an empirically derived classification of parenting practices

16
This article was downloaded by: [Adams State University] On: 31 October 2014, At: 08:51 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 An Empirically Derived Classification of Parenting Practices Viktor Brenner a & Robert A. Fox b a Wisconsin School of Professional Psychology , b Department of Counseling and Educational Psychology , Marquette University , Published online: 30 Mar 2010. To cite this article: Viktor Brenner & Robert A. Fox (1999) An Empirically Derived Classification of Parenting Practices, The Journal of Genetic Psychology: Research and Theory on Human Development, 160:3, 343-356, DOI: 10.1080/00221329909595404 To link to this article: http://dx.doi.org/10.1080/00221329909595404 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.

Upload: robert-a

Post on 07-Mar-2017

216 views

Category:

Documents


2 download

TRANSCRIPT

Page 1: An Empirically Derived Classification of Parenting Practices

This article was downloaded by: [Adams State University]On: 31 October 2014, At: 08:51Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

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

An Empirically Derived Classificationof Parenting PracticesViktor Brenner a & Robert A. Fox ba Wisconsin School of Professional Psychology ,b Department of Counseling and Educational Psychology ,Marquette University ,Published online: 30 Mar 2010.

To cite this article: Viktor Brenner & Robert A. Fox (1999) An Empirically DerivedClassification of Parenting Practices, The Journal of Genetic Psychology: Research and Theoryon Human Development, 160:3, 343-356, DOI: 10.1080/00221329909595404

To link to this article: http://dx.doi.org/10.1080/00221329909595404

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 warrantieswhatsoever as to the accuracy, completeness, or suitability for any purpose of theContent. Any opinions and views expressed in this publication are the opinions andviews of the authors, and are not the views of or endorsed by Taylor & Francis. Theaccuracy of the Content should not be relied upon and should be independentlyverified with primary sources of information. Taylor and Francis shall not be liablefor any losses, actions, claims, proceedings, demands, costs, expenses, damages,and other liabilities whatsoever or howsoever caused arising directly or indirectly inconnection 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. Anysubstantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,systematic supply, or distribution in any form to anyone is expressly forbidden.

Page 2: An Empirically Derived Classification of Parenting Practices

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

Dow

nloa

ded

by [

Ada

ms

Stat

e U

nive

rsity

] at

08:

51 3

1 O

ctob

er 2

014

Page 3: An Empirically Derived Classification of Parenting Practices

The Journal of Genetic Psychology, 1999,160(3), 343-356

An Empirically Derived Classification of Parenting Practices

VIKTOR BRENNER Wisconsin School of Professional Psychology

ROBERT A. FOX Department of Counseling and Educational Psychology

Marquette Universiry

ABSTRACT. Cluster analysis was used to empirically determine if parenting practices would reflect distinct patterns of parental behavior in a sample of 1,056 mothers. Hierar- chical agglomerative clustering was performed with J. H. Ward’s (1963) method. R. Moje- na’s (1977) criterion determined that a 4-cluster solution was optimal; the resulting pat- terns are described. Simplified rules for classifying new cases also were generated. The clusters differentiated the mothers by age, education, number of children, socioeconomic status, and frequency of reported behavior problems. Theoretical and methodological issues, links to D. Baumrind’s (1991) parenting styles, and the heuristic potential of clas- sifying parenting practices are discussed.

BAUMRIND’S (1966, 1991) seminal work on the classification of parenting styles has profoundly influenced research on parenting and its effects on children. Her early work identified three parenting styles: (a) authoritarian-parents who are punitive and focus on gaining a child’s obedience to parental demands rather than responding to the demands of the child; (b) permissive-parents who are more responsive to their children but do not set appropriate limits on their behav- ior; and (c) authoritative-parents who are flexible and responsive to the child’s needs but still enforce reasonable standards of conduct. Later, a fourth style, neglecting, was added, describing parents who are underinvolved with their chil- dren and respond minimally to either the child’s needs or the child’s behavior.

This four-part classification system has had heuristic value in that specific parenting styles have been related to differential outcomes in children in such

This project was supported in part by grants from Robert Z Foote and the Todd Wehr Foun- dation.

Address correspondence to Robert A. Fox, Marquette University Parenting Center; 749 N. 17th Street, l? 0. Box 1881, Milwaukee, WI 53201-1881: e-mail may be sent to vik- tor Brenner at [email protected].

