Optimism as a Mediator of the Relation Between Perceived Parental Authoritativeness and Adjustment Among Adolescents: Finding the Sunny Side of the Street
Abstract
Authoritative parenting has been associated with positive outcomes for children and adolescents, but less is known about the mechanisms responsible for such effects. Two longitudinal studies examined the hypothesis that the relation between authoritative parenting and adolescents’ adjustment is mediated by adolescents’ level of dispositional optimism. In Study 1, university students’ perceptions that their parents were authoritative predicted higher self-esteem, lower depression, and better university adjustment during the students’ transition into, and throughout, university. Importantly, these relations were mediated by students’ levels of optimism. In Study 2, high school students’ perceptions that their parents were authoritative predicted higher self-esteem and lower depression six years later when they were young adults, and these relations again were mediated by students’ level of dispositional optimism.
As the old song would have it, there are advantages to learning to walk on the sunny side of life's pathways. One of the more pervasive findings in the recent research literature on ‘positive psychology’ in adulthood (e.g., Seligman & Csikszentmihalyi, 2000) has been evidence that personal optimism is linked to better health (e.g., Danner, Snowden, & Friesen, 2001) and more effective adaptation and adjustment (e.g., Andersson, 1996). Less is known about the potential developmental bases of such personal optimism, however. What patterns of family background or socialization might be linked to a disposition to be optimistic in viewing the world, to ‘walk on the sunny side’? Research in developmental psychology suggests that certain patterns of parenting are linked to better subsequent adjustment (Steinberg, 2001), though the basis for these relations is not fully established. One novel integrative account of these observations across development might be that the experience of effective parenting in one's childhood fosters the growth of an optimistic view of the world in the child and then later in the adult that he or she becomes. The present paper presents evidence for this view, that more adaptive, ‘authoritative’ parenting is a basis for positive adjustment in adolescence and adulthood through the mediator of dispositional optimism. Below, we review the research literature on the various elements of this account.
Parenting Style and Adjustment
Parenting that combines emotional warmth with reasonable firmness in the control of children's behavior is often argued to be optimal, in that it facilitates the development of healthy adjustment in children (e.g., Baumrind, 1966, 1967, 1991). Maccoby and Martin (1983) proposed that parental warmth and control can be represented as two dimensions of parenting, producing a fourfold classification of parenting styles. According to their model, authoritative (reciprocal) parents, by definition, are high on both warmth toward, and control of, children; authoritarian (autocratic) parents exercise control with little warmth; indulgent (permissive) parents are warm but tend to de-emphasize control; and indifferent (uninvolved) parents reveal little of either warmth or control. Subsequent work has indicated that parental control and warmth are not fully independent, as this scheme implies, because the parents’ degree of warmth influences how control is manifested and how it is construed by children (Darling & Steinberg, 1993). Moreover, effective parents of adolescents may be particularly adept at balancing control with autonomy granting (Steinberg, 2001). Nevertheless, the combination of warmth and control is thought to foster healthy development in children (Lamborn, Mounts, Steinberg & Dornbusch, 1991).
Indeed, children with authoritative parents tend to have better psychosocial competence and academic achievement, as well as fewer emotional and behavioral problems, than do children whose parents are not authoritative (for reviews, see Darling & Steinberg, 1993; Parker & Gladstone, 1996). These effects are often driven by the relatively poor adjustment that seems to exist among children with non-authoritative parents (e.g., Glasgow, Dornbusch, Troyer, Steinberg & Ritter, 1997; Weiss & Schwarz, 1996). The relations between parents’ authoritativeness and the adjustment of their children exist not only among young children, but persist into the offsprings’ adolescence (e.g., Baumrind 1991; Brown, Mounts, Lamborn & Steinberg, 1993; Hein & Lewko, 1994; Lamborn et al., 1991; Slicker, 1998; Steinberg, Lamborn, Darling, Mounts & Dornbusch, 1994; Weiss & Schwarz, 1996). Although the research is quite clear that authoritative parenting does predict positive adjustment generally among children and adolescents, the strength of this relation varies cross-culturally and across subcultures in North America (Darling & Steinberg, 1993).
Of concern to the present research is the question of how or why authoritative parenting produces positive outcomes for children and adolescents. According to Steinberg (2001), authoritative parenting is effective for three primary reasons. He suggests that children of authoritative parents develop competence in social and cognitive realms, healthy self-regulation, and a receptivity to parental guidance. Relevant research has examined relations between authoritative parenting and skills and habits in adolescents that promote both their academic achievement and personal well-being.
With regard to relations between authoritative parenting and particular skills that may foster competent academic achievement, Steinberg, Elmen and Mounts (1989) found that parental characteristics associated with authoritativeness (acceptance, psychological autonomy, and behavioral control) predicted early adolescents’ grades via their impact on adolescents’ work orientation (skills, aspirations, and enjoyment of work). Adolescents with authoritative parents have also been found, relative to others, to anticipate success, focus on relevant tasks, and make self-enhancing attributions (Aunola, Stattin & Nurmi, 2000; see also Glasgow et al., 1997). Parental involvement in adolescents’ education has also been implicated as a mediator of the relation between authoritative parenting and school success among adolescents (Steinberg, Lamborn, Dornbusch & Darling, 1992). For example, Pratt, Green, MacVicar and Bountrogianni (1992) showed that authoritative parents were more effective tutors in helping with long-division homework, showing more successful patterns of scaffolding children's skill acquisition in an observational study.
With respect to personal-social adjustment, there is some indication that particular attitudinal and behavioral patterns that develop in children may help to explain relations between parental authoritativeness and adjustment of their offspring. Mackey, Arnold and Pratt (2001) reported that parental authoritativeness corresponded to adolescents’ appropriation of parental advice, suggesting an openness to parental guidance in their personal development. In turn, adolescents who are more responsive to parent views in stories of their own lives are more likely to be well adjusted several years later as young adults (Arnold, Pratt & Hicks, 2004). Perhaps correspondingly, adolescents with more authoritative parents have reported using more effective techniques for coping with stressors than have adolescents with less authoritative parents (McIntyre & Dusek, 1995; see also Dusek & Danko, 1994). In addition, because adolescents of authoritative parents display relatively positive behaviors, they tend to attract, and be attracted to, networks of other well-adjusted individuals who may reinforce positive behaviors (e.g., Brown et al., 1993; Fletcher, Darling, Steinberg & Dornbusch, 1995).
Parenting Style, Optimism, and Adjustment
As a complement to work identifying specific skills and behavior patterns that mediate relations between authoritativeness and particular forms of adjustment, and consistent with Steinberg's (2001) suggestion that authoritative parenting is effective, in part, because it fosters healthy self-regulation, in the present research we explored the possibility that authoritative parenting may give rise to a relatively enduring, broad-based tendency in children—that of dispositional optimism. Dispositional optimism is a form of generalized outcome expectancy (see Rotter, 1954) involving a chronic tendency to focus on, expect, and hence strive toward positive outcomes and experiences (Scheier & Carver, 1985). As such, it is theoretically and empirically linked with healthy self-regulation (Carver & Scheier, 1982, 2001). Consistent with the colloquial use of the term, Scheier and Carver (1985) suggested that ‘optimists expect things to go their way, and generally believe that good rather than bad things will happen to them’ (p. 219). If authoritative parenting gives rise to optimism in children, and indeed continues to reinforce it in adolescents, it may foster both strong academic and healthy personal adjustment.
Several reasons exist to expect that authoritative parenting promotes optimism. First, optimism may initially be an outgrowth of the early development of trust. According to Erikson (1950, 1982), trust forms when caregivers consistently respond to infants’ needs. Authoritative parenting is defined, in part, by relatively high levels of responsiveness. The resultant trust in children, in essence, involves optimism that one's needs will be met – early in life by parents or caregivers, and perhaps later in life by other people and situations. Certainly, such an analysis is compatible with the attachment literature (e.g., Ainsworth, Blehar, Waters & Wall, 1978), in which caregiver responsiveness is theorized to give rise to more secure forms of attachment in the child that later predict an array of positive socioemotional variables.
Second, children raised in a context of interpersonal optimism may learn to be similarly optimistic through modeling and reinforcement. This is likely in families with authoritative parents because such parents are inclined to be optimistic. Parental authoritativeness has been found to be correlated with generativity (Peterson, Smirles & Wentworth, 1997), a construct that includes the optimistic belief in the goodness of the human species (McAdams & de St. Aubin, 1992; see also Erikson, 1950, 1982). This relation is particularly notable among mothers (Pratt, Danso, Arnold, Norris & Filyer, 2001). Suggestive of the possibility that authoritative parents model optimism, in a narrative analysis, Pratt, Norris, van de Hoef and Arnold (2001) found that more generative and more authoritative parents spoke in especially positive terms about their role as parents of adolescents, focusing more on the positive potential of their children in comparison with less authoritative parents. Thus, modeling by parents, and perhaps also reinforcement, may further support optimism during later childhood and adolescence.
