Research paper-How parenting style affect child development

R

A m c

M N

a

A R R 1 A A

K C A S R S

I

s c m 2 r s

0

Child Abuse & Neglect 51 (2016) 263–275

Contents lists available at ScienceDirect

Child Abuse & Neglect

esearch article

uthoritarian parenting attitudes and social origin: The ultigenerational relationship of socioeconomic position to

hildrearing values�

ichael Friedson ew York University, Department of Sociology, 295 Lafayette Street, 4th Floor, New York, NY 10012, USA

r t i c l e i n f o

rticle history: eceived 6 April 2015 eceived in revised form 2 September 2015 ccepted 2 October 2015 vailable online 14 November 2015

eywords: orporal punishment uthoritarian parenting ocioeconomic status ace ocioeconomic disadvantage

a b s t r a c t

Support for authoritarian approaches to parenting, including corporal punishment, is known to be elevated among individuals with low current levels of socioeconomic attain- ment. The objectives of this study are: (1) to determine whether authoritarian parenting dispositions are related to disadvantages in one’s social background, in addition to one’s present socioeconomic standing; and (2) to distinguish, in this regard, between support for spanking and other authoritarian parenting dispositions. Ordered logit models, applied to General Social Survey data concerning a nationally representative sample of US adults, are used to examine relationships of authoritarian parenting dispositions to the socioecono- mic positions that respondents currently occupy and in which they were raised. It is found that support for spanking (N = 10,725) and valuing of obedience (N = 10,043) are inversely related to the socioeconomic status (SES) of one’s family of origin, and that these associ- ations are robust to controls for one’s current SES. A disadvantaged family background is found to increase support for spanking most among those with high current SES. Strong associations (robust to controls for SES indicators) are additionally found between African- American racial identity and support for authoritarian parenting. Prior research indicates that authoritarian parenting practices such as spanking may be harmful to children. Thus, if the parenting attitudes analyzed here translate into parenting practices, then this study’s findings may point to a mechanism for the intergenerational transmission of disadvantages.

© 2015 Elsevier Ltd. All rights reserved.

ntroduction

Harsh parenting practices may help to transmit socioeconomic disadvantages across generations of a family. Numerous tudies have asked how an individual’s current socioeconomic status (SES) influences the manner in which she raises her hildren. A near-consensus exists that lower levels of SES are associated with a greater propensity to support authoritarian ethods of parenting including corporal punishment (Flynn, 1996; Hoff, Laursen, & Tardif, 2002; Kohn, 1969; Lareau, 2002,

011; Luster, Rhoades, & Haas, 1989; Rosier & Corsaro, 1993; Tudge, Hogan, Snezhkova, Kulakova, & Etz, 2000). The present esearch goes beyond this existing consensus to ask whether influences persist from the SES of one’s family of origin to hape her attitudes toward corporal punishment and authoritarian parenting.

� This research was funded by a grant from the William T. Grant Foundation. This funding source had no involvement in the preparation of this manuscript.

http://dx.doi.org/10.1016/j.chiabu.2015.10.001 145-2134/© 2015 Elsevier Ltd. All rights reserved.

264 M. Friedson / Child Abuse & Neglect 51 (2016) 263–275

Two factors may work together to give rise to the relationship between an individual’s current socioeconomic position and childrearing preferences. First, SES is associated with one’s current traits and living conditions, which shape one’s needs and worldview, which in turn influence parenting (Conger, Ge, Elder, Lorenz, & Simons, 1994; Hoff et al., 2002; Lareau, 2011). For example, an individual accustomed to performing low status jobs, which tend to require conformity to established procedures, may find it natural to expect the same from her children (Curtner-Smith, Bennett, & O’Rear, 1995; Erlanger, 1974; Kohn, 1963, 1969; Lareau, 2011). The second factor potentially influencing the relationship between current SES and childrearing preferences has to do with how one was raised rather than one’s current living conditions. An individual’s SES in adulthood tends to reflect that of the family in which she grew up (Hill & Duncan, 1987; Sharkey, 2013). Low SES adults are likely to have been raised in a low SES household and to have internalized values, expectations, and practices associated with their parents’ socioeconomic position.

Studies showing that current SES is related to childrearing preferences, without taking into account the social conditions in which an individual was raised, cannot distinguish the influences of these two factors on childrearing. This distinction is of more than esoteric interest because of its implications for understanding how socioeconomic disadvantages are transmitted through multiple generations of a family (Hill & Duncan, 1987; Jencks & Mayer, 1990; Sharkey, 2013). One reason posited for why lower SES parents tend to produce lower SES children is that an authoritarian upbringing, by teaching obedience and conformity, best prepares children for lower SES careers (Lareau, 2002, 2011). Harsh corporal punishment has itself been associated with a litany of negative developmental outcomes that could help to perpetuate socioeconomic disadvantages. While it is beyond this study’s scope to summarize the vast literature on these outcomes, it is worth noting that they have been posited to include reduced cognitive and academic performance and increased propensities toward aggressive and antisocial conduct, substance abuse, depression, and other mental illnesses (Bugentall, Martorell, & Barraza, 2003; Christie-Mizell, Pryor, & Grossman, 2008; Durant & Ensom, 2012; Gershoff, Lansford, Sexton, Davis-Kean, & Sameroff, 2012; Grogan-Kaylor, 2004, 2005; MacKenzie, Nicklas, Waldfogel, & Brooks-Gunn, 2013; MacKenzie, Nicklas, Waldfogel, & Brooks-Gunn, 2015; MacMillan et al., 1999; Strassberg, Dodge, Pettit, & Bates, 1994).

It is known that low SES adults tend to prefer authoritarian parenting methods that may disadvantage their children (Flynn, 1996; Kohn, 1969; Lareau, 2011; Tudge et al., 2000). Does this also apply to adults who come from a disadvantaged family background but achieve upward mobility? Does their upward mobility override authoritarian preferences passed down from their families of origin? Or do these preferences persist and potentially allow the disadvantages they experienced while growing up, but have now overcome, to “skip a generation” and affect their children? To answer these questions, this study analyzes how adults’ attitudes toward parenting are related to the socioeconomic positions they currently occupy and in which they were raised.

Aims of the Present Study in Light of Past Findings on Socioeconomic Disadvantage and Authoritarian Childrearing

Among scholars who have studied the relationship between SES and childrearing preferences, a consensus exists that, relative to their middle class peers, working class and poor individuals tend to place greater value on obedience (Bluestone & Tamis-Lemonda, 1999; Conger et al., 1994; Dornbusch, Ritter, Leiderman, Roberts, & Fraleigh, 1987; Kohn, 1963, 1969; Lareau, 2011; Lipset, 1959; Lipsitz, 1965; Luster et al., 1989; McLloyd, 1990; Rosier & Corsaro, 1993; Tudge et al., 2000). In testing the relationship of current SES to various indices of individuals’ orientations toward social institutions, Kohn (1969, p. 86) finds it is most strongly associated, “by a very wide margin,” with the valuing of conformity to the dictates of authority. Working class subjects, according to this author, tend to believe that “the most important thing to teach children is absolute obedience to parents” (Kohn, 1969, pp. 78–79). Lareau’s recent study (2011) similarly finds that children in working class households are expected to respond to parents’ commands with unquestioning obedience. Middle class children, in contrast, are taught to be self-directed and to promote their interests by negotiating with authority figures (Curtner-Smith, Bennett, & O’Rear, 1995; Hoff et al., 2002; Lareau, 2011; Luster et al., 1989; Tudge et al., 2000).

The corporal punishment of children is consistent with an approach to childrearing that values obedience to authorities and conformity to established norms. Past studies have shown that parents harboring authoritarian worldviews, embodied in religious beliefs emphasizing absolute obedience to God, use or support spanking more than others (Ellison, Bartkowski, & Segal, 1996; Flynn, 1994; Grogan-Kaylor & Oti, 2007). The application of corporal punishment demonstrates to children, in blunt and visceral terms, who has authority and power and who is subordinate in the parent-child relationship. The establishment of such clear and definitive boundaries, as well as the threat of physical sanctions, provide key foundations for the unquestioning obedience found by Lareau (2011, p. 139) to prevail in lower SES households.