343

Dow

nloa

ded

by [

Ada

ms

Stat

e U

nive

rsity

] at

08:

51 3

1 O

ctob

er 2

014

Page 4: An Empirically Derived Classification of Parenting Practices

344 The Journal of Genetic Psychology

areas as school performance, delinquency, and drug and alcohol use (Steinberg, Lamborn, Darling, Mounts, & Dornbusch, 1994); social responsiveness and inde- pendence (Baumrind, 197 1); internalized distress (Lambom, Mounts, Steinberg, & Dornbusch, 1991); and disruptive playground behavior (Hart, DeWolf, Woz- niak, & Burts, 1992). In this article, the term outcomes is used to refer to these and any of the ways in which children can differ later in life as a result of parental influences.

Recently, Darling and Steinberg (1993) made an important theoretical ad- vance by distinguishing between parenting styles and parenting practices. Par- enting style is defined as a stable complex of attitudes and beliefs that form the context in which parenting behaviors occur. For instance, an authoritarian father may believe that his children should obey his commands out of respect for their elders. Parenting practices, on the other hand, are “specific, goal-directed behav- iors through which parents perform their parental duties” (Darling & Steinberg, 1993, p. 488). Spanking a child to induce compliance or giving compliments to enhance the child’s self-esteem are examples of parenting practices.

The most important implication of the distinction between parenting style and parenting practices is how each affects the outcomes of children. In Dar- ling and Steinberg’s (1993) model, parenting practices have a direct effect on children’s outcomes: Because parental behavior has immediate consequences for a child, it has a strong effect on a child’s life. Conversely, the role of par- enting style is indirect; Darling and Steinberg suggested that it acts as a mod- erator of the link between parenting practices and children’s outcomes. Thus, the often-cited reciprocal determinism (Maccoby & Martin, 1983) between parental and child behavior-that parental behavior affects children’s behav- iors, which in turn elicit certain responses from parents-is specifically relat- ed to parenting practices.

The primary importance of parenting practices also has been noted in the research on determinants of parenting practices and their relation to externaliz- ing disorders. The term determinants refers to any of the multiple demographic and psychological factors that have been correlated with parental behavior (Bel- sky, 1990) and thus can be used to predict it. Factors such as marital satisfaction, beliefs about discipline, parental abuse history, parental depression, level of spousal support (Simons, Beaman, Conger, & Chao, 1993), maternal age and edu- cation (Kelley, Power, & Wimbush, 1992), and family economic stress (Takeuchi, Williams, & Adair, 1991) have been identified as determinants of parenting prac- tices. Several of these determinants are also known risk factors for externalizing disorders in children (Patterson, DeBaryshe, & Ramsey, 1989). Patterson et al. suggested that the mechanism through which these determinants exert their influ- ence on children’s outcomes is through their effect on parenting practices. Thus, parenting practices are again implicated as having direct effects on a child’s out- comes.

If parenting practices have a direct link to a child’s behavior and long-term

Dow

nloa

ded

by [

Ada

ms

Stat

e U

nive

rsity

] at

08:

51 3

1 O

ctob

er 2

014

Page 5: An Empirically Derived Classification of Parenting Practices

Brenner & Fox 345

outcomes, a system of describing and classifying parenting practices could prove to be very valuable. First, because parenting practices are observable behaviors, they can be precisely defined and reliably assessed. Second, if clusters of par- enting practices are derived, we may be able to relate them to children’s out- comes. Third, if those clusters represent stable parental behavior patterns, we may be able to use them to identify parents whose children are at risk for poor out- comes, and thus provide preventive interventions. Furthermore, it may be possi- ble to use the clusters to tailor interventions to the needs of a given family. Thus, a system of classifying parenting practices could have much utility in research and clinical intervention with parents of young children.

In the present study, parenting practices were assessed with the Parent Behavior Checklist (PBC; Fox, 1994). The PBC provides norm-referenced measures of parenting behavior on three independent, factorially derived dimensions: discipline (responding to difficult behavior with verbal or corpo- ral punishment), nurturing (parental behaviors that promote psychological growth), and expectations (developmental tasks that the parent believes the child should be capable of performing). To date, use of the PBC has been lim- ited to the assessment of parenting practices in mothers and has been based on the elevations of individual scales. Fox, Platz, and Bentley (1995). for instance, showed that scale elevations differed as a function of maternal age, marital sta- tus, education, and family socioeconomic status. In that study, younger, single mothers from lower educational and socioeconomic backgrounds reported rel- atively more frequent use of verbal and corporal punishment with their young children. Similar results also were found in Mexico with a Spanish version of the PBC (Solis-Camara & Fox, 1996). Mothers in both the United States (Fox & Bentley, 1992) and Mexico (Solis-Camara & Fox, 1995) also were found to adjust their parenting practices to the age of the child, resulting in different scale elevations by the age of the child. However, cluster analysis can be used to discern consistent patterns within these scores. Pattern-based interpretation may increase the utility of PBC scores by capturing the potential interactive effects inherent in score patterns. For instance, the effects of parental use of frequent verbal and corporal punishment may differ depending on the level of nurturing present in the home. Thus, if parenting-practice patterns can be derived, such information could be useful in the assessment of parents of young children.

