Quality & Quantity 35: 107–115, 2001.
© 2001 Kluwer Academic Publishers. Printed in the Netherlands.
Using Social Desirability Scales in Research among
, JOHANNES H. SMIT
and HANNIE C. COMIJS
Department of Social Research Methodology,
Department of Sociology and Social Gerontology,
Institute of Extramural Health Research, Vrije Universiteit, Amsterdam, The Netherlands
Abstract. A consistent ﬁnding is that elderly people obtain higher scores on social desirability
(SD) scales than younger ones. It is usually assumed that elderly people are more eager to present
themselves in a favourable way. Hence, especially in survey-research among the elderly, it is common
practice to include SD-scales to correct for spurious relationships. In this article it is argued that the
assumptions underlying such practice are doubtful, because SD-scores may be affected by a variety
of factors but the tendency to present oneself favourably. Two such factors related to the question-
answering process are discussed in more detail. Empirical support is provided that these factors may
explain the relationship between the score on a SD-scale and age. It is argued that using SD-scales
to correct for relationships between variables, may lead to erroneous results.
Key words: social desirability scales, data collection, elder respondents, need for approved cognitive
To measure attitudes, behaviour and opinions in gerontological research, much
use is made of standardised questionnaires as method of data collection. Unfor-
tunately, the validity of the data obtained by this method may be hampered by
several factors. A well-known disturbing factor is the tendency to present oneself
in a more favourable way. If the possible responses on a survey question differ
with respect to social desirability, respondents tend to select those alternatives that
present themselves in a more favourable way than is actually the case.
It is it is usually assumed that this tendency reﬂects a more or less stable per-
sonality trait like ‘need for approval’. This assumption has led to the construction
of so-called social desirability scales, or SD-scales for short (Crowne and Mar-
lowe, 1964; Edwards, 1990; Eysenck and Eysenck, 1975; Eysenck and Eysenck,
1990). SD-scales measure the tendency to give socially desirable answers by us-
ing questions with response alternatives that differ strongly with respect to their
social desirability, the most desirable alternative being highly unlikely to apply