between specific temperament characteristics and specific play interaction behaviors across infancy (3 to 20 months). This study differed from Rothbart’s 1986 study in a number of ways. First, a longer interval separated the assessments, i.e., 3 to 20 months vs. 3 to 9 months. Second, traditional temperament scales were used, scales that featured more global tempera- ment dimensions. Third, the face-to-face interaction play context was employed at 3 months because this is the most widely investigated interaction situation at this stage of infancy, and a floorplay interaction play context was studied at 20 months for the same reason. And, finally, the specific behaviors coded were those traditionally observed in the face-to-face and floorplay contexts. Thus, while specific play interac- tion behaviors were coded instead of rated, these behaviors were less global than the traditional temperament ratings, which presumably diminished the probability of noting convergence between play behavior and temperament ratings, as has been mentioned already. Similarly, the long interval between assessments reduces the chances of finding stability. Nonetheless, because the protocol of many longitudinal studies (like this one) on both normal and atypical infants featured the use of tradi- tional temperament scales and the coding of specific play behaviors during early face- to-face interactions and later floorplay interactions, those data bases may provide an important opportunity for examining stability and convergence of temperament ratings and play interaction behaviors across infancy. Thus, at 3 months and at 20

T. Field, S . Adler, N . Vega-Lahr, F. Scafidi and S . Goldstein 159

months infants and mothers were videotaped in play interactions together, and the mothers completed temperament ratings on their infants.

METHOD Sample

The sample was comprised of 26 normal infants (13 males and 13 females) and their middle-SES, college-educated mothers (M age = 31 years). The sample was recruited from the local birth register.

Procedure

For the mother-infant interactions at 3 months (this being an optimal age for mother-infant face-to-face play), the mother and infant were filmed in an infant videotaping laboratory. The infant was positioned in a semi-upright infant seat on a table opposite the mother in an en face position with approximately 18 inches be- tween their faces. Two videocameras, positioned 6 feet from the dyad in opposite corners of the room, and a split-screen generator enabled a mixing of the images of the mother’s face and torso and the entire body of the infant. Five-minute face-to- face interactions then were videotaped for each of the infant-mother dyads. For these interactions, the mothers simply were asked to pretend that they were at home play- ing with their infant.

For the mother-infant interactions at 20 months (an optimal age for mother-infant floorplay interactions) the same dyads were filmed in the same laboratory. The playroom floor was furnished with a selection of age-appropriate toys, which included puppets, a Jack-in-the-box, a busy box, a musical ball, stacking and connecting toys, and preschool Legos. Two videocameras were positioned in opposite corners of the room such that the entire room could be filmed and the images mixed on a split- screen recorder. The mothers then were asked to play with their infants for 10 minutes.

The videotapes from each of these visits subsequently were coded by two research assistants,,who were naive to the purpose of the study and who used event recorders to code the absolute duration of either the mother’s or the infant’s behaviors. The following behaviors were coded for the 3-month infant-mother dyads: Infant smil- ing, vocalizing, motor activity (limb or trunk movements), distress brow (knitted brow), gaze aversion, and crying, and for the mother, smiling, exaggerated facial expressions, vocalizing, looking at infant, touching, and moving the infant’s limbs. For the 20-month infant-mother dyads the following behaviors were coded for the infant and the mother: Looking at the other, vocalizing, affectionate play (smiling, laughing), and constructive play (play behavior that involved a toy).

Intercoder reliability was determined by the sjmultaneous coding of one-third of the videotapes from each observation period and was calculated by Cohen’s Kappa (Bartko & Carpenter, 1976), a chance-corrected statistic. For the 3-month observa- tions, the intercoder reliabilities for the infant behaviors averaged .87 (R = .82-.93), and for the mother behaviors they averaged .92 (R = .82-.97). For the 20-month observations the intercoder reliabilities for the infant behaviors averaged .88 (R = 33-.93), and for the mother behaviors they averaged .92 ( R = .87-.97).