The present study was designed to determine whether consistent patterns of maternal parenting practices would emerge from a large-scale sampling of report- ed parenting behaviors. We hypothesized that maternal parenting practices would show a multiple-cluster structure reflecting consistent patterns in parental behav- ior. Furthermore, because there are several known determinants that are correlat- ed with parenting practices, we hypothesized that the heuristic potential of a clas- sification system could be assessed concurrently by its ability to differentiate mothers on these variables.

Dow

nloa

ded

by [

Ada

ms

Stat

e U

nive

rsity

] at

08:

51 3

1 O

ctob

er 2

014

Page 6: An Empirically Derived Classification of Parenting Practices

346 The Journal of Genetic Psychology

Method

Participants

A sample of 1,056 mothers of young children were recruited as part of a large validation study of the PBC. The sample was stratified by children’s ethnicity and gender to ensure that it would be representative of the large, midwestern, urban population from which it had been selected. The average age of the participating mothers was 3 1.6 years (SD = 6.0); 80% were married. The mothers had an aver- age of 14.7 years of education (SD = 2.16). Middle- and upper-middle-income families were somewhat overrepresented in the final sample, as measured by a sim- plified version of Stevens and Cho’s (1985) Occupational Classification Scheme, a system for deriving socioeconomic status based on a parent’s occupation.

Procedures

The mothers were identified through 57 urban day-care centers and preschools and then invited to participate in the study. All who agreed to partic- ipate received a packet containing a letter of explanation, a consent form, a demo- graphic form, and a test pool of 232 parenting-behavior items from which 100 items were eventually selected for the PBC. The demographic questionnaire also included a 9-item screening measure for behavior problems, adapted from Rich- man and Graham (1971). The mothers completed the packets and returned them to the day-care centers; in return, each received an appreciation package that included informational parenting pamphlets and a storybook or coloring book and crayons.

Special considerations were made for day-care centers located in demo- graphically low-income neighborhoods. To encourage low-income mothers to participate, we provided some centers with financial incentives for participation. A research assistant was available to read the packet to participants in an inter- view format, so that less educated mothers could participate.

Measures

The PBC (Fox, 1994) is a 100-item rating scale that measures the parenting practices of mothers of 1- to 5-year-old children. The PBC consists of three sub- scales, empirically derived through factor analyses. The discipline scale (30 items) assesses parental responses to children’s challenging behaviors with ver- bal and corporal punishment (e.g., “I yell at my child for spilling food”). The nur- turing scale (20 items) measures specific behaviors that promote a child’s psy- chological growth (e.g., “I read to my child at bedtime”). The expectations scale (50 items) measures developmental expectations (e.g., “My child should be able to feed hidherself”). Respondents rate each item on 4-point frequency scales

Dow

nloa

ded

by [

Ada

ms

Stat

e U

nive

rsity

] at

08:

51 3

1 O

ctob

er 2

014

Page 7: An Empirically Derived Classification of Parenting Practices

Brenner & Fox 347

(4 = almost always/always, 3 =frequently, 2 = sometimes, and 1 = almost never/ never). Because PBC scores are affected by children’s behaviors and abilities at various ages, raw scores are transformed into normalized T scores in relation to other parents of children the same age (defined in 6-month blocks between 1 and 5 years of age).

Good internal consistency alphas have been reported (Fox, 1992) for each subscale: Expectations = .97, Discipline = .91, and Nurturing = .82. Test-retest reliabilities determined through two administrations of the PBC separated by at least 1 week were as follows: Expectations = .98, Discipline = .87, and Nurtur- ing = .8 1. Additional studies have established the PBC’s developmental sensitiv- ity (ability to distinguish between parents with different-aged children; Fox & Bentley, 1992), interrater reliability (Bentley & Fox, 1991; Platz, Pupp, & Fox, 1994), unrelatedness to social desirability (Peters & Fox, 1993), correlation with known determinants of parenting practices (Fox, Platz, & Bentley, 1995), and cor- relation with behavior problems in children (Brenner & Fox, 1998).