On this basis, it is unsurprising that past studies have repeatedly found that working or lower class individuals in the US are more likely to favor or use corporal punishment (Erlanger, 1974, p. 81; Flynn, 1994, p. 318; Flynn, 1996; Fox & Solis- Camara, 1997; Giles-Sims, Straus, & Sugarman, 1995, p. 173; Lareau, 2011, p. 211; Smith Blau, 1965). One study has also shown this to apply in Mexico (Fox & Solis-Camara, 1997). Agreement on the link between SES and support for corporal punishment, however, is not universal (Ellison et al., 1996; Horn, Cheng, & Joseph, 2004). One study has purported to show that associations between SES and attitudes toward spanking are specific to particular regions of the US (Flynn, 1994).

While there is widespread agreement that lower SES individuals are more prone to favoring corporal punishment and placing greater value on children’s obedience, there is much disagreement over what causes these differences. Some authors accredit them to disparities in levels of educational attainment (Hoff et al., 2002; Kohn, 1969, p. 472; Lareau, 2011; Lipset, 1959, p. 489; Lipsitz, 1965, p. 104). Formal education may promote a sense of one’s entitlement or ability to reason with

t a g K

l 1 o h h (

a a t t o e h c p e f

r t i M i 2 a P p 1 T t

i i p t o w v a a t i

D

O

o T a s p 1

M. Friedson / Child Abuse & Neglect 51 (2016) 263–275 265

hose in power and enables greater access to research on the harmfulness of spanking. These differences have also been ttributed to disparities in the characteristics of middle and working/lower class jobs, with the latter being subjected to reater “standardization and direct supervision” (Kohn, 1963, p. 477; see also Curtner-Smith et al., 1995; Erlanger, 1974; ohn, 1969; Lareau, 2011; Luster et al., 1989).

Finally, these differences may be brought about by the greater financial, occupational, or psychological stress faced by ower SES individuals and families (Bluestone and Tamis-LeMonda, 1999; Conger et al., 1994; Giles-Sims et al., 1995; Lipset, 959; McLloyd, 1990). Conformity at home may help these families to better coordinate their efforts to meet the challenges f their precarious financial circumstances. Spanking may provide an expedient response to domestic problems for parents ard-pressed by the demands of working class jobs. Parental depression or distress, induced by economic or occupational ardship, may cause parents to respond with less understanding or warmth and greater harshness toward their children Bluestone and Tamis-LeMonda, 1999; Conger et al., 1994; McLloyd, 1990).

Although the influences of traits related to current SES on parenting preferences have been widely considered, less ttention has been given to how these preferences are affected by the social position of one’s family of origin. The nalysis of Kohn (1969, p. 1370) provides that the magnitude of the relationship between a person’s present class posi- ion and the degree to which that person harbors conservative authoritarian values is about three times greater than hat between class origin and level of conservative authoritarianism. Smith Blau (1965) holds that the degree to which ne’s social background continues to influence one’s present values is contingent on the amount of social pressure xerted on an individual to conform to the values of her present class. A study on the intergenerational transmission of arsh parenting concludes that subjects’ parents’ SES only influences subjects’ childrearing dispositions by shaping their urrent SES (Simons, Whitbeck, Conger, & Chyi-In, 1991). This study, however, neglects to test for more direct influences of arental SES in its multivariate models (Simons et al., 1991, p. 165). In order to better understand the potential for intergen- rational transmission of disadvantage via parenting preferences, it is necessary to more fully analyze how disadvantages rom one’s family of origin may influence these preferences.

It is also necessary, toward this end, to consider the influences of disadvantages from one’s social background that are not educible to household-level SES. Racial inequality is one potential source of such disadvantages. It has been demonstrated hat an individual’s prospects for socioeconomic attainment are shaped not only by her parents’ SES, but also by disadvantages n the neighborhood contexts in which that individual and her parents were raised (Burdick-Will et al., 2011; Jencks &

ayer, 1990; Leventhal & Brooks-Gunn, 2000; Sharkey, 2013). African-American households are more likely to be situated n neighborhoods with high levels of poverty and interpersonal violence than white households of a similar SES (Massey, 004; Massey & Denton, 1993; Peterson & Krivo, 2010). These disparities, which are rooted in racial residential segregation, re among the many burdens of racial stereotyping and discrimination faced by blacks (Massey & Denton, 1993; Pager, 2003). ast research shows that, controlling for present SES, African-Americans favor corporal punishment and harbor authoritarian arenting values to a greater degree than whites (Berlin et al., 2009; Day, Peterson, & McCracken, 1998; Erlanger, 1974; Flynn, 994, 1996, 1998; Gershoff et al., 2012; Giles-Sims et al., 1995; Grogan-Kaylor & Oti, 2007; Hawkins et al., 2010; Kohn, 1969). hese dispositions may constitute defenses to the harsher social environment faced by blacks (Carr, 2014), or may reflect he influence of multiple generations of socioeconomic disadvantage within black families.

In analyzing the relationship between social origins and parenting preferences, this study builds on past research show- ng associations between these preferences and present SES or racial minority status. The key question framing the study’s nquiry concerns whether socioeconomic disadvantages in one’s family background have a durable relationship to parenting ractices that persists notwithstanding improvements in one’s current SES. To address this question, the study asks whether he association between present SES and parenting preferences is moderated by the socioeconomic environment in which ne was raised, or if the latter is related to parenting preferences in a manner independent of present SES. It further analyzes hether relationships between racial minority status and parenting attitudes are robust to controls for socioeconomic disad-

antages in minority individuals’ households of origin. Finally, it explores whether leanings toward authoritarian parenting re unique to African-Americans among disadvantaged minority groups, or shared by other groups such as Hispanics that re affected by processes of segregation and discrimination (Hawkins et al., 2010; Peterson & Krivo, 2010). By addressing hese topics, a fuller picture emerges of the relationship of social origins to parenting preferences, and of the potential for ntergenerational transmission of disadvantage via such preferences.

ata and Methods

utcome, Predictor, and Control Variables

This study uses logistic regression techniques, applied to General Social Survey (GSS) data, to determine the relationship f an individual’s social background to the degree to which her attitudes toward parenting reflect authoritarian preferences. he GSS is administered biannually to a full probability sample intended to be representative of non-institutionalized US

dults (ages 18 and over) who speak English or Spanish. The substantive breadth of the GSS allows for the implementation of tatistical controls for a broad range of potentially confounding traits of respondents. In order to increase sample sizes and rovide for more robust results than would be possible with a single survey wave, results from all survey waves between 994 and 2010 are combined to constitute the study’s data set (Treiman, 2009, p. 222).

266 M. Friedson / Child Abuse & Neglect 51 (2016) 263–275

The outcome variables consist of responses to questions about approval of spanking and valuing of obedience in children. The former asks respondents to designate whether they “strongly agree, agree, disagree, or strongly disagree” with the notion “that it is sometimes necessary to discipline a child with a good, hard spanking” (GSS Codebook, 2013, p. 427). The latter asks respondents to rank how important it is for a child to learn “to obey,” versus four other dispositions, in order “to prepare him or her for life” (GSS Codebook, 2013, p. 369). The other four dispositions are learning “to be well-liked or popular,” “to think for himself or herself,” “to work hard,” and “to help others when they need help” (GSS Codebook, 2013, p. 369). Among GSS participants, rates of non-response are minimal (<1.5%) to the questions from which the spanking and obedience outcome variables are derived.

The predictor variables consist of indicators of respondents’ present socioeconomic status and social origins. The former is operationalized via respondents’ scores on the 1989 socioeconomic index (SEI) of occupations (Nakao & Treas, 1990, 1992), which uses procedures originally developed by Duncan (1961) to classify respondents on a continuous scale of occupational prestige. To reduce difficulties involved in classifying respondents who are currently in school, the sample is limited to those ages 24 and over. The prestige of an individual’s occupation may not accurately capture the socioeconomic standing of his or her current household when that individual’s spouse works in a higher-status field. Accordingly, when the SEI of a respondent’s spouse exceeds that of the respondent, the current SEI variable is adjusted to equal the spouse’s SEI. Regressions in the appendix (Table A) test whether results are sensitive to this adjustment by repeating the analysis with unadjusted current SEI values.

Two aspects of respondents’ social origins are captured by predictor variables, their racial backgrounds and the socio- economic positions of the households in which they were raised. The latter is operationalized via a metric of respondents’ parents’ socioeconomic standing, namely the greater of the “masei” and “pasei” variables in the GSS data. These variables indicate, respectively, the SEIs of subjects’ mothers and fathers. Where a value for one of these variables is missing from the GSS data, the metric of parental socioeconomic standing is assigned the value of the other variable. Cases are excluded where values of both variables are missing.

