This article was downloaded by: [Monash University Library] On: 11 April 2015, At: 17:29 Publisher: Routledge Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

The Journal of Genetic Psychology: Research and Theory on Human Development Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/vgnt20

How Do Only Children Differ from Other Children? Steven Mellor

a

a

Department of Psychology , Pennsylvania State University , USA Published online: 06 Jul 2010.

To cite this article: Steven Mellor (1990) How Do Only Children Differ from Other Children?, The Journal of Genetic Psychology: Research and Theory on Human Development, 151:2, 221-230, DOI: 10.1080/00221325.1990.9914656 To link to this article: http://dx.doi.org/10.1080/00221325.1990.9914656

PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden.

Downloaded by [Monash University Library] at 17:29 11 April 2015

Terms & Conditions of access and use can be found at http://www.tandfonline.com/ page/terms-and-conditions

The Journal of Genetic Psychology, 151(2), 221-230

Downloaded by [Monash University Library] at 17:29 11 April 2015

How Do Only Children Differ fkom Other Children? Steven Mellor Department of Psychology Pennsylvania State University

ABSTRACT. Developmental outcomes of only and non-only children, categorized by birth order and by family size, were investigated. Multiple hypotheses based on meta-analyses of the only-child literature (Falbo & Polit, 1986) were tested with independent comparison techniques. Results indicated that developmental outcomes of only children were similar to outcomes for firstborns and children from two-child families but dissimilar to outcomes for later borns and children from larger families. Furthermore, outcomes were more positive for only children, firstborns, and children from two-child families than for all other comparison groups. In agreement with Falbo and Polit, only-child deprivation and only-child uniqueness explanations for outcome differences were not supported in favor of an explanation that emphasizes qualities of the parent-child relationship. Results suggest that future comparisons to onlychild outcomes should preserve the independence of contrast results and expand the only-child category to include data from firstborns and others from two-child families. HOW DO ONLY CHILDREN differ from other children? Falbo and Polit (1986) reported the results of six meta-analyses of the research literature on the only child since 1925 and concluded that several of the developmental outcomes of only children were similar to the outcomes for firstborn children and children from two-child families but dissimilar to outcomes for children from three- or four-child families and children from five- or more child families. Despite dissimilarities between only children and a group of non-only The “we” referenced throughout the article expresses my appreciation of support to student research assistants who helped collect, enter, and analyze data for the study. Most notable were Anne Cordaro, Kathy MacGregor, Barbara Masteller, and Kelly Schmidt. Appreciation is also extended to Denise F. Polit, who provided helpful comments on an earlier draft of the article. Requests for reprints should be sent to Steven Mellor at the Department of Psychology, 41 7 Moore Building, Pennsylvania State University, University Park, PA 16802. 22 I

Downloaded by [Monash University Library] at 17:29 11 April 2015

222

The Journal of Genetic Psychology

children, Falbo and Polit ruled out explanatory mechanisms relating to onlychild deprivation and only-child uniqueness in favor of theoretical propositions that emphasized qualities of the parent-child relationship. Although Falbo and Polit are probably correct in their conclusion that the developmental paths of only children have more in common with the paths of children from smaller families, we believe that theoretical propositions based on metaanalyses are tenuous at best and require support from independent comparison techniques designed to test multiple hypotheses. Falbo and Polit (1986) searched through the sibling relationship, birth order, and family size literature for studies that included only-child comparison groups. These studies variously defined the only child as fi-rstborn, last born or as a one-child family. Of 200 studies reviewed for inclusion in the meta-analyses, 115 survived a 6-point quality rating system based on sample size, controls for extraneous variables, and other indices of methodological rigor. Other studies were omitted because effect sizes (based on Cohen's d) could not be computed from reported correlations or because data for only borns were combined with data for firstborns or others from small families. On the basis of included studies, with total sample sizes ranging from 50 to over 6OO,OOO,Falbo and Polit formed a series of contrasts between only children and a non-only comparison group (anyone who had a sibling), and between only children and comparison groups defined in terms of birth order (firstborns or later borns from multichild families) and family size-small (two-child), medium (three- or four-child), or large (five- or more child) families. A broad variety of developmental outcomes found in the literature were classified into 14 categories (e.g., academic progress/grades, self-esteem, personal controllautonomy, IQhtandardized ability tests, and affiliation need/ extraversion). Because of the uneven number of studies represented by these categories, outcomes were grouped into five general developmental categories: achievement, adjustment, character, intelligence, and sociability. The outcome groups were used as dependent variable measures for five separate meta-analyses involving contrasts between only children and non-only children and comparison groups defined by birth order and by family size. l Mean effect sizes reliably differed from zero (based on nonorthogonal t tests) for achievement, character, and intelligence outcomes. For achievement and intelligence, only children had more desirable outcomes than comparison ' A sixth meta-analysis was conducted on 19 studies that included various ratings of the quality of the parent-child relationship (keyed toward more positive relationships). Results indicated mean effect sizes reliably different from zero (based on multiple t tests) for comparisons between only children and all non-only borns, and between only children and children from large families. Because mean effect sizes were positive, Falbo and Polit suggested that only children had more positive relationships with their parents than did other children.