Results

Cluster Analysis

Clusters of parenting practices were derived with Ward’s (1963) method of agglomerative hierarchical clustering. * The optimal number of clusters was deter- mined with Mojena’s (1977) stopping rule as recommended by Aldenderfer and Blashfield (1984). Figure 1 shows the four resulting clusters of parenting prac- tices. The first cluster was characterized by low to moderate discipline, low nur- turing, and low to moderate expectations. Graphically, its shape is best described as a flat line. This was the modal cluster, with 328 (31%) cases fitting this pat- tern. The second cluster combined low to moderate discipline, high nurturing, and high expectations; this cluster is best described as a line with an upward slope. This pattern fit 226 (21%) cases. The third cluster had an inverted-V shape because of the combination of low discipline, high nurturing, and low expecta- tions; 279 cases (27%) fit this pattern. The final pattern was made up of high to very high discipline, low nurturing, and moderate to high expectations, resulting in a V-shaped configuration; 223 (21%) cases fit this pattern.

To identify prototypical mothers from each cluster for future research or clin- ical practice, we generated simplified rules for classifying PBC profiles from the mean profiles in Figure 1. The goal of the resulting algorithm (see Figure 2) was to maximize the percentage of cases correctly classified within each cluster, a

‘An agglomerative hierarchical clustering procedure starts with each case in separate clus- ters, then combines those that are most similar according to some criterion until all cases are part of one super-cluster. Ward’s method uses minimization of the error sum of squares as the criterion.

Dow

nloa

ded

by [

Ada

ms

Stat

e U

nive

rsity

] at

08:

51 3

1 O

ctob

er 2

014

Page 8: An Empirically Derived Classification of Parenting Practices

348 The Journal of Genetic Psychology

40

35

- -

- -

PBC Scale

Clusterl, N = 3 2 8 - - Cluster 3, N = 279 - - - - Cluster 2, N = 226

Cluster 4, N = 223 I - - - FIGURE 1. Empirically derived parenting practice clusters as a function of Parent Behavior Checklist T scores.

measure of classification utility also known as predictive value (defined as the ratio of true positives to true positives plus false positives). However, focusing on predictive value tends to attenuate the sensitivity (the ratio of true positives to true positives plus false negatives) of classification, which in practice means that not all cases will be classified. With the results of the hierarchical agglomerative clus- ter analysis as a criterion, the percentages of cases correctly classified in each cluster were as follows: Cluster 1, 92.63%; Cluster 2, 90.8%; Cluster 3, 71.6%; and Cluster 4,98.7%. Kappa agreement between classifications made by the sim- plified rules and by the cluster analysis was good (K = .817), f(9) = 39.9, p c .001. The sensitivities of the individual clusters were as follows: Cluster 1,60%; Clus- ter 2,74%; Cluster 3,62%; and Cluster 4, 69%. Overall, the algorithm success- fully classified 75.9% of the mothers into one of the four clusters; the others were not classified.

Dow

nloa

ded

by [

Ada

ms

Stat

e U

nive

rsity

] at

08:

51 3

1 O

ctob

er 2

014

Page 9: An Empirically Derived Classification of Parenting Practices

Brenner 8c Fox 349

Nurturing, Expectations, and Discipline are less than 55 and Discipline greater than 45

Nurturing is 5+ points higher than Discipline I

Yes * Cluster 1

Discipline L 57 and Yes

Discipline is 5+ points greater than nurturing and

Expectations is 2+ points greater than nurturing

* Cluster 4

No V

No I

Yes

V Yes

Expectations > 50 * Cluster 2

No

* Cluster 3

Does not tit one of the common parenting practice clusters

FIGURE 2. Algorithm for classifying parents using Parent Behavior Check- list clusters.

Dow

nloa

ded

by [

Ada

ms

Stat

e U

nive

rsity

] at

08:

51 3

1 O

ctob

er 2

014

Page 10: An Empirically Derived Classification of Parenting Practices

350 The Journal of Generic Psychology

Determinants

A multiple analysis of variance (MANOVA) was performed to determine if the four parenting-practice patterns were related to known demographic deter- minants (socioeconomic status [SES], parent’s education, parent’s age, number of children, marital status, and sex of child) and level of reported child behavior problems. There was a statistically significant multivariate effect (h = .770, F approximation = 12.51, p < .OOOl) and, as shown in Table 1, each individual variable except for the sex of the child differed significantly by pattern. Roy- Bargman Stepdown-F tests (Tabachnick & Fidell, 1996) were performed to reduce this data set by determining which variables demonstrated unique vari- ance. Problem behaviors, parent’s education, parent’s age, and number of chil- dren remained significant with shared variance removed.

