where R stands for randomization (which is employed to achieve 

 equivalent groups and is a direct attempt to control for the above 

 mentioned internal validity problems); X is the treatment (CETA 

 program experience); and O^ represents the groups for comparative 

 purposes. This design examines external validity in that the 

 introduction of groups O5 and Oe (no pretest) allow for comparisons 

 between groups d to O4 with respect to the main effects of testing 

 and the interaction of testing and X (CETA program experience). Its 

 major advantage is that it allows for a control group (O3, O4) to 

 compare with the treatment group. In fact, the first two lines repre- 

 sent the classic experimental design, a luxury not common to most 

 social science research. 



Upon examination of the structure of the design, four different 

 comparisons of the effect of X are possible: (1) 02>0i; (2) 02>04; 

 (3) 05>06; and (4) 05>03. According to Campbell and Stanley 

 (1963: 25): 



The actual instabilities of experimentation are such 

 that if these comparisons are in agreement, the 

 strength of inference is greatly increased. 



In the analysis of the effects of this design, a 2 x 2 analysis of 

 variance is possible: 



NOX X 



Pretest O4 O2 



No Pretest Oe O5 



To perform the statistical test one would estimate the main effects of 

 X from the column means, the main effect of pretesting from the row 

 means, and the interaction of testing with X from the cell means. 



The differences between the two designs (one-Group Pretest- 

 Post-test vs. Solomon Four-Group Design) are apparent. Control is a 

 key element of the experimental designs whereas in the One Group 

 Pretest-Post-test Design such control is lacking. Causality, in its 

 true sense, is difficult to achieve at any time. However, if we are 

 interested in speaking to this question, the One-Group Pretest-Post- 

 test Design is clearly not applicable. With the Solomon Four-Group 

 Design, the comparative measures necessary to overcome the 

 internal validity threats characteristic of the One Group Pretest- 

 Post-test Design are available. 



THE ISSUES OF QUESTIONNAIRE CONSTRUCTION 



While experimental design issues are a major problem in the 

 Follow-Up Evaluation Project, they are not the only pressing 

 issues which need to be addressed. Of primary importance among 

 these latter issues are problems which relate to the construction of 

 the actual instruments (questionnnaires). Some of the most im- 

 portant issues which will be discussed include tracking, length. 



