structural Effects in Education 117

structural Effects in Education 117
difficulties that would subsequently emerge: when a variety of school properties are treated
simultaneously with aggregate status under the rubric of climate, it is impossible to tell whether
school climate or other conditions correlated with it influence the outcome.
In contrast, Rogoff's [Rams0y] (1961) analysis of the same data was an attempt to identify
the "social process," sequence of events, and "social mechanisms" (p. 241) by which
amount of education is directly related to adult social status. She considered three explanations:
family influences on academic motivation, the positive scholastic impact of schools on
able students whatever their class origins, and community compositional influences both in
support of schools as cultural institutions and manifest in school normative climate (pp. 242-
243). With respect to structural effects, she noted the importance of the classroom, through the
mechanism of its rewards and punishments, as an influence on ability and motivations and as
resulting in differing levels of achievement; that is, interaction effects between talent and the
school's system of rewards. She also considered the contextual effects of town, suburban, and
city school location, presumably operating through normative climates. However, why climates
should vary by type of community is not clear. Rogoff was clearly interested in the
problem of mechanisms and saw them residing in the "educational experience" (p. 250) that
schools provide. However, in the absence of direct evidence about the classroom and its "reward-
punishment system," and of the normative impact of communities and schools, mechanisms
remained conjectural.
Boyle (1966a), reviewing earlier studies of socioeconomic composition effects, asked
why there should be contextual effects and by what mechanisms they occurred: peer group
culture, the pedagogical characteristics of schools, values and attitudes, or scholastic abilities
(p. 628). He devised two explanatory arguments: first, schools differ in their success at developing
knowledge and skills; second, they influence attitudes, values, and motivation to attend
college. Concerning the first, Boyle drew upon Rogoff's (1961) idea that the structure of
national educational systems (p. 631) should be taken into account (e.g., the decentralized
character of the American system, allowing variation in educational quality school-by-school,
in contrast to the more provincially centralized Canadian system). Second, he alluded to the
normative pressure generated among students related to the social-class composition of schools.
Limitations of his data prevented B oyle from employing satisfactory measures of instructional
adequacy. He ended up interpreting his evidence as Wilson would have ("students in generally
middle-class schools who lack . . . [a strong academic background] appear to be carried along
by the majority" [p. 634]; note the similarity to Berelson and associates' concept of the "breakage
effect" [1954, pp. 98-101]). In so doing he presented a structural effects argument that
spanned several levels of educational systems including the knowledge-imparting curricular
function of schools in addition to normative climate transmitted through interaction.
By the late 1960s, the drift of investigations revealed the widespread currency of structural
effects arguments, but Sewell and Armer (1966a) expressed doubt about them. Their
position resembled Wilson's: neighborhood contexts represent subcommunities reflected in
the composition of student populations within which normative climates form that influence
the aspirations of all youth, irrespective of individual social status and ability (pp. 161-162).
Sewell and Armer maintained, however, that neighborhood context made little contribution to
explaining college plans independent of "traditional variables" (p. 167), an assessment affirmed
by Brown and House (1967) and by Hauser (1970, 1971). Their conclusion also took
a poke at recent claims about neighborhood effects, which "may be traced to popular assessments
of American education by various educational authorities" (p. 160), a snide reference to
Conant (1961) and his book. Slums and Suburbs.
Proponents of contextual analysis (Boyle, 1966a; Michael, 1966; Turner, 1966) leapt to



118 Robert Dreeben
their own defense; Sewell and Armer (1966b) rejoined. Most of this debate addressed technicalities
of definition, of method, of design, of inference, and of causal order. Little of it addressed
the substantive meaning of context and what happens in neighborhoods and schools,
which, according to Sewell and Armer, "need to be measured directly and appropriately rather
than inferred from the social-class composition of the school or the neighborhood" (p. 711;
my italics). Michael (1966), taking umbrage at Sewell and Armer's (1966b) dig at educational
authorities, thought that sociological paternity should prevail because Durkheim's Suicide
was published sufficiently long ago "to elevate contextual analysis from the status of fad to
tradition" (1966, p. 706). Aside from the parties to this debate airing technical matters, Sewell
and Armer (1966b) raised an important point about structural effects arguments (but without
supplying substantive remedies): the need for more direct measurement of what context means
and less reliance on conjectural inference from compositional measures.
The period from the late 1960s through the 1980s witnessed further developments in the
area of structural effects, some extending and elaborating it, others questioning its conceptual
and methodological soundness. Earlier, Robinson (1950) aimed a salvo at its underpinnings.
He doubted "whether ecological correlations can validly be used as substitutes for individual
correlations" (p. 357), except under unlikely conditions, because a correlation between two
characteristics of individuals in a population will not necessarily be the same when calculated
in subgroups. Hauser (1970) extended this line of criticism with a broader methodological
critique in "Context and Consex: A Cautionary Tale" and in a more complete and detailed
statement in Socioeconomic Background and Educational Performance (1971). The logic of
his argument was similar, the intent different. He was more concerned with how to establish
structural conditions as causes of individual conduct than with whether ecological correlation
will serve as a defensible expedient when individual-level data are unavailable.
The contextual fallacy occurs when residual differences among a set of social groups, which remain
after the effects of one or more individual attributes have been partialed out, are interpreted in terms
of social or psychological mechanisms correlated with group levels of one of the individual attributes.
(1970, p. 659; my italics)
Criticizing Blau, Hauser (1970) noted that equating group differences with "the social"
and individual differences with "the psychological" represented "a misunderstanding of statistical
aggregation and of social process" (p. 13). He was mainly concerned with properly
modeling social process, jointly implicating both individual and structural considerations,
though he did not construct a substantive argument about school organization and schooling.
(Blau's analysis, however, did not really depend on a sharp distinction between the "social"
and the "psychological.") Hauser questioned the defensibility of contextual arguments: they
can be arbitrary, because selecting one contextual variable does not rule out the appropriateness
of others or of individual explanations. They fail to identify internal mechanisms because
they assign school-level properties to all students and do not distinguish selection based on
the dependent variable from a contextual effect (Hauser, 1971, p. 32, 1974). A key issue was
whether mechanisms can be identified, a point where Hauser joined Sewell and Armer.
Campbell and Alexander (1965) also found difficulty with structural effects arguments
over the issue of mechanisms. They drew on earlier research that emphasized the significance
of personal interaction with school peers. " . . . [AJnalyses of 'structural effects,"' they maintained,
must move "from the characteristics of the total system to the situation faced by the
individual due to the effects of these characteristics and then from the social situation confronting
the individual to his responses to it" (pp. 284-285). Context represented opportunities
for interaction with friends planning to attend college. The larger the pool of high socioeconomic
status peers in the school, the greater the chance of finding such friends (an argument