Combining Love and Limits in Authoritative Parenting: A Conditional Sequence Model of Disciplinary Responses

Popular Books

Current popular childrearing books often emphasize either nurturance and communication on the one hand or firm control on the other. Take, for example, the three best-selling popular books from my bookshelf. Thomas Gordon's (1975) book on Parent Effective Training emphasizes communication. Seven of his 16 chapters deal with communication. Not only does he advocate good communication, he is against forceful disciplinary tactics. "One thing [to learn] from this book," he said, is that "each and every time they force a child to do something by using their power or authority, they deny that child a chance to learn self-discipline" (p. 158). At the other extreme is James Dobson's (1970) book, Dare to Discipline. Four of his 7 chapters are on disciplinary responses to misbehavior. He is not against nurturance or communication, but the first of his five key elements is the following: "Developing respect for parents is the critical factor in child management." Elsewhere he said, "When a youngster tries . . . stiff-necked rebellion, you had better take it out of him, and pain is a marvelous purifier" (p. 16).

A third best-seller presents more of a balance between the two extremes represented by Gordon and Dobson. Fitzhugh Dodson emphasizes both nurturance and control in his title, How to Discipline with Love. Although 6 of his 9 chapters are on discipline responses, his balance is represented by the following quote: "I believe it is far better to solve a conflict by negotiation and agreement rather than through power. However, in extreme cases . . . I believe we have to fall back on sheer power" (p. 92).

Social Science

The same polarization exists in empirical, social scientific literature. Cognitive developmental psychology and behavioral parent training have had the most sustained series of empirical studies on parental discipline. They complement each other in many important ways, as shown in Table 1. However, they often hold contradictory views about optimal disciplinary responses. Cognitive developmentalists recommend reasoning as a disciplinary response (Grusec & Kuczysnki, 1997; Grusec & Goodnow, 1994; Hoffman, 1977). Their measures of parental discipline often reflect this by including reasoning and gentle discipline tactics at one end of a continuum and harsh punishment at the other end (e.g., Kochanska, 1991; Weiss, Dodge & Bates, 1992). In contrast, behavioral parent trainers feature consistent use of punishment as a response to misbehavior in training parents to manage their disruptive children more effectively (Barkley, 1987; Eyberg & Boggs, 1989; Forehand & McMahon, 1981; Patterson, 1982). They generally regard reasoning as an ineffective discipline tactic (e.g., Blum, Williams, Friman & Christopherson, 1995), except for descriptions of the contingencies of punishment and reinforcement.

A few experts have bridged this conceptual gap, most notably Diana Baumrind in her work on parenting styles. She contrasted three major parenting styles, authoritarian, authoritative, and permissive. Authoritarian parents emphasized firm control, permissive parents emphasized nurturance, whereas authoritative parents emphasized both. In addition, authoritative parents emphasized communication with their children and encouraged age-appropriate skills and autonomy. The children of authoritative parents generally showed more individual initiative and social responsibility in Baumrind's studies, compared to children of the other two types of parents.

The contradictory recommendations of cognitive developmental vs. clinical behavioral research concerning reasoning vs. punishment is an important puzzle that needs solved to have a good scientific foundation for advising parents. Solving that puzzle has been a major goal of my own research program. Something is wrong when the most recommended disciplinary tactic in developmental psychology is totally ignored in clinical behavioral work with parents and when the centrality of effective punishment in the clinical work is directly contradicted in the developmental literature. The two literatures generally ignore each other on this matter, with a few exceptions.

This strange situation is reflected in the kinds of research questions asked and the measures used, which often assume the correctness of the author's implicit recommendations. For example, few studies investigate differences between effective and counterproductive use of a particular disciplinary tactic, whether reasoning or punishment. Instead, the preferred disciplinary tactic is assumed to be invariably effective and the other one invariably ineffective, thus thwarting finer discriminations about their respective effectiveness. In contrast, my research program attempted to compare the effectiveness of reasoning and punishment with each other and with other disciplinary responses.

