This study brings empirical data to the debate about the efficacy of s
tandardized tests. A teacher questionnaire containing 81 Likert-type i
tems was mailed to 450 upper elementary school teachers in 48 schools.
Questions of interest included (a) What are the effects of these test
s on the teaching and learning process in schools? (b) What is the mea
ning of test score gains? and (c) Does such testing have differential
effects in schools serving students at different socioeconomic status
(SES) levels? Approximately 75% of the teachers (n = 341) completed th
e questionnaires. Results of Cronbach's alpha and principal components
analyses indicated that items within specified subscales were interna
lly consistent. Results of a confirmatory factor analysis showed that
the interitem correlations within subscales were stronger than those a
cross subscales. The results of multiple discriminant analysis (DA) us
ing categories of TREND (schools with increasing vs. stable or decreas
ing test scores) as the grouping variable indicated that only two subs
cales were significantly different across TREND categories. However, t
he results of a second multiple DA showed that many subscale scores si
gnificantly discriminated groups formed by student SES levels. The DA
by SES findings suggest that teachers of lower SES students are under
greater pressure to improve test scores and focus more on test content
in their classroom instruction than teachers of higher SES students.