| Comparative |
A type of research that compares the results
of one group receiving a treatment (or intervention)
to another receiving either no treatment or a
different one. (See quasi-experimental and experimental.) |
| Convenience Samples |
A sample developed not from a
random selection of a given population but from
an available group, e.g., ‘man-in-the-street
surveys.’ Likely occurs when the full extent
and characteristics of the population of interest
are unknown or random sampling is not feasible.
Results from a convenience sample cannot be generalized
to represent a larger or different population. |
| Experimental |
The most rigorous method of evaluating
effects of some treatment or intervention: by
selecting two groups of people, comparable on
various potentially relevant characteristics,
then randomly assigning them to participate in
a particular program or not. The results are compared
by researchers who are blinded to (unaware of)
which people participated. If no other relevant
common factors influence the group of participants
during the research period, any measurable changes
after the program to the ‘participant’ group
alone, can be viewed as a program effect. Randomly
matched experimental studies require certain resources
and expertise and sufficient numbers in each group
to meet statistical tests for reliability of results. |
| Formative |
A type of evaluation
resulting from a compilation of data
from various sources over time to show development
of the person
or thing being evaluated. An example
would be an employment skills program that evaluated
students
based on a series of practice lessons,
work assignments, in-class participation and teacher
observations
based on interaction.
|
| Longitudinal |
A study in which
the same group of people are studied
periodically over a long time and compared with
themselves
to determine the occurrence of change
in relation to specific variables.
|
| Matching |
Process of making the
study group and comparison group comparable
with respect to certain factors, e.g. age, sex.
|
| Non-comparative |
A study or examination
of subjects that draws no comparisons
with others outside the group. It may examine
aspects from
historical data (retrospective), describe
a current single-time situation or track subjects
periodically
over time (prospective).
|
| Non-experimental |
Similar to non-comparative
studies, having none of the attributes
of experimental studies, e.g., random selection
into matched groups
that are then treated differently, with
blinded comparative analysis of effects.
|
| Quartiles |
A set of values that mark dividing
points in a set of numerical data, used
to help describe the dispersion of the data. The
first
or lower quartile has 25% of the values
equal to it or lesser in value. The top quartile
has
75% of the values either at or less than
it in value. The first quartile may be approximately
calculated by placing a group of values
in order
from smallest to largest, determining
the median of the values, then finding the median
for the
values on either side of that mid-point
value. (e.g. 2,3,3, 4,5,5,6,7,7,9,10) For an odd
number
of values, exclude the median when calculating
the quartiles, for an even number, include
it.
More precise calculation for small data
sets:
To calculate the position of the quartile
values.
- Rank the data set observations
from smallest to largest.
- To find the
lowest quartile, add
one to the number of data values, rounding
to the nearest whole
number, then divide by four, (n+1/4).
If the resulting number
is halfway between two whole numbers, round
up to the next number. This
gives you the rank position
of the lowest
quartile value.
- To calculate
the highest quartile, first
multiply the number of data
plus one by three then
repeat the above step. The observation
with this rank represents
the upper quartile. If 3(n+1)/4 falls
halfway between two integers,
round down.
Based on UNESCO site
for computer training at
http://203.162.7.85/unescocourse/statistics/35.htm
|
| Quasi-experimental |
Studies that
generally compare groups of participants
and non-participants that share some important
similar characteristics
but are not matched and randomly selected
e.g., one classroom of children in a specific
grade
in one school participate in a program
and are compared to a second class in the same
grade and
school who did not participate. This
is done more frequently when standardized instruments
(questionnaires
or scales) are used to collect data about
a specific characteristic or question such as
self-esteem,
substance use habits or bullying. Such
standardized instruments have accepted levels
of reliability
and validity. Such studies may or may not
be blinded
|
| Single-time evaluation |
A one-time
snapshot assessment of a program, or
intervention, to provide a description of the
state of a particular
indicator (or characteristics of the
program or participants). Often used at the conclusion
of
a program to determine levels of satisfaction.
|
| Standard deviation |
The size
of the standard deviation is a measure of
the variability of quantitative results.
With regard to ordinal,
interval and ratio scale data, it is
a numerical expression of the average
of how far each value
differs from the mean of a data set.
Simple steps:
- Calculate the difference
between each data value and the
mean of the set. (individual
value – mean value.)
- Square each difference,
then sum them and divide
the total by the number of
data values to find the variance.
- Take the square root
of the variance to find
the standard deviation.
|
| Statistical samples |
The selection
of a smaller group to represent a larger
known group (the population) using random selection
so that everyone in the population has
a chance
of being selected. The sample group should
then be representative of the whole population.
|
| Summative |
A type of single-time
evaluation showing the final outcome
from a single measurement or assessment. A single
test at the
end of a course to pass the level is
a summative evaluation. |