Punishment Enhances Reasoning

Another advantage of a reasoning-punishment combination is that it enhances the subsequent effectiveness of reasoning when used by itself. Three different analyses of my data showed that disciplinary reasoning with 2- and 3-year-olds was ineffective unless it was backed up with punishment periodically. The children whose behavior improved the most over the next 20 months were those whose mothers frequently used reasoning alone (i.e., without punishment), but also backed up the reasoning with punishment when necessary. In contrast, the children whose behavior deteriorated the most had mothers who frequently used reasoning alone, but rarely backed it up with punishment.

The same kind of conditional sequencing of disciplinary tactics has been studied with a stronger research design by Mark Roberts and his colleagues. Behavioral parent training has been shown to be one of the most effective treatments for disruptive children (Kazdin, 1987). Roberts did a series of studies to identify the components of that treatment that accounted for its effectiveness. The treatment includes five major components: positive reinforcement of appropriate behavior, clear instructions, a single warning of impending time out, time out, and a back-up for time-out noncompliance, traditionally a 2-swat spank for 2- to 6-year-olds. With particularly noncompliant children, time out was often ineffective without a back-up procedure to enforce compliance with the time out. The spank back-up was effective in enforcing compliance with the time out, thus making the time out effective and minimizing the subsequent need for the spank back-up. Roberts and his colleagues investigated 3 alternative back-up procedures, but a brief room isolation was the only back-up that proved to be as effective as the original spank back-up. Similar to my research, contingent use of a more severe back-up procedure enhanced the subsequent effectiveness of a less severe disciplinary tactic, in this case time out.

Predictions of the Model

The conditional sequence model makes a variety of predictions that are consistent with the evidence to date, some of it counter-intuitive. First, it predicts that the correlational superiority of reasoning over punishment will be more evident among easily managed children than among difficult children. This is consistent with Grusec and Goodnow's (1994) assessment of the cognitive developmental literature. Disciplinary reasoning was associated with better child outcomes than was punishment, but mostly in samples of middle-class mothers. Samples of fathers, working-class families, preschoolers, boys, and temperamentally difficult children have generally failed to find such associations. Kochanska (1991) used a measure that contrasted power assertion at one extreme with rational growth encouragement at the other extreme. She found that power assertion predicted less conscience development in high-anxiety children, but not in low-anxiety children. She concluded that there were different paths to conscience development in the two types of children. An alternative conclusion is that parents could phase out the punishment aspects of the conditional sequence model faster for high-anxiety children than for low-anxiety children. A more equal balance between reasoning and punishment would then have been optimal for the low-anxiety children, which her measure would have missed (because an equal balance was represented in the muddled middle of her measure).

A second prediction is that the superiority of punishment over reasoning will be most evident among particularly difficult children. Patterson (1982) initially tried to help parents manage their disruptive children better by emphasizing reinforcement of appropriate behaviors. By 1982, he concluded, "If I were allowed to select only one concept to use in training parents of antisocial children, I would teach them how to punish more effectively" (p. 111), referring to time out as the punishment of choice.

Thirdly, the conditional sequence model predicts that children's aggression may be decreased more by conditional recommendations against spanking than by universal recommendations against spanking. Universal anti-spanking advice may make gentler disciplinary tactics less effective unless an equally effective back-up replaces spanking. Some evidence from Sweden is consistent with that prediction. Figure 4 shows the frequency of assaults by minors against minors since Sweden banned spanking in 1979. Figure 5 shows a possible reason for that: Swedish parents are less likely to use recommended alternative disciplinary tactics such as reasoning and time out, but they are more likely to use yelling and restraining (Palmerus & Scarr, 1995). Instead, the conditional sequence model implies that spanking should be discouraged as a first response to misbehavior and as a second response to continued misbehavior, but it should not be prohibited as a back-up enforcer for nonphysical punishment, unless a replacement enforcer is used.

