school psychology for high-risk populations: gleanings from the chicago longitudinal study

7
Journal of School Psychology, Vol. 37, No. 4, pp. 457–463, 1999 Copyright 2000 Society for the Study of School Psychology Printed in the USA 0022-4405/99 $–see front matter Pergamon PII S0022-4405(99)00021-7 School Psychology for High-Risk Populations: Gleanings From the Chicago Longitudinal Study Jeffery P. Braden University of Wisconsin–Madison This series in the Journal of School Psychology presents four articles stemming from the Chicago Longitudinal Study (CLS). These studies address auto- biographical retrospective perceptions of resilience and risk, and explore the role of parent and teacher expectations, parent participation, and mo- bility on the academic achievements of African American urban youths. The virtues of these research approaches are elaborated in the series, and need not be repeated here. These lines of research are essential for addressing what has become a major social problem: the lack of academic and social success of urban mi- nority children. The problems of urban minority youth are described in the series. However, the sense of urgency for solving these problems bears re- peating, as does society’s frustration with the failure to solve these prob- lems. The optimism of the 1960s, in which the United States declared a “War on Poverty,” has given way to pessimism regarding the malleability of these enduring (and increasing) social and educational problems. The direct and indirect costs of failure in this regard are significant. The direct costs of social and educational failure can be calculated in terms of dropping out, criminal behavior, lost wages, increased burden on social welfare programs, high rates of teenage pregnancy, and many other indica- tors of economic, social, and physical well-being. The indirect costs are also significant. Perhaps the greatest indirect cost is the lost opportunity due to doing ineffective programs. That is, ineffective programs have two costs that undermine the social weal: the lost opportunity cost (Branham, 1989) for effectively changing outcome trajectories through educational and so- cial intervention, and the lost public trust and increased public cynicism in the effectiveness of prevention programs. These costs, although indirect, affect school psychologists’ ability to garner support for and engage in pre- vention. Received July 15, 1999; accepted July 15, 1999. Address correspondence to Dr. Jeffery P. Braden, University of Wisconsin–Madison, De- partment of Educational Psychology, 1025 W. Johnson Street, Madison, WI 53706-1796. Phone: (608) 262-4586; fax: (608) 265-4559; E-mail: [email protected] 457

Upload: jeffery-p-braden

Post on 02-Jul-2016

213 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: School Psychology for High-Risk Populations: Gleanings From the Chicago Longitudinal Study

Journal of School Psychology, Vol. 37, No. 4, pp. 457–463, 1999Copyright 2000 Society for the Study of School Psychology

Printed in the USA0022-4405/99 $–see front matter

Pergamon

PII S0022-4405(99)00021-7

School Psychology for High-Risk Populations:Gleanings From the Chicago Longitudinal Study

Jeffery P. BradenUniversity of Wisconsin–Madison

This series in the Journal of School Psychology presents four articles stemmingfrom the Chicago Longitudinal Study (CLS). These studies address auto-biographical retrospective perceptions of resilience and risk, and explorethe role of parent and teacher expectations, parent participation, and mo-bility on the academic achievements of African American urban youths.The virtues of these research approaches are elaborated in the series, andneed not be repeated here.

These lines of research are essential for addressing what has become amajor social problem: the lack of academic and social success of urban mi-nority children. The problems of urban minority youth are described in theseries. However, the sense of urgency for solving these problems bears re-peating, as does society’s frustration with the failure to solve these prob-lems. The optimism of the 1960s, in which the United States declared a“War on Poverty,” has given way to pessimism regarding the malleability ofthese enduring (and increasing) social and educational problems.

The direct and indirect costs of failure in this regard are significant. Thedirect costs of social and educational failure can be calculated in terms ofdropping out, criminal behavior, lost wages, increased burden on socialwelfare programs, high rates of teenage pregnancy, and many other indica-tors of economic, social, and physical well-being. The indirect costs are alsosignificant. Perhaps the greatest indirect cost is the lost opportunity due todoing ineffective programs. That is, ineffective programs have two coststhat undermine the social weal: the lost opportunity cost (Branham, 1989)for effectively changing outcome trajectories through educational and so-cial intervention, and the lost public trust and increased public cynicism inthe effectiveness of prevention programs. These costs, although indirect,affect school psychologists’ ability to garner support for and engage in pre-vention.

