This preview shows page 1. Sign up to view the full content.
Unformatted text preview: andeﬀect pattern. A strong regression relation
neither implies “X causes Y” nor implies “Y causes X”.
• Additional information (often through controlled experiments) is needed to draw causeandeﬀect conclusions.
• Does it make sense to conclude “good midterm score” causes “good ﬁnal score”? Note that, a strong regression relation
also holds when taking the ﬁnal score as X variable and the
midterm score as Y variable.
• A data on size of vocabulary (X ) and writing speed (Y ) for a group of children aged 510 shows a positive relationship.
This relation does not imply that an increase in vocabulary
causes a faster writing speed. Other variables (confounders),
such as age and education level, may aﬀect both X and Y ....
View Full
Document
 Spring '08
 Hsieh,F
 Linear Regression, Regression Analysis

Click to edit the document details