measures and hence their trustworthiness. Validity should be seen more as an issue of
validity of inference or generalizability rather than as a property of a scale per se. It is the
appropriateness of the interpretation of measurement results which is important.
Researchers and other readers may over-interpret results, that is, generalize findings
beyond the context of a particular study. Researchers when writing reports and papers
should be aware of this and use coefficients of validity, reliability and standard errors
when reporting data derived from tests and measurement scales.
One point often overlooked is that validity and reliability are not synonymous; it
should be realized that valid measures are likely to be reliable but reliable measures may
not always be valid. For example, a clock may be reliably five minutes fast, but it is not
giving you the true time. Similarly a measurement scale may be consistent in measuring
something it was not designed to measure.
Finally, for a more detailed discussion of educational measurement the reader is
referred to Ebel and Frisbie’s (1986) text on Essentials of Educational Measurement, and
useful reviews of criterion and norm referenced testing are provided by Pilliner (1979)
and Murphy and Torrance, (1988). An accessible and introductory guide to the use of
psychological tests is The British Psychological Society’s Psychological Testing Guide
(1990) which contains an introduction to testing, practical advice on what to look for in
tests and further information on how to proceed.
Statistical analysis for education and psychology researchers 32