Received July 15, 1999; accepted July 15, 1999.Address correspondence to Dr. Jeffery P. Braden, University of Wisconsin–Madison, De-

partment of Educational Psychology, 1025 W. Johnson Street, Madison, WI 53706-1796.Phone: (608) 262-4586; fax: (608) 265-4559; E-mail: [email protected]

457

Page 2: School Psychology for High-Risk Populations: Gleanings From the Chicago Longitudinal Study

458 Journal of School Psychology

The key questions to ask of this series are:

1. How are circumstances and outcomes linked?2. How strongly do circumstances affect outcomes?3. How can school psychologists use prevention research to improve the

academic and social well-being of the children they serve?

How Are Circumstances and Outcomes Linked?

There are two approaches to establishing links between circumstances(e.g., interventions, family characteristics) and outcomes (e.g., academicsuccess or failure). The first is retrospective studies, in which individualswho have or have not experienced positive outcomes are studied to deter-mine (after the fact) what might have influenced their status. For example,finding that incarcerated males have low rates of literacy suggests poor liter-acy may lead to criminal behavior. In contrast, prospective studies follow acohort (often selected for risk factors) forward to evaluate their outcomes.Whereas retrospective studies are useful for identifying circumstances thatmay discriminate between successful and unsuccessful groups (i.e., generat-ing hypotheses), prospective studies are important for evaluating the con-nections between circumstances and outcomes (i.e., testing hypotheses).Three of the articles in this series are prospective studies that follow cohortsforward through time to link circumstances and outcomes. One of the stud-ies (Smokowski, Reynolds, & Bezruczko, this issue) is retrospective, and isuseful primarily for generating hypotheses about circumstances that influ-ence students toward more or less optimistic outcomes.

Research linking circumstances to outcomes is susceptible to misunder-standing in at least three ways: (a) correlations are misconstrued as causes,(b) related variables are treated as if they are unrelated, and (c) hypothesisgeneration is confused with hypothesis confirmation. Each of these issuesis discussed in the following paragraphs.

The first issue is the confusion between causation and correlation. Socialscientists may wrongly attribute a causal relationship to a statistical relation-ship. For example, the number of firemen sent to fight a fire is positivelycorrelated with the amount of fire damage. Few people would suggest thatfewer firemen should be sent to fight fires to reduce fire damage. Rather,they understand that the link between the number of firemen fighting afire and the amount of fire damage is not causally related, but is instead cor-related due to the real cause: the severity of the fire.

However, such errors occur in social research. For example, the relation-ship between socioeconomic status (SES) and achievement is frequently in-terpreted as evidence that circumstances available to children in higherSES homes cause better educational outcomes. This may be true, but it is

Page 3: School Psychology for High-Risk Populations: Gleanings From the Chicago Longitudinal Study

459Braden

also possible that this relationship is due to a third factor (e.g., the intelli-gence of the parents), which influences parental social status and the intel-ligence of their children. Thus, prevention science must be cautious aboutinterpreting correlations as evidence of causality.

The CLS series occasionally succumbs to this problem. For example, Gilland Reynolds (this issue) suggest that higher parental expectations maybring about higher levels of student achievement. However, it is equallyplausible to argue that higher levels of student achievement bring abouthigher parental expectations. Thus, the cause may be reciprocal or in theopposite direction than predicted. It is possible to test competing accountsof causality by using statistical relationships, but the methods for doing sorequire meeting some important assumptions.

The first assumption is that appropriate factors are measured in themodel. That is, one cannot relate variations in outcomes to factors thathave not been measured and included in the statistical model. This is re-ferred to as specificity of the model; incomplete specification leads to inac-curate estimation of effects. A second problem is the adequacy of statisticaltools and data to meet assumptions for use. For example, the analysis of ex-pectancy effects on student achievement uses structural equation model-ing, which is a more powerful tool than regression for isolating directionaleffects. However, the assumptions behind structural equation modeling arerarely met in data collected in naturalistic settings. Furthermore, tools areonly as good as the information that is put into them (e.g., a single item forassessing children’s perceptions may poorly measure those perceptions).Readers must be cautious in inferring causation from correlations, evenwhen the logic for doing so is appealing.

