The following papers
on measuring change can be downloaded as pdf
files, by clicking on their respective titles. The
papers have been presented in reverse chronological order.
In some cases,
you can also access the website of the associated journal by clicking
on
the
journal name.
A
lengthier version of Willett (1994), produced for a different
audience, describing the benefits of using individual growth
modeling in the analysis of change.
A
short non-technical overview of, and general introduction
to, the benefits of individual growth modeling as the
preferred method for the measurement of change.
A
short paper showing how the reliability of of individual
change measurement depends dramatically on the frequency
and spacing of the waves of data used in the analysis. The
paper illustrates the extent of the improvement in reliability
for the measurement of change that can be achieved by
adding one or two extra waves of data collection to a
traditional two-wave design.
Willett,
J. B.(1989).Questions
And Answers In The Measurement Of Change.In
Ernest Z. Rothkopf (Ed.), Review
of Research in Education, Volume 15.Washington,
D.C.: American Education Research Association, 345-422.
A
lengthy (and tediously complete) overview of issues,
problems, and misconceptions in the measurement of change. The
paper contrasts data-analytic approaches using two-wave
and multi-wave data, and illustrates how the latter wins
out in the end. It also debunks several of the
standard myths about the measurement of change.
A
technical exposition of how the relationship between
individual change and its correlates/predictors can be represented
in a multilevel model, and how this specification illuminates
traditional (and often inappropriate) approaches to the
problem of investigating change. The paper also
comments on the relationship between change and initial
status,
on the notion of linear and non-linear change, and on
the flaws that exist in a variety of common residual
change
strategies.
Markedly
different types of growth (learning) curves may generate
indistinguishable covariance structures. We illustrate
with an example of a 5x5 covariance matrix representing
longitudinal measurements at five occasions. This examples
appears to conform closely to a simplex correlation pattern,
and a simplex covariance structure provides an excellent
fit with LISREL V. However, the known structure of this
example differs greatly from a simplex model. In addition
to illustrating that the excellent fit of a simplex structure
can be misleading, this example provides an opportunity
to question common uses of covariance structure models
for the study of growth.
The
study of the stability of teacher behavior over time is
formulated through two questions: Is the behavior
of an individual teach consistent over time? and, Are individual
differences among teachers consistent over time? The first
question has rarefly been addressed in previous investigations
of the stability of teacher behavior, and
empirical research on the second question has been confused.
We develop statistical procedures for addressing both questions.
The approaches of previous studies of temporal
stability are re-evaluated and methods for
assessing the stability of teacher behavior across contexts
are described and illustrated
with data.
