| Summary: | 碩士 === 東海大學 === 統計學系 === 91 === This research focused on the building of a quantitative model for click through of Web advertisements. To clarify and quantify the effects of design parameters, a factorial experiment and associated questionnaire survey will be used to collect the related data. Ad size（medium rectangle, full banner, and button 2）, Ad position (top, intermediate , and bottom zones), and Ad contents (commodities information, sales promotion, and entertainment activity) are three independent design variables in this experiment. Considering the effects of gender and the design factors, various logistic regression models were selected. The results indicated the two-factor interactions of Ad size and Ad contents as well as the interaction of gender and Ad position were verified to be significantly affecting the likelihood of a visitor clicking on a Web ad. From the fitted logistic regression model, the estimated odds and the probability of click through were computed. Fixed the levels of gender and Ad position, using commodities information as Ad contents, subjects will have much more click through on medium rectangle size than full banner and button 2. In addition, subjects will have more click through on full banner than button 2. It seems to be shown that larger Ad size will have better advertising effectiveness as the odds of largest medium rectangle size （75,000 pixels）is higher than full banner size（28,080 pixels）and button 2 size（7,200 pixels）. Similarly, using sales promotion activity as Ad contents, advertising effectiveness of medium rectangle size is better than full banner and full banner is better than button2. However, using entertainment activity as Ad contents, we found advertising effectiveness of button 2 is better than full banner. We also found intermediate Ad position will result in gender difference. Male subjects tend to much more click through than female.
|
|---|