A second problem of investigation is the additive versus nonadditive ef-fects of variables. That is, although mobility, parent participation, parentexpectations, and other variables may be coded and entered separately intoa regression equation, it is not clear that they are truly independent vari-ables. For example, parents who are more stable may also tend to havehigher expectations and participate more in their children’s education.These constellations of characteristics may violate the fundamental assump-tions for teasing apart the relative impact of circumstances on outcomes.Although circumstances can be thought of as independent, they may co-occur in constellations or patterns. If so, variables entered early into theequation will account for more outcome variance than those entered later,simply because the predictor variables are related to each other. Anotherproblem with related variables is that it may mislead researchers regardingintervention targets. That is, if mobility, expectations, and participation arerelated, one might wrongly conclude that the prevention target should bemobility. However, it may be that the most important factor is parental ex-pectations; those parents with higher expectations value student perfor-mance, and so they move less. Misconceptions may be encouraged by say-ing a given circumstance “accounted for” some outcome. This is accuratein the statistical, but not in the real or causal, sense.

Page 4: School Psychology for High-Risk Populations: Gleanings From the Chicago Longitudinal Study

460 Journal of School Psychology

The third issue is the confusion between hypothesis generation and hy-pothesis confirmation. This is critical for the Smokowski et al. article, be-cause perceptions are subject to many influences. First, attribution theorysuggests people tend to attribute success to characteristics under their con-trol, but attribute failure to characteristics beyond their control. Thus, au-tobiographical narratives from successful individuals may elicit responsesthat suggest protective mechanisms mediated by the individual, more sothan other factors (cf. Graham, 1994). A second influence is awareness offactors. Students must recognize and be aware of factors to identify themas influences. For example, none of the autobiographies mention the auto-nomic nervous system in maintaining breathing and circulation, yet the au-thors could not have survived without it. The autonomic nervous system isnot within one’s conscious awareness, and so is not identified as important.Other circumstances, too, may be absolutely necessary, and yet remain out-side the individual’s awareness. A third problem with autobiographies is theassumption that the unit of analysis (the author) is self-defining. Qualitativemethods that define individual outcomes relative to the individual (i.e.,methods eschewing interpersonal statistical methods in favor of self-defin-ing qualitative methods) are logically inconsistent with prevention. If indi-viduals are truly self-defining, knowledge of their circumstances cannot illu-minate processes or outcomes for other individuals. By definition,prevention programs presume common circumstances affect outcomesacross individuals.

In summary, the methods used in prevention research influence whatone can learn from them. The articles in this series are careful to identifythe problems associated with the methods that they employ. Even so, thelinks between these factors and those outcomes are at times difficult to de-fine and interpret, and yet they beg for definition and interpretation. In-voking statistical and procedural caveats is difficult in a climate that de-mands action.

How Strongly Do Circumstances Affect Outcomes?

Having considered some of the problems with interpretation embedded inprevention approaches, it is now possible to ask, What are the relationshipsbetween circumstances and outcomes? The articles in this series concludethat mobility depresses academic achievement, that parental expecta-tions do not appear to exhibit significant independent effects on studentachievement after considering prior student achievement (but may influ-ence early achievement), and that parent involvement has some positive in-fluence on student outcomes. A number of factors relating to personal,family, and community variables are suggested as potential influences on

Page 5: School Psychology for High-Risk Populations: Gleanings From the Chicago Longitudinal Study

461Braden

success from autobiographical accounts. One of the biggest problems con-fronted in reporting these results is to distinguish between statistical sig-nificance and clinical meaningfulness. That is, effects can be statistically sig-nificant (i.e., the relationship between two variables is likely to be differentfrom 0), but the degree of relationship may have little if any meaning ina practical sense. This problem is confounded when people use the term“significant” and do not clearly report and discuss the degree of relation-ship. Sadly, some of the articles in this series succumb to this problem, butothers do not. When the statistical significance is separated from clinicalmeaningfulness, it becomes evident that many of the patterns that are reli-ably not due to chance do not have especially strong associations with out-comes. That is, mobility, expectations, and parental involvement are clearlyrelated to outcomes in a way that is not accidental. However, the strengthof these associations vary substantially, to the point where it is not clear thatlarge changes in one (e.g., parental expectation) results in meaningfulchanges in outcomes.

Finally, the autobiographical research suggests protective factors that willpromote resilience and positive outcomes. These could be useful in gaininginsight that cannot be provided by statistical measures, but again must betested in prospective studies rather than self-validated through retrospec-tive accounts.

What Does This Mean for School Psychologists?