I60 Infant Mental Health Journal

After the 3-month interaction observation, the mothers were asked to complete the Revised Infant Temperament Questionnaire (RITQ; Carey & McDevitt, 1978). Although it is conceivable that the infants’ behavioral style during the interactions may have biased the mothers’ subsequent assessment of the infants’ temperament, this was not considered a problem inasmuch as mothers typically base their assessments of infant temperament on their interactions with their infants, and presumably this interaction situation was representative of their typical interactions. In addition, if the mothers did base their assessments on this interaction, a clearer test could be made of the convergence between mother and observer assessments of infant behavior. Scores on the temperament questionnaire then were derived for the nine tempera- ment dimensions: (a) activity - amount and rigor of motor activity; (b) rhythmicity - regularity of eating, sleeping, and toileting patterns; (c) adaptability - adjust- ment to new routines and places; (d) approach - responsivity to novel objects and friendliness to strangers; (e) threshold - responses to intense stimulation and to changes in stimulation; (f) intensity - intensity of responses to stimulation; (8) mood - affective reactions to people and to daily routines; (h) distractability - degree to which child can be distracted from ongoing activity, such as crying; and (i) per- sistence - degree to which activities, such as play, are sustained.

After the 20-month interaction session, the mothers were given the toddler version of the temperament scale designed by the same authors and called the Toddler Temperament Scale (TTS; Fullard, McDevitt, & Carey, 1979). This scale yields the same dimensions as the Revised Infant Temperament Questionnaire.

RESULTS

The data first were submitted to Hotelling’s T2 (multivariate r test) to determine any sex differences. Because the Hotelling’s T2 yielded no significant sex differences, the data for both sexes were combined in the subsequent correlation analyses.

To determine the stability of temperament ratings, the correlation coefficients for the 3- and 20-month ratings were examined. As can be seen in Table 1, of the nine

Table 1 Correlation Coeficients for Relations Between 3- and 20-month Infant Temperament Ratings

Dimension 3 - 20 months

Activity

Rhythmicity Adaptability

Approach

Threshold

Intensity

Mood Distractability

Persistence

.39*

. 51* *

. I 1

.09

.11

.39*

.53**

.06

.57**

*p< .05. **p< .01.

T. Field, S. Adler, N . Vega-Lahr, F. Scafidi and S . Goldstein 161

possible correlations between 3- and 20-month temperament dimensions, five dimen- sions were correlated significantly (activity, rhythmicity, intensity, mood, and per- sistence), which suggests stability on at least five of the nine dimensions.

To determine stability of interaction behaviors, the correlation coefficients for the 3- and 20-month interaction behaviors were examined. (See Table 2.) Only those behaviors that were labeled the same or could be considered similar at the 3- and 20-month periods were entered into the analysis as follows: (a) for the infant, vocaliz- ing at each period, smiling at 3 months and affectionate play at 20 months and gaze aversion at 3 months and looking at 20 months; (b) for the mother, vocalizing and looking at infant for each period and smiling, exaggerated facial expressions and touching at 3 months and affectionate play at 20 months. Of the eight possible cor- relations, all were significant. Stability was noted in the following infant and mother behaviors: (a) infant smiling at 3 months and affectionate play (smiling, laughing) at 20 months; (b) infant vocalizing at 3 and 20 months; (c) infant gaze aversion at 3 months and looking at mother at 20 months (a negative relationship); (d) mother smiling, displaying exaggerated faces, and touching infant at 3 months and affec- tionate play at 20 months; (e) mother vocalizing at 3 and 20 months; and (f) mother looking at infant at 3 and 20 months.

Table 2 Correlation Coeficients for Significant Relations Between 3- and 20-month Play Interaction Behaviors

Infant behaviors

3 months 20 months r ~

Smiling Vocalizing Gaze aversion

~ ~

Affectionate Play (Smiling/Laughing) .52**

Vocalizing .49**

Looking at mother - .40* ~~

Mother behaviors

3 months 20 months r

Smiling Affectionate play Exaggerated facial expressions Affectionate play Touching infant Affectionate play Vocalizing Vocalizing Looking at infant Looking at infant

.57**

.61**

.49**

.8l***

. 51* *

*p< .05. **p< .01. ***p<.oo1.

