HANDOUT 6.1: TRADEOFFS INVOLVED IN USING DIFFERENT TYPES OF MEASURES

                  Self-report         Disguised         Non-verbal        Physiological    
                                      self-report                                           
VALIDITY          Reactivity biases   May not tap       Often, good. Be   May not tap       
                  (demand             construct.        wary of           construct.        
                  characteristics     However,          observer biases.  However, free     
                  and social          relatively free                     from some         
                  desirability),      from reactivity                     reactivity        
                  and response sets   biases and                          biases and        
                  may pose            observer biases.                    often free from   
                  problems.                                               observer biases.  
                  However,                                                                  
                  generally                                                                 
                  reliable and                                                              
                  often has good                                                            
                  validity.                                                                 
SENSITIVITY       GOOD                Often, poor       Sometimes hurt    Sometimes hurt    
                                      because measure   by limited        by poor           
                                      is affected by    range of scale    reliability.     
                                      so many other     points or by                        
                                      factors           poor                                
                                                        reliabi-lity.                       
SCALE OF         Interval?            Ordinal           Often, ordinal    Often, ordinal    
MEASUREMENT                                                                               
ETHICS            Good.               Raises informed   Depends.          Usually okay.     
                                      consent           Privacy issues                      
                                      problems.         may be raised.                      
PRACTICAL         Good.               Okay.             Depends, may be   Usually fine.     
                                                        too difficult     Expense may be    
                                                                          a problem         


Go back to Chapter 6 Main Menu