These results, and other results stemming from prevention science, suggestpsychologists should have equal measures of humility, optimism, and cau-tion. Prevention is an integral part of the practice of school psychology(e.g., Doll & Lyon, 1998; Nastasi, 1998). School psychologists are in aunique position to work with a wide range of the population at a time whenthat population is arguably most malleable. Thus, the promise of preven-tion for school psychology is high, but this promise must be tested againstour ability to deliver. To deliver, we need three things: humility, optimism,and caution.

School psychologists need humility. The factors identified as influencingoutcomes are frequently beyond the control of most current school psy-chologists, given their position within schools and their relatively infre-quent contact with students, families, and communities. Many of the envi-ronmental attributes that affect students are neither well understood, noreasily changed from school psychologists’ typical work settings. Perhapsmore importantly, the vast majority of variance in outcomes is not ex-plained within statistical models. That is, even when a number of circum-stances are considered, most of the variability in outcomes is unrelated to

Page 6: School Psychology for High-Risk Populations: Gleanings From the Chicago Longitudinal Study

462 Journal of School Psychology

those circumstances. Even if school psychologists could control the circum-stances investigated in this series, substantial variety of variability in out-comes would remain.

School psychologists also need optimism. Although pessimists might viewthe inability to account for significant portion of outcome variance as aproblem, optimists might note that this leaves tremendous opportunitiesyet unexplored. That is, something must account for the variability in out-comes that is not captured in the circumstances studied here. Some of thecircumstances that are frequently overlooked in prevention research arequality of instruction, engaged time on task, and other circumstances thatfall within the purview of school psychologists (see Durlak, 1997). There-fore, although we need a strong dose of humility with respect to our limitedknowledge and ability to control outcomes, we also can embrace a strongdose of optimism in hopes that much of the variability in outcomes has yetto be explained, and thus could be associated with structural and other vari-ables within the school. Even when research shows limited relationships be-tween circumstances and outcomes, school psychologists can use thatknowledge to make modest improvements in the lives of some students.Making things a little better is better than letting things get worse.

School psychologists must be cautious. There is a tendency to overstateor exaggerate the impact of interventions on the lives of children. Exagger-ation may help garner support and mobilize action. However, if psycholo-gists promise more than they can deliver, the result can be cynicism in our-selves, our colleagues, and the children and families we serve. Onedefinition of a cynic is an optimist who failed. Thus, we must respect ourlimitations for both professional and scientific reasons. The ability of pre-vention programs to affect outcomes should be carefully stated. Some maybe discouraged at the relatively small effects for some programs and thecosts of these programs. However, by staying close to the data and by usingresearch to guide our estimation of expected effects, we can avoid the long-term costs of cynicism, failed expectations, and lost opportunities.

Finally, school psychologists must advocate for prevention as applied re-search. Schools emphasize the delivery of services to children, and fre-quently view research as an expensive diversion from service delivery. By be-ing clear about the limitations of our knowledge, and the limitations of ourinterventions, school psychologists could invite the public to view preven-tion programs in the spirit of applied inquiry. Inquiry-based approaches toeducational practices have garnered support in general education, as edu-cators must search for “what works” (e.g., Oborn, 1996). This series offersstarting points and models for beginning prevention-based inquiry with ur-ban minority youths; school psychologists would be wise to capitalize onthis work with their own inquiry-based applications of prevention sciences,guided by humility, optimism, and caution.

Page 7: School Psychology for High-Risk Populations: Gleanings From the Chicago Longitudinal Study

463Braden

REFERENCES

Branham, R. J. (1989). Roads not taken: Counterplans and opportunity costs. Argumenta-tion and Advocacy, 25(4), 246–255.

Doll, B., & Lyon, M. A. (1998). Resiliance applied: The promise and pitfalls for school-based resilience programs. School Psychology Review, 27(3), 346–347.

Durlak, J. A. (1997). Successful prevention programs for children and adolescents. New York:Plenum.

Graham, S. (1994). Motivation in African Americans. Review of Educational Research,64(1), 55–117.

Nastasi, B. K. (1998). A model for mental health programming in schools and communi-ties: Introduction to the mini-series. School Psychology Review, 27(2), 165–174.

Oborn, C. S. (1996, August). The integration of a theoretical and epistemological leadership styleto the practical realm of school administration: A discussion of inquiry based leadership and thedevelopment of a research based school (ERIC Document No. 401610). Paper presentedat the Annual Meeting of the National Council of Professors of Educational Adminis-tration, Corpus Christi, TX.