In order to better understand the differences between the empirical clusters, we used Tukey’s Honestly Significant Difference (HSD) to determine which groups differed from each other on each dependent variable. The V-shaped Clus- ter 4 (high discipline, low nurturing, moderate to high expectations) was most clearly differentiated: These mothers were the youngest, had the least education and lowest SES, were least likely to be married, and reported the most behavior

TABLE 1 Univariate Tests of Demographic and Behavior Problem Differences,

by Parenting Practices Cluster

Stepdown Contrasts: Clusters that Variable Univariate F(3, 967) F df differ significantly

Problem behaviors

Parental education

Social class (lower

Parent’s age

Number of children Marital status (married

(vs. unmarried) Sex of child

vs. higher)

35.1*** 35.1*** 3, 967 1,2,3<4 1<2

48.8*** 34.66*** 3,966 1, 2, 3 > 4 1<2,3

13.6*** 1.69 3, 965 1,2,3>4 1c3

23.9*** 9.32*** 3,964 1,2,3>4 2<3

3.17* 4.52** 3, 963 2 < 4 13.5*** 1.33 3, 962 1,2,3<4

1 > 3 1.18 1.10 3,961 NIA

Nore. Cluster 1 = low to moderate discipline, low nurturing, and low to moderate expectations. Clus- ter 2 = low to moderate discipline, high nurturing. and high expectations. Cluster 3 = low discipline, high nurturing, and low expectations. Cluster 4 = high to very high discipline, low nurturing, and moderate to high expectations. *p c .05. **p c . O l . ***p c .OOOI.

Dow

nloa

ded

by [

Ada

ms

Stat

e U

nive

rsity

] at

08:

51 3

1 O

ctob

er 2

014

Page 11: An Empirically Derived Classification of Parenting Practices

Brenner & Fox 35 I

problems in their children. The flat Cluster 1 (low to moderate scores on all three scales) was the next lowest in education and social class and next most likely to report behavior problems. Clusters 2 (low discipline, high nurturing, and high expectations) and 3 (low discipline, high nurturing, low expectations) differed only in that Cluster 2 mothers were younger than Cluster 3 mothers.

Discussion

This empirical derivation revealed four clusters of parenting practices that characterized maternal parenting behaviors as measured by the PBC. Further- more, the heuristic potential of these clusters was shown by their ability to dif- ferentiate parents by known determinants of parenting practices and by reported behavior problems. In this regard, the V-shaped Cluster 4 (high to very high dis- cipline, low nurturing, and moderate to high expectations) was the most distinc- tive. Cluster 4 mothers frequently punished their children for misbehavior and had a tendency to expect more than the children were developmentally capable of doing, yet they spent relatively little time in positive nurturing behaviors such as reading to their children. These mothers were the youngest, least educated, and lowest in SES, all of which are risk factors for externalizing disorders (Patterson et al., 1989). Perhaps not surprisingly, these mothers were the most likely to report behavior problems in their very young children.

The most common profile was a flat profile characterized by low to moder- ate scores on all three dimensions of the PBC. In light of both its form and its fre- quency, this is probably best thought of as an average profile. Mothers with flat profiles used an average amount of punishment, had moderate developmental expectations, and spent a moderate amount of time in positive nurturing activi- ties with their children. Mothers with flat profiles were second only to Cluster 4 parents in having the least education and lowest SES. They also reported the sec- ond most frequent behavior problems. It is interesting to note that the two pat- terns that indicated the most behavior problems also reported the most frequent use of verbal and corporal punishment. This is consistent not only with the reci- procal determinism view of parent and child behavior (Maccoby & Martin, 1983) but also with the recent finding (Brenner & Fox, 1998) that the PBC Discipline subscale is a better predictor of child behavior problems than several demo- graphic determinants combined.

The remaining clusters were rather similar in both form and determinants. Cluster 2 is best characterized as an upward slope (low to moderate discipline, high nurturing, and high expectations), and Cluster 3 is an inverted V character- ized by low discipline, high nurturing, and low expectations. Thus, both types of mothers seldom used punishment to influence their children’s behavior and fre- quently engaged in positive nurturing behavior with their children. However, the key difference between them was in what they expected their children to be capa- ble of doing. Cluster 2 mothers had the highest expectations of any cluster; Clus-

Dow

nloa

ded

by [

Ada

ms

Stat

e U

nive

rsity

] at

08:

51 3

1 O

ctob

er 2

014

Page 12: An Empirically Derived Classification of Parenting Practices

352 The Journal of Genetic Psychology

ter 3 mothers had the lowest. Conceptually, extreme parental expectations in either direction could result in either positive or negative outcomes for a child. Very high expectations could result in encouragement of achievement (positive) or in excessive pressure to succeed (negative); very low expectations could be associated with giving a child room to learn (positive) or parental laxness, which may lead to behavior problems related to lack of parental guidance (negative). Investigating differences between these parenting practices clusters may help elu- cidate the specific role of parental expectations in determining children’s out- comes.

