For each of my 210 subjects, I have:
Parameter: Cell density (y1) at time T1; parameter: Cell Density (y2) at time T2; parameter: Cell Density (y3) at time T3; parameter: Cell Density (y4) at time T4; Parameter: Tube distance (g1) at time T1; Parameter: Tube distance (g2) at time T2; Parameter: Tube distance (g3) at time T3; Parameter: Tube distance (g4) at time T4; Time between T1 en T2 isn't the same as time between T2 and T3 etc. I want to know: Is there a correlation betweet Cell density and time Is there a correlation between Tube distance and time Is there a correlation between Cell density and Tube Distance I have no clue how to do this in SPSS -- Sent from: http://spssx-discussion.1045642.n5.nabble.com/ ===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD |
Administrator
|
Hello Betty. When you say "correlation", do you mean Pearson's r (i.e., a
measure of linear association), or do you mean statistical association more generally? Thanks for clarifying. betty wrote > For each of my 210 subjects, I have: > > Parameter: Cell density (y1) at time T1; > > parameter: Cell Density (y2) at time T2; > > parameter: Cell Density (y3) at time T3; > > parameter: Cell Density (y4) at time T4; > > Parameter: Tube distance (g1) at time T1; > > Parameter: Tube distance (g2) at time T2; > > Parameter: Tube distance (g3) at time T3; > > Parameter: Tube distance (g4) at time T4; > > Time between T1 en T2 isn't the same as time between T2 and T3 etc. > > I want to know: > > Is there a correlation betweet Cell density and time > > Is there a correlation between Tube distance and time > > Is there a correlation between Cell density and Tube Distance > > I have no clue how to do this in SPSS > > > > -- > Sent from: http://spssx-discussion.1045642.n5.nabble.com/ > > ===================== > To manage your subscription to SPSSX-L, send a message to > LISTSERV@.UGA > (not to SPSSX-L), with no body text except the > command. To leave the list, send the command > SIGNOFF SPSSX-L > For a list of commands to manage subscriptions, send the command > INFO REFCARD ----- -- Bruce Weaver [hidden email] http://sites.google.com/a/lakeheadu.ca/bweaver/ "When all else fails, RTFM." NOTE: My Hotmail account is not monitored regularly. To send me an e-mail, please use the address shown above. -- Sent from: http://spssx-discussion.1045642.n5.nabble.com/ ===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD
--
Bruce Weaver bweaver@lakeheadu.ca http://sites.google.com/a/lakeheadu.ca/bweaver/ "When all else fails, RTFM." PLEASE NOTE THE FOLLOWING: 1. My Hotmail account is not monitored regularly. To send me an e-mail, please use the address shown above. 2. The SPSSX Discussion forum on Nabble is no longer linked to the SPSSX-L listserv administered by UGA (https://listserv.uga.edu/). |
In reply to this post by betty
Bruce's reply suggests this - in terms of Pearson r, you can have sets of those r's
for each one of your 210 cases, looking at those correlations across 4 time periods.
What I should want to produce would be some sort of average for those correlations.
The way that I know to do that uses discriminant function (if DF will allow 210 "groups").
Write the data out to "long" format, with four records (times) for each case, including
two "time" variables - one of them goes 1-4, and the other uses your actual interval Times.
Then you do the DF with Subjects as groups, using the Density, Distance, Time1 and Time2;
and ask for the "within groups correlation matrix" - that has the version of r that you want.
It uses pooled variances, and subtracts out effects of mean-differences between subjects.
I'm suggesting two Time variables because, when data are collected at four times, studies
USUALLY pick time intervals in which they expect equal amounts of change, so the 1-4
better represents an Expected linear trend than using the actual Times. If you don't have
a linear trend, then you might consider showing your results with something other than r.
--
Rich Ulrich
From: SPSSX(r) Discussion <[hidden email]> on behalf of betty <[hidden email]>
Sent: Tuesday, October 30, 2018 11:13 AM To: [hidden email] Subject: Correlations for Repeated measured variables For each of my 210 subjects, I have:
Parameter: Cell density (y1) at time T1; parameter: Cell Density (y2) at time T2; parameter: Cell Density (y3) at time T3; parameter: Cell Density (y4) at time T4; Parameter: Tube distance (g1) at time T1; Parameter: Tube distance (g2) at time T2; Parameter: Tube distance (g3) at time T3; Parameter: Tube distance (g4) at time T4; Time between T1 en T2 isn't the same as time between T2 and T3 etc. I want to know: Is there a correlation betweet Cell density and time Is there a correlation between Tube distance and time Is there a correlation between Cell density and Tube Distance I have no clue how to do this in SPSS -- Sent from: http://spssx-discussion.1045642.n5.nabble.com/ ===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD |
In reply to this post by Bruce Weaver
Hello Bruce,
Thanks for the reply! I indeed meant a Pearson's r correlation. -- Sent from: http://spssx-discussion.1045642.n5.nabble.com/ ===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD |
In reply to this post by Rich Ulrich
Thanks for te reply!
Which kind of grapgh would be best to pot this kind of data into? -- Sent from: http://spssx-discussion.1045642.n5.nabble.com/ ===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD |
I think the CurveFit command might be easiest to use and the most informative. A correlation imposes a linear relationship, which is what you are interested in, but the actual relationship may not be so linear. If CurveFit doesn't seem useful enough, I think you have to go to the graphing routines. I'd do a scatterplot and add a fit line. My opinion is that the Graph command (Graph x with y.) is easiest to use but the output has to be edited to add a regression line. The alternative is Ggraph and GPL, which, I admit, is easiest done through the Chart Builder menu under the Graph dropdown menu.
Gene Maguin -----Original Message----- From: SPSSX(r) Discussion <[hidden email]> On Behalf Of betty Sent: Saturday, November 3, 2018 4:25 AM To: [hidden email] Subject: Re: Correlations for Repeated measured variables Thanks for te reply! Which kind of grapgh would be best to pot this kind of data into? -- Sent from: http://spssx-discussion.1045642.n5.nabble.com/ ===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD ===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD |
I should point out the STATS REGRESS PLOT (Graphs > Regression Variable Plots) extension command. It provides a way to do compact plots with many variables and optionally include up to four fit lines (linear, quadratic, cubic, and LOESS). It can be installed from the Extension menu (or Utilities in older versions) if it is not already present. On Mon, Nov 5, 2018 at 7:46 AM Maguin, Eugene <[hidden email]> wrote: I think the CurveFit command might be easiest to use and the most informative. A correlation imposes a linear relationship, which is what you are interested in, but the actual relationship may not be so linear. If CurveFit doesn't seem useful enough, I think you have to go to the graphing routines. I'd do a scatterplot and add a fit line. My opinion is that the Graph command (Graph x with y.) is easiest to use but the output has to be edited to add a regression line. The alternative is Ggraph and GPL, which, I admit, is easiest done through the Chart Builder menu under the Graph dropdown menu. |
Free forum by Nabble | Edit this page |