Regardless of its developmental origins, it is clear that optimism predicts positive adjustment. This occurs both directly and through behavioral self-regulation. For example, some evidence indicates that optimistic individuals reveal an implicit attentional bias in favor of positively valenced stimuli (Segerstrom, 2001). Hence optimism may lead directly to healthy adjustment through positive construals of life events. Optimism also predicts positive adjustment because of its relation with coping skills. For example, optimistic individuals tend to persist at tasks until goals are met because they expect success, and they also tend to use active, problem-focused coping mechanisms when faced with stressors (see Armor & Taylor, 1998, for a related discussion). A considerable literature supports these relations (see Andersson, 1996; Scheier & Carver, 1985, 1992; Scheier, Carver & Bridges, 2001). Importantly for the present research, evidence exists which suggests that optimism predicts effective coping specifically among older adolescents. Personal adjustment and adaptation to university have been demonstrated to be predicted by both dispositional optimism (Aspinwall & Taylor, 1992; Brissette, Scheier & Carver, 2002; Chemers, Hu & Garcia, 2001; Davis, Hanson, Edson & Ziegler, 1992; Segerstrom, Taylor, Kemeny & Fahey, 1998) and by optimism in specific expectations about university (Jackson, Pancer, Pratt & Hunsberger, 2000).
We reasoned that parental authoritativeness may maintain its positive influence on adolescents, even as they move toward independence and young adulthood, in part because it gives rise to and supports an optimistic approach to life that maintains and furthers positive adjustment. In the present research, we were able to test this possibility in two longitudinal datasets. The first (Study 1) included adolescents who were involved in the transition into and throughout university, and the second (Study 2) involved adolescents undergoing the transition from high school to the next phase of their lives, such as university or work, during the period of ‘emerging adulthood’ (Arnett, 2000).
Study 1
The transition to university is a significant life adjustment, and it is experienced as a challenge by many students (Cantor, Norem, Brower, Niedenthal & Langston, 1987; Cutrona, 1982; Fisher & Hood, 1988; Gerdes & Mallinckrodt, 1994; Lamothe, Currie, Alisat, Sullivan, Pratt, Pancer & Hunsberger, 1995; Levitz & Noel, 1989; Rickinson & Rutherford, 1995, 1996; Zirkel & Cantor, 1990). During life transitions, individual differences in coping and adjustment may be particularly salient (e.g., Cowan, 1991). Variations in students’ family experiences, particularly the authoritativeness of their parents, and corresponding levels of optimism in adolescents, may predict this differential adjustment. This may be true for both personal and academic forms of adjustment, because in both domains, optimism fosters reasonable persistence and positive construals of events. We were able to assess this possibility by exploring the relations among perceived parenting style, optimism, self-esteem, depression, and university adjustment among university students from prior to their first year of university, until their fourth and typically final year of study. We hypothesized that the relations between perceived parental authoritativeness and personal and university adjustment would be mediated by optimism.
Method
Participants
All students who planned to commence studies at a medium-sized university in Ontario, Canada, in the fall of 1993 immediately following their graduation from high school were contacted during the summer of 1993 (N=1,147). Of those contacted, 356 individuals (234 women, 117 men, 5 gender unspecified; M age 19) agreed to participate (31%). Approximately 30% of the original sample remained in the study until its completion four years later (N=107; 77 women, 30 men). Comparison of participants who remained in the study versus those who dropped out before the final measurement period indicated no difference as a function of initial age (both Ms=19, t < 1), or the proportion of men and women, chi-square (1, N=351) < 1.
Of the original 356 participants, the majority was white, and 86% were English speaking only. Nevertheless, a total of 16 other languages were spoken by small numbers of students, representing backgrounds from various European countries, the Mid- and Far East, and Africa. Eighty-two percent of the students planned to live in residence at the university, 11% planned to remain at home, 5% intended to live off-campus, and the remaining 2% either had other plans or did not indicate where they intended to live. The majority planned to study arts (41%), followed by business and economics (38%), science (9%), music (5%), and physical education (3%), and some were undecided (4%). Participants had relatively high academic standing during their final year of high school: The average reported final high-school year grade was an A minus.
Procedure and Measures
Individuals who agreed to participate in this research were sent a questionnaire through the mail, just prior to the beginning of their studies (in August, 1993), and during their first, second, and fourth years of university. Prior to starting university, measures of personal adjustment (self-esteem, depression) and optimism were taken. During year 1, the adjustment measures were repeated, along with a measure of university adjustment. In year 2 we measured perceived parental authoritativeness and personal and academic adjustment, and in year 4 we measured optimism, personal adjustment, and academic adjustment. At various points during students’ time at university, they also completed various measures not relevant to the present research (e.g., scales of religiosity and various social attitudes, see Hunsberger, Pancer, Pratt & Alisat, 1996). Specific constructs assessed at the various points in time, and sample sizes for each measurement period, are summarized in Table 1. Each questionnaire required approximately 45 to 60 minutes to complete.1
| Aug. 1993 (Pre-university) | Feb. 1994 (Year 1) | March 1995 (Year 2) | Nov. 1996 (Year 4) | March 1997 (Year 4) | |
|---|---|---|---|---|---|
| Authoritativeness | X | ||||
| Optimism | X | X | |||
| Self-esteem | X | X | X | X | |
| Depression | X | X | X | ||
| SACQ | X | X | X | ||
| N | 356 | 261 | 218 | 137 | 107 |
- Note: SACQ refers to the university adjustment scale.
Parenting Style. Students’ perceptions of their parents’ authoritativeness were assessed using an adapted version of Lamborn et al.'s (1991) measure of parental responsiveness and demandingness. The items were phrased in the past tense, because many of our participants no longer lived with their parents and thus probably had less frequent contact with them. Ten items measured parental responsiveness, or warmth (e.g., ‘My parents spent time just talking with me’), and six items assessed parental demandingness, or strictness (e.g., ‘My parents tried to know where I went at night’). The scale items were rated using a 9-point response format, ranging from −4 (very strongly disagree) to +4 (very strongly agree), with higher scores indicating more warmth and strictness.
Ratings on the warmth and strictness items were converted to a 1 to 9 format, and scale totals for warmth and strictness were computed for each participant. Participants reported that their parents were moderately warm. Given a possible score range of 10 to 90, the average reported parental warmth was 69.36, SD=14.44. Participants also reported that their parents were moderately strict. Given a possible score range of 6 to 54, the average reported strictness was 42.55, SD=8.48. Warmth and strictness were substantially correlated with each other, r(218)=.43, p < .001, somewhat stronger than the correlation reported by Lamborn et al. (1991) with younger adolescents.
Because there were different numbers of items measuring warmth and strictness, we standardized warmth and strictness scores and then computed an index of authoritativeness. Our continuous measure of perceived parental authoritativeness was the standardized sum of warmth (standardized) and strictness (standardized).2 Coefficient alpha for the authoritativeness measure in this sample was .88.
Although both parents’ and adolescents’ reports of parenting style are interesting, as Steinberg et al. (1992) noted, at least two reasons exist to focus on adolescents’ reports. First, parents may tend, due to social desirability concerns, to present themselves as effective, leading to less accurate assessments. Second, adolescents’ construals of their parents may be as important as parents’ actual behaviors. Indeed, although adolescents’ perceptions of parental authoritativeness may differ somewhat from their parents’ perceptions (e.g., Smetana, 1995), Paulson (1994) found that adolescents’ reports of parental authoritativeness better predicted adolescent outcomes than did parents’ reports of their own parenting style. Moreover, in comparing adolescents’ and mothers’ ratings of maternal parenting behaviors to those of independent raters, Gonzales, Cauce and Mason (1996) found adolescent reports to be the more valid indicators.
Optimism. Students’ dispositional optimism was assessed using Scheier and Carver's (1985) Life Orientation Test (LOT). This scale assesses the extent to which people expect positive outcomes in life (e.g., ‘In uncertain times, I usually expect the best,’‘Things never work out the way I want them to,’ reverse scored). The scale's 8 items were rated with a 9-point response format ranging from −4 (very strongly disagree) to +4 (very strongly agree), with higher scores indicating more optimism. Coefficient alphas in the present sample at the two measurements were .84 and .88.
