SPSS 教程 回归分析.ppt

Now, do you remember we talked about measuring attitudes? Satisfaction is a type of attitude. We therefore need a scale measurement and also a weight measurement… Physical beauty Score Girl1 Girl2 Lifestyle Personality 6 6 1 Likert 1-7 1 4 6 You are so lucky! You have two girls (or boys if you are offended) to compare 11 13 Total Note that there are two dimensions to every attitude scale, weight and scale. E.g., Physical beauty Importance Score Total Lifestyle Personality Constant sum, 10 Physical beauty Importance Girl1 Girl2 Total Lifestyle Personality 3 2 5 Note that there are two dimensions to every attitude scale, weight and scale. E.g., Physical beauty Importance Girl1 Girl2 Total Lifestyle Personality 3 2 5 Note that there are two dimensions to every attitude scale, weight and scale. E.g., 12 3 18 1 6 6 4 6 1 8 5 30 Girl1 Girl2 Total 41 35 11 13 You can assess the weights either directly, through asking, or through inference. Direct measurement has just been illustrated, inference might use regression… Instead of asking for weights, ask for overall attitude plus rating of components, then regress O/a=f(const+?1looks+?2lstyle+?3personality) Measuring attitude… Now, do you remember we talked about measuring attitudes? Satisfaction is a type of attitude. We therefore need a scale measurement and also a weight measurement… By using regression analysis we can get both at once, and more accurately… We can regress the individual (value-adding) items against the overall score, thus: Overall assessment = F (objs.+ relevance etc) This will tell us how much variations in each element (e.g.”handouts”) contributes to variations in the overall judgment. So each item will have: A (mean) score (how well we did), an impact (Beta) value (how important it is to the respondents’ satisfaction). Measuring satisfaction… 1.327 .774 1.714 .099 .482 .122 .601 3.961 .001 .348 .141 .413 2.458 .021 .172 .117 .194 1.468 .155 .156 .102 .255 1.