Using gender as an outcome variable in logistic regression
Locked
 
|
Hi,
I was wondering if it's acceptable to use gender as an outcome variable in a logistic regression since it can be treated as a dichotomous variable?
Thanks very much for your patience and help. I am thankful for both.
Joesy
|
Re: Using gender as an outcome variable in logistic regression
|
Sounds weird at first, but I once had survey where
the gender question was at the end of a self-completion questionnaire
(administered in a school) and some respondents didn't get that far as there was
a major disruption in one class. Missing gender was accurately predicted
using Discriminant Function Analysis on other items in the
questionnaire.
|
|
|
In reply to this post by Java Joe
Joesy,
I suppose there could be situations where using gender as the dependent variable in a binary logistic regression analysis may be appropriate (i.e. using a set of variables to predict the gender of a child). For a more precise response, you'll need to give more details regarding your research question(s). Ryan
|
|
|
In reply to this post by John F Hall
I have got this material where one way to represent the data
is an ordinary cross table with values expressed in row-wise percentages. The
goal is to get a graph with stacked bar segments and where the height of each
bar is 100%. There must be a clever way to achieve this, but so far, I haven't
found it.
The table which is the result from a CTABLE
sequence:
(%) A
B C
Total
1
11 23 66
100%
2 23 24 43
100%
Put horizontally, the graph would look like the
following:
1
___|_____|____________|
2
_____|______|_________|
0-----------------------------------100%
This should be a straightforward thing, shouldn't it?
Suggestions are more than welcome.
Robert
Robert Lundqvist
|
Re: Using gender as an outcome variable in logistic regression
|
|
In reply to this post by Ryan
I've carried out a logistic regression with gender as the dependent
variable once. Even though gender can't be a dependent variable in the strict sense (except in some rare examples, such as the one Ryan gave), there may be reasons to use it in a logistic regression as a dependent variable. For example, suppose you want to know whether men and women differ on a large number of background variables such as income, age, education etc. One could carry out a large number of t-tests for the continuous variables and chi-square tests for the categorical ones, to compare men and women on each of these variables. However, in this way you don't correct for the relations among the background variables. Moreover, since you don't perform one overall test, you don't correct for capitalization of chance either. Thus, it may be argued that a logistic regression is most appropriate here, even though gender can't be a dependent variable in the strict sense. Besides, how bad is it that gender can't really be a dependent variable? The statistical technique itself doesn't care about the content of the variables, it just computes associations and corrects these associations for the associations among the background variables. The bottom line is that I think it would not necessarily be a problem to do a logistic regression using gender as a dependent variable. Best regards, Joost van Ginkel Joost R. Van Ginkel, PhD Leiden University Faculty of Social and Behavioural Sciences Data Theory Group PO Box 9555 2300 RB Leiden The Netherlands Tel: +31-(0)71-527 3620 Fax: +31-(0)71-527 1721 -----Original Message----- From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of rblack Sent: 15 March 2010 14:41 To: [hidden email] Subject: Re: Using gender as an outcome variable in logistic regression Joesy, I suppose there could be situations where using gender as the dependent variable in a binary logistic regression analysis may be appropriate (i.e. using a set of variables to predict the gender of a child). For a more precise response, you'll need to give more details regarding your research question(s). Ryan Java Joe wrote: > > Hi, > > I was wondering if it's acceptable to use gender as an outcome > variable in a logistic regression since it can be treated as a > dichotomous variable? > > Thanks very much for your patience and help. I am thankful for both. > > Joesy > > -- View this message in context: http://old.nabble.com/Using-gender-as-an-outcome-variable-in-logistic-re gression-tp27903356p27904425.html Sent from the SPSSX Discussion mailing list archive at 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 ********************************************************************** This email and any files transmitted with it are confidential and intended solely for the use of the individual or entity to whom they are addressed. If you have received this email in error please notify the system manager. ********************************************************************** ===================== 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 |
Re: Using gender as an outcome variable in logistic regression
Administrator
|
Frank Harrell describes exactly this use of logistic regression in his book on regression models. Here's the relevant passage:
--- Start of excerpt (p. 227) --- One unique design that is interesting to consider in the context of logistic models is a simultaneous comparison of multiple factors between two groups. Suppose, for example, that in a randomized trial with two treatments one wished to test whether any of 10 baseline characteristics are maldistributed between the two groups. If the 10 factors are continuous, one could perform a two-sample Wilcoxon-Mann-Whitney test or a t-test for each factor (if each is normally distributed). However, this procedure would result in multiple comparison problems and would also not be able to detect a combined effect of small differences across all the factors. A better procedure would be a multivariate test. The Hotelling T^2 test is designed for just this situation. It is a k-variable extension of the one-variable unpaired t-test. The T^2 test, like discriminant analysis, does assume multivariate normality of the k factors. This assumption is especially tenuous when some of the factors are polytomous. A better alternative is the global test of no regression from the logistic model. This test is valid because it can be shown that H0 : mean X is the same for both groups (= H0 : mean X does not depend on group = H0 : mean X | group = constant) is true if and only if H0 : Prob{group | X} = constant. Thus, k factors can be tested simultaneously for differences between the two groups using the binary logistic model, which has far fewer assumptions than does the Hotelling T^2 test. The logistic global test of no regression (with k d.f.) would be expected to have greater power if there is nonnormality. Since the logistic model makes no assumption regarding the distribution of the descriptor variables, it can easily test for simultaneous group differences involving a mixture of continuous, binary, and nominal variables. In observational studies, such models for treatment received or exposure (propensity score models) hold great promise for adjusting for confounding. [several references given] --- End of excerpt --- Harrell, Frank E. Regression Modeling Strategies, with Applications to Linear Models, Survival Analysis and Logistic Regression. Springer, 2001. ISBN 0-387-95232-2.