Self-esteem. Self-esteem was measured with Rosenberg's (1965) 10-item Self-esteem scale. Items (e.g., ‘I feel that I am a person of worth, at least on an equal basis with others,’‘I certainly feel useless at times,’ reverse scored) were rated with a 9-point response format ranging from −4 (very strongly disagree) to +4 (very strongly agree), with higher scores indicating higher self-esteem. Coefficient alphas ranged across time from .83 to .90.
Depression. Depression was assessed using the 20-item Center for Epidemiological Studies Depression scale (Radloff, 1977). This scale assesses the frequency with which people experience depressive symptomatology, such as loss of sleep and altered eating patterns. The frequency of depressive symptoms during the last week was rated with a 4-point response format ranging from 1 [rarely or none of the time (less than one day)] to 4 [most or all of the time (five to seven days)]. Thus, higher scores indicate more depressive symptoms. Coefficient alphas ranged across time from .89 to .91.
University Adjustment. To measure university adjustment, Baker and Siryk's (1986) Student Adaptation to College Questionnaire (SACQ) measure was used. The 67 items in this scale were rated with a 9-point response format, ranging from 1 (doesn’t apply to me at all) to 9 (applies very closely to me), with higher scores indicating better adjustment. Coefficient alphas in the present sample ranged across time from .94 to .95. The SACQ contains subscales that assess academic, social, and personal-emotional adjustment, as well as attachment to the university. Analyses involving the subscales and the full scale provided parallel findings. Hence, results involving the full scale only are reported.
Results
Preliminary Analyses
First, to explore any potential implications of participant attrition in our data, we compared baseline levels of adjustment and optimism between students who did and did not remain in the study at least until the second year measurement period. No differences emerged between these groups on self-esteem, depression, or optimism, all Fs < 1. Similarly, students who remained in the study until its completion after four years did not differ on pre-university levels of depression, self-esteem, optimism, or on perceived parental authoritativeness (measured in year 2), from those who dropped out prior to the fourth year measurement period, all Fs < 1.3. Thus, attrition was not selective with regard to pre-university levels of personal adjustment, optimism, or perceived parenting style.
Unsurprisingly, however, attrition was related to university adjustment. For example, students who dropped out before the second year measurement period reported somewhat poorer university adjustment in Year 1 (M=401.00, SD=62.05) than did those who did not drop out (M=421.21, SD=58.32), F (1,271) =6.34, p < .05. Similarly, students who dropped out between the second and fourth year of the study reported poorer university adjustment in Year 2 (M =406.89, SD=57.73) than did those who remained in the study until its fourth year (M=429.29, SD=62.75), F(1,214)=7.40, p < .01. Thus, the only implication of attrition for the present study seemed to be a slightly attenuated range of overall university adjustment scores.
Second, before testing the predicted mediational models, we assessed the possibility that participant gender and socioeconomic status related to adjustment. To determine whether participant gender predicted any of the adjustment measures, a participant gender by time of measurement repeated measures ANOVA was computed on self-esteem, depression, university adjustment, and optimism. Relevant means for each analysis are provided in Table 2. In the analysis on depression, neither the main effect of participant gender, nor the main effect of time, was significant. The participant gender by time interaction was marginally significant, F(2,226) =2.45, p=.09. The pattern of means indicated a weak tendency for women's depression levels to remain stable, and men's to increase slightly over time. In the analysis of self-esteem, the main effect of participant gender was not significant. However, both the main effect of time, F(3,309)=3.13, p < .05, and the participant gender by time interaction, F(3,309)=4.17, p < .01, were significant. Post-hoc comparisons indicated that at no point in time did women's and men's levels of self-esteem differ significantly, however (using either the conservative Scheffe test, or the more liberal Tukey test). Nevertheless, the pattern of means indicated a weak tendency for women's self-esteem, but not men’s, to gradually increase over time. In the analysis of university adjustment, only the effect of time was significant, F(2,176)=11.53, p < .001, indicating increases in university adjustment over time. Neither the main effect of gender, nor the time by gender interaction, was significant. No significant effects emerged in the analysis on optimism, all Fs < 1. Because there were slight differences in patterns of depression and self-esteem over time for men and women, participant gender was controlled in subsequent tests of hypotheses (i.e., the mediational analyses).
| Baseline N=356 | Year 1 N=261 | Year 2 N=218 | Year 4 Fall N=137 | Year 4 Spring N=107 | |
|---|---|---|---|---|---|
| Women | |||||
| Optimism | 50.89 (11.07) |
51.93 (10.72) |
|||
| Self-esteem | 69.43 (12.84) |
70.22 (14.41) |
70.13 (14.68) |
73.24 (13.17) |
|
| Depression | 34.12 (9.90) |
35.96 (10.74) |
34.06 (9.53) |
||
| SACQ | 413.18 (59.37) |
416.85 (64.05) |
449.91 (57.51) |
||
| Men | |||||
| Optimism | 49.91 (9.87) |
50.31 (12.34) |
|||
| Self-esteem | 71.84 (11.48) |
71.40 (13.89) |
73.72 (11.82) |
71.45 (14.68) |
|
| Depression | 32.01 (7.94) |
35.13 (10.37) |
35.30 (11.53) |
||
| SACQ | 420.72 (61.21) |
422.85 (55.02) |
444.69 (65.59) |
||
- Note: Standard deviations are provided in parentheses. Baseline=August 1993; Year 1=February, 1994; Year 2=March, 1995; Year 4 fall=November, 1996; Year 4 spring=March, 1997. SACQ=University adjustment. Possible scale ranges are: optimism 9–72; Self-esteem 10–90; Depression 20–80; SACQ 67–603.
We also examined whether parent education needed to be controlled in our tests of hypotheses by computing the Pearson correlations between mothers’ and fathers’ education and students’ self-esteem, depression, and university adjustment. However, all of the correlations were small and non-significant, all ps > .05. Because parental education was unrelated to students’ adjustment, this variable was not analyzed further in this study.
Finally, we computed intercorrelations among key variables within the baseline and final (fourth year) samples. As shown in Table 3, the variables were related in the theoretically consistent manner. We also examined the stability of individual differences by computing the correlation between the baseline and final measure for each variable. These correlations, also shown in Table 3, reveal a moderate to high degree of stability across time.
| Optimism | Self-esteem | Depression | |
|---|---|---|---|
| Optimism | .62 | .65 | −.51 |
| Self-esteem | .60 | .64 | −.58 |
| Depression | −.60 | −.69 | .43 |
| SACQ | .45 | .63 | −.53 |
- Note: Correlations among baseline measures are provided above the diagonal and correlations among measures in the final (fourth year) sample are provided below the diagonal. Correlations between the first and last measure for each variable (stability correlations) are provided on the diagonal. SACQ = Student Adaptation to College Questionnaire (University Adjustment). The SACQ was not measured at baseline and thus appears below the diagonal only. All correlations are significant at p < .001.
Tests of Hypotheses
We tested the mediational models in which perceived parental authoritativeness (assessed in year 2) and subsequent adjustment (assessed in year 4) would be mediated by optimism (as assessed in year 4). According to Baron and Kenny (1986), mediation is indicated when: (1) a predictor variable correlates with both the mediator and the criterion; (2) the mediator correlates with the criterion; and (3) when regressing the criterion on the predictor and the mediator together, the relation between the predictor variable and the criterion is considerably reduced, but the relation between the mediator and the criterion is not. If the relation between the predictor and criterion is reduced to zero, this indicates complete mediation by the variable controlled. Partial mediation—the situation in which a measured mediator is one of several intervening variables—is indicated when the relation between the predictor and criterion is reduced, but not to zero. This is reflected in a significant reduction in the path from the predictor to the criterion, controlling for the mediator.