The correlation matrix then was examined to determine relations between tempera- ment ratings and interaction behaviors at 3 and 20 months and concurrent relations between mother and infant behavior. At 3 months, of the 60 possible correlations between the 10 temperament ratings and 6 interaction behaviors, 7 were significant (a greater than chance number of relations); (a) higher activity was related to more frequent motor activity (r = - .53, p < .01) (this correlation is negative because higher activity on the temperament scale is assigned a lower rating); (b) higher threshold

162 Infant Mental Health Journal

was related to more frequent crying ( r = - .42, p < .05) (this correlation is negative because higher threshold is assigned a lower rating on temperament scale); (c) higher intensity was related to more frequent vocalizations ( r = - .50, p < . O l ) ; (d) more negative mood was related to more frequent distress brow (r = .51, p < .Ol); (e) less distractability was related to more frequent gaze aversion (r = .52, p <.01); and (f) generally easier temperament (summary rating) was related to more frequent vocalizing (r = - .40, p < .05) and less frequent crying ( r = .51, p < .01). At 20 months, 3 of the possible 40 correlations were significant (a marginally greater than chance number of relations); (a) lower activity level was related to more frequent looking at the mother (r = .44, p < -05); (b) greater distractability was related to more frequent vocalizing (r = - .48, p < .Ol); and (c) greater persistence was related to more frequent constructive play (r = - .52, p < .01). No significant relations were noted between mother and infant play behaviors at 3 or at 20 months.

DISCUSSION

The absence of sex differences in infant temperament and interaction behavior is consistent with other literature. Rothbart and her colleagues, for example, consistently have reported no significant sex differences in infant temperament (Rothbart, 1986; Rothbart, Furby, Kelly, & Hamilton, 1977) or in infant interaction behaviors (Rothbart, 1986). Although Rothbart (1986) speculated that sex differences may emerge after the first 9 months of infancy that she studied, there were no apparent sex differences in the infants of the present study as late as 20 months. It is possible, however, that a microanalysis of a larger sample of interaction variables on a larger sample of infants would reveal sex differences.

The moderate stability of temperament ratings is also consistent with reports by other investigators, including Rothbart and Derryberry (1981), who reported a me- dian of .48 for stability in temperament dimensions across the first 9 months, and McDevitt and Carey (1981), who noted a median of .38 for stability in temperament ratings from 4 to 23 months. Based on the same scales as those used by McDevitt and Carey (1981) over a similar time span (3 to 20 months), stability of similar tempera- ment dimensions was noted in the present study. McDevitt and Carey (1981) reported that activity level, rhythmicity, mood, and persistence were at or above the median, the same dimensions (plus intensity) that were temporally stable in the present study. Stability has been reported for similar dimensions on different scales, including ac- tivity level for the first 9 months on the Infant Behavior Questionnaire (Rothbart & Derryberry, 1981) and activity level, emotionality, and persistence from 4 to 18 months on the Colorado Childhood Temperament Inventory (Field, Vega-Lahr, Scafidi, & Goldstein, 1987b). Thus, data from at least four studies converge to sug- gest moderate stability on the same temperament dimensions across infancy.

The stability of infant play interaction behaviors across the 3- to 20-month period is somewhat surprising given that very little stability has been reported in infant behavior across early infancy and over shorter intervals (Field et al., 1987a; Fish & Crockenberg, 1981; Moss, 1967). It is possible that because major motor milestones develop during the second half of the first year (e.g., sitting, crawling, walking) or because individual differences occur in the rate of organization of social responses,

T. Field, S. Adler, N . Vega-Lahr, F. Scajdi and S. Coldstein 163

such as attachment behavior and stranger fear, this may be a transitional period dur- ing which stable social characteristics (e.g., an attentive, smiley, vocal infant) are masked. How attentive, smiley, and vocal an infant is during this period may de- pend in part on how he is coping with and performing these motor tasks and the rate at which he is developing attachment and stranger fear behaviors. Social behavior during this transitional period may not be consistent with the sociability that is noted before and after this period. Thus, despite the longer 3- to 20-month interval, in- traindividual stability of social behavior may appear to be greater across these two points in development, which are relatively free of frustrating developmental milestones.

