Objectives. To examine the replicability and generalizability of findi
ngs suggesting that mood-dependent negative thinking in depression ref
lects changes in the schematic mental models through which the world i
s interpreted, rather than a generalized increase in accessibility of
negative constructs. Design. Depressed and non-depressed samples were
compared on a sentence completion task concerning anticipated outcomes
of social approval or success. This task was designed so that the sch
ematic mental models view predicted more positive completions from dep
ressed participants, the construct accessibility view predicting the o
pposite. Methods. A total of 98 depressed participants (scoring > 15 o
n the Center for Epidemiologic Studies Depression Scale; CESD) from a
depressive self-help organization, and 50 non-depressed controls (CESD
< 16) completed the CESD, Sentence Completion Task and the Dysfunctio
nal Attitude Scale (DAS). Results. Depressed participants made more po
sitive completions in the Sentence Completion Task, and scored higher
on the DAS than controls. Subanalyses suggested that the depressed gro
up's increased positive completions were accounted for largely by thos
e currently in psychiatric treatment; these participants also showed g
reater evidence of dysfunctional schematic models on DAS. Conclusions.
For depressed participants in psychiatric treatment, results replicat
e previous findings, supporting the view that negative depressive thin
king reflects a change in schematic mental models through which the wo
rld is interpreted. Results are inconsistent with predictions from the
construct accessibility view. The results have methodological implica
tions for the use of the sentence completion task in future studies.