Downloaded by [Monash University Library] at 17:29 11 April 2015

Mellor

223

groups of all non-only borns, children from medium and large families, and later borns, but had similar outcomes to groups of firstborns and children from small families. In the character area, only children had more desirable outcomes than groups of all non-only borns and groups of children from medium and large families but similar outcomes to all other comparison groups. None of the mean effect sizes for sociability and adjustment outcomes were reliably different from zero (p < .05). Meta-analysisresults were used by Falbo and Polit (1986) to suggest that only borns had higher levels of achievement and intelligence than children from medium and large families, had more desirable personalities, and were as sociable and adjusted as peers with siblings. Because no data were uncovered that revealed a single significant disadvantage for only borns across developmental outcomes, and because only borns were not found to be reliably different from all other comparison groups, Falbo and Polit argued that explanatory mechanisms that employ only-child deprivation and uniqueness were less feasible than a mechanism that emphasizes the parent-child relationship. Indirect support for this mechanism came from the finding that, across all meta-analyses, comparisons between only borns and firstborns, and comparisons between only borns and children from small families, never yielded reliable outcome differences. Falbo and Polit suggested that, because both only borns and firstborns were their parents’ first child, at least for a while, both were only borns, and concluded that the parent-child mechanism that produces only child outcomes was probably at work in producing similar firstborn and small-family outcomes. Our central concern was with Falbo and Polit’s (1986) theoretical prop osition that two explanatory mechanism have been excluded in favor of one that can explain how only children differ from others on general developmental outcomes. At the very least, Falbo and Polit’s use of nonorthogonal t tests (36 at p < .05) to evaluate contrasts between comparison groups (when independent mean comparison tests are preferable) calls for replication; a task we address with a new sample in our study. But, we also believed that theoretical propositions ruling out other mechanisms must find support from procedures designed to test multiple hypotheses and preserve outcome probability levels. I l y s , our study was designed to retest hypotheses that follow from Falbo and Polit’s meta-analyses and to further clarify a number of the similar and dissimilar developmental outcomes of only and non-only borns.

Method Subjects

The sample consisted of 434 students (239 males and 195 females) recruited from six junior and senior high schools and one community college in the Los

Downloaded by [Monash University Library] at 17:29 11 April 2015

224

The Journal of Genetic Psychology

Angeles area. Subjects’ ages ranged from 11 to 19 years, with a mean age of 15.9 years. Adolescents and young adults were targeted as subjects to approximate the mean age of subjects (16.6 years) and the majority of age groups (56%) included in Falbo and Polit’s (1986) sample. Seventy-nine percent were Caucasian, 11% were Black, 8% were Hispanic, and 2% were Asian, an ethnic mix proportional to current national percentages (“The Nation,” 1988). Subjects were first categorized by birth order and then by family size. Eight percent were only borns, 32% were firstborns, and 60% were later borns. In terms of family size, 8% percent were from one-child families, 38% were from small (two-child) families, 38% were from medium (three- or fourchild) families, and 15% were from large (five- or more child) families. The distribution of gender within categories for birth order and family size was proportional to the distribution in the total sample, and mean ages did not significantly differ from one category to the next (p < .05). Procedure Measures were administered in random order to students in “home” classrooms by undergraduates earning course credits. Informed consent forms were used to brief subjects on the nature of the project, to request voluntary participation, and to assure subjects of anonymity. Completed measures were received from 89% of the students contacted. Feedback on group results were sent to subjects through their home classroom instructors. Measures The Erikson Psychosocial Stage Inventory (EPSI; Rosenthal, Gurney, & Moore, 1981) was selected to measure general developmental outcomes included in Falbo and Polit’s (1986) meta-analyses. The EPSI is a measure of negative and positive resolutions to identity crises as defined by Erikson, and many of the resolutions are thought to play a critical role in influencing the outcomes grouped by Falbo and Polit. For example, outcomes related to affiliation need, personal controVautonomy, and academic progresdgrades overlap with resolutions to the crises of trust versus mistrust, autonomy versus shame, and industry versus inferiority (cf. Hamachek, 1988). One important shortcoming for this claim is in the area of intelligence. The EPSI does not include standardized items designed to measure ability per se, but arguments can be launched that suggest that achievement and intelligence are highly correlated with respect to developmental outcomes (cf. Fitzgerald & Mellor, 1988). The EPSI is a 72-item self-report inventory consisting of six 12-item subscales that assess degrees of negative to positive resolutions to six of Er-