Where the respondent did not live with her mother or father, the GSS assigns the relevant parental SEI variable with the SEI of a mother- or father-substitute (if any). The SEIs of these parent-substitutes are thus reflected in the study’s metric of parental socioeconomic standing. Regressions in the appendix (Table B) test whether results are robust to instead calculating parental SEI without considering parent substitutes.

In characterizing respondents’ social origins, racial/ethnic background is considered, in addition to parental SEI, because the former captures aspects of disadvantage that are not reducible to household-level SES. Respondents are categorized by their racial origins according to whether they are designated as black or white in the GSS data. Respondents not designated as black or white only constitute about 6% of GSS subjects for whom values are available of the predictor and outcome variables. Classification of these subjects’ race and ethnicity is inconsistent, vague, or missing across the GSS waves included in the data set. Given these considerations, respondents who are not black or white are excluded from the analysis.

A consistent measure of Hispanic origin is not provided until the 2000 GSS, which prevents controlling for Hispanic origin in the regressions within the study’s main body. However, the appendix (Table E) includes regressions using GSS data from 2000 to 2010, which demonstrate that results are robust to controls for Hispanic origin. These regressions also make it possible to ascertain whether relationships between African-American racial status and the outcome variables apply to other disadvantaged minorities or are particular to blacks.

Controls are included in the models for a variety of factors that may influence childrearing preferences and values. The experience of raising children may shape attitudes toward doing so. Indeed, existing research indicates that those with greater numbers of children display more favorable attitudes toward spanking (Flynn, 1994, 1996; see also Ellison et al., 1996). The models thus control for the number of children (capped at eight, as in the GSS data) to whom the respondent is a parent. They also control for the age of the respondent, to account for the possibility that dispositions toward childrearing change as individuals grow older. Given that the survey data and past research indicate that males show a stronger preference for corporal punishment, the models control for respondents’ gender (Flynn, 1994, 1998). Multiple years of survey data are combined in the study, and the approval of spanking and the valuing of obedience have declined slightly (according to GSS data) over the period under analysis. To capture these trends, a control is included for the survey year.

Given past findings indicating that regionally based cultural differences affect attitudes toward spanking, controls are included for dummy variables indicating region of residence (Flynn, 1994). For this purpose, the country is divided into four regions—the Northeast, Mid-West, South, and West—following US Census definitions. Most states in the South (and few in other regions) allow corporal punishment in public schools, suggesting a cultural environment especially conducive to approval of spanking (Strauss, 2014). Given that cultural differences may also be structured along urban/rural lines, controls are adopted for whether a respondent resides in a large city, rural area, or neither. The former is defined as a city with at least 250,000 residents. A rural area is defined, following the GSS Codebook (2013, p. 190), as a county outside of the country’s 112 largest SMSAs not containing a town with 10,000 or more residents.

Analytic Approach

Ordered logit regressions are used to model the relationship of the outcome to the predictor variables. Ordered logit is an extension of logistic regression techniques. While the latter applies to dichotomous outcomes, the former instead applies to ordinal variables—such as the outcome variables in this study—with more than two potential values (Borooah, 2002).

M. Friedson / Child Abuse & Neglect 51 (2016) 263–275 267

Table 1 Estimated population values for key variables.

Estimated population mean Standard error of estimated mean Estimated standard deviation

(a) Socioeconomic status Current SEI 54.52 0.26 19.75 Parental SEI 48.92 0.25 19.35 Current SEI × parental SEI 2781.52 23.67 1725.04

Variable Response category Estimated percentage of population Standard error of estimate

(b) Outcome variables Spanking (N = 10,725)

1—Strongly agree 27.02 0.55 2—Agree 46.14 0.60 3—Disagree 19.61 0.46 4—Strongly disagree 7.23 0.29

Obedience (N = 10,043) 1—Top priority 16.75 0.46 2—Second priority 12.22 0.36 3—Third priority 16.74 0.43 4—Fourth priority 39.60 0.56 5—Lowest priority 14.69 0.39

Cells show: estimated row percentage (standard error) Tercile of parental SEI

Lowest Middle Highest

(c) Cross-tabulation of current SEI and parental SEI Tercile of current SEI

Lowest 57.77 (0.91) 22.71 (0.74) 19.51 (0.76) Middle 37.23 (1.00) 28.52 (0.90) 34.25 (0.94) Highest 26.77 (0.88) 26.28 (0.79) 46.95 (0.93)

Notes. 1. A full summary of estimated population values for the study’s regressors is given in Table F of the Appendix. 2. The estimates in (a) of this table are derived using the sample from Table 2(a)’s regressions of the spanking variable (N = 10,725). Table F in the Appendix additionally provides estimated values of the SEI variables that are derived using the sample from Table 2(b)’s regressions of the obedience variable. 3. The estimates in (c) of this table are derived using the sample from Table 2(a)’s regressions of the spanking variable (N = 10,725). 4. A test of the likelihood ratio for the crosstab in (c) of this table (Ho : no relationship between current SEI tercile and parental SEI tercile)—adjusted for survey design parameters—indicates that F (3.94, 3602.88) = 175.40 (p = 0.00). The unadjusted likelihood ratio �2 = 879.00 (d.f. = 4, p = 0.00).

T r

o p p p s e T

s i r T t p

a d l a y t w

his study’s outcome variables are ordinal because their values can be ordered according to the strength or weakness of espondents’ support for spanking or valuing of obedience in children.

The models’ purpose is to analyze the relationship of individuals’ childrearing preferences to their current SES and social rigins. Toward this end, each outcome variable is first regressed on the respondent’s current SEI, the SEI of the respondent’s arents, the respondent’s race, and a set of control variables. This approach treats current and parental SEI as discrete redictors of parenting preferences, which facilitates straightforward interpretation of results, but does not account for how resent SES and social origins may interact to predict outcomes. A second set of regressions tests whether respondents’ ocial origins moderate the relationship of their current SES to their parenting preferences. On this basis, an interaction ffect between the respondent’s current SEI and that of the respondent’s parents is added to this second set of regressions. able 2 summarizes the results of these regressions.

The regressions of the obedience variable make it possible to ascertain whether relationships involving attitudes toward panking apply more broadly to authoritarian parenting dispositions or are specific to spanking. Toward this end, the spank- ng and obedience outcomes are regressed on identical sets of variables in the models in Table 2, enabling comparison of the elationships of each outcome to the predictors. An additional set of regressions of the spanking variable, summarized in able 4, adds controls for respondents’ attitudes toward obedience. This makes it possible to determine the extent to which he predictors’ relationships to the spanking variable are mediated by or independent of their relationships to authoritarian arenting values more broadly, as embodied by the obedience variable.

GSS data spanning 1994–2010 are used in the regressions because this timeframe encompasses all years for which data re available on both paternal and maternal SEI, in addition to respondents’ current SEI. Within individual years of GSS ata, sample sizes are small for subpopulations of interest in this study, such as African-Americans and individuals with

ow current and high parental SES. Combining multiple years of GSS data, to form the study’s data set, helps to overcome nalytic limitations that would be imposed by these small sample sizes. It is valid to combine data measured across multiple ears in a single regression, as long as the underlying phenomena being measured in the regression are consistent across

hese years (Treiman, 2009, p. 222). The regressions in Table E of the appendix, which use GSS data from 2000 to 2010, test hether results are sensitive to the specification of the study’s timeframe.

268 M. Friedson / Child Abuse & Neglect 51 (2016) 263–275

Table 2 Ordered logit regressions of attitudes toward spanking and obedience.