Relationship to Parenting Style

In light of the quantity of research that has been done on parenting styles, it may be useful to compare the present results with Baumrind’s (1991) most recent taxonomy. When her definitions of styles are applied to the parenting practice pat- terns each style might be expected to generate, it appears that Cluster 2 corre- sponds to the authoritative parenting style, Cluster 3 corresponds to the permis- sive style, and Cluster 4 corresponds to the authoritarian parenting style. Thus, the clusters of maternal parenting practices may show some convergence with parenting styles. This finding may reflect the role of parenting style as a moder- ator of parenting practices in children’s outcomes.

Alternatively, the similarity between parenting practices and parenting styles may reflect specific features of the present study. For instance, parenting prac- tices are thought to be relatively situational in nature; when combined to form a scale, they may reflect the more consistent elements of behavior and thus be influ- enced by parenting style. The fact that the PBC is a self-report measure, albeit a report of parenting behaviors, also may result in these data being influenced by both parenting styles and practices. If the empirical clusters are valid, however, they should be replicable regardless of method used to assess parenting practices (e.g., videotaped parent-child interaction, in vivo observation).

One key difference between the present findings and those in the parenting- styles literature is that the modal cluster, characterized by low to moderate disci- pline, low nurturing, and low to moderate expectations, has not been previously described. Although Baumrind (1991) used the term good-enough parenting to describe a generally flat parenting profile, that term was specifically related to parental control of adolescents and may not be generalizable to parenting of very young children. Also, the good-enough profile tended to reside above the mean on all dimensions, whereas the modal profile in the present study was at or below the mean on all three PBC scales. Thus, the modal profile appears to describe a type of parent not previously described. One possible reason for this omission is that parenting style research has tended to heighten the prototypicality of contrast groups through such methods as excluding parents who fall into the middle ranges on classification variables (Lamborn et al., 1991; Steinberg et al., 1994), such that

Dow

nloa

ded

by [

Ada

ms

Stat

e U

nive

rsity

] at

08:

51 3

1 O

ctob

er 2

014

Page 13: An Empirically Derived Classification of Parenting Practices

Brenner & Fox 353

large numbers of essentially average parents may have been overlooked. Alterna- tively, this could reflect a point of departure between parenting styles and practices.

Another difference between the parenting-styles literature and the present results is that in the present study, we did not find a cluster that would correspond to the neglectful parenting style. A likely explanation is that in this study, we used an active consent procedure, which has been shown to underrepresent families at the lowest levels of functioning (Weinberger, Tublin, Ford, & Feldman, 1990). Researchers who have focused on neglectful families have adopted passive con- sent procedures, to maximize the inclusion of such families; the present study may have included insufficient numbers of neglectful parents to be detected as a separate cluster.

Limitations

Although we attempted to use the best performing clustering methods avail- able, cluster analysis is a less exact procedure than statistics that are based on the general linear model. Some researchers have endorsed Ward’s (1963) method as the best hierarchical agglomerative clustering procedure (cf. Mojena, 1977). but it has a known tendency to create hyperspheroid (n-dimensional spherical) clus- ters as opposed to other possible shapes (Aldenderfer & Blashfield, 1984). Use of a different clustering algorithm might have yielded different cluster centers and, relatedly, different relative cluster sizes. Similarly, Mojena’s stopping rule is one of the most highly regarded of many possible ways of determining the opti- mal number of clusters. It is, however, based on fusion coefficients (the value at which a new cluster is formed); stopping rules based on other criteria, such as explained versus unexplained variance, might have suggested that a different number of clusters was optimal. At the same time, the observed convergence with the extensive literature on parenting styles increases confidence in the generaliz- ability of this cluster solution.

The Simplified rules for classification according to the four-cluster solution could facilitate the usage of the clusters clinically or in research. It was disap- pointing, though not surprising, that only 75% of the mothers were classified according to the algorithm, because increasing the predictive value of a classifi- cation system tends to decrease its sensitivity. In part, this result reflects a prob- lem inherent in cluster analysis: All cases are classified regardless of goodness of fit with the overall clusters. Given that the PBC has three scales, even simple divi- sion into low, medium, and high scores results in 27 possible cluster patterns. As a result, any cases that show 1 of the 23 possible noncluster patterns may not be classified with this algorithm. For example, a parent with low discipline, nurtur- ing, and expectations-corresponding to Baunuind’s neglectful parent-would not be classified.

