It is of great importance for advertising industry to accurately predict how influential the carry-over effect will be on advertising effectiveness. In this study, the predictability of the three models (linear, modified exponential and logistic) for ...
It is of great importance for advertising industry to accurately predict how influential the carry-over effect will be on advertising effectiveness. In this study, the predictability of the three models (linear, modified exponential and logistic) for the influence of carry-over effect on the effectiveness of TV advertising was comparatively evaluated by gender-age (four groups) and industry (ten sectors), respectively.
It is investigated in the study if the advertising effect in the third month after launching an advertisement is directly influenced by the advertising effects from the first and the second month. To quantitatively measure the carry-over effect for a given advertisement, the models were developed in the time point of the second and the third months respectively. The carry-over coefficients of the second month and the third month were compared to see if there is any statistical difference in the carry-over effects between the second and the third month. Here are the main results.
First, the linear model showed that the advertising effect generated from the second month has meaningful influence on the advertising effect of the third month in all four sex-age groups. However, it was simulated that the influence from the first month effect has little on the third month.
Second, the linear model revealed better predictabilities in most cases in simulating the carry-over effect of the second month on the third month in comparison with other models tested in the study. The R-square values obtained from the model were in the reasonably high range, showing 0.474~0.830. In addition, the carry-over coefficient for the entire data set was 0.518, meaning that approximately 51.8% of the second month advertising effect has influence on the third month effect. On the other hand, it was found that the logistic model was useful in describing the carry-over effects in “Male group aged 10~29” and “Transportation” sector.
Third, in terms of the carry-over effect of the first month on the second month, the linear model also showed better description abilities in most cases, showing R-square values in range of 0.501~0.817. The only exception was in the sector of “Cosmetics, Sanitary goods and Detergent” of which R-square is 0.320. Based on the linear model, the carry-over coefficient for the entire data was modeled to be 0.768, which is comparatively higher than the coefficient for the second month effect on the third month. The logistic model was found to be the best for “Female group aged 10~29” in gender/sex, and “Food” and “Beverage, Favorite food” in business sector.
Fourth, it was found in all four gender-age groups that the extent of carry-over effects between the second month and the third month were statistically different. Seen by industry, the coefficients were in meaningful difference in four sectors, “Household Electric Appliances”, “Finance / Insurance / Securities”, “Service” and “Beverage/ Favorite Food”- at the significance level of 0.05. However, meaningful differences were not found in other six industries.
In conclusion, the linear model was determined to be widely applicable to most cases because of its superior ability to other models tested in the study in describing the carry-over effects regardless of gender/age groups and industry sectors. Even though the logistic model showed better simulating result in a few cases, it might be good to use this model with greater care because of its lack of consistency.