Although greater stability has been noted for parental social behaviors than for infant behaviors over the first 9 months (Belsky et al., 1984) and from 6 to 13 months (Pettit & Bates, 1984), several behaviors of both the mothers and infants were moderately stable in this study even though no relations were noted between mother and infant behavior at either the 3- or 20-month stages. The similarity of behaviors that were stable for both mothers and infants (smiling, vocalizing, and attentiveness to each other) suggest that consistency in the mothers’ behaviors may have contributed to stability of infant behaviors or that their interactions or the relationship that derived from their interactions facilitated the intraindividual stability of their behaviors (Stevenson-Hinde, 1985). Unfortunately, there is no way to determine from these data how much the individual members of the dyad or their relationship contributed to the variance in the stability of their play behaviors across age.

For at least two of the infant interaction behaviors that were reliably stable across infancy in this study (i.e., smiling/laughing and vocalizing), Rothbart (1986) also reported significant stability across the first 9 months. Visual attentiveness, the other stable play interaction behavior noted in this study, is similar to the persistence/at- tentiveness temperament rating for which this study and that of McDevitt and Carey (1981) have noted temporal stability. A priori, one might expect very little intrain- dividual stability in play behaviors over infancy because of the growth of social and motor skills and the change in functional significance of these behaviors. However, if temperament dimensions are viewed as composites of specific behaviors that a per- son observes and then formulates an impression of a more global characteristic (called a temperament dimension), there may be no reason to expect less intraindividual stability in the components than in the global descriptor. Thus, for example, an in- fant who consistently is rated as more positive in mood is probably an infant who consistently shows less distress brow.

Significant relations among temperament dimensions and interaction behaviors in this study lend support to the convergent validity between play behaviors observed in the home and those reported by parents on the Infant Behavior Questionnaire in the Rothbart (1986) study. Unlike the Rothbart (1986) study, in which the same behaviors were tapped by both play observations and parent questionnaires, the present study employed traditional temperament rating scale dimensions and com- monly observed play interaction behaviors, dimensions and behaviors that are labeled differently and differed in their specificity, thus diminishing the probability of finding convergent validity. Nonetheless, some interesting relations emerged among tempera- ment dimensions and play interaction behaviors that are conceptually similar and

I64 Infant Mental Health Journal

intuitively reasonable including: (a) greater activity ratings and more frequent motor activity; (b) higher threshold and more frequent crying; (c) greater intensity and more frequent vocalizing; (d) more negative mood and more frequent distress brow; (d) less distractability and more frequent gaze aversion; (f) greater persistence and more frequent constructive play; and (g) generally easier temperament and less frequent crying, but more frequent vocalizing. Thus, it appears that several of the dimensions tapped by the Revised Infant Temperament Questionnaire and the Toddler Tempera- ment Scale are congruent with observable interaction behaviors and that mothers may be relatively unbiased in their reports on infant temperament.

The moderate stability of infant temperament ratings and play interaction behaviors in this relatively small sample (one-half the size of the Rothbart, 1986, sample and one-quarter the size of the McDevitt and Carey, 1981, sample) and the convergent validity of traditional temperament ratings by mothers and interaction behaviors coded by observers (despite the different behavior labels) may relate to the homogeneity of the socioeconomic and educational level of this sample. Both Rothbart (1986) and McDevitt and Carey (1981) studied broad socioeconomic status (SES) samples, for which larger samples would be required to show stability and convergent validity. In contrast, this sample was exclusively middle SES and college educated, which prob- ably contributed to stability and convergent validity. Stability of personality develop- ment might be expected in more stable and optimal socioeconomic environments, and unbiased, objective reports might be expected from more educated parents. Con- tinuities in development might be greater in environments that consistently foster in- fants’ temperament traits and play interaction behaviors.

REFERENCES

Bartko, J. J. , &Carpenter, W. T. (1976). On the methods of reliability. Journalof Nervousand Mental Disease, 163, 307-317.

Bates, J. E., Freeland, C. A. B., & Lounsbury, M. L. (1979). Measure of infant difficultness. Child Develop- ment, 50, 794-803.