Another limitation is that the simplified classification algorithm has diffi- culty distinguishing between Cluster 2 and Cluster 3, in large part because of the

Dow

nloa

ded

by [

Ada

ms

Stat

e U

nive

rsity

] at

08:

51 3

1 O

ctob

er 2

014

Page 14: An Empirically Derived Classification of Parenting Practices

354 The Journal of Genetic Psychology

bipolar nature of the expectations scale. As discussed earlier, excessively high or excessively low expectations may be indicative of a potential parenting problem, whereas the low predictive value of Cluster 3 stems from difficulty classifying those cases that overlap with Cluster 2 at the midpoint of the scale; the majority of errors in Cluster 3 are misclassified as Cluster 2. In practice, there is probably no difference between parents who score just above or just below the mean on the scale; it is as one moves farther to either extreme that differences emerge. When the two clusters are combined, such that these midpoint classification errors disappear, the resulting pattern shows high specificity (90.1%) and the overall classification system is improved (kappa rises to 237). Therefore, it is rec- ommended that if a more prototypical parenting pattern is desired, the decision rule cutoffs in Figure 2 be raised to at least 55 for Cluster 2 classification and less than 45 for classification as Cluster 3. Use of the more stringent criteria may also increase the likelihood of finding differences between these parenting clusters on demographic or outcome variables.

In the present study, we relied on self-reports of behavior as measured by the PBC. Cross-validation of these clusters with behavioral observation data is there- fore important. Also, the present study was limited to mothers. Some initial data have been collected using the PBC with fathers, but not as yet in sufficient num- bers for cluster analysis. An initial report contrasting 52 mothers and fathers on the PBC showed that the fathers had significantly lower elevations on the nur- turing subscale but similar scores on the expectations and discipline subscales (Bentley & Fox, 1991). These findings suggest that maternal and paternal par- enting practices are different, and separate norms should be established for using the PBC with fathers. As a result, it is an unanswered, empirical question whether paternal parenting practices show the same patterns as the maternal parenting practice patterns described herein.

Implications

The present study indicates that parenting practices aggregate in clusters. The specific cluster solution needs to be replicated with other methods of assessing parenting practices. If replication is successful, further research would then be needed to determine if this classification system for parenting practices will improve our prediction of children’s outcomes. One advantage of this classifica- tion system is measurement; because these clusters are based on behaviors rather than attitudes, they can be assessed with high reliability. However, the potential value of this system is dependent on its predictive validity. If it is possible to show differential children’s outcomes by cluster, this classification may be used to iden- tify at-risk families-perhaps those associated with Cluster 4. Furthermore, the different parenting patterns may be found to reflect different parental interven- tion strategies and foci; this information could be used to match preventive inter- ventions with the specific needs of a given family. This parenting practice clas-

Dow

nloa

ded

by [

Ada

ms

Stat

e U

nive

rsity

] at

08:

51 3

1 O

ctob

er 2

014

Page 15: An Empirically Derived Classification of Parenting Practices

Brenner & Fox 355

sification system may be an important step in increasing the effectiveness of par- enting-based assessment and intervention with children.

REFERENCES

Aldenderfer, M. S., & Blashfield, R. K. (1984). Cluster analysis (Sage University Paper series on Quantitative Applications in the Social Sciences Number 07-044). Beverly Hills, CA: Sage.

Baumrind, D. (1966). Effects of authoritative parental control on child behavior. Child Development, 37, 887-907.

Baumrind, D. (197 1). Current patterns of parental authority. Developmental Psychology Monographs, 4( I , Part 2).

Baumrind, D. (1991). Parenting styles and adolescent development. In J. Brooks-Gunn, R. Lemer, & A. C. Petersen (Eds.), The encyclopedia of adolescence (pp. 746-758). New York: Garland.

Belsky. J. (1990). Parental and nonparental child care and children’s socioeconomic devel- opment: A decade in review. Journal of Marriage and the Family, 52, 885-903.

Bentley, K. S., & Fox, R. A. (1991). Mothers and fathers of young children: Comparison of parenting styles. Psychological Reports, 69, 320-322.

Brenner. V., & Fox, R. A. (1998). Parental discipline and behavior problems in young chil- dren. The Journal of Genetic Psychology, 159, 251-256.

Darling, N., & Steinberg, L. (1993). Parenting style as context: An integrative model. Psy- chological Bulletin, 113, 487496.

