Lecture 17 Flashcards
why meta-analysis is at the top of the pyramid?
from this type of research we get the best evidence and results
steps of meta-analysis
• Collect data from publications
• Apply statistical methods
Interpret the results obtained
is it combinable, we must answer these questions:
• Do the studies estime a sufficiently similar effect?
• Is the study result sufficiently homogeneous (measurement of the result)?
Do the studies deal with sufficiently comparable/homogeneous populations?
if not combinable, the meta-analysis will
- analyse apples and pears
- GIGO (garbage-in, garbage-out) +vb method
- quality of a study will be omitted
- Nor should objective conclusions be expected if only published/statistically significant results will be analyzed
fixed meta-analysis assumption:
The observed variation in treatment effects in the different studies is due entirely to sampling variation
• The underlying treatment effect is the same in all the study populations.
Sometimes too simple in reality.
what happens to the way treatment affects individuals in random-effect model?
the treatment effects of individual studies may differ from each other
what is assumed about treatment effect?
the treatment effect is universal, and the meta-analysis provides the best available estimate of it.
what is the estimate in random-effect meta-analysis?
the estimate is of a mean effect about which it is assumed that the true study effects vary.
what is assumed in fixed-effect meta-analysis?
the true effect is the same in each study and the only reason for variation in the estimates between studies is sampling error. It is assumed that the treatment effect is universal, and the meta-analysis provides the best available estimate of it.
what cluster analysis does with data?
Cluster analysis divides data into groups (clusters) that are meaningful, useful, or both.
what is cluster analysis?
Cluster analysis groups data objects based only on information found in the data that describes the objects and their relationships.
what factors produce better clusters?
the greater the similarity (or homogeneity) within a group and the greater the difference between groups, the better or more distinct the clustering.
define hierarchical cluster
a set of nested clusters that are organized as a tree
define partitional cluster
a division of the set of data objects into nonoverlapping subsets (clusters) such that each data object is in exactly one subset.
what K-means method defines?
defines a prototype in terms of a centroid, which is usually the mean of a group of points, and is typically applied to objects in a continuous n-dimensional space.
what K-medoid defines?
defines a prototype in terms of a medoid, which is the most representative point for a group of points, and can be applied to a wide range of data since it requires only a proximity measure for a pair of objects.
do centroid and medoid correspond to an actual data point?
centroid does not, but medoid does
K-means issues
• Number of clusters is predefined • Initial centroids • Restricted to data for which there is a notion of a centre (centroid) • Empty clusters Outliers
why we use factor analysis?
to reduce the number of variables to explain and to interpret the results
how factor analysis reduces data?
It does this by seeking underlying unobservable (latent) variables that are reflected in the observed variables (manifest variables)
do we need large sample size for factor analysis?
yes, because this model is based on correlation matrix, which requires large sample size
is 50 cases for factor analysis sufficient?
no, to have an excellent analysis, we need to have 1000 + cases
explain exploratory (EFA) factor analysis
A method to explore the underlying structure of a set of observed variables
explain confirmatory (CFA) factor analysis
A method to verify a factor structure that has already been defined
what is the basic assumption of factor analysis
for a collection of observed variables there are a set of underlying variables called factors (smaller than the observed variables), that can explain the interrelationships among those (original) variables.
how the total variance is split
common variance and unique variance
define common variance
the amount of variance that is shared among a set of items. Items that are highly correlated will share a lot of variance.
define unique variance
Unique variance is any portion of variance that’s not common
define specific variance
is variance that is specific to a particular item.
define error variance
comes from errors of measurement and basically anything unexplained by common or specific variance.
what is the main difference between common factor analysis and principle components
The main difference between common factor analysis and principal components is that factor analysis assumes total variance can be partitioned into common and unique variance, whereas principal components assumes common variance takes up all of total variance (i.e., no unique variance).
what is the value of total variance for common factor analysis and principle components
total variance is 1