Belsky, J., Gilstrap, B., & Rovine, M. (1984). The Pennsylvania Infant and Family Development Project, I: Stability and change in mother-infant and father-infant interaction in a family setting at one, three, and nine months. Child Development, 55, 692-705.

Carey, W. B., & McDevitt, S. C. (1978). Revision of the infant temperament questionnaire. Pediatrics,

Field, T., Vega-Lahr, N., Goldstein, S., & Scafidi, F. (1987a). Face-to-face interaction behavior across early infancy. Infant Behavior and Development, IO, 111-116..

Field, T., Vega-Lahr, N., Scafidi, F., & Goldstein, S. (1987b). Reliability, stability, and relationships be- tween infant and parent temperament. Infant Behavior and Development, IO, 117-122.

Fish, M., & Crockenberg, S. (1981). Correlates and antecedents of nine-month infant behavior and mother- infant interaction. Infant Behavior and Development, 4, 69-81.

Fullard, W., McDevitt, S. C., & Carey, W. B. (1979). The Toddler Temperament Scale. Unpublished manuscript. (Available from Dr. W. B. Carey, 319 West Front Street, Media, PA 19063.)

Goldsmith, H. H., & Campos, J. J. (1986). Fundamental issues in the study of early temperament: The Denver twin temperament study. In M. Lamb & A. Brown (Eds.), Advances in developmental psychology. Hillsdale, NJ: Erlbaum.

Greenberg, R., & Field, T. (1982). Temperament ratings of handicapped infants during classroom, mother and teacher interactions. Journal of Pediatric Psychology, 7, 387-405.

61, 735-739.

T. Field, S . Adler, N . Vega-Lahr, F. Scajidi and S . Goldstein 165

Hagekull, B., & Bohlin, G. (1981). Individual stability in dimensions of infant behavior. Infant Behavior and Development, 4, 97-108.

McDevitt, S. C., & Carey, W. B. (1981). Stability of ratings vs. perceptions of temperament from early infancy to 1-3 years. American Journal of Orthopsychiatry, 51, 342-345.

Moss, H. A. (1967). Sex, age, and state as determinants of mother-infant interaction. Merrill-Palmer Quarter- ly, 13, 19-36.

Pettit, G. S., & Bates, J. E. (1984). Continuity of individual differences in the mother-infant relationship from 6 to 13 months. Child Development, 55, 729-739.

Plomin, R. (1982). Behavioral genetics and temperament. In R. Porter & G. M. Collins (Eds.), Tempera- mental differences in infants and young children (Ciba Foundation Symposium 89). London: Pitman.

Roth, K., Eisenberg, N., &Sell, E. R. (1984). The relation of preterm and full-term infants’temperament to test-taking behaviors and developmental states. Infant Behavior and Development, 7, 495-506.

Rothbart, M. K. (1986). Longitudinal observation of infant temperament. Developmental Psychology,

Rothbart, M. K., & Derryberry, D. (1981). Development of individual differences in temperament. In M. E. Lamb & A. L. Brown (Eds.), Advances in developmentalpsychology (pp. 383-400). New York: Medical and Scientific Books.

Rothbart, M. K . , Furby, L., Kelly, S. R., & Hamilton, J . S. (1977, March). Development of a caretaker report temperament scale for use with 3-, 6-, 9-, and 12-month old infants. Paper presented at the meeting of the Society for Research in Child Development, New Orleans.

Sroufe, L. A. (1985). Attachment classification from the perspective of infant-caregiver relationships and infant temperament. Child Development, 56, 1-14.

Stevenson-Hinde, J. (1985, April). Q-sort attachment data and temperament. Paper presented at the meeting of the Society for Research in Child Development, Toronto.

Vaughn, B. E., Taraldson, B. J., Crichton, L., & Egeland, B. (1981). The assessment of infant tempera- ment: A critique of the Carey Infant Temperament Questionnaire. Infant Behavior and Develop- ment, 4, 1-17.

Wilson, R. S., & Matheny, A. (1983). Assessment of temperament in infant twins. Developmental

22, 356-365.

Psychology, 19, 172- 183.