One possibility (which is easily checked by examining the scatterplot of all data with the two groups visually coded) is this. In a scatterplot of all the data, lets say group 1's data shows a positive correlation, but most of the points are in the upper left quadrant. Group 2's data, which also shows a positive correlation by itself, is clustered mostly in the lower right quadrant. A linear best fit line of _all_ the data and the result is an overall negative ... visually, it would look like:
R R R B B B X axis --------------------------------> ========================== John W. Kulig, Ph.D. Professor of Psychology Coordinator, University Honors Plymouth State University Plymouth NH 03264 ========================== ----- Original Message ----- From: "Arlie Belliveau" <[email protected]> To: "Teaching in the Psychological Sciences (TIPS)" <[email protected]> Sent: Monday, February 27, 2012 1:37:05 PM Subject: Re: [tips] Statistical question-correlations Hi Marie, Have you calculated your effect sizes? It could be that the positive correlations of .23 and .16 are so small that, when the groups are combined, the error (or noise) turns them into negative correlations. At a glance, the scatter plots don't looks as though the relationships between variables are very significant. Just a thought. Cheers, Arlie -- Arlie R. Belliveau, BA (Hons), MA History & Theory of Psychology, PhD2 York University Department of Psychology 059 Behavioural Science Building 4700 Keele St. Toronto, ON [email protected] On Mon, Feb 27, 2012 at 11:30 AM, Helweg-Larsen, Marie < [email protected] > wrote: I have a simple statistical question. I have a sample of 307 people. 111 are in the red group and 196 are the blue group. The correlation between variables x and y in the red group is r= .226 (n=111), p <.05 and in the blue group r=.164 (n=196), p<.05. However, when I run the correlation between x and y in the entire sample (red and blue combined, no missing data) I get a negative correlation, r=-.142 (n=307), p < .05. Now what doesn’t make sense to me that two groups individually have positive and significant correlations but the two groups combined can have a negative and significant correlation. So you stats tipsters. Is that statistically possible? I have checked everything I possibly can in terms of errors in the data or the analyses and have found none. Some suggestions about what I ought to look at? Marie Marie Helweg-Larsen, Ph.D. Associate Professor l Department of Psychology Kaufman 168 l Dickinson College Phone 717.245.1562 l Fax 717.245.1971 Office Hours: Mondays and Tuesdays 2:00-3:30 http://users.dickinson.edu/~helwegm/index.html --- You are currently subscribed to tips as: [email protected] . To unsubscribe click here: http://fsulist.frostburg.edu/u?id=13091.5fbcba6b94d471416a45a34246e4403b&n=T&l=tips&o=16294 (It may be necessary to cut and paste the above URL if the line is broken) or send a blank email to leave-16294-13091.5fbcba6b94d471416a45a34246e44...@fsulist.frostburg.edu --- You are currently subscribed to tips as: [email protected] . To unsubscribe click here: http://fsulist.frostburg.edu/u?id=13338.f659d005276678c0696b7f6beda66454&n=T&l=tips&o=16303 (It may be necessary to cut and paste the above URL if the line is broken) or send a blank email to leave-16303-13338.f659d005276678c0696b7f6beda66...@fsulist.frostburg.edu --- You are currently subscribed to tips as: [email protected]. To unsubscribe click here: http://fsulist.frostburg.edu/u?id=13090.68da6e6e5325aa33287ff385b70df5d5&n=T&l=tips&o=16304 or send a blank email to leave-16304-13090.68da6e6e5325aa33287ff385b70df...@fsulist.frostburg.edu