Cells show: coefficient (S.E.) (a) Spanking outcome variable (N = 10,725)

(b) Obedience outcome variable (N = 10,043)

Model 1 Model 2 Model 1 Model 2

Current SEI 0.00811*** 0.00186 0.0150*** 0.0163***

(0.00105) (0.00283) (0.00103) (0.00268) Parental SEI 0.00561*** −0.00184 0.0100*** 0.0116***

(0.00119) (0.00342) (0.00109) (0.00339) Current SEI × parental SEI 0.000127** −0.0000276

(0.0000538) (0.0000518) Survey year 0.0126*** 0.0128*** 0.0140*** 0.0140***

(0.00411) (0.00411) (0.00373) (0.00372) Age 0.00312 0.000257 −0.00762*** −0.00761***

(0.00138) (0.00138) (0.00140) (0.00140) # of children −0.0359*** −0.0373*** −0.0736*** −0.0734***

(0.0140) (0.0140) (0.0134) (0.0135) Race Black −0.635*** −0.644*** −0.649*** −0.647***

(0.0645) (0.0643) (0.0648) (0.0650) White (default) – – – – Gender Male −0.419*** −0.418*** −0.0525 −0.0526

(0.0410) (0.0411) (0.0385) (0.0385) Female (default) – – – – Region Midwest −0.396*** −0.396*** −0.151** −0.151**

(0.0680) (0.0681) (0.0598) (0.0598) South −0.854*** −0.857*** −0.502*** −0.501***

(0.0645) (0.0645) (0.0576) (0.0576) West −0.397*** −0.397*** −0.0506 −0.0510

(0.0692) (0.0692) (0.0635) (0.0635) Northeast (default) – – – – Type of place Big city 0.257*** 0.258*** 0.230*** 0.230***

(0.0624) (0.0624) (0.0582) (0.0582) Rural −0.339*** −0.343*** −0.282*** −0.282***

(0.0731) (0.0732) (0.0715) (0.0716) Other (default) – – – – Cut point 1 −1.092*** −1.450*** −1.180*** −1.104***

(0.122) (0.190) (0.111) (0.188) 2 1.042*** 0.685*** −0.416*** −0.339

(0.123) (0.191) (0.110) (0.188) 3 2.658*** 2.303*** 0.379*** 0.445**

(0.126) (0.193) (0.111) (0.188) 4 (obedience only) – – 2.453*** 2.529***

(0.116) (0.190)

F-statistic 47.17 45.46 70.09 64.66

Notes. 1. The interaction of current and parental SEI is excluded in model 1 and included in model 2. 2. Greater values of the logit and coefficients are associated with lesser approval of spanking or valuing of obedience. 3. The p-values of the variables are indicated as follows: * p < 0.05

** p < 0.02 *** p < 0.01.

4. For all regressions, prob > F = 0.000. 5. Results are presented with ordered logistic regression coefficients, rather than odds ratios, to facilitate comparisons of how various amounts of change in the continuous independent variables affect the outcome. Results with odds ratios are available from the author upon request. 6. Coefficients are not standardized to avoid ambiguities in defining the “standard deviation” of the y variable in an ordered logit model (see Menard, 2004), as well as difficulties in interpreting the standard deviations of binary independent variables.

7. The survey year variable equals the difference between the year in which the respondent was surveyed and 1994, the initial year of GSS data in the study’s timeframe.

Additional Methodological Considerations

Two potential indicators of SES contained in GSS data consist of measurements of occupational prestige (SEI) and educa- tional attainment. The former is chosen as this study’s primary metric of SES for substantive and practical reasons. In order to provide a substantively meaningful metric of SES, levels of educational attainment must be classified into a relatively small

M. Friedson / Child Abuse & Neglect 51 (2016) 263–275 269

Table 3 Dichotomized values of outcome variables by current and parental SEI.

Cells show: estimated percentage of population (standard error) Tercile of parental SEI

Lowest Middle Highest

(a) Percentage agreeing or strongly agreeing that spanking is necessary Tercile of current SEI Lowest 80.92 (1.01) 74.51 (1.68) 74.21 (1.81)

Middle 78.55 (1.35) 72.65 (1.60) 70.14 (1.45) Highest 75.31 (1.64) 67.72 (1.84) 62.58 (1.35)

(b) Percentage ranking obedience among the top three traits (of five) they value in children Tercile of current SEI Lowest 63.35 (1.21) 50.05 (1.98) 46.48 (2.10)

Middle 55.45 (1.52) 44.51 (1.81) 37.32 (1.67) Highest 44.12 (1.88) 35.69 (1.76) 26.93 (1.16)

Note. 1. The estimates in (a) of this table are derived using the sample from the regressions of the spanking variable that are summarized in Table 2(a) (N = 10,075). 2. A test of the likelihood ratio for the joint association of the row and column variables in (a) of this table with the dichotomized spanking outcome (Ho : neither the row variable, nor the column variable, nor their interaction is associated with the outcome)—adjusted for survey design parameters—indicates that F (7.88, 7199.01) = 19.45 (p = 0.00). The unadjusted likelihood ratio �2 = 201.78 (d.f. = 8, p = 0.00). 3. The estimates in (b) of this table are derived using the sample from the regressions of the obedience variable that are summarized in Table 2(b) (N = 10,043). 4 n t

n a S v v b

a t S o i y v

“ o t r g a a

R

l A m s s

fi d o o p h o i

. A test of the likelihood ratio for the joint association of the row and column variables in (b) of this table with the dichotomized obedience outcome (Ho : either the row variable, nor the column variable, nor their interaction is associated with this outcome)—adjusted for survey design parameters—indicates hat F (7.90, 6797.38) = 62.03 (p = 0.00). The unadjusted likelihood ratio �2 = 601.61 (d.f. = 8, p = 0.00).

umber of categories (indicating, for example, whether respondents dropped out of high school, graduated high school, ttended college, received a bachelor’s degree, etc.). These categories do not provide as fine-tuned of a measurement of ES as respondents’ positions on the continuous socioeconomic index. Furthermore, interaction effects between categorical ariables—such as metrics of educational attainment—are difficult to interpret when there are numerous categories for each ariable. Use of educational attainment as the indicator of SES would render onerous the interpretation of the interaction etween current and parental SES in relation to the spanking and obedience outcomes.

Nonetheless, it is imperative to assure that the study captures the relationship of SES to the outcome variables, rather than relationship specific to the metric through which SES is operationalized (SEI) (Krieger, Williams, & Moss, 1997). Toward his end, regressions are repeated in the appendix (Table D) with educational attainment substituted for SEI as the metric of ES. Using educational attainment as the SES metric also helps assure that the study’s results are not artifacts of the omission f subjects for whom current or parental SEI data is missing. The SEI of a respondent or respondent’s parent is not provided n the data source when that individual is not employed or looking for work and has not previously worked for at least a ear. (Those who are not employed, but are looking for a job or previously worked for at least a year, are assigned an SEI alue, by the GSS, based on their occupations when they were employed.)

It might be argued that the distinction between the responses indicating support for spanking (“strongly agree” versus agree”) or opposition to spanking (“disagree” versus “strongly disagree”) is arbitrary and does not reflect the actual strength f respondents’ attitudes. If this were true, the spanking outcome variable would not be ordinal, and it would be invalid o analyze it via ordered logit models. To account for this possibility, the appendix (Table C) provides results for logistic egressions of a dichotomized spanking variable, indicating whether subjects support or oppose spanking. Results are also iven for logistic regressions of a dichotomized obedience variable. By comparing results from these logistic regressions nd the ordered logit models in the study’s main body, it is possible to confirm that the study’s findings are not artifacts of ssumptions specific to ordered logistic regression techniques.

esults

Estimated population values of the SES and outcome variables are presented, respectively, in Table 1(a) and (b). The atter shows that a great majority of the population favors spanking, while few give top priority to the obedience of children. pproximately 73% of the population agrees or strongly agrees that spanking is necessary. Despite this widespread endorse- ent of spanking, ambivalence toward this practice also may be commonplace. The proportion expressing agreement with

panking is 1.7 times greater than that expressing strong agreement. Over half of the population ranks obedience as last or econd-to-last in importance among the five childhood traits listed in the survey, while less than 30% ranks it first or second.

To understand how one’s current SES and that of one’s family of origin combine to predict one’s parenting attitudes, it is rst necessary to understand how these SES variables are related to each other. To gauge the degree of correspondence or ivergence between the socioeconomic attainment of respondents and of their parents, the sample was divided into terciles f roughly equal population size by respondents’ current SEI. It was then divided into such terciles a second time by the SEI f respondents’ parents. Table 1(c) displays the estimated proportion of the population in each current SEI category with

arents in each SEI category. For individuals in the lowest category of current SEI, it is about three times more common to ave parents in the lowest than in the highest category. Disparities are not as great for individuals in the highest category f current SEI, with about 27% having parents in the lowest tercile and 47% having parents in the highest tercile. Among ndividuals in the middle category of current SEI, similar proportions fall into the highest and lowest parental SEI terciles.