--
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/). |
Fwd: Using gender as an outcome variable in logistic regression
|
|
In reply to this post by Joost van Ginkel
I am asking if there is a way to read some variables from a file,
then write them in a different order.
I mean re arrange the order of the variables in the file in different order than it is existed.
regards
ahmed
---------- Forwarded message ---------- From: Ginkel, Joost van <[hidden email]> Date: Mon, Mar 15, 2010 at 4:21 PM Subject: Re: Using gender as an outcome variable in logistic regression To: [hidden email] I've carried out a logistic regression with gender as the dependent variable once. Even though gender can't be a dependent variable in the strict sense (except in some rare examples, such as the one Ryan gave), there may be reasons to use it in a logistic regression as a dependent variable. For example, suppose you want to know whether men and women differ on a large number of background variables such as income, age, education etc. One could carry out a large number of t-tests for the continuous variables and chi-square tests for the categorical ones, to compare men and women on each of these variables. However, in this way you don't correct for the relations among the background variables. Moreover, since you don't perform one overall test, you don't correct for capitalization of chance either. Thus, it may be argued that a logistic regression is most appropriate here, even though gender can't be a dependent variable in the strict sense. Besides, how bad is it that gender can't really be a dependent variable? The statistical technique itself doesn't care about the content of the variables, it just computes associations and corrects these associations for the associations among the background variables. The bottom line is that I think it would not necessarily be a problem to do a logistic regression using gender as a dependent variable. Best regards, Joost van Ginkel Joost R. Van Ginkel, PhD Leiden University Faculty of Social and Behavioural Sciences Data Theory Group PO Box 9555 2300 RB Leiden The Netherlands Tel: +31-(0)71-527 3620 Fax: +31-(0)71-527 1721 -----Original Message----- From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of rblack Sent: 15 March 2010 14:41 To: [hidden email] Subject: Re: Using gender as an outcome variable in logistic regression Joesy, I suppose there could be situations where using gender as the dependent variable in a binary logistic regression analysis may be appropriate (i.e. using a set of variables to predict the gender of a child). For a more precise response, you'll need to give more details regarding your research question(s). Ryan Java Joe wrote: > > Hi, > > I was wondering if it's acceptable to use gender as an outcome > variable in a logistic regression since it can be treated as a > dichotomous variable? > > Thanks very much for your patience and help. I am thankful for both. > > Joesy > > -- View this message in context: http://old.nabble.com/Using-gender-as-an-outcome-variable-in-logistic-re gression-tp27903356p27904425.html Sent from the SPSSX Discussion mailing list archive at 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 This email and any files transmitted with it are confidential and intended solely for the use of the individual or entity to whom they are addressed. If you have received this email in error please notify the system manager. ********************************************************************** ===================== 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 -- Magda Abdel Hamid PHD Demographer |
Re: Using gender as an outcome variable in logistic regression
|
|
In reply to this post by Bruce Weaver
Ok, well it's nice to see that my thought was correct.
