Conclusions from the ARF Copy Research Validity Project
重新分析ARF广告文案研究有效性项目的数据，发现多数广告测试指标不可通用，广告喜爱度的重要性被夸大，而前后说服力是更优的预测指标。
This study reanalyzes the results from the ARF's Copy Research Validity Project (Haley and Baldinger, JAR, April/May 1991). The CRVP examined the ability of 35 copy-testing measures to predict TV commercials' sales results, concluding that none could be rejected and implying that ad likability is the best single measure. The reanalysis suggests very different conclusions. Despite the CRVP's placatory conclusion, a majority of the copy-test measures examined in the CRVP clearly can be rejected for general use. Liking's importance has been overstated: of those measures that survive the decision rule of 90 percent correct predictions and significantly above-chance prediction at the 10 percent level of significance, about half involve ad likability but half do not. Moreover, additional tests following the CRVP suggest that Liking, too, fails as a general measure and that a better candidate appears to be Pre/Post Persuasion. Guidelines are offered for future use of the measures in ad tests.