270 M. Friedson / Child Abuse & Neglect 51 (2016) 263–275

Although there is some correspondence between the SEI of respondents and of their parents, these variables are sufficiently independent to allow both to be included as predictors in the regressions. The Pearson’s coefficient (R) for the correlation of these variables is about 0.3, meaning that the linear relationship between them explains less than 10% of their respective variances.

Table 2 shows results for the regressions of attitudes toward spanking (part [a]) and obedience (part [b]) on current and parental socioeconomic attainment. As discussed above, the regression is run twice for each outcome variable, first (model 1) with current and parental SEI as independent predictors, and a second time (model 2) with the addition of an interaction effect between these SEI variables.

The regressions of the obedience variable provide evidence for highly significant relationships (p < 0.01) between current SEI and attitudes toward obedience, as well as between parental SEI and these attitudes. The value of the obedience outcome variable increases as respondents assign less importance to obedience. The positive coefficients of the current and parental SEI variables thus indicate that one’s valuing of obedience declines as current and parental SEI increase. The coefficient of parental SEI is one-third less than that of current SEI in model 1 of the obedience variable. Model 2 does not provide evidence of an interaction between current and parental SEI in predicting attitudes toward obedience. However, the current and parental SEI variables remain highly significant (p < 0.01) in predicting attitudes toward obedience with the addition of this interaction effect.

Results from model 1 of the spanking outcome variable (which lacks the interaction term) show highly significant (p < 0.01) relationships between current SEI and attitudes toward spanking, and between parental SEI and these attitudes. The value of the spanking variable increases as subjects exhibit less favorable attitudes toward spanking. The positive coefficients of the current and parental SEI variables thus indicate that one’s favoring of spanking decreases as current and parental SEI increase. The coefficient of parental SEI is about 31% less than that of current SEI in model 1 of the spanking variable. With the addition of the interaction effect in model 2, the current and parental SEI variables drop out of statistical significance as independent predictors of attitudes toward spanking. The interaction between current and parental SEI is, however, highly significant (p < 0.02), with a positive coefficient.

In all models in Table 2, the dummy variable for black racial identity is strongly associated (p < 0.01) with more favorable attitudes toward spanking or obedience, and the coefficient of this dummy variable is great in magnitude. In model 1 of the spanking outcome, the logit of a black subject with the maximum current SEI (97.2), all else being equal, is approximately the same as that of a white subject with the minimum current SEI (17.1). In model 1 of the obedience outcome, being black decreases the logit by roughly the same amount as a decline in current SEI of two standard deviations (or about half the range of the socioeconomic index).

The control variables in the regressions display a consistent pattern of statistical significance, with coefficients in the expected directions (where applicable). Only the control for age lacks significance (at the p < 0.05 level) in the regressions of the spanking variable. As expected (see Flynn, 1994, 1998), residing in the West and being male are significant (p < 0.01) in predicting greater support for spanking (versus residing in the Northeast and being female, respectively). These traits are not, however, significant (at the p < 0.05 level) in predicting attitudes toward obedience. The variable indicating the number of children to whom the respondent is a parent is significant (p < 0.01) in predicting greater support for spanking and greater valuing of obedience. Note that regression results are robust to the substitution of a dichotomous variable indicating whether the respondent has any children for the continuous variable indicating number of children.

Living in a rural area is associated with greater support for spanking and valuing of obedience (p < 0.01), while residing in a big city predicts the opposite (p < 0.01). Analysis carried out by the author indicates that the study’s findings are robust to controls for additional categories of subject’s place of residence. Results are also robust to a variety of specifications of the family structure in which subjects were raised. These controls are not included in the study’s regressions, due to substantial collinearity with the variables indicating race and parental SEI. For example, 25.15% (std. error = 1.27%) of blacks and 9.32% (std. error = 0.32%) of whites in the survey population were living with a single mother at age sixteen. Results of regressions with controls for the family structure in which subjects were raised, or additional categories of respondents’ place of residence, are available from the author upon request.

The GSS questions regarding the valuing of spanking and obedience were asked to non-identical but overlapping subsets of GSS respondents (GSS Codebook, 2013, Appendix U). As a result, the sample sizes (N) in Table 2’s regressions of attitudes toward spanking and obedience are not equal. Nonetheless, the survey is designed to allow the subsets of respondents asked various questions (including those on spanking and obedience) to be representative of the survey population. Additionally, the regressions’ main results are maintained when the sample is limited to subjects for whom data on attitudes toward both spanking and obedience are available. Results with the sample limited in this fashion are available from the author upon request.

Table 3(a) and (b) are provided to aid in the interpretation of the interaction of current and parental SEI in predicting attitudes toward spanking and obedience. As in Table 1(c), respondents are divided into categories of high, middle, and low current SEI, and into categories of high, middle, and low parental SEI. Tables show the estimated proportion of the population, within each combination of the current and parental SEI categories, that agrees or strongly agrees with the

necessity of spanking or that ranks obedience in the top three among the five childhood virtues listed in the survey.

The results in Table 3(b) show that current SEI and parental SEI are independently related to the valuing of obedience. Among those with high SEI parents, roughly 20% separates the proportions in the high current SEI category and in the low current SEI category who rank obedience among their top three childhood virtues. The same applies in regard to individuals

M. Friedson / Child Abuse & Neglect 51 (2016) 263–275 271

Table 4 Ordered logit regressions of attitudes toward spanking, with controls for attitudes toward obedience.

Cells show: coefficient (S.E.) Spanking outcome variable (N = 4969)

Model 1 Model 2

Current SEI 0.00823*** 0.000523 (0.00157) (0.00405)

Parental SEI 0.00564*** −0.00353 (0.00171) (0.00503)

Current SEI × parental SEI 0.000157* (0.0000795)

Survey year 0.0111 0.0111 (0.00614) (0.00613)

Age 0.00527*** 0.00528***

(0.00198) (0.00198) # of children −0.0209 −0.0228

(0.0225) (0.0224) Obedience Top priority −1.128*** −1.130***

(0.116) (0.116) Second priority −0.890*** −0.888***

(0.117) (0.118) Third priority −0.702*** −0.696***

(0.116) (0.116) Fourth priority −0.554*** −0.553***

(0.0987) (0.0985) Lowest priority (default) – – Race Black −0.478*** −0.488***

(0.0906) (0.0904) White (default) – – Gender Male −0.447*** −0.445***

(0.0600) (0.0600) Female (default) – – Region Midwest −0.292*** −0.294***

(0.101) (0.102) South −0.830*** −0.832***

(0.0952) (0.0955) West −0.442*** −0.439***

(0.102) (0.102) Northeast (default) – – Type of place Big city 0.0969 0.0963

(0.0880) (0.0881) Rural −0.222* −0.223*

(0.102) (0.103) Other (default) – – Cut point 1 −1.702*** −2.137***

(0.199) (0.290) 2 0.640*** 0.206

(0.196) (0.286) 3 2.433*** 2.002***

(0.201) (0.289)

S

w p

d t p l p s

F-statistic 24.94 23.53

ee the notes for Table 2.

ith low SEI parents. Similarly, for individuals in each current SEI category, there is a difference of about 17–18% in the roportions with high SEI parents and with low SEI parents who rank obedience among their top three virtues.

The results in Table 3(a), in contrast, show that the relationship of current SEI to attitudes toward spanking is quite ifferent for individuals in different categories of parental SEI. The same applies in regard to the relationship of parental SEI o these attitudes for those in different categories of current SEI. Among those with high SEI parents, almost 12% separates the roportions in the high and low current SEI categories who favor spanking. Among those with low SEI parents, in contrast,

ess than 6% separates these proportions. Similarly, among those in the high current SEI category, almost 13% separates the roportions with high and low SEI parents who favor spanking. Among those in the low current SEI category, less than 7% eparates these proportions. Attitudes toward spanking are more sensitive to parental SEI for high current SEI individuals

272 M. Friedson / Child Abuse & Neglect 51 (2016) 263–275

than for low current SEI individuals, and these attitudes are more sensitive to current SEI for those with high SEI parents than for those with low SEI parents.

Patterns similar to those in Table 3(a) and (b) are obtained when the sample is limited to whites (contact author for details). This provides evidence that, though blacks disproportionately originate from lower-SES families, parental SEI is not serving as a proxy for race in its interactions with current SEI. Black subjects’ attitudes toward spanking are less differentiated than those of whites by current SEI (contact author for details). This could be due to error resulting from smaller N values for blacks, especially in the higher-SES range, rather than differences found in the population.