Joost van Ginkel Joost R. Van Ginkel, PhD Leiden University Faculty of Social and Behavioural Sciences Data Theory Group PO Box 9555 2300 RB Leiden The Netherlands Tel: +31-(0)71-527 3620 Fax: +31-(0)71-527 1721 -----Original Message----- From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Bruce Weaver Sent: 15 March 2010 15:37 To: [hidden email] Subject: Re: Using gender as an outcome variable in logistic regression Frank Harrell describes exactly this use of logistic regression in his book on regression models. Here's the relevant passage: --- Start of excerpt (p. 227) --- One unique design that is interesting to consider in the context of logistic models is a simultaneous comparison of multiple factors between two groups. Suppose, for example, that in a randomized trial with two treatments one wished to test whether any of 10 baseline characteristics are maldistributed between the two groups. If the 10 factors are continuous, one could perform a two-sample Wilcoxon-Mann-Whitney test or a t-test for each factor (if each is normally distributed). However, this procedure would result in multiple comparison problems and would also not be able to detect a combined effect of small differences across all the factors. A better procedure would be a multivariate test. The Hotelling T^2 test is designed for just this situation. It is a k-variable extension of the one-variable unpaired t-test. The T^2 test, like discriminant analysis, does assume multivariate normality of the k factors. This assumption is especially tenuous when some of the factors are polytomous. A better alternative is the global test of no regression from the logistic model. This test is valid because it can be shown that H0 : mean X is the same for both groups (= H0 : mean X does not depend on group = H0 : mean X | group = constant) is true if and only if H0 : Prob{group | X} = constant. Thus, k factors can be tested simultaneously for differences between the two groups using the binary logistic model, which has far fewer assumptions than does the Hotelling T^2 test. The logistic global test of no regression (with k d.f.) would be expected to have greater power if there is nonnormality. Since the logistic model makes no assumption regarding the distribution of the descriptor variables, it can easily test for simultaneous group differences involving a mixture of continuous, binary, and nominal variables. In observational studies, such models for treatment received or exposure (propensity score models) hold great promise for adjusting for confounding. [several references given] --- End of excerpt --- Harrell, Frank E. Regression Modeling Strategies, with Applications to Linear Models, Survival Analysis and Logistic Regression. Springer, 2001. ISBN 0-387-95232-2. Ginkel, Joost van wrote: > > I've carried out a logistic regression with gender as the dependent > variable once. Even though gender can't be a dependent variable in the > strict sense (except in some rare examples, such as the one Ryan > gave), there may be reasons to use it in a logistic regression as a > dependent variable. For example, suppose you want to know whether men > and women differ on a large number of background variables such as > income, age, education etc. One could carry out a large number of > t-tests for the continuous variables and chi-square tests for the > categorical ones, to compare men and women on each of these variables. > However, in this way you don't correct for the relations among the background variables. > Moreover, since you don't perform one overall test, you don't correct > for capitalization of chance either. Thus, it may be argued that a > logistic regression is most appropriate here, even though gender can't > be a dependent variable in the strict sense. Besides, how bad is it > that gender can't really be a dependent variable? The statistical > technique itself doesn't care about the content of the variables, it > just computes associations and corrects these associations for the > associations among the background variables. The bottom line is that I > think it would not necessarily be a problem to do a logistic > regression using gender as a dependent variable. > > Best regards, > > Joost van Ginkel > > > Joost R. Van Ginkel, PhD > Leiden University > Faculty of Social and Behavioural Sciences Data Theory Group PO Box > 9555 2300 RB Leiden The Netherlands > Tel: +31-(0)71-527 3620 > Fax: +31-(0)71-527 1721 > > > -----Original Message----- > From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf > Of rblack > Sent: 15 March 2010 14:41 > To: [hidden email] > Subject: Re: Using gender as an outcome variable in logistic > regression > > Joesy, > > I suppose there could be situations where using gender as the > dependent variable in a binary logistic regression analysis may be > appropriate (i.e. > using a set of variables to predict the gender of a child). For a more > precise response, you'll need to give more details regarding your > research question(s). > > Ryan > > > Java Joe wrote: >> >> Hi, >> >> I was wondering if it's acceptable to use gender as an outcome >> variable in a logistic regression since it can be treated as a >> dichotomous variable? >> >> Thanks very much for your patience and help. I am thankful for both. >> >> Joesy >> >> > > -- > View this message in context: > http://old.nabble.com/Using-gender-as-an-outcome-variable-in-logistic- > re > gression-tp27903356p27904425.html > Sent from the SPSSX Discussion mailing list archive at 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 > > ********************************************************************** > This email and any files transmitted with it are confidential and > intended solely for the use of the individual or entity to whom they > are addressed. If you have received this email in error please notify > the system manager. > ********************************************************************** > > ===================== > 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 [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. -- View this message in context: http://old.nabble.com/Using-gender-as-an-outcome-variable-in-logistic-re gression-tp27903356p27905044.html Sent from the SPSSX Discussion mailing list archive at 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 |
Re: Using gender as an outcome variable in logistic regression
|
|
In reply to this post by Joost van Ginkel
Taxonomists and evolutionary biologists often do this sort of thing.
Many species of animals with 2 sexes can't be distinguished except by dissection or genetic testing. Think of something like larval insects, where no sex organs or external sex characteristics are shown in this immature stage. A suite of morphometric measures and a PCA or DFA is used to predict -- usually very accurately -- the sex of individuals. It's certainly a useful predictive tool, even if sex isn't actually a dependent variable in the classic definition. regards, Ian Ian D. Martin, Ph.D. Aquatic Ecