statistical test with data which are means of percentages

[Caroline Wilson]

Hello,

I have a dataset with a variable called A. For each person in the dataset, the variable for A is a percentage, e.g. person 1 had .75 (75%), person 2 had .23, etc.  When I summarize this variable across 5 groups (A-E), I get the following:

               Variable A
Group A 0.55
Group B 0.62
Group C 0.79
Group D 0.84
Group E 0.79

Where, for example, .55 is the mean percentage across all the people in group A.

I would like to see whether there are overall statistically significant differences between the 5 groups, and then compare the different categories with each other, for example see if Group A is statistically significantly lower than Group E.

Is there a way to do this in SPSS?

Any help would be much appreciated!

Thank you,
Carrie

=====================
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

[Caroline Wilson-2]

Hello,

I posted the previous question awhile back and got the advice to use logistic regression.

If I were to use regression, it would be the categorical (nominal) variable that would be the dependent variable and variable A (the proportion) which would be the independent variable. I would run a multinomial logit model.

However, I'm really just trying to look at association between these two variables and not build a regression model for predictive purposes. Is it still recommended that I use a regression model with one independent variable to get the association or is there another test for association that would be better? I was thinking something like a chi-square, but when one variable is a percentage and another is nominal.

Thanks,
Carrie

--------------------------------------------
On Sun, 6/8/14, Caroline Wilson <[hidden email]> wrote:

 Subject: statistical test with data which are means of percentages
 To: [hidden email]
 Date: Sunday, June 8, 2014, 6:04 PM
 
 Hello,
 
 I have a dataset with a variable called A. For each person
 in the dataset, the variable for A is a percentage, e.g.
 person 1 had .75 (75%), person 2 had .23, etc.  When I
 summarize this variable across 5 groups (A-E), I get the
 following:
 
            Variable
 A
 Group A    0.55
 Group B    0.62
 Group C    0.79
 Group D    0.84
 Group E    0.79
 
 Where, for example, .55 is the mean percentage across all
 the people in group A.
 
 I would like to see whether there are overall statistically
 significant differences between the 5 groups, and then
 compare the different categories with each other, for
 example see if Group A is statistically significantly lower
 than Group E.
 
 Is there a way to do this in SPSS?
 
 Any help would be much appreciated!
 
 Thank you,
 Carrie


=====================
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]

Caroline, if you use the "events-var OF trials-var" (or R of N) option for GENLIN, you can estimate what I would call a binomial (as opposed to binary) logistic regression.  Take a look at the syntax chart for GENLIN.  You want something like this:

GENLIN
 events-var OF trials-var { or "OF N" if number of trials is constant }
    [BY factor-list
    [WITH covariate-list]
[/MODEL  [effect-list]
    [DISTRIBUTION = {BINOMIAL        }]
    [LINK = {LOGIT           }
etc.

This requires a data file with one row per person with variables for the number of events (or successes) and the number of trials.  If the number of trials is the same for everyone, you can plug in that constant number of trials where you see trials-var above.  In your case, the (main) factor in the model is Group.

HTH.

Caroline Wilson-2 wrote

Hello,

I posted the previous question awhile back and got the advice to use logistic regression.

If I were to use regression, it would be the categorical (nominal) variable that would be the dependent variable and variable A (the proportion) which would be the independent variable. I would run a multinomial logit model.

However, I'm really just trying to look at association between these two variables and not build a regression model for predictive purposes. Is it still recommended that I use a regression model with one independent variable to get the association or is there another test for association that would be better? I was thinking something like a chi-square, but when one variable is a percentage and another is nominal.

Thanks,
Carrie

--------------------------------------------
On Sun, 6/8/14, Caroline Wilson <[hidden email]> wrote:

 Subject: statistical test with data which are means of percentages
 To: [hidden email]
 Date: Sunday, June 8, 2014, 6:04 PM
 
 Hello,
 
 I have a dataset with a variable called A. For each person
 in the dataset, the variable for A is a percentage, e.g.
 person 1 had .75 (75%), person 2 had .23, etc.  When I
 summarize this variable across 5 groups (A-E), I get the
 following:
 
            Variable
 A
 Group A    0.55
 Group B    0.62
 Group C    0.79
 Group D    0.84
 Group E    0.79
 
 Where, for example, .55 is the mean percentage across all
 the people in group A.
 
 I would like to see whether there are overall statistically
 significant differences between the 5 groups, and then
 compare the different categories with each other, for
 example see if Group A is statistically significantly lower
 than Group E.
 
 Is there a way to do this in SPSS?
 
 Any help would be much appreciated!
 
 Thank you,
 Carrie


=====================
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