Results for regressions of the spanking outcome, with additional controls for respondents’ attitudes toward obedience, are provided in Table 4. With these controls, the predictor variables for current SEI and parental SEI remain highly significant (p < 0.01) in predicting attitudes toward spanking in model 1 (which lacks the interaction term), while the interaction term is statistically significant (p < 0.05) in model 2. The directionality of these predictors is the same as in Table 2; greater current and parental SEI continue to be associated with less favorable attitudes toward spanking. These results provide evidence of a relationship of the SEI variables to attitudes toward spanking that is not mediated by the valuing of obedience. The categorical variables indicating the ranking of obedience are also highly significant (p < 0.01), relative to the default category (obedience ranked last), in predicting support for spanking. As expected, greater valuing of obedience is associated with more favorable attitudes toward spanking.

The regression results reported in the appendix demonstrate that the main findings reported in Table 2, with regard to the significance and directionality of the SEI variables and interaction effects, are robust to a variety of potential modifications of the regressions. As shown in Table A, these findings are substantially confirmed when the current SEI variable is not adjusted, in cases where the SEI of the respondent’s spouse is greater than that of the respondent, to equal that of this spouse. Likewise, as shown in Table B, they are confirmed when the SEIs of parent substitutes are not considered in determining the value of the parental SEI variable.

Table C reports results for logit models of dichotomized versions of the outcome variables, and shows that the regression results in Table 2 are not specific to the ordered logit models employed there. Table C also demonstrates that results regarding attitudes toward spanking are not dependent on measuring the strength of subjects’ agreement or disagreement with spanking. The regressions in Table D use educational attainment rather than SEI as the metric of socioeconomic status. Their results demonstrate that the associations of greater current and parental SES with less favorable attitudes toward spanking and obedience are not specific to the use of SEI as the measurement of socioeconomic status.

Finally, Table E demonstrates that the regression results are robust to the employment of an alternative timeframe for the analysis and to controls for whether respondents are of Hispanic origin. The Hispanic dummy variable, like that indicating African-American racial identity, is highly significant (p < .01) in predicting greater valuing of obedience but is also, in contrast to the African-American dummy, highly significant (p < .01 and p < .02, respectively, in model 1 and model 2 of the spanking outcome) in predicting less favorable attitudes toward spanking. This result demonstrates that the strong support for spanking found among African-Americans is not generalizable to all minority groups that are subjected to racial residential segregation and other forms of discrimination.

Discussion

The results of the analysis provide evidence that childrearing attitudes are related not only to one’s current SES, but also to the SES in which one was raised. After controlling for current SES, growing up in a low-SES household remains associated with greater approval of spanking and valuing of obedience. While current and parental SES are independently related to the valuing of obedience, they compound each other in regard to dispositions toward spanking. The relationship of parental SES to attitudes toward spanking is most pronounced among those with high current SES. Thus, rather than muting the association of lowly social origins with endorsement of spanking, attainment of upward mobility amplifies this association. Approval of spanking is common across the socioeconomic spectrum, and it may require multiple generations of socioeconomic privilege for individuals to deviate from this pattern.

In addition to being related to household-level socioeconomic disadvantage, preferences for authoritarian parenting are associated with disadvantage defined in terms of race. African-Americans in US society face disadvantages not reducible to household-level SES (Massey, 2004; Massey & Denton, 1993; Peterson & Krivo, 2010; Sharkey, 2013). The results of this study show that childrearing attitudes may constitute an important element of these additional disadvantages. Strong relationships of black racial identity with the valuing of obedience and support for spanking appear in regressions with controls for current and parental SES. This demonstrates that these relationships are not strictly a product of the elevated prevalence of socioeconomic disadvantage found among African-American households. A black racial identity is associated with greater support for spanking to a degree that would cancel out the reduction in support for spanking commensurate with the greatest possible gain in current SES. It is likewise associated with greater valuing of obedience to a degree comparable to an extreme decrease in current SES. These parenting values do not, however, appear to be shared universally among disadvantaged minority groups, as indicated by the results in the appendix regarding Hispanics.

The bulk of the study’s findings nonetheless point to a common conclusion, namely that coming from a disadvantaged background is related to greater support for authoritarian modes of childrearing. This is true whether parenting values are defined in terms of valuing of obedience or support for spanking, and whether disadvantage is defined in terms of SEI, edu- cational attainment, or racial identity. If parenting values translate into parenting practices, and if spanking and an excessive

e g f

c l t B o

u u i o s

o h h o

t o t i c t

I

a i e m 2 h e h N

b t t i a d

i c d T c M f a u a T d o

M. Friedson / Child Abuse & Neglect 51 (2016) 263–275 273

mphasis on obedience are harmful to children, then these findings may have important implications regarding the inter- enerational transmission of disadvantage. Elevated support for harsh parenting approaches among high-SES individuals rom low-SES origins may lend itself to the reappearance of their former disadvantages in future generations.

The findings outlined here should be considered in light of the study’s limitations, which are primarily driven by data onstraints. The data source provides attitudinal and not behavioral data regarding childrearing. Support is provided for a ink between parenting values and harsh parenting behavior by a past study showing that beliefs about spanking mediate he inverse relationship between current SES and the severity of punitive parenting practices (Pinderhughes, Dodge, Zelli, ates, & Pettit, 2000; see also Luster et al., 1989). Nonetheless, if it is harsh parenting behavior that directly harms children r induces disadvantage, it is crucial for this study to be replicated with outcome variables measuring such behavior.

This study is additionally constrained in its ability to represent parenting attitudes. The self-report data used here may nderestimate subjects’ actual approval of spanking, owing to the same social and normative pressures that contribute to nderreporting of spanking behavior (see Giles-Sims et al., 1995; Holden, Williamson, & Holland, 2014). Subjects here were

nterviewed no later than 2010, prior to the highly publicized sanctioning of Adrian Peterson for using harsh corporal means f child discipline. This case, which generated public outrage, may have induced or indicated shifts in public approval of panking too recent to be reflected in the study’s data.

It is not possible, with the available data, to distinguish between genetic influences on childrearing and those shaped by ne’s current or past social environments. Both childrearing approaches and SES may be partly determined by genetically eritable traits such as intelligence, self-control, and aggression. Genetics may thus help to account for relationships revealed ere of childrearing values to current or parental SES. To better parcel out effects of current or former social environments n childrearing, future studies should seek to control for genetic influences.

A final limitation imposed by the data source is that it does not facilitate controls for whether respondents were subjected o corporal punishment or physical abuse as children. An extensive literature documents the intergenerational transmission f violent behavior (see, e.g., Simons et al., 1991). It is possible that respondents from lower SES backgrounds are more likely o support spanking because they were more likely to have been corporally punished or physically abused as children. Even f true, this possibility helps to explain, rather than negates, the associations revealed in this study. Whatever the cause, it is ritical to understand that low levels of SES are associated with preferences for potentially detrimental modes of parenting hat may persist across generations.

mplications of the Persisting Associations of Social Origins with Parenting Values

The sociological significance of the relationship between disadvantage and authoritarian parenting should be assessed s an aspect of the greater punitiveness to which disadvantaged individuals may be subjected through the life course. Adult ncarceration rates have increased fivefold in the US since the 1970s (Wildeman & Western, 2009, p. 227), and these rates are xceptionally elevated among poor and minority populations. An extensive literature documents severe collateral harms of ass incarceration that reinforce the disadvantages faced by underprivileged individuals, families, and communities (Clear,

009; Defina & Hannon, 2013; Fagan & Meares, 2008; Wildeman & Western, 2010). A wave of media and scholarly attention as focused upon the school suspension of minority students at disproportionate levels, starting as early as preschool and xtending through secondary education (Arcia, 2007; Carr, 2014; Rich, 2014; Rios, 2010). These greater suspension levels ave been shown to contribute to the gap in educational achievement impacting minority populations (Gregory, Skiba, & oguera, 2010).