Fox, R. A. (1992). Development of an instrument to measure the behaviors and expecta- tions of parents of young children. Journal of Pediatric Psychology, 17, 231-239.

Fox, R. A. (1994). Parent Behavior Checklist manual. Austin, T X : Pro-Ed. Fox, R. A., & Bentley, K. S. (1992). Validity of the Parenting Inventory: Young Children.

Psychology in the Schools, 29, 101-107. Fox, R. A.. Platz, D. L., & Bentley, K. S. (1995). Maternal factors related to parenting

practices, developmental expectations, and perceptions of child behavior problems. The Journal of Genetic Psychology, 156, 431-441.

Hart, C. H., DeWolf, D. M., Wozniak, P., & Burts, D. C. (1992). Maternal and paternal disciplinary styles: Relations with preschoolers’ playground behavioral orientations and peer status. Child Development, 63, 879-892.

Kelley. M. L.. Power, T. G., & Wimbush, D. D. (1992). Determinants of disciplinary prac- tices in low-income Black mothers. Child Development, 63, 573-582.

Lambom. S . D.. Mounts, N. S., Steinberg, L., & Dombusch, S. M. (1991). Patterns of com- petence and adjustment among adolescents from authoritative, authoritarian, indulgent, and neglectful families. Child Development, 62, 1049-1065.

Maccoby, E. E.. & Martin, J. A. (1983). Socialization in the context of the family: Parent- child interaction. In P. H. Mussen (Series Ed.) & E. M. Hetherington (Vol. Ed.), Hand- book of child psychology: Vol. 4. Socialization, personality, and social development (4th ed., pp. 1-101). New York: Wiley.

Mojena, R. (1977). Hierarchical grouping methods and stopping rules: An evaluation. Computer Journal, 20, 359-363.

Patterson, G. R.. DeBaryshe, B. D., & Ramsey, E. (1989). A developmental perspective on antisocial behavior. American Psychologist, 44, 329-335.

Peters, C., & Fox, R. A. (1993). Parenting Inventory: Validity and social desirability. Psy- chological Reports, 72, 683-689.

Platz, D. L., Pupp, R. P., & Fox, R. A. (1994). Raising young children: Parental percep- tions. Psychological Reports. 74, 643-646.

Dow

nloa

ded

by [

Ada

ms

Stat

e U

nive

rsity

] at

08:

51 3

1 O

ctob

er 2

014

Page 16: An Empirically Derived Classification of Parenting Practices

356 The Journal of Genetic Psychology

Richman, M., & Graham, P. J. (1971). A behavioral screening questionnaire for use with three-year old children. Journal of Child Clinical Psychology and Psychiatry, 12, 5-33.

Simons, R. L., Beaman, J., Conger, R. D., & Chao, W. (1993). Childhood experience, con- ceptions of parenting, and attitudes of spouse as determinants of parental behavior. Jour- nal of Marriage and the Family, 55, 91-106.

Solis-Camara, P., & Fox, R. A. (1995). Parenting among mothers with young children in Mexico and the United States. The Journal of Social Psychology, 135, 591-599.

Solis-Camara, P., & Fox, R. A. (1996). Parenting practices and expectations among Mex- ican mothers with young children. The Journal of Genetic Psychology, 157, 465476.

Steinberg, L., Lamborn, S. D., Darling, N., Mounts, N. S., & Dornbusch, S. M. (1994). Over-time changes in adjustment and competence among adolescents from authorita- tive, authoritarian, indulgent and neglectful families. Child Development, 65, 754-770.

Stevens, G., & Cho, J. H. (1985). Socioeconomic indexes and the new 1980 census occu- pational classification scheme. Social Science Research, 14, 142-168.

Tabachnick, B. G., & Fidell, L. S. (1996). Using multivariate statistics (3rd ed.). New York: HarperCollins.

Takeuchi, D. T., Williams, D. R., & Adair, R. K. (1991). Economic stress in the family and children’s emotional and behavior problems. Journal of Marriage and the Family, 53,

Ward, J. H. (1963). Hierarchical grouping to optimize an objective function. Journal of the American Statistical Association, 58, 236-244.

Weinberger, D. A., Tublin, S. K., Ford, M. E., & Feldman, S. S. (1990). Preadolescents’ social-emotional adjustment and selective attrition in family research. Child Develop- ment, 61, 1374-1386.

103 1-1 04 I .

Received November 30, 1998

Dow

nloa

ded

by [

Ada

ms

Stat

e U

nive

rsity

] at

08:

51 3

1 O

ctob

er 2

014