Downloaded by [Monash University Library] at 17:29 11 April 2015

Mellor

225

ikson’s identity crises in childhood, adolescence, and early adulthood (trust vs. mistrust, autonomy vs. shame, initiative vs. guilt, industry vs. inferiority, identity vs. identity confusion, and intimacy vs. isolation). Items are rated on a 5-point scale ranging from hardy ever true (1) to almost always true (5). Subscale scores are represented by mean item scores. High mean scores for each subscale indicate positive resolutions to crises. Rosenthal et al. (1981) reported internal reliability (alpha) coefficientsfor subscales ranging from .57 to .8 1. Alpha coefficients for the present sample ranged from .74 to .86. Supportive validity can be found in Rosenthal, Moore, and Taylor (1983), and further subscale uses can be found in Moore and Rosenthal (1984), Pickar and Tori (1986), and Mellor (1989)? Sociobiographic measures also were included to assess subject gender, age, educational level (indexed by grade level), ethnic mix, vocabulary, and socioeconomic status (SES). A variety of other measures were used to determine birth order and family size.3Form Z of the Basic Word Vocabulary Test (BWVT; Dupey, 1974) was used as the measure of vocabulary, and SES was based on Hollingshead’s (1975) Four-Factor Index of Social Status, adjusting for combined parental income. Along with age and educational level, the latter two measures were given to check for covariation affects on birth-order and family-size differences reported in the only-child literature (cf. Claudy, 1984; Falbo, 1984; Gecas & Pasley, 1983; Mednick, Baker, & Hocevar, 1985).

Results The results of primary interest are the outcomes of sets of orthogonal planned comparions computed on the data of all subjects. Comparison sets were constructed within Gender x Birth Order (only borns, firstborns, later borns) and Gender x Family Size (one-child, small, medium, large), analyses of variance. Analyses were repreated for EPSI subscales as separate dependent variable measures. Weights were assigned to birth-order and family-size groups to ensure that contrast results were independent within sets, despite the fact that some contrast results were redundant within analyses. Four sets of contrasts (two per analysis) were performed to retest contrasts indicated by Falbo and Polit (1986). Comparison weights used in these analyses are presented in Tables 1 and 2. Also indicated in Tables 1 and 2 are contrasts with sets expected to be significant and predicted mean differences (high-low) between groups. 2Additional psychometric information on the EPSI, including factor structure results and intersubscale correlationsmay be found in Rosenthal(l982). ’A summary table of means, standard deviations, and ranges of sample descriptives is available on request.

TABLE 1 Comparison Weights and Predicted Mean Differences for Birth Order

Downloaded by [Monash University Library] at 17:29 11 April 2015

Contrast

Only borns

Birth-Order Group Firstborns

Later borns

(n= 35)

(n = 140)

(n= 259)

a

ba

a

ba Note. (High) (Low) within contrast indicates predicted mean differences. 'Contrast expected to be significant (JJ < .05).

Comparison results indicated that only predicted contrasts within sets yielded reliable differences between group means or group mean combinations on dependent variable measures. All other group contrasts failed to reach significance (p < .05).Furthermore, reliable mean differences were consistent with high-low predictions. Within the first contrast set for birthorder groups, reliable differences were not found between only borns and firstborns, but differences were found between only borns and later borns on autonomy, F( 1,411) = 4.84, p < .03, initiative, F( 1,405) = 5.00, p < .03, and industry, F(1, 413) = 3.79, p < .05.Mean differences for these groups indicated that only borns had significantly higher mean scores (4.12, 3.93, 3.88) than later borns (3.87, 3.69, 3.64). Within the second set for birthorder groups, group combination differences were not detected between means for only borns and average means for firstborns and later borns, but differences were detected between average means for only borns and firstborns and means for later borns on autonomy, F(1, 411) = 4.26, p < .04, initiative, F( 1, 405) = 3.86, p < .05,industry, F( 1 , 413) = 5.36, p < .02, and identity, F ( l , 404) = 3.97, p < .05.Mean differences for these latter contrasts indicated that mean scores for only borns and firstborns were significantly higher (4.02,3.82,3.82,3.84)than mean scores for later borns (3.87, 3.69, 3.64, 3.69). For the third and fourth contrast sets, we noted a consistent pattern of comparison results. Within the third contrast set for family-size groups, only