The present research helps to demonstrate that the focus on punitiveness, in the treatment of disadvantaged populations, egins before encounters with school officials or the police in the domestic sphere of childrearing. This suggests that punitive reatment may be as persistent through the lives of the underprivileged as are preferences for authoritarian childrearing hrough generations of disadvantaged families. The bulk of existing research on the effects of corporal punishment analyzes ts psychological harms to individuals. The focus of this research should be extended to consider how corporal punishment nd authoritarian parenting may work in conjunction with other punitive interventions through the life course to the etriment of underprivileged individuals, families, and communities.

Disadvantaged individuals face punitive extremes in their treatment by public authorities and may also harbor internal- zed preferences for excessively punitive or controlling modes of self- or family-regulation. A recent Atlantic Monthly article laims that underprivileged New Orleans high school students and their parents seek out charter schools structured around raconian disciplinary regimes because they believe such discipline is required for academic and career success (Carr, 2014). his piece, and its academic counterparts, portray preferences for strict discipline as adaptations to the demands of lower lass individuals’ social environments (Baldwin, Baldwin, & Cole, 1990; Furstenberg, 1993; Gonzales, Cauce, Friedman, & ason, 1996). These individuals can least afford disciplinary lapses because they have fewer resources and opportunities,

ace heightened exposure to violent crime and other social ills, and are subjected to greater scrutiny by law enforcement uthorities. Extremes in disciplinary regulation may, however, be detrimental to underprivileged individuals’ prospects for pward mobility. This is shown to be the case in regard to the aforementioned New Orleans schools, where an excess of rules

nd sanctioning impairs students from developing the drive and self-directedness necessary for future success (Carr, 2014). he present study demonstrates a different type of potentially counterproductive excess in regard to preferences for strict isciplinary regulation, where such preferences persist beyond the immediacy of low-SES environments (in households of rigin) to which they may be adaptive.

274 M. Friedson / Child Abuse & Neglect 51 (2016) 263–275

While past analyses have argued that preferences for strict discipline are adaptations to a contemporaneous social context, the findings here instead suggest that these preferences may be ingrained among those from low-SES backgrounds. It does not necessarily follow, however, that such preferences are an aspect of a “culture of poverty” in the conventional sense of this concept (Bowles, Durlauf, & Hoff, 2006; Lewis, 1969; Moynihan, 1965). According to this concept’s proponents, members of poor communities tend to develop a set of self-defeating beliefs and practices that serve to perpetuate and intensify their poverty and social marginality. A culture of poverty is typically thought to include such traits as an excessive emphasis on instant gratification and violent modes of conflict resolution that are antithetical to traits associated with strict disciplinary controls. An emphasis on strict and punitive regulation may be a means through which disadvantaged individuals attempt to overcome the marginalizing inertia of a hostile social environment. Yet when taken to an excess, either in terms of its severity or durability, this emphasis may have an opposite and self-defeating outcome.

Acknowledgements

The author would like to thank Patrick Sharkey and Mikaila Arthur for excellent feedback on an earlier draft of this manuscript.

References

Arcia, E. (2007). A comparison of elementary/k-8 and middle schools’ suspension rates. Urban Education, 48(5), 456–469. Baldwin, A. L., Baldwin, C., & Cole, R. E. (1990). Stress resistant families and stress resistant children. In J. Rolf, A. S. Masten, D. Cicchetti, & K. H. Nuechterlein

(Eds.), Risk and protective factors in the development of psychopathology (pp. 257–280). New York: Cambridge University Press. Berlin, L. J., Ispa, J. M., Fine, M. A., Malone, P. S., Brooks-Gunn, J., Brady-Smith, C., & Bai, Y. (2009). Correlates and consequences of spanking and verbal

punishment for low-income White, African American, and Mexican American toddlers. Child Development, 80, 1403–1420. Bluestone, C., & Tamis-LeMonda, C. S. (1999). Correlates of parenting styles in predominantly working- and middle-class African-American mothers. Journal

of Marriage and the Family, 61, 881–893. Borooah, V. (2002). Logit and probit: Ordered and multinomial models. Thousand Oaks: Sage. Bowles, S., Durlauf, S., & Hoff, K. (2006). Poverty traps. Princeton: Princeton University Press. Bugentall, D. B., Martorell, G. A., & Barraza, V. (2003). The hormonal costs of subtle forms of infant maltreatment. Hormones and Behavior, 43(1), 237–244. Burdick-Will, J., Ludwig, J., Raudenbush, S., Sampson, R., Sanbonmatsu, L., & Sharkey, P. (2011). Converging evidence for neighborhood effects on children’s

test scores: An experimental quasi-experimental and observational comparison. In G. Duncan, & R. Murnane (Eds.), Whither opportunity? Rising inequality and the uncertain life chances of low income children (pp. 255–276). New York, NY: Russell Sage Foundation.

Carr, S. (2014, November 17). How strict is too strict? The backlash against no-excuses discipline in high school. The Atlantic Monthly, http://www.theatlantic.com/magazine/archive/2014/12/how-strict-is-too-strict/382228/

Christie-Mizell, C. A., Pryor, E. M., & Grossman, E. R. (2008). Child depressive symptoms, spanking, and emotional support: Differences between African American and European American youth. Family Relations, 57(3), 335–350.

Clear, T. (2009). Imprisoning communities: How mass incarceration makes disadvantaged neighborhoods worse. New York, NY: Oxford University Press. Conger, R. D., Ge, X., Elder, G. H., Lorenz, F. O., & Simons, R. L. (1994). Economic stress, coercive family process, and developmental problems of adolescents.

Child Development, 65, 541–561. Curtner-Smith, M. E., Bennett, T. L., & O’Rear, M. R. (1995). Fathers’ occupational conditions, values of self-direction and conformity, and perceptions of

nurturant and restrictive parenting in relation to young children’s depression and aggression. Family Relations, 44(3), 299–305. Day, R. D., Peterson, G. W., & MacCracken, C. (1998). Predicting spanking of younger and older children by mothers and fathers. Journal of Marriage and the

Family, 60, 79–94. Defina, R., & Hannon, L. (2013). The impact of mass incarceration on poverty. Crime & Delinquency, 59(4), 562–586. Dornbusch, S. M., Ritter, P. L., Leiderman, P. H., Roberts, D. F., & Fraleigh, M. J. (1987). The relation of parenting style to adolescent school performance. Child

Development, 58, 1244–1257. Duncan, O. D. (1961). A socioeconomic index for all occupations. In Albert Reiss (Ed.), Occupations and social status (pp. 109–138). New York, NY: Free Press

of Glencoe. Durant, J., & Ensom, R. (2012). Physical punishment of children: Lessons for 20 years of research. Canadian Medical Association Journal, 184, 1373–1377. Ellison, C. G., Bartkowski, J. P., & Segal, M. L. (1996). Conservative Protestantism and the parental use of corporal punishment. Social Forces, 74, 1003–1028. Erlanger, H. S. (1974). Social class and corporal punishment in childrearing: A reassessment. American Sociological Review, 39, 68–85. Fagan, J., & Meares, T. (2008). Punishment, deterrence, and social control: The paradox of punishment in minority communities. Ohio State Journal of Criminal

Law, 6, 173–229. Flynn, C. P. (1994). Regional differences in attitudes toward corporal punishment. Journal of Marriage and the Family, 56, 314–324. Flynn, C. P. (1996). Normative support for corporal punishment: Attitudes, correlates, and implications. Aggression and Violent Behavior, 1(1), 47–55. Flynn, C. P. (1998). To spank or not to spank: The effect of situation and age of child on support for corporal punishment. Journal of Family Violence, 13(1),

21–37. Fox, R., & Solis-Camara, P. (1997). Parenting of young children by fathers in Mexico and the United States. Journal of Social Psychology, 137(4), 489–495. Furstenberg, F. (1993). How families manage risk and opportunity in dangerous neighborhoods. In P. W. J. Wilson (Ed.), Sociology and the Pubic Agenda (pp.

231–258). Newbury Park, CA: Sage. (2013). General Social Surveys, 1972–2012: Cumulative Codebook. Chicago, IL: National Opinion Research Center. http://publicdata.norc.org:41000/

gss/documents/BOOK/GSS Codebook.pdf Gershoff, E. T., Lansford, J. E., Sexton, H. R., Davis-Kean, P., & Sameroff, A. (2012). Longitudinal links between spanking and children’s externalizing behavior

in a national sample of Black, White, Hispanic, and Asian American families. Child Development, 83(3), 838–843. Giles-Sims, J., Straus, M., & Sugarman, D. (1995). Child, maternal, and family characteristics associated with spanking. Family Relations, 44, 170–176. Gonzales, N. A., Cauce, A. M., Friedman, R. J., & Mason, C. A. (1996). Family, peer, and neighborhood influences on academic achievement among African-

American adolescents: One-year prospective effects. American Journal of Community Psychology, 24, 365–387. Gregory, A., Skiba, R., & Noguera, P. (2010). The achievement gap and the discipline gap: Two sides of the same coin? Educational Researcher, 39(1), 59–68. Grogan-Kaylor, A. (2004). The effect of corporal punishment on antisocial behavior in children. Social Work Research, 28(3), 153–162.