Following the procedure outlined by Baron and Kenny (1986), we first confirmed that the predictor variable (authoritativeness, year 2) correlated with the mediator (optimism, year 4), r(114)=.28, p < .01, and all outcome variables [self-esteem, year 4, r(114)=.21, p < .05; depression, year 4, r(114)=−.28, p < .01; university adjustment, year 4, r(93)=.32, p < .01]. Second, we confirmed that the mediator (optimism, year 4) correlated with each outcome variable [self-esteem, year 4, r(136) =.60, p < .001; depression, year 4, r(136)=−.60, p < .001; university adjustment, year 4, r(104)=.45, p < .001]. Third, to determine whether optimism accounted for the relations between authoritativeness and adjustment, we performed hierarchical regression analyses in which we predicted the fourth year adjustment variables by entering gender in the first step (to control for its relation with adjustment), parental authoritativeness (year 2) in the second step, and optimism (year 4) in the third step.3 Regression coefficients are provided in Table 4. Of interest was whether relations between authoritativeness and adjustment revealed in step 2 would be significantly reduced by entering optimism into the equation in step 3.
| DV | IV | Before Mediator Added | After Mediator Added | ||||||
|---|---|---|---|---|---|---|---|---|---|
| B | beta | t | R2 | B | beta | t | R2 | ||
| Self-esteem | Gender | 2.25 | .08 | .89 | .05 | 1.70 | .06 | .81 | .36*** |
| PS | 2.82 | .22 | 2.67* | .76 | .06 | .74 | |||
| OP | .63 | .58 | 7.29*** | ||||||
| CESD | Gender | −1.94 | −.10 | −1.05 | .09** | −1.55 | −.08 | −1.00 | .37*** |
| PS | −2.80 | −.30 | −3.24** | −1.35 | −.14 | −1.78 | |||
| OP | −.44 | −.55 | −6.99*** | ||||||
| SACQ | gender | 22.10 | .18 | 1.73 | .14** | 20.85 | .17 | 1.73 | .25*** |
| PS | 22.69 | .40 | 3.86*** | 16.56 | .29 | 2.85** | |||
| OP | 1.69 | .34 | 3.48** | ||||||
- Note: DV=criterion, IV =predictor, CESD=depression, SACQ=university adjustment, PS=parenting style (authoritativeness), OP=optimism.
- * p < .05;
- ** p < .01;
- *** p < .001.
Self-esteem. As shown in Table 4, controlling for gender, the relation between authoritativeness and self-esteem was significant, pr=.22, t(112)=2.37, p < .05. In support of mediation, when controlling for optimism in the third step of the analysis, the relation between authoritativeness and self-esteem was no longer significant, pr=.07, t(111)=.74, ns, whereas the relation between optimism and self-esteem remained significant, pr =.64, t(111)=7.29, p < .001. A test of the indirect effect of authoritativeness on self-esteem via optimism (see Sobel, 1982) was significant, Z=2.81, p < .01. The final equation accounted for 44% of the variance in students’ self-esteem, and was highly significant, F(3,89)=23.32, p < .0001. In addition, controlling for pre-university levels of self-esteem did not alter these findings, suggesting that optimism predicted changes in self-esteem: When the regression was recomputed, entering pre-university levels of self-esteem in the first step, the relation between year 4 optimism and year 4 self-esteem remained significant in the final equation, pr=.43, p < .0001.
Depression. As shown in Table 4, controlling for gender, the relation between authoritativeness and depression was significant, pr=−.29, t(112)=3.24, p < .01. In support of mediation, when controlling for optimism in the third step of the analysis, the relation between authoritativeness and depression was no longer significant, pr=−.17, t(111)= 1.78, ns, whereas the relation between optimism and depression remained significant, pr=−.62, t(111)=6.99, p < .001. A Sobel test of the indirect effect of authoritativeness on depression via optimism was significant, Z=2.77, p < .01. This final equation accounted for 45% of the variance in students’ depression, and was highly significant, F(3,89) =24.27, p < .0001. In addition, the relation between year 4 optimism and year 4 depression appeared to be dynamic: When the regression was recomputed controlling for pre-university levels of depression in the first step, the relation between year 4 optimism and year 4 depression remained significant in the final step, pr =−.52, p < .0001.
University Adjustment. As shown in Table 4, controlling for gender, the relation between authoritativeness and university adjustment was significant, pr =.38, t(90)=2.85, p < .01. When controlling for optimism in the third step of the analysis, the relation between authoritativeness and university adjustment was reduced, but remained significant, pr=.29, t(89)= 2.85, p < .01, and the relation between optimism and university adjustment also remained significant, pr=.35, t(89)=3.48, p < .01. A Sobel test of the indirect effect of authoritativeness on depression via optimism was significant, Z=2.22, p < .05. This pattern indicates that optimism partially mediated the relation between authoritativeness and university adjustment. The final equation accounted for 15% of the variance in students’ university adjustment, F(2,90)=7.98, p < .001. When the regression was recomputed controlling for baseline levels of university adjustment in the first step, the relation between optimism and university adjustment in the final step was no longer significant, pr=.17, ns. Thus, optimism did not predict changes in university adjustment over time.
Mediational analyses supported our predictions: Optimism fully mediated relations between authoritative parenting and self-esteem and depression, and it partially mediated the relation between authoritative parenting and university adjustment. Because the data are correlational, alternative models are of course possible. Thus, we tested two additional possibilities with regard to temporal relations between authoritativeness, optimism, and adjustment.
Alternative Model 1
According to alternative model 1, more optimistic individuals construe their parents as more authoritative, and this perception gives rise to healthy adjustment. To assess this model, we examined the possibility that perceptions of parenting style mediated relations between optimism and adjustment. First, the predictor in this alternative model (optimism, pre-university) correlated with both the mediator [perceived authoritativeness, r(216)=.30, p < .001], and with each fourth year outcome variable [self-esteem, r(131)=.48, p < .001; depression, r(131) =−.37, p < .001; university adjustment, r(106)=.33, p < .01]. Second, as reported in the tests of the hypothesized model, perceived parenting style (here, the mediator) correlated with each of the fourth year outcome variables. Third, to determine whether authoritativeness accounted for the relations between optimism and adjustment, we performed hierarchical regression analyses in which we predicted the fourth year adjustment variables by entering gender in the first step, optimism (pre-university) in the second step, and authoritativeness (year 2) in the third step. Regression coefficients are provided in Table 5. Of interest was whether relations between optimism and adjustment revealed in step 2 would be reduced to non-significance by entering authoritativeness into the equation in step 3. In no case did this occur.
| DV | IV | Before Mediator Added | After Mediator Added | ||||||
|---|---|---|---|---|---|---|---|---|---|
| B | beta | t | R2 | B | beta | t | R2 | ||
| Self-esteem | Gender | 1.64 | .06 | .75 | .26*** | 2.19 | .08 | .98 | .26*** |
| OP | .59 | .50 | 6.17*** | .57 | .48 | 5.68*** | |||
| PS | 1.23 | .10 | 1.13 | ||||||
| CESD | Gender | −1.04 | −.05 | −.60 | .16*** | −1.90 | −.10 | −1.10 | .20*** |
| OP | −.35 | −.40 | −4.56*** | −.30 | −.34 | −3.90*** | |||
| PS | −1.96 | −.21 | −2.32* | ||||||
| SACQ | Gender | 8.50 | .07 | .70 | .13** | 20.65 | .17 | 1.68 | .21*** |
| OP | 1.91 | .36 | 3.63*** | 1.63 | .31 | 3.19** | |||
| PS | 17.54 | .31 | 3.08** | ||||||
- Note: DV=criterion, IV =predictor, CESD=depression, SACQ=university adjustment, OP=optimism, PS= parenting style (authoritativeness).
- * p < .05;
- ** p < .01;
- *** p < .001.
Self-esteem. Controlling for gender, optimism assessed prior to university did predict year 4 self-esteem, pr=.50, t(112)=6.17, p < .001. Adding year 2 authoritativeness to the regression reduced the correlation between pre-university optimism and year 4 self-esteem trivially, pr=.48, t(111)=5.68, p < .001, but reduced the relation between authoritativeness and self-esteem to non-significance, pr=.14, t(111)=1.13, ns. Moreover, the Sobel test for the indirect effect of optimism on self-esteem via perceived authoritativeness was not significant, Z=1.07, ns. This pattern failed to support the mediation indicated by alternative model 1.
Depression. Controlling for gender, optimism assessed prior to university did predict year 4 depression, pr=−.40, t(112)=4.56 p < .001. Adding year 2 authoritativeness to the regression reduced the correlation between pre-university optimism and year 4 depression slightly, pr=−.35, t(111)= 3.90, p < .001. The relation between authoritativeness and depression remained significant, pr=−.27, t(111)=2.32, p < .05. The Sobel test for the indirect effect of optimism on depression via perceived authoritativeness was not significant, Z=1.62, ns. Thus, mediation as described by alternate model 1 was not indicated.
