Using the Theory of Signal Detection to Improve Ad Recognition Testing
指出当前广告识别测试中的记忆分数受回答偏差影响，引入信号检测理论来分离记忆与偏差，从而改进测试方法。
Recognition tests are a very popular means of assessing the memory effectiveness of advertisements.Unfortunately the recognition scores obtoined by current methods reflect both the memory for an advertisement and the response biases of the respondents.The outhors introduce the theory of signol detection (TSD) v/hich can be used to secure independent estimates of memory and response bias in recognition tests.They discuss hov/ TSD can be used to improve ad recognition testing. Using the Theory of Signal Detection to Improve Ad Recognition TestingOf the several methods for measuring the impact of printed advertisements, one of the most widely used is the recognition method (Singh and Rothschild 1983)."The critical thing that defines a recognition task is that the person is given a copy of the information he or she needs to find in memory" (Glass.Holyoak, and Santa 1979, p. 59).In a typical recognition test, such as performed by Starch/INRA/Hooper, subjects are shown a series of advertisements, one at a time.As each advertisement appears, the subjects are to respond "yes" if they think they have seen the ad earlier and "no" if not.In spite of their popularity, these recognition tests are shrouded in controversy.Appel and Blum (1961) and Marder and David (1961) pointed out long ago, for example, that a large percentage of respondents will claim recognition of bogus ads (ads respondents could not have seen before) contained in magazines when real ads are also being tested.In some studies, the claimed level of recognition for bogus ads has been almost as high as that for real ads (Simmons 1961).The tendency to "recognize ads," irrespective of prior exposure to them, may be due to "acquiescence response