Grogan-Kaylor, A. (2005). Corporal punishment and the growth trajectory of children’s antisocial behavior. Child Maltreatment, 10(3), 283–292. Grogan-Kaylor, A., & Oti, M. D. (2007). The predictors of parental use of corporal punishment. Family Relations, 56(1), 80–91. Hawkins, A. O., Danielson, C. K., de Arellano, M. A., Hanson, R. F., Ruggiero, K. J., Smith, D. W., Saunders, B. E., & Kilpatrick, D. G. (2010). Ethnic/racial differences

in the prevalence of injurious spanking and other child physical abuse in a national survey of adolescents. Child Maltreatment, 15(3), 242–249. Hill, M. S., & Duncan, G. J. (1987). Parental family income and the socioeconomic attainment of children. Social Science Research, 16(1), 39–73.

H

H

H

J

K K K

L L L

L

L L L

M

M

M

M M M

M M N

N

P P P

R

R R S S S S

S

T T

W W

A

j

M. Friedson / Child Abuse & Neglect 51 (2016) 263–275 275

off, E., Laursen, B., & Tardif, T. (2002). Socioeconomic status and parenting. In M. H. Bornstein (Ed.), Handbook of parenting: Vol. 2. Biology and ecology of parenting (pp. 231–252). Mahwah, NJ: Lawrence Erlbaum Associates.

olden, G. W., Williamson, P. A., & Holland, G. W. (2014). Eavesdropping on the family: A pilot study of corporal punishment in the home. Journal of Family Psychology, 28(3), 401–406.

orn, I. B., Cheng, T. L., & Joseph, J. (2004). Discipline in the African American community: The impact of socioeconomic status on beliefs and practices. Pediatrics, 113(5), 1236–1241.

encks, C., & Mayer, S. E. (1990). The social consequences of growing up in a poor neighborhood. In L. Lynn, & M. McGeary (Eds.), Inner-city poverty in the United States (pp. 111–186). Washington, DC: National Academy Press.

ohn, M. (1963). Social class and parent-child relationships: An interpretation. American Journal of Sociology, 68(4), 471–480. ohn, M. (1969). Class and conformity: A study in values. Homewood, IL: Dorsey Press. rieger, N., Williams, D., & Moss, N. (1997). Measuring social class in US public health research: Concepts, methodologies, and guidelines. Annual Review of

Public Health, 18, 341–378. areau, A. (2002). Invisible inequality: Social class and childrearing in black families and white families. American Sociological Review, 67(5), 747–776. areau, A. (2011). Unequal childhoods: Class, race and family life. Berkeley, CA: UC Press. eventhal, T., & Brooks-Gunn, J. (2000). The neighborhoods they live in: The effects of neighborhood residence on child and adolescent outcomes. Psycho-

logical Bulletin, 126, 309–337. ewis, O. (1969). Culture of poverty. In D. P. Moynihan (Ed.), On understanding poverty: Perspectives from the social sciences (pp. 187–220). New York, NY:

Basic Books. ipset, S. M. (1959). Democracy and working class authoritarianism. American Sociological Review, 24, 482–501. ipsitz, L. (1965). Working-class authoritarianism, a re-evaluation. American Sociological Review, 30, 103–109. uster, T., Rhoades, K., & Haas, B. (1989). The relation between parental values and parenting behavior: A test of the Kohn hypothesis. Journal of Marriage

and the Family, 51, 139–147. acKenzie, M. J., Nicklas, E., Waldfogel, J., & Brooks-Gunn, J. (2013). Spanking and child development across the first decade of life. Pediatrics, 132(5),

1118–1125. acKenzie, M. J., Nicklas, E., Waldfogel, J., & Brooks-Gunn, J. (2015). Spanking and children’s externalizing behaviors across the first decade of life: Evidence

for transactional processes. Journal of Youth and Adolescence, 44(3), 658–669. acMillan, H. L., Boyle, M. H., Wong, M. Y., Duku, E. K., Fleming, J. E., & Walsh, C. A. (1999). Slapping and spanking in childhood and its association with

lifetime prevalence of psychiatric disorders in a general population sample. Canadian Medical Association Journal, 161(7), 805–809. assey, D. (2004). Segregation and stratification. Du Bois Review, 1, 1–19. assey, D., & Denton, N. (1993). American apartheid: Segregation and the making of the underclass. Cambridge, MA: Harvard University Press. cLloyd, V. C. (1990). The impact of economic hardship on black families and children: Psychological distress, parenting, and socioemotional development.

Child Development, 61, 311–346. enard, S. (2004). Six approaches to calculating standardized regression coefficients. American Statistician, 58(3), 218–223. oynihan, D. P. (1965). The Negro family: The case for national action. Washington, DC: Office of Policy Planning and Research, US Department of Labor. akao, K., & Treas, J. (1990). Computing 1989 occupational prestige scores. General Social Survey Methodological Report No. 70.

http://publicdata.norc.org:41000/gss/documents/reports/Methodological Reports/MR070.pdf akao, K., & Treas, J. (1992). The 1989 socioeconomic index of occupations: Construction from the 1989 occupational prestige scores. General Social Survey

Methodological Report No. 74. http://publicdata.norc.org:41000/gss/documents/reports/Methodological Reports/MR074.pdf ager, D. (2003). The mark of a criminal record. American Journal of Sociology, 108, 937–975. eterson, R., & Krivo, L. (2010). Divergent social worlds: Neighborhood crime and the racial–spatial divide. New York, NY: Russell Sage Foundation. inderhughes, E., Dodge, K., Zelli, A., Bates, J., & Pettit, G. (2000). Discipline responses: Influences of parents’ socioeconomic status, ethnicity, beliefs about

parenting, and cognitive-emotional processes. Journal of Family Psychology, 14(3), 380–400. ich, M. (2014, March 21). School data finds pattern of inequality across racial lines. New York Times,

http://www.nytimes.com/2014/03/21/us/school-data-finds-pattern-of-inequality-along-racial-lines.html ios, V. (2010). Punished: Policing the lives of Black and Latino boys. New York, NY: New York University Press. osier, K. B., & Corsaro, W. A. (1993). Competent parents, complex lives: Managing parenthood in poverty. Journal of Contemporary Ethnography, 22, 171–204. harkey, P. (2013). Stuck in place: Urban neighborhoods and the end of progress toward racial equality. Chicago, IL: University of Chicago Press. imons, R., Whetbeck, L., Conger, R., & Chyi-In, W. (1991). Intergenerational transmission of harsh parenting. Developmental Psychology, 27(1), 159–171. mith Blau, Z. (1965). Class structure, mobility, and change in child rearing. Sociometry, 28, 210–219. trassberg, Z., Dodge, K., Pettit, G., & Bates, J. (1994). Spanking in the home and children’s subsequent aggression toward kindergarten peers. Development

and Psychopathology, 6(3), 445–461. trauss, V. (2014, September 18). 19 states still allow corporal punishment in school. Washington Post, http://www.washingtonpost.com/

blogs/answer-sheet/wp/2014/09/18/19-states-still-allow-corporal-punishm ent-in-school/ reiman, D. J. (2009). Quantitative data analysis: Doing social research to test ideas. San Francisco, CA: John Wiley & Sons. udge, J. R., Hogan, D. M., Snezhkova, I. A., Kulakova, N. N., & Etz, E. E. (2000). Parents’ child-rearing values and beliefs in the United States and Russia: The

impact of culture and social class. Infant and Child Development, 9, 105–121. ildeman, C., & Western, B. (2009). The black family and mass incarceration. Annals of the American Academy of Political and Social Science, 621, 221–242. ildeman, C., & Western, B. (2010). Incarceration in fragile families. Future of Children, 20(2), 157–177.

ppendix A. Supplementary data

Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/ .chiabu.2015.10.001.