The four kinds of evaluation strategies are scientific-experimental models, qualitative-anthropological models, management-oriented systems models and participant-oriented models. Scientific-experimental evaluations focus on the desirability of impartiality, objectivity, accuracy, and the validity of the facts. Qualitative-anthropological models rely on observation and subjective interpretation. Management-oriented systems models often utilize the "Program Evaluation and Review Technique" and the "Critical Path Method." Participant-oriented models depend on clients/ users of a particular program or technology for consumer evaluations.
Formative evaluations include needs assessments, evaluability assessments, structured conceptualization, implementation evaluation and process evaluation. Summative evaluations involve outcome evaluations, income evaluations, secondary analysis, cost effectiveness/cost benefit analysis and meta analysis. Meta analysis requires the integration of outcome approximations from various studies to determine a general summary regarding an evaluation matter.
An example of a descriptive research study would be a survey that determined how many people in a certain area would vote for a Democrat in an election. Relational studies examine the relationship between two or more variables. A survey that compared how many males vs. females would vote for a Democrat would be an example of a relational research study. The two variables would be gender and voting. A causal research study looks at whether one or more variables causes one or more outcomes. Using the election example again, a causal study would determine whether a certain political advertising campaign had an affect on how people voted.
A longitudinal study is designed to occur over time whereas a cross-sectional research study is designed to occur at one single point in time. Longitudinal studies may involve repeated measures or a time series measurement method. The use of two or more waves of measurement indicates repeated measures while many waves of measurement indicate the use of a time series. Time series studies typically involve twenty or more observations with as many as twenty waves of measurement.