A DESIGN FOR EVALUATION (CIPP)

Here is another alternative to evaluation. This one is not as
detailed as the one described above.

There are many designs for evaluation and the CIPP model has been
around for many years. Here are the main components of CIPP:

C - context or things in the environment

I - input or resources consumed

P - Process, the flow of activities and the decisions made
	(Same as formative evaluation)

P - Product, the output
	(Same as summative evaluation)

The model was proposed by Dan Stufflebeam when he was
at Ohio State U.

          Stufflebeam,   et    al.,   EDUCATIONAL    EVALUATION    AND
     DECISION-MAKING IN EDUCATION.  Itasca, IL.: Peacock, 1971

In a design for evaluation based on this model (or any model you
select), a designer must expand these dimensions with standards 
and criteria to judge the worth or value of something (the project 
in your case).

Standards are broad categories of statements about the thing we are
judging, for example:

1.  The system's  COST/EFFECTIVENESS

	criteria 1.a: Cost must be at or less than other methods of
		      presentation or the additional value must be
		      justified with enhanced learning.

2.  The system's  EASE OF USE 

	criteria 2.a: Users must not have to learn a special language
		      or complex commands to make full use of the
		      system's capabilities.

As you can see, the criteria are SPECIFIC statements where we can judge
whether or not a standard has been met. You can go as far as you need
to on building standards and criteria for your particular system.

It helps to have other benchmarks or baseline data from other systems
with which to compare your system's performance, but in a new design
this may not be possible.

You build these "standards/criteria" sets within the CIPP dimensions,
and soon your evaluation design is living and breathing.  Mix this with
the procedures for performing the actual evaluation (and testing the
results) and you are really rolling in the fast lane of evaluation.
Here is where your statistical analyses (if appropriate would come
into play).

This is just one way to go on evaluation, and there are many others. 
For instance Mike Scriven (ex of Berkeley) developed a model of 
"Goal Free" evaluation, where the evaluator was to keep an open mind 
for "unanticipated" outcomes. Another is based on an adversarial 
approach where feedback is solicited from two or more parties (with 
opposing views).