[full image: 15K]

Figure 4. Frequency of child abuse and assaults by minors against minors, 1981 to 1994 (Statistics Sweden, 1995).

[full image: 20K]

Figure 5. Discipline practices reported by parents in Sweden and the United States (from Palmerus & Scarr, 1995).

This is consistent with behavioral parent training, which discourages spanking except to enforce time-out compliance. Outcome studies support the effectiveness of this treatment for reducing disruptive behavior (Kazdin, 1995). In addition, four studies found that behavioral parent training reduced parental spanking. Two of them used the spank back-up for noncompliance with time out (Eyberg, 1993; Roberts, 1984), and two of them used alternative back-ups for time out (McNeil, Clemens-Mowrer, Gurwitch & Funderburk, 1994; Webster-Stratton, 1990). How spanking is used is more critical than whether it is used.

A fourth prediction of the conditional sequence model is that early prevention programs that discourage the use of parental punishment will be more effective for general populations than for at-risk populations. Guterman (1997) reviewed the 18 most rigorous evaluations of early parenting programs to prevent child abuse. He found that 8 of the 10 programs targeted to general populations were effective in reducing subsequent child abuse, but only 1 of the 8 programs for at-risk populations showed similar effectiveness. The programs all featured support services and parent education. Presumably the support services would help any family, but the parent education component may have helped parents deal with easily managed children more than it helped with at-risk children. I am assuming that the parent education components often discouraged punishment, but I cannot document that.

Finally, the conditional sequence model predicts that features of the model will minimize the risk of escalation within discipline incidents. Some support for this is shown by the evidence summarized above that behavioral parent training decreases subsequent use of spanking. It also seems consistent with the escalation processes differentiating aggressive from non-aggressive families, as identified by Snyder and his colleagues (Snyder, Edwards, McGraw, Kilgore & Holton, 1994). They found that parents of aggressive sons tended to match their sons' level of aversiveness, whereas parents of non-aggressive sons were more likely to respond at a slightly lower level of aversiveness. Further, parents of aggressive sons were more likely to respond to any de-escalation of the conflict by escalating the incident themselves. In contrast, parents of non-aggressive sons were likely to respond to the child's de-escalation by bringing the incident to a close. Thus parents of non-aggressive sons out-persist their children, but do not out-escalate them. In one sense, this may seem to contradict the conditional sequence model, implying that parents should never escalate to spanking. As discussed above, however, the spank back-up enhances subsequent effectiveness of less severe disciplinary tactics, helping persistence to work later without resorting to spanking. Parents should typically persist until either compliance is achieved or the child de-escalates the situation in an appropriate way (e.g., specifying when he will do the requested chore). They should prefer the least aversive tactics that will accomplish that end, but be willing to resort to more aversive, but non-abusive disciplinary tactics when necessary.

Expanding the Model

To this point, this paper has focused on a simple three-step conditional sequence model. Although it suggests a way to reconcile contradictory recommendations and successfully makes several innovative predictions, actual parental practices are more varied. This section uses empirical evidence to suggest helpful ways to expand the simple model. First, there is some evidence that a single warning can dramatically reduce the need to move to the next level of severity without compromising the effectiveness of the overall disciplinary strategy. Roberts (1982) found that a single time-out warning reduced the need to use time out by 74% without diminishing the effectiveness of the overall parental disciplinary sequence. Multiple warnings reduce the effectiveness of the overall discipline somewhat (Sherrill, O'Leary & Kendziora, 1992).

The tactics used can also be expanded beyond the three featured in the simple conditional sequence model. My major research was designed to investigate a wide range of disciplinary responses with 2- and 3-year-olds. As indicated above, the combination of reasoning and punishment was associated with the longest delay until the next recurrence of misbehavior. Some other disciplinary responses, however, were nearly as effective as the reasoning-punishment combination. Four of 13 other discipline responses were never significantly worse in delaying misbehavior recurrences. Three were combinations of an aversive verbal disciplinary response with some kind of action. Aversive verbal tactics included scolding or shaming, threatening, and yelling. The actions were either punishment, forced compliance, or promoting positive behavior. Aversive models without one of these action components were significantly less effective than other discipline responses. The remaining effective disciplinary response was a combination of reasoning and forced compliance. Thus five of six possible combinations of a verbal disciplinary tactic with an action component were the five most effective disciplinary responses at delaying the next misbehavior recurrence.

The most promising of these is the combination of reasoning and forced compliance. Like the reasoning-punishment combination, it tended to delay the next misbehavior recurrence more than did either forced compliance or reasoning alone. However, the combination of forced compliance and reasoning did not enhance the subsequent effectiveness of reasoning as consistently as the reasoning-punishment combination did. One possible reason is that punishment communicates to the child that he or she should have made a better behavior choice. The reasoning component clarifies what that choice should have been and why. In contrast, forced compliance communicates that the parent will rescue the child from difficulty whenever he or she gets entangled in a problem, thus minimizing the need for the child to make a better choice in the future. Nonetheless, the combination of reasoning with forced compliance seems to be an effective disciplinary response even though it is limited in enhancing the subsequent effectiveness of reasoning.

The combinations of aversive verbal tactics with actions seem problematic despite their effectiveness. Taken to excess, aversive verbal tactics should not be recommended. Some findings that affective intensity enhances the effectiveness of reasoning might help account for this somewhat (Chamberlain, Reid, Ray, Capaldi & Fisher, 1997; Pfiffner & O'Leary, 1989). Zahn-Waxler, Radke-Yarrow, and King (1979) explored the affective intensity of maternal explanations to 15- to 29-month-old preschoolers in some detail. They found that verbal explanations were used in 40% of disciplinary incidents. In most of these cases, the mothers communicated intensely with moralizing and judgmental components. Affectively and morally charged explanations were positively associated with child reparations and altruism. Neutrally communicated explanations were the only form of explanations that failed to correlate positively with child reparations. Similarly, one of our earlier studies found that, when reasoning alone resulted in no child distress, it was a less effective disciplinary response than when it resulted in some child distress for about a minute or two (Larzelere & Merenda, 1994).

Perhaps the affective intensity is the aspect of aversive verbal tactics that make them effective when combined with an action component. Further research is needed to clarify when such verbal intensity becomes counterproductive.

Sather (1992) examined whether the effectiveness of other verbal disciplinary tactics was enhanced by being backed up with punishment. He found some evidence that each verbal tactic was subsequently more effective after being combined with punishment. The verbal tactics included scold/shame or yell, threaten, label behavior bad, and commanding. The effectiveness of a punishment back-up for enhancing subsequent use of the verbal tactic was found for both disobedience and fighting incidents for labeling the behavior bad, but only for one type of incident for the other four verbal tactics.

I have recently investigated whether other forceful backup procedures would also enhance the subsequent effectiveness of reasoning. One obvious alternative is forced compliance. As summarized above, it rarely enhanced the subsequent effectiveness of reasoning by itself, perhaps because of different motivational implications than is the case for punishment.

Another conclusion from my studies concerns the level of child distress associated with maximal effectiveness. Research on punishment has found that the stronger the punishment, the more effective it is. That suggests that effectiveness will be maximized when the child is most distressed. That is what we found when punishment was used alone (Larzelere & Merenda, 1994). However, when reasoning was used, whether by itself or in combination with punishment, an intermediate level of child distress was associated with optimal effectiveness. This is consistent with previous research that found that adding reasoning to punishment changed two important parameters of punishment's effectiveness. Once reasoning was added to punishment, its effectiveness no longer depended on its severity. Second, punishment's effectiveness no longer depended upon its precise timing once reasoning was added to it (Cheyne &Walters, 1969; Parke, 1969). As will be seen, this pattern is also consistent with Hoffman's (1977) theory of moral internalization.