Statistics question

[Deepa Bhat]

Hi, I thought I would ask a statistics question on this listserve.

 

In my baseline data, I have 75 injection observations. In my follow-up data, I have 266 injection observations.

Technically, there should have been about the same number of injection observations in the baseline and follow-up, but there isn't.

We usually do a chi-square to see if there is a difference between the baseline and follow-up. The chi-square is usually a 2*2 table (time vs. variable, time= baseline and follow-up, the variable is usually a yes or no question).

Given the difference in the number of observations , is it still correct/okay to do a chi-square?

If not, then how do we compare the baseline and follow-up?

 

Thank you once again,
Deepa

[Burleson,Joseph A.]

Deepa:

When you say you do a chi-square to "see if there is a difference," do you mean to see if there is a difference between the two proportions? Without data from the second time point, those cases cannot be in the analysis. So, regardless, they cannot be utilized.

The chi-square test is best thought of as a measure of association (analogous to r in correlation), except that the variables are dichotomous.

If you want to know if there is a differential change among individuals from time 1 to time 2 on their status on the dichotomous variable, you need to look at the McNemar statistic (in crosstabs). This assesses whether the number of people who go from one stat to the other (e.g., zero to 1) are significantly different from those who change conversely (1 to zero), and it ignores those who do not change. Think of it as a paired t-test in which the change scores can range from -1, to zero, to +1.

Joe Burleson

________________________________

From: SPSSX(r) Discussion on behalf of Deepa Bhat
Sent: Tue 4/14/2009 8:21 AM
To: [hidden email]
Subject: Statistics question


Hi, I thought I would ask a statistics question on this listserve.

In my baseline data, I have 75 injection observations. In my follow-up data, I have 266 injection observations.
Technically, there should have been about the same number of injection observations in the baseline and follow-up, but there isn't.
We usually do a chi-square to see if there is a difference between the baseline and follow-up. The chi-square is usually a 2*2 table (time vs. variable, time= baseline and follow-up, the variable is usually a yes or no question).

Given the difference in the number of observations , is it still correct/okay to do a chi-square?
If not, then how do we compare the baseline and follow-up?

Thank you once again,
Deepa

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

[Melissa Ives]

How is it that you have >3.5x the baselines at follow-up.  Are you repeating the same respondents over multiple time points or adding respondents who don't have baselines or just an accidental duplication of some of the data or ...??

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Burleson,Joseph A.
Sent: Tuesday, April 14, 2009 9:02 AM
To: [hidden email]
Subject: Re: [SPSSX-L] Statistics question

Deepa:

When you say you do a chi-square to "see if there is a difference," do you mean to see if there is a difference between the two proportions? Without data from the second time point, those cases cannot be in the analysis. So, regardless, they cannot be utilized.

The chi-square test is best thought of as a measure of association (analogous to r in correlation), except that the variables are dichotomous.

If you want to know if there is a differential change among individuals from time 1 to time 2 on their status on the dichotomous variable, you need to look at the McNemar statistic (in crosstabs). This assesses whether the number of people who go from one stat to the other (e.g., zero to 1) are significantly different from those who change conversely (1 to zero), and it ignores those who do not change. Think of it as a paired t-test in which the change scores can range from -1, to zero, to +1.

Joe Burleson

________________________________

From: SPSSX(r) Discussion on behalf of Deepa Bhat
Sent: Tue 4/14/2009 8:21 AM
To: [hidden email]
Subject: Statistics question


Hi, I thought I would ask a statistics question on this listserve.

In my baseline data, I have 75 injection observations. In my follow-up data, I have 266 injection observations.
Technically, there should have been about the same number of injection observations in the baseline and follow-up, but there isn't.
We usually do a chi-square to see if there is a difference between the baseline and follow-up. The chi-square is usually a 2*2 table (time vs. variable, time= baseline and follow-up, the variable is usually a yes or no question).

Given the difference in the number of observations , is it still correct/okay to do a chi-square?
If not, then how do we compare the baseline and follow-up?

Thank you once again,
Deepa

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

PRIVILEGED AND CONFIDENTIAL INFORMATION
This transmittal and any attachments may contain PRIVILEGED AND
CONFIDENTIAL information and is intended only for the use of the
addressee. If you are not the designated recipient, or an employee
or agent authorized to deliver such transmittals to the designated
recipient, you are hereby notified that any dissemination,
copying or publication of this transmittal is strictly prohibited. If
you have received this transmittal in error, please notify us
immediately by replying to the sender and delete this copy from your
system. You may also call us at [phone] for assistance.

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

[David Greenberg]

Contrary to the posting below, chi-square is not a measure of association comparable to r. It is a distribution that can be used to determine whether an observed association is statistically significant (a question of inferential statistics). It is thus comparable to a t-test or F-test. In a two-by-two table there are various measures of association, such as phi, which is, in fact, the value of r in a two-by-two table.
  - David Greenberg, Sociology Department, New York University

----- Original Message -----
From: "Burleson,Joseph A." <[hidden email]>
Date: Tuesday, April 14, 2009 10:06 am
Subject: Re: Statistics question
To: [hidden email]

> Deepa:
>
> When you say you do a chi-square to "see if there is a difference," do
> you mean to see if there is a difference between the two proportions?
> Without data from the second time point, those cases cannot be in the
> analysis. So, regardless, they cannot be utilized.
>
> The chi-square test is best thought of as a measure of association
> (analogous to r in correlation), except that the variables are dichotomous.
>
> If you want to know if there is a differential change among
> individuals from time 1 to time 2 on their status on the dichotomous
> variable, you need to look at the McNemar statistic (in crosstabs).
> This assesses whether the number of people who go from one stat to the
> other (e.g., zero to 1) are significantly different from those who
> change conversely (1 to zero), and it ignores those who do not change.
> Think of it as a paired t-test in which the change scores can range
> from -1, to zero, to +1.
>
> Joe Burleson
>
> ________________________________
>
> From: SPSSX(r) Discussion on behalf of Deepa Bhat
> Sent: Tue 4/14/2009 8:21 AM
> To: [hidden email]
> Subject: Statistics question
>
>
> Hi, I thought I would ask a statistics question on this listserve.
>
> In my baseline data, I have 75 injection observations. In my follow-up
> data, I have 266 injection observations.
> Technically, there should have been about the same number of injection
> observations in the baseline and follow-up, but there isn't.
> We usually do a chi-square to see if there is a difference between the
> baseline and follow-up. The chi-square is usually a 2*2 table (time
> vs. variable, time= baseline and follow-up, the variable is usually a
> yes or no question).
>
> Given the difference in the number of observations , is it still
> correct/okay to do a chi-square?
> If not, then how do we compare the baseline and follow-up?
>
> Thank you once again,
> Deepa
>
> =====================
> 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