Glossary - Evaluation Methods in Biomedical Informatics - page 369

of dividing either the inputs or expected outputs of their use, so true mar-
ginal costs cannot be calculated. (See also Direct costs, Indirect costs).
Measurement study: Study to determine the extent and nature of the errors
with which a measurement is made using a specific instrument (cf. Demon-
stration study). (See Chapters 5 and 6).
Member checking: In subjectivist investigation, the process of reflecting
preliminary findings back to individuals in the setting under study; one way
of confirming that the findings are truthful.
Meta-analysis: a set of statistical techniques for combining quantitative
study results across a set of completed studies of the same phenomenon to
draw conclusions more powerful than those obtainable from any single
study of that phenomenon. Used in many systematic reviews or overviews.
Meta regression: the use of systematic review (qv.), meta-analysis (qv.) and
regression (qv.) techniques on a large body of primary studies to uncover
significant associations between aspects of the study design, intervention
etc., and a single outcome variable.
Nominal variable: Variable that can take a number of discrete values but
with no natural ordering or interval properties.
Non-parametric tests: Class of statistical tests (such as the chi squared and
Mann Whitney U tests) that requires few assumptions about the distribu-
tions of values of variables in a study (e.g., that the data follow a normal
distribution).
Null hypothesis: In inferential statistics, the hypothesis that an intervention
will have no effect: that there will be no differences between groups and no
associations or correlations among variables.
Object (of measurement): Entity on which a measurement is made and to
which a measured value of a variable is assigned.
Objective: (1) Noun: state of practice envisioned by the designers of an
information resource, usually stated at the outset of the design process.
Specific aims of an information resource. (2) Adjective: a property of an
observation or measurement such that the outcome is independent of the
observer (cf. Subjective).
Objectivist approaches: Class of evaluation approaches that makes use of
descriptive, correlational or comparative designs and emphasizes statistical
analyses of quantitative data.
Observational study (naturalistic study): Approach to study design
that entails no experimental manipulation. Investigators typically draw con-
clusions by carefully observing users with or without an information
resource.