University Adjustment. Controlling for gender, optimism assessed prior to university did predict year 4 university adjustment, pr=.36, t(91)=3.63 p < .001. Adding year 2 authoritativeness to the regression reduced the correlation between pre-university optimism and year 4 university adjustment slightly, pr=.32, t(90)=3.19, p < .01. The relation between authoritativeness and university adjustment remained significant, pr=.34, t(90)=3.08, p < .01. The Sobel test for the indirect effect of optimism on university adjustment via perceived authoritativeness was not significant, Z=1.80, ns, indicating no mediation.
Alternative Model 2
A second alternative model involved the possibility that optimism is an outgrowth of the good adjustment fostered by authoritative parenting: That is, that the adjustment variables mediated relations between authoritativeness and optimism. In testing this model for self-esteem and depression as mediators, we used the fourth year measures. Because the final university adjustment questionnaire was completed subsequent to the fourth year optimism measure, in the regression with university adjustment as the mediator we use the second year university adjustment variable. First, the predictor, authoritativeness, correlated with each potential mediator [fourth year self-esteem, r(114)=.21, p < .05 and depression, r(114)=−.28, p < .01, second year university adjustment, r(215)=.20, p < .01] as well as with the outcome variable, fourth year optimism, r(114) =.28, p < .01. Second, the mediators (the adjustment variables) each correlated with the outcome measure, fourth year optimism [self-esteem, r(136)=.60, p < .001; depression r(136)=−.60, p < .001; university adjustment r(113)=.49, p < .001]. Third, to determine whether the adjustment variables accounted for the relations between authoritativeness and optimism, we performed hierarchical regression analyses in which we predicted fourth year optimism by entering gender in the first step, authoritativeness (year 2) in the second step, and the adjustment variables in the third step. Regression coefficients are provided in Table 6. Of interest was whether relations between authoritativeness and optimism revealed in step 2 would be reduced to non-significance by entering the adjustment variables into the equation in step 3.
| DV | IV | Before Mediator Added | After Mediator Added | ||||||
|---|---|---|---|---|---|---|---|---|---|
| B | beta | t | R2 | B | beta | t | R2 | ||
| OP | Gender | 1.00 | .04 | .43 | .08** | −.22 | −.01 | −.12 | .37*** |
| PS | 3.32 | .29 | 3.07** | 1.85 | .16 | 2.02* | |||
| SE | .51 | .56 | 7.21*** | ||||||
| OP | Gender | 1.00 | .04 | .43 | .08** | −.35 | −.01 | −.18 | .36*** |
| PS | 3.32 | .29 | 3.07** | 1.39 | .12 | 1.46 | |||
| CESD | −.69 | −.56 | −6.96 | ||||||
| OP | Gender | 1.00 | .04 | .43 | .08** | .44 | .02 | .21 | .27** |
| PS | 3.32 | .29 | 3.07** | 2.25 | .19 | 2.28 | |||
| SACQ | .08 | .45 | 5.41 | ||||||
- Note: DV=criterion, IV =predictor, OP=optimism, PS=parenting style (authoritativeness), SE=self-esteem, CESD=depression, SACQ=university adjustment.
- * p < .05;
- ** p < .01;
- *** p < .001.
Self-esteem. Controlling for gender, authoritativeness predicted year 4 optimism, pr =.28, t(111)=3.07, p < .01. Adding self-esteem to the model reduced the relation between authoritativeness and optimism, pr=.19, but it remained significant, t(110)=2.01, p < .05. The relation between self-esteem and optimism was also significant, pr=.57, t(110)=7.21) p < .001. The Sobel test for the indirect effect of authoritativeness on optimism via self-esteem was significant, Z=2.26, p < .05, indicating that self-esteem partially mediated the relation between authoritativeness and optimism.
Depression. The relation between authoritativeness and optimism revealed in step 2 of the regression was reduced to non-significance when controlling for depression in step 3, pr =.14, t(110)=1.46, ns. The relation between depression and optimism in step 3 was significant, pr=−.55, t(110)=6.96, p < .001. In support of mediation, the Sobel test of the indirect effect of authoritativeness on optimism via depression was significant, Z=2.90, p < .01.
University Adjustment. The relation between authoritativeness and optimism revealed in step 2 of the regression remained significant when controlling for university adjustment in step 3, pr =.21, t(110)=2.28, p < .05. The relation between university adjustment and optimism in step 3 was also significant, pr=.46, t(110)=5.41, p < .001. The Sobel test for the indirect effect of authoritativeness on optimism via university adjustment was not significant, Z=1.85, ns. Mediation was thus not supported.
Discussion
The present data provided quite clear evidence that perceived parental authoritativeness predicted adjustment during students’ time at university, lending further support to the body of evidence showing that parenting style predicts adjustment among older adolescents (see Baumrind, 1991; Slicker, 1998; Weiss & Schwarz, 1996). Moreover, the unique position articulated in this paper—that parenting style predicts adjustment, in part, because it is related to adolescents’ optimism—was supported, in that the relations between perceived parental authoritativeness and adjustment were mediated by students’ optimism. Mediation was complete for self-esteem and depression, indicating that optimism is a critical mechanism by which parenting style produces healthy personal adjustment. Mediation was partial for university adjustment, indicating that optimism is likely one of several mechanisms by which parenting style influences adjustment in an academic context: No doubt various more cognitive pathways are also important to this relation (e.g., Glasgow et al., 1997; Pratt et al., 1992).
No support was found for the alternative model in which optimism levels predicted adjustment via construals of their parents’ authoritativeness. This suggests that our participants’ perceptions of their parents’ authoritativeness were not best understood as simply a reflection of their own positivity. Nevertheless, we did find some support for a second alternative model in which authoritative parenting predicts optimism via good adjustment: Self-esteem partially, and depression fully, mediated the relation between authoritative parenting and optimism. Thus, there was support for the notion that the healthy adjustment fostered by authoritative parenting might give rise to optimism. This finding does not negate the importance of support for the predicted model, but rather suggests that both dynamics may operate. Perhaps in terms of development, for example, authoritative parenting gives rise to optimism that fosters healthy adjustment, and individuals who enjoy positive adjustment find their optimistic world views reinforced and further cultivated.
Although our results are correlational, they certainly suggest that optimism may have facilitated the personal and social adjustment of our participants during their university careers. The relations between optimism and adjustment appeared to be dynamic because optimism predicted changes in self-esteem and depression, from prior to entry at university to the end of participants’ education. The finding that optimism predicted improvements in personal adjustment over time is consistent with a conceptualization of optimism as a coping resource that can be effectively drawn upon by people who are undergoing a potentially stressful experience (e.g., Scheier & Carver, 1985; Taylor & Brown, 1988), such as the transition to university.
In general, Study 1 supported our reasoning that parental authoritativeness may predict healthy adjustment among adolescents because it has earlier helped to shape an enduring and adaptive personality characteristic—dispositional optimism. Moreover, the data are suggestive of the substantial long-term benefits of optimism in that optimism predicted adjustment during a major, stressful life transition among late adolescents, many of whom had been living away from parents for a number of years. Of course, these findings would be substantiated with further replication; hence we performed a second study in which we attempted to replicate the basic model in which optimism mediates the authoritativeness—adjustment relations among adolescents across an even longer time frame.
Study 2
Using a second longitudinal dataset, we explored the possibility that optimism would mediate relations between parenting style and adjustment over a substantial period of development, from ages 17 to 23. It seemed likely that optimism would mediate relations between parenting style and adjustment over this time period, given our reasoning that authoritative parenting generates optimism early in life that fosters subsequent adjustment, and in light of evidence that optimism does indeed predict positive adjustment in young adolescents in high school (Boman & Yates, 2001). Specifically, we tested the hypothesis that, among high-school students, perceptions of authoritative parenting would predict higher self-esteem and lower depression six years later, and that these relations would be due to adolescents’ optimism levels. Given the literature cited above on academic achievement and authoritative parenting (e.g., Steinberg et al., 1989), we also tested the possibility that authoritative parenting would predict educational attainment six years later, perhaps as a function of dispositional optimism.