Mellor

227

TABLE 2 Comparison Weights and Predicted Mean Differences for Family She Family-Size Group

Downloaded by [Monash University Library] at 17:29 11 April 2015

Contrast

Only child (n = 35)

Small

(n = 167)

Medium

(n = 167)

Large

(n = 65)

a ba C

Nore. (High) (Low) within contrast indicates predicted mean differences. *Contrast expected to be significant @ < .05).

contrasts between only-child and medium family groups, and between onlychild and large family groups, yielded reliable differences on dependent variable measures. Significant mean differences between only-child and medium family groups were recorded on autonomy, F(1,409) = 4.03, p < .04,and initiative, F( 1 , 403) = 4.26, p < .04.Mean differences for these contrasts indicated that only-child scores were significantly higher (4.12, 3.93) than medium-family scores (3.88,3.70). Reliable mean differences between onlychild and large-family groups were recorded on initiative, F(1,403) = 3.75, p < .05, and identity, F(1, 402) = 6.20, p < .01. Mean differences for the latter contrasts indicated that only-child scores wre significantly higher (3 -93, 3.88) than large-family scores (3.57, 3.53). The most reliable differences between comparison groups occurred with the fourth set for family size. Although no other contrasts between group combinations were significant, average means for only-child and smallfamily groups were reliably different from average means for medium- and large-family groups on trust, F(1, 404) = 3.99, p < .05, autonomy, F(1,409) = 5.95, p < .01, initiative, F(1,403) = 7.92, p < .005, industry, F( 1,411) = 7.39, p < .007, and identity, F( 1,402) = 4.54, p < .03. Mean

Downloaded by [Monash University Library] at 17:29 11 April 2015

228

The Journal of Generic Psychology

differences for these combined groups indicated that only-child and smallfamily mean scores were significantly higher (3.55, 4.02, 3.83, 3.82, 3.82) than medium- and large-family mean scores (3.39, 3.83, 3.63, 3.59, 3.65). Within all contrast sets, we also found reliable differences between male and female groups on intimacy, F( 1, 408) = 27.28, p < .001. Mean differences for these groups indicated that females’ mean scores were significantly higher (3.86) than males’ mean scores (3.52). In addition, we repeated analyses for all sets, controlling for the influence of age, grade level, SES, and vocabulary. Results for separate and multiple analyses of covariance with adjusted probability levels (p < .05) produced similar results, and reported significant effects were unchanged. Furthermore, across all analyses, we failed to find a single significant interaction effect between gender and birth order or gender and family size.

Discussion The data of this study support Falbo and Polit’s (1986) conclusion that the developmental paths of only children are similar to the paths of firstborns and children from two-child families but dissimilar to paths of children from larger families. In fact, on the basis of developmental outcomes related to resolutions of developmental crises in childhood, adolescence, and early adulthood, the crucial outcome differences appear precisely between the only child, whether defined as only born or firstborn, and the later-born child on the birth-order variable, and precisely between children from two-child families and three-child families on the family-size variable. We believe that our results provide much clearer support for the hypotheses revealed in Falbo and Polit’s meta-analyses. Although the procedures they used to construct the analyses appeared sound, the use of non-orthogonal t tests to test comparison group differences failed to preserve the independence of contrast results and protect against inflated outcome probability levels. As such, the theoretical propositions offered by Falbo and Polit were tenuous at best. In contrast with techniques used by Falbo and Polit (1986), we retested multiple hypotheses related to only-child uniqueness and deprivation, and the parent-child relationship, with orthogonal contrasts that ensured that comparison results were independent and outcome probability levels were fixed as stated. The only-child uniqueness hypothesis was not supported. Onlychildren resolution outcomes were not reliably different from outcomes for all non-only comparison groups, whether defined by birth order or by family size. Comparisons revealed that, across outcomes, the positive-negative resolutions of only children did not significantly differ from the resolutions of firstborns or children from small families.

Downloaded by [Monash University Library] at 17:29 11 April 2015

The only-child deprivation hypothesis also was not supported. Only children had significantly higher positive outcomes to crises than did later borns and children from medium and large families. Futhermore, group means indicated that the positive outcomes of only borns were higher across the board than the positive outcomes for all other comparison groups, including firstborns and children from small families4 Our findings were consistent with the parent-child hypothesis simply because we failed to find significant outcome differences between group combinations of only children and firstborns or between group combinations of only children and children from small families. In contrast, we did find significant outcome differences between combinations of only children and later borns and others from medium and large families. Although we did not directly test the parent-child hypothesis, qualities of the parent-child relationship are apparently similar for only children, firstborns, and children from two-child families but dissimilar to qualities of the parent-child relationship for later borns and others from larger families. On the basis of the outcome differences found in this study, we would suggest that the definition of the only child be expanded to include firstborns and others from two-child families in future studies that use the parent-child mechanism to explain the outcome differences of only children. We found differences in resolution outcomes between birth-order groups and between family-size groups. Findings that suggest that only children, compared with non-only children, are generally more autonomous in terms of personal control, have higher levels of initiative or personal aspiration or motivation, are more industrious in terms of educational or occupational achievement, and have stronger identities (as indexed by self-esteem or adjustment levels) are not new in the only-child literature (cf. Falbo, 1984, 1987 for review^).^ Inconsistencies often appear with respect to each of these findings, however, and a consensus of thought about only-child outcome advantages and disadvantages is far from established. We suggest that profitable meta-analyses for future research include sets of orthogonal comparisons of developmental outcomes of an only-child group (defined as only borns, firstborns, and others from two-child families) with outcomes of various non-only child groups. We believe that outcome differences will be consistent with the present results and that a stronger consensus can be reached in terms of the similarities and dissimiliarities of the developmental paths of the only child.

4A complete table of means (including adjusted means) and F results is available on

request. A detailed explanation for male-female outcome differences relating may be found in Mellor (1989).

to

intimacy

230

The Journal of Generic Psychology

Downloaded by [Monash University Library] at 17:29 11 April 2015

REFERENCES Claudy, J. G. (1984). The only child as a young adult: Results from Project Talent. In T. Falbo (Ed.). , The single-child family (pp. 21 1-252). New York: Guilford Press. Dupey, H. J. (1974). The rationale, development, and standardization of the Basic Word Vocabulary Test. Vital and Health Statistical Series 2 , 60, 1-7 1. Falbo, T. (1984). Only children: a review. In T. Falbo (Ed.). , The single-child family (pp. 1-24). New York: Guilford Press. Falbo, T. (1987). Only children in the United States and China. In S. Oskamp (Ed.), Annual social psychology (Vol. 7, pp. 159-183). Beverly Hills: Sage. Falbo, T., & Polit, D. F. (1986). Quantitative review of the only child literature: Research evidence and theory development. Psychological Bulletin, 100, 176189. Fitzgerald, J. M., & Mellor, S. (1988). How do people think about intelligence? Multivariate Behavioral Research, 23, 143-1 57. Gecas, V., & Pasley, K. (1983). Birth order and self-concept in adolescence. Journal of Youth and Adolescence, 12, 521-535. Hainachek, D. E. (1988). Evaluating self-concept and ego development within Erikson's framework: A formulation. Journal of Counseling and Development, 66, 354-360. Hollingshead, A. B. (1975). Four-Factor Index of Social Sratus. Unpublished manuscript, Department of Sociology, Yale University, New Haven, CT. Mednick, B. R., Baker, R. L., & Hocevar, D. (1985). Family size and birth order correlates of intellectual, psychosocial, and physical growth. Merrill-Palmer Quarterly, 31, 67-84. Mellor, S. (1989). Gender differences in identity formation as a function of selfother relationships. Journal of Youth and Adolescence, 18, 361-375. Moore, S. M., & Rosenthal, D. A. (1984). Balance versus main effects androgyny: Their relationship to adjustment in three ethnic groups. Psychological Reports, 54, 823-831. Pickar, D. B., & Tori, C. D. (1986). The learning disabled adolescent: Eriksonian psychological development, self-concept, and delinquent behavior. Journal of Youth and Adolescence, 15, 429-440. Rosenthal, D. A., (1982). The Erikson Psychosocial Stage Inventov. Unpublished manuscript. University of Melbourne, Department of Psychology, Victoria, Australia. Rosenthal, D. A., Gurney, R. M., & Moore, S. M. (1981). From trust to intimacy: A new inventory for examining Erikson's stages of psychosocial development. Journal of Youth and Adolescence, 10, 525-537. Rosenthal, D. A., Moore, S. M., &Taylor, M. J. (1983). Ethnicity and adjustment: A study of the self-image of Anglo-, Greek-, and Italian-Australian working class adolescents. Journal of Youth and Adolescence, 12, 117-135. The Nation. (1988, September). The Chronicle of Higher Education Almanac, p. 3.

Received August 9, 1989

How do only children differ from other children?

Developmental outcomes of only and non-only children, categorized by birth order and by family size, were investigated. Multiple hypotheses based on m...
565KB Sizes 0 Downloads 0 Views