Method
Participants
Participants at time 1 were 938 high-school students (380 boys; 558 girls) from 16 high schools (12 public and 4 Roman Catholic) in southwestern Ontario, Canada. The majority of participants were in their fourth year of studies, although smaller numbers were in either their third or fifth year of high school (at the time this study was conducted, the Ontario high-school system had a five year university preparation program). Ages ranged from 13 to 21 (M=17.5). Students participated on a volunteer basis during school time. All schools involved in the research were provided with an honorarium of $2.00 for each student from that school who participated. The majority of students were white, and born in Canada (88%). Students born outside of Canada came from a wide variety of countries (N=39) in Europe, the Far East, and the Caribbean. The majority of participants (82%) spoke English (only) at home, but 36 other languages were represented across the remaining participants. Seventy-three percent of participants came from an intact family, and eighty-three percent had either one or two siblings. At time 2, 337 adolescents remained in the study. The mean age of the participants at time 2 was 19.28 (and the sample was 71% female). At time 3, the sample included 288 participants (32% of the original sample), of mean age 23 (69% female). Although the final sample contained a reasonable number of young men (N=78), a greater proportion of female participants remained in the study until completion than did male participants, chi square (1, N=939)=24.18, p < .001. A comparison of age at baseline between participants who did and did not complete all three sessions indicated that those who remained in the study until completion were very slightly younger at baseline (M=17.34, SD=.78) than were those who later dropped out of the study (M=17.57, SD=.82), t(934)=3.87), p < .001. Thus, participants in the final sample were slightly younger and more likely to be female as a function of attrition.
Measures and Procedure
The first phase of the research was conducted as part of a larger study (Alisat, Pancer, Pratt, & Hunsberger, 1998; Pratt, Hunsberger, Pancer & Alisat, 2003) during the spring of 1997. The majority of students who volunteered to participate completed a survey during regular class time. Because of scheduling restrictions, in two schools questionnaires were completed during a separate session outside of class time. Questionnaires were administered with standardized instructions by either a graduate student or faculty member. The questionnaire included demographic and family background information, and measures of social support, optimism, depression, self-esteem, parenting style, family functioning, family and peer interactions, and identity status. Scales were identical to those used in Study 1, with one minor exception. During this first measurement period, the items in the parenting style measure were worded in the present tense (e.g., ‘My parents spend time just talking with me’). The questionnaire typically took a total of 60 to 75 minutes to complete, spread across two sessions.
For phase 2 of the research, participants were contacted during the spring of 1999, initially by mail and then with a follow-up telephone call. The mailing included an information letter that introduced the enclosed questionnaire, and indicated that returning the questionnaire would be treated as consent to participate. Two draws using a monetary prize were conducted in order to motivate participation. The questionnaire included scales measuring optimism, depression, self-esteem, and parenting style.
For phase 3, participants completed a 12-page follow-up survey questionnaire in the spring of 2003. These individuals were first contacted by telephone, often through our contact information for parents from 1999. They were offered an honorarium of $5 each for completion of the questionnaire that was mailed to them, and reminder phone calls were made when the questionnaires had not been received over one month after mailing. A letter enclosed with the questionnaire again indicated that returning the questionnaire would be treated as consent to participate. The questionnaires included standard scales measuring optimism, depression, and self-esteem, as well as a report of years of completed educational attainment.
Coefficient alphas in the present sample were: time 1 authoritativeness .80, time 2 authoritativeness, .89; time 1 optimism .83, time 2 optimism, .90, time 3 optimism, .89; time 1 self-esteem .83, time 2 self-esteem .90; time 3 self-esteem, .89; time 1, time 2, and time 3 depression, .89.
As with Study 1, participants reported that their parents were moderately warm and strict. Given a scale range of 10 to 90 for warmth, averages were 65.56, SD =16.03 at time 1 and 68.37, SD=14.14 at time 2. Given a scale range of 6 to 54 for strictness, averages were 40.55, SD=10.33 at time 1 and 41.54, SD=8.97 at time 2. In addition, the warmth and strictness subscales of the perceived authoritativeness measure were substantially positively related to each other (time 1, r(933)=.50, p < .001; time 2, r(331)=.54, p < .001). Given the different number of items in the warmth and strictness scales, the warmth and strictness scales were standardized, and an authoritativeness measure was computed as the standardized sum of warmth (standardized) and strictness (standardized).
Results
Preliminary Analyses
First, to explore any implications of attrition in our data, we compared baseline levels of optimism, adjustment, and perceived parental authoritativeness between participants who completed all three sessions, versus those who dropped out prior to completion. No differences emerged between the groups on any of these variables (all ts < 1.2). Thus, attrition did not alter the nature of our sample in terms of the variables involved in our hypothesized model.
Second, we assessed the possibility that participant gender and socioeconomic status related to adjustment and educational attainment. A participant gender by time of measurement repeated measures ANOVA was computed on each of optimism, self-esteem, and depression. Relevant means for each analysis are provided in Table 7. In each analysis, the only significant effect was the effect of time, reflecting improved adjustment over time. As shown in Table 7, participants became more optimistic over the course of the study, F(2,230) =5.77, p < .01, their self-esteem increased, F(2,230)=5.36, p < .01, and their levels of depression lowered, F(2,232)=9.69, p < .001. In no case was the effect of gender or the time by gender interaction significant, all Fs < 1. Additionally, a comparison of the educational attainment of male and female participants revealed no difference between these groups, F < 1. Overall, participants completed between 0 and 7 years of higher education (M=2.7). Although no effects involving gender emerged, because the gender composition of the sample changed somewhat over time, in tests of hypotheses we control for participant gender.
| Optimism | Self-esteem | Depression | |
|---|---|---|---|
| Time 1 | 47.16 (11.20) | 68.13 (14.84) | 17.63 (10.17) |
| Time 2 | 49.02 (13.02) | 69.46 (11.46) | 16.19 (9.68) |
| Time 3 | 51.86 (12.24) | 73.67 (13.23) | 12.87 (8.85) |
- Note: Standard deviations are provided in parentheses.
Mothers’ and fathers’ education levels were highly correlated, r =.52, p < .001; thus a parental education index was computed by averaging their education levels. Correlations between parents’ education level and adolescents’ adjustment were small, but significant in the first two measurement periods, in which the samples were larger: Parental education correlated positively with adolescents’ self-esteem [time 1, r(930)=.07, p < .05; time 2, r(338)=.12, p < .05; time 3, r(268)=.06, ns] and optimism [time 1, r(929)=.11, p < .01; time 2, r(339)=.12, p < .05, time 3, r(268)=.05, ns], and negatively with depression [time 1, r(926)=−.07, p < .05; time 2, r(337)=−.14, p < .01, time 3, r(269) =−.01, ns.]. Finally, parental education correlated positively with adolescents’ educational attainment, r(267)=.35, p < .001. Based on these relations, in subsequent tests of our hypotheses, parental education is controlled.
Third, we computed Pearson correlations among all key variables in the baseline and final sample. As shown in Table 8, the variables were interrelated in the expected directions. Given the substantial duration of our study, we also examined the stability of individual differences in adjustment over time by computing the Pearson correlations between the first and last measurement for each variable. As shown in Table 8, a moderate to strong degree of stability was revealed. Encouraging were the substantial over time correlations revealed for the measures of perceived parental authoritativeness and dispositional optimism, suggesting that our assumption of some persistence over time in these variables is valid.
| Authoritativeness | Optimism | Self-esteem | Depression | |
|---|---|---|---|---|
| Authoritativeness | 0.64*** | 0.34*** | 0.37*** | −0.33*** |
| Optimism | 0.13 | 0.49*** | 0.69*** | −0.59*** |
| Self-esteem | 0.12 | 0.68*** | 0.54*** | −0.71*** |
| Depression | −0.10 | −0.53*** | −0.67*** | 0.47*** |
| Educational Attainment | 0.16* | 0.09 | 0.10 | −0.18** |
- Note: Correlations above the diagonal are based on time 1 measures. Correlations below the diagonal are based on time 3 measures for optimism, self-esteem, depression, and educational attainment and the time 2 measure for authoritativeness (because for authoritativeness, this
- was the last measurement of interest). Correlations on the diagonal reflect correlations between the first and last measurement for each variable (stability correlations).
- * p < .05;
- ** p < .01;
- *** p < .001.
Hypothesized Models
We tested the hypothesis that relations between authoritative parenting (measured at time 1) and adjustment (measured at time 3) would be mediated by optimism (measured at time 2). We also assessed whether this model would be supported for educational attainment. Following the procedure for testing mediation recommended by Baron and Kenny (1986), we first confirmed that the predictor variable (authoritativeness, time 1) correlated with the mediator (optimism, time 2), r(337)=.26, p < .001 and all outcome variables [self-esteem, time 3, r(274)=.26, p < .001; depression, time 3, r(274)=−.29, p < .001; educational attainment, time 3, r(271)=.23, p < .001]. Second, we tested whether the mediator (optimism, time 2) correlated with the outcome variables. The relation was significant for both adjustment variables [self-esteem time 3, r(198) =.47, p < .001; depression time 3, r(199)=−.40, p < .001], but not for educational attainment, r(198)=.08, ns. Thus, we tested our hypothesized mediational models for self-esteem and depression, but not educational attainment. To determine whether time 2 optimism accounted for relations between time 1 authoritativeness and time 3 adjustment, we performed hierarchical regression analyses in which we predicted the time 3 adjustment variables by entering demographic variables requiring control (gender and parental education) in the first step, parental authoritativeness (time 1) in the second step, and optimism (time 2) in the third step. Regression coefficients are provided in Table 9. Of interest was whether relations between time 1 authoritativeness and time 3 adjustment revealed in step 2 would be reduced significantly by entering time 2 optimism into the equation in step 3.
| DV | IV | Before Mediator Added | After Mediator Added | ||||||
|---|---|---|---|---|---|---|---|---|---|
| B | beta | t | R2 | B | beta | t | R2 | ||
| Self-esteem | Gender | −2.89 | −.09 | −1.68 | .06*** | −3.91 | −.12 | −3.20** | .53*** |
| Educ | .85 | .08 | 1.54 | .24 | .02 | .62 | |||
| PS | 3.13 | .20 | 3.69*** | .58 | .04 | .94 | |||
| OP | .79 | .71 | 18.29*** | ||||||
| CESD | Gender | 2.68 | .13 | 2.36* | .09*** | 3.24 | .15 | 3.52*** | .41*** |
| Educ | −.69 | −.10 | −1.89 | −.35 | −.05 | −1.19 | |||
| PS | −2.79 | −.27 | −4.99*** | −1.39 | −.13 | −2.98 | |||
| OP | −.43 | −.58 | −13.17*** | ||||||
- Note: DV=criterion, IV =predictor, CESD=depression, Educ=parental education, PS=parenting style (authoritativeness), OP=optimism.
- * p < .05;
- ** p < .01;
- *** p < .001.
Self-esteem. As shown in step 2 of the regression analysis, controlling only for demographic variables, the relation between authoritativeness and self-esteem was significant, pr=.19, t(191)=2.71, p < .01. In support of mediation, when controlling also for optimism in the third step of the analysis, this relation was no longer significant, pr=.09, t(190)=1.29, ns. Further, the Sobel test of the indirect effect of authoritativeness on self-esteem via optimism was significant, Z=3.70, p < .001, indicating clear support of this hypothesized model. Overall, the predictors in the model accounted for 23% of the variance in time 3 self-esteem, which was significant, F(4,190)=14.32, p < .001. Finally, when the latter equation was recomputed, entering time 1 levels of self-esteem as well, the relation between time 2 optimism and time 3 self-esteem remained significant, pr=.23, p < .001, suggesting that prior optimism predicted variations in self-esteem over time.
Depression. As shown in step 2 of the regression analysis on depression, controlling only for demographic variables, the relation between authoritativeness and depression was significant, pr=−.23, t(192)=3.29, p < .01. Controlling also for optimism in the third step of the analysis reduced the relation, although it remained significant, pr=−.16, t(191)=2.18, p < .05. A Sobel test of the indirect effect of authoritativeness on self-esteem via optimism was significant, Z=3.27, p < .01. This pattern indicates partial mediation, suggesting that optimism is one of several mediators of the relation between authoritativeness and depression. Overall, the predictors accounted for 17% of the variance in depression scores, which was significant, F(4,191)=10.03, p < .001. Finally, when the last equation was recomputed, controlling for time 1 depression as well, the relation between time 2 optimism and time 3 depression remained significant, pr=−.20, p < .01, suggesting that prior levels of optimism predicted variations in depression over time.
Alternative Model 1
According to alternative model 1, more optimistic individuals construe their parents as particularly authoritative, and this perception contributes to healthy adjustment. The predictor in this model, time 1 optimism, did correlate significantly with the presumed mediator, time 2 authoritativeness, r(336)=.18, p < .01, as well as with both outcome variables [time 3 self-esteem, r(274)=.35, p < .001; time 3 depression, r(275)=−.29, p < .001]. However, the mediator, time 2 authoritativeness, did not correlate significantly with either outcome measure [time 3 self-esteem, r(196)=.12, ns; time 3 depression, r(197)=−.10, ns]. Because this precondition of mediation was not upheld, this alternative model was not plausible, and further testing of it was not warranted.4
Alternative Model 2
According to alternative model 2, more authoritative parenting generates healthy adjustment that, in turn, gives rise to optimism. Thus, we tested the possibility that time 2 self-esteem and depression mediated the relation between time 1 authoritativeness and time 3 optimism. First, the predictor, time 1 authoritativeness, did correlate significantly with both mediators [time 2 self-esteem, r(337)=.22, p < .001; time 2 depression, r(336)=−.27, p < .001] as well as with the outcome variable, time 3 optimism, r(273)=.27, p < .001. Second, the two mediators both correlated with the outcome variable, time 3 optimism [time 2 self-esteem, r(198)=.50, p < .001; time 2 depression, r(197)=−.41, p < .001]. Third, to determine whether self-esteem and depression accounted for relations between authoritativeness and optimism, we performed hierarchical regression analyses in which we predicted time 3 optimism by entering demographic variables requiring control (gender and parental education) in the first step, authoritativeness (time 1) in the second step, and self-esteem or depression (time 2) in the third step. Regression coefficients are provided in Table 10.
| DV | IV | Before Mediator Added | After Mediator Added | ||||||
|---|---|---|---|---|---|---|---|---|---|
| B | beta | t | R2 | B | beta | t | R2 | ||
| OP | Gender | 1.29 | .05 | .83 | .07*** | 3.12 | .11 | 2.84** | .54*** |
| Educ | .77 | .08 | 1.55 | .23 | .03 | .65 | |||
| PS | 3.23 | .23 | 4.25*** | 1.24 | .09 | 2.26* | |||
| SE | .63 | .70 | 18.29*** | ||||||
| OP | Gender | 1.31 | .05 | .84 | .07*** | 3.46 | .12 | 2.73** | .39*** |
| Educ | .78 | .08 | 1.56 | .23 | .02 | .56 | |||
| PS | 3.26 | .23 | 4.27*** | 1.03 | .07 | 1.60 | |||
| CESD | −.80 | −.59 | −13.17*** | ||||||
- Note: DV=criterion, IV =predictor, OP=optimism, Educ=parental education, PS=parenting style (authoritativeness), SE=self-esteem, CESD=depression.
- * p < .05;
- ** p < .01;
- *** p < .001.
Self-esteem. Controlling for demographic variables, time 1 authoritativeness predicted time 3 optimism, pr=.22, t(191)=3.02, p < .01. Controlling also for time 2 self-esteem in the third step of the regression reduced the correlation between authoritativeness and optimism to marginal significance, pr=.13, t(190)=1.78, p=.07. In support of partial mediation, the Sobel test of the indirect effect of time 1 authoritativeness on time 3 optimism via time 2 self-esteem was significant, Z=3.57, p < .001. This pattern indicates that self-esteem levels partially explain relations between perceived authoritativeness and optimism.
Depression. As noted above, controlling for demographic variables, time 1 authoritativeness predicted time 3 optimism. In support of mediation, controlling also for time 2 depression in the third step of the regression reduced the correlation between time 1 authoritativeness and time 3 optimism to non-significance, pr=.11, t(189)=1.56, ns. Further, the Sobel test of the indirect effect of authoritativeness on optimism via depression was significant, Z=3.73, p < .001. Thus, this model was supported for depression.
Discussion
Consistent with Study 1 and previous research by others (e.g., Baumrind, 1991; Brown et al., 1993; Hein & Lewko, 1994; Lamborn et al., 1991; Steinberg et al., 1994; Weiss & Schwarz, 1996), greater perceived parental authoritativeness predicted better adjustment and higher educational attainment in a sample of adolescents. Specifically, high-school students who reported at age 17 that their parents were more authoritative, reported at age 23 having higher self-esteem, lower depression, and more years of post-secondary education than did those who did not experience their parents as authoritative.
Supportive of the novel hypothesis advanced in this paper, the relations between authoritative parenting and self-esteem and depression were mediated by students’ levels of dispositional optimism. Mediation was complete in the model on self-esteem, and partial in the model on depression, indicating that optimism may be a central mechanism by which authoritative parenting facilitates self-esteem, and one of several mechanisms by which it prevents depression. Parental authoritativeness did predict the educational attainment of our participants, but this relation was not mediated by optimism. Thus, optimism seems to be more powerfully linked to personal and social forms of adjustment than to academic attainment.
As with Study 1, no support was found for the alternative model in which optimism leads to perceptions of authoritative parenting, which in turn directly foster positive adjustment. However, again consistent with Study 1, we did find some support for the alternative model in which authoritative parenting predicts healthy adjustment, which then gives rise to optimism, suggesting that the optimism and healthy adjustment fostered by authoritative parenting are probably mutually reinforcing.
General Discussion
The present research on optimism provides some integrative evidence regarding lacunae in two distinct fields of psychology, developmental research on parenting as well as social-personality research on adaptation. We first discuss the developmental aspects of this work and then turn to issues in the social psychology of adult adjustment.
First, the present studies contribute to our understanding of the mechanisms responsible for the relations between parenting style and adjustment outcomes in adolescents. Although such mechanisms have been discussed (e.g., Steinberg, 2001), clear evidence for them is needed. This research provides evidence that an index of healthy self-regulation—dispositional optimism—may be one such mechanism. In Study 1, we found that dispositional optimism mediated both personal adjustment and adaptation in university. With respect to the former case, the degree of mediation indicated that optimism is a central mechanism, whereas with regard to the latter case, partial mediation indicated that optimism is likely only one of several mediators of the effects of parenting style. Indeed, consistent with the finding that optimism is most relevant to personal-social forms of adjustment, in Study 2, optimism mediated between authoritative parenting and self-esteem and depression, but it did not mediate between optimism and educational attainment. Optimism may contribute most powerfully to personal and social forms of adjustment because optimistic people are likely to focus on the positive characteristics of their social environment, for example. In this regard, optimism may underlie, or work in concert with, other known mediators of good adjustment. For example, optimistic people may, in focusing on the positive attributes of others, readily make and maintain friendships which further their good adjustment. Optimism may similarly generate positive expectations in academic domains, a tendency characteristic of adolescents with authoritative parents (Aunola et al., 2000), yet educational outcomes may be more proximally determined by cognitive and familial variables, such as academic skills (Steinberg et al., 1989) and parental involvement and assistance (Pratt et al., 1992; Steinberg et al., 1992).
Of course, these findings are based on students’ rather than parents’ reports of parenting style. It is possible that different results would emerge using parents’ self-reports, because adolescents and their parents do not always agree on how the parents’ style is best characterized (Glasgow et al., 1997; Paulson, 1994; Smetana, 1995). However, as noted earlier, there is evidence suggesting that adolescents’ perceptions of parental behaviors may more accurately reflect those behaviors than do parental reports (Gonzales et al., 1996). Nevertheless, the use of self-report data does reflect the fact that our results speak only to adolescents’perceptions of family parenting style. The self-report nature of these data need not be viewed as a problematic limitation, however. Rather, consistent with a constructivist tradition in psychology, they indicate that adolescents’experiences of their parents’ behaviors—which likely correspond reasonably to parental behaviors—predict their well-being vis-à-vis optimism. Most convincing is the fact that these experiences were not, in this research, simply a reflection of adolescents’ own optimism, in that no support was found in either study for the alternative model in which optimism gives rise to perceptions of parenting style and subsequent adjustment.
Our data did indicate some support for an alternative model in which parenting style gives rise to good adjustment, which in turn fosters optimism. Although further research will be necessary to untangle issues of temporal patterning fully, we suggest that authoritative parenting may foster optimism quite early in life, and that this optimism subsequently contributes to, and generates a reinforcement loop with, healthy adjustment. As such, we see optimism as distinct from the several positive outcomes of authoritative parenting.
To be sure, the variables in our models were highly correlated, and one might reasonably wonder if optimism is better construed as simply one of several adjustment-related outcomes of parenting style, rather than as a mediator of parenting style ‘effects.’ The correlations notwithstanding, for two reasons we think it is reasonable to make a distinction between optimism and adjustment. First, although similar, optimism is not conceptualized as redundant with adjustment. Dispositional optimism refers to very generalized positive expectations that tend to generate effective goal-directed behavior. Consequently, positive adjustment is conceptualized as an outcome of this process (Scheier & Carver, 1985). Second, with regard to the overlap between optimism and adjustment, one logically might have anticipated a negative relation between optimism and adjustment, as an alternative to a positive relation. For example, some theorists have reasoned that people who are more optimistic are likely to be disappointed and disillusioned when life experiences are not as positive as were their expectations (for a review, see Armor & Taylor, 1998). Because this is conceivable in principle (although not generally supported in fact), it seems problematic to equate optimism with adjustment. Indeed, dispositional optimism has been distinguished from adjustment (neuroticism) empirically as well (Scheier, Carver, & Bridges, 1994). For these reasons, we think that it is useful to distinguish between these constructs.
It is noteworthy that we found the same pattern of results among university and high-school students in Studies 1 and 2, because this indicates that the relations between parenting style and adjustment in later adolescence likely do not depend upon consistent immediate interactions between parents and their children. Although the majority of high-school students lived with their parents at the beginning of the study, the majority of university students did not. The latter group thus would likely have less frequent contact with parents than would the former. Thus, parenting style may continue to predict adjustment among older adolescents (e.g., university students), even when interactions with parents are clearly reduced in scope, because it has previously facilitated the development of adaptive characteristics. This is also noteworthy because transitions that involve movement away from one's family and social environment are likely to be quite disruptive. Indeed, the transition to university is a major life event for many late adolescents. Recently, Slicker (1998) has pointed to the need to determine whether parenting style influences late adolescents who are moving away from home, or whether ‘all those years of parental influence prior to (high school) graduation quickly dissipate upon launching into young adulthood’ (p. 370). The present results of Study 1 are consistent with a more ‘optimistic’ possibility for socializers—that parents may, indeed, have an important, albeit indirect, influence on the quality of this sort of late adolescent transition experience. Moreover, the results of Study 2, extending across six years into the period of emerging adulthood (Arnett, 2000), are also fully consistent with this interpretation.
The present research also contributes importantly to the extant literature on optimism by pointing to its possible developmental origins. Several years of research in social and personality psychology with adults have clearly indicated that optimism predicts many positive outcomes, such as physical health and psychological well-being (see Andersson, 1996). Whether optimism has been conceived as a stable dispositional trait (e.g., Scheier & Carver, 1985), or a contextually based type of positive expectation (e.g., Armor & Taylor, 1998; Jackson et al., 2000), it has been conceptualized as a predictor of positive adjustment. Despite the substantial evidence of the centrality of optimism to adjustment, however, little is known about the developmental origins of optimism (Gillham, Reivich, & Shatte, 2001). The present studies indicate that this precursor of positive adjustment may in turn have at least part of its roots in authoritative parenting during childhood and adolescence. Future research may address the specific processes by which, and time frames during which, parenting produces optimism. Similarly, additional work may identify the relative contributions of parental warmth and strictness, the components of authoritative parenting, to optimism and subsequent well-being.
It seems certain that a range of earlier contextual factors are also likely to influence adult optimism, including other social and ecological settings, such as the peer group and the school, as well as contemporary experiences in adult roles. And there may be other pathways of transmission within the family context (e.g., genetic or temperamental factors) that cannot be tested within the present dataset and surely deserve examination as well.
Future research should also address contextual constraints on the ability of optimism to mediate relations between authoritative parenting and adjustment. Our sample was relatively homogeneous, and thus we were not able to test for group differences in the relations identified. Existing research does indicate that cross-cultural variations exist in the implications of both parenting style (see Darling & Steinberg, 1993), and optimism (Chang, 1996; Hein & Lehman, 1995), although in the former case the cross-cultural variations may be less notable than the similarities (Steinberg, 2001). Thus, as the contextual influences on the impact of parenting style continue to be examined and articulated, researchers may wish to consider whether the mechanisms responsible for the relations identified also vary cross-culturally. One possibility is that optimism may be functional for different reasons, or have different meanings, in individualistic and collectivist cultural contexts. For example, in individualistic contexts that particularly encourage and reward autonomy and self-reliance, belief in one's unique potential may generate positive outcomes, whereas in more collectivist cultures, to a greater degree it may be confidence in one's support network that may foster the same. Cultural variations in self-enhancement tendencies suggest that such differences are plausible (e.g., Kitayama, Markus, Matsumoto, & Norasakkunkit, 1997), and such broad contextual factors deserve attention in models of the growth and persistence of personal optimism as well.
References
Acknowledgment
This research was supported by a research grant to the second, third, and fourth authors from the Social Sciences and Humanities Research Council of Canada. We thank Philip Cowan for his helpful comments on an earlier version of this manuscript, and Susan Alisat for her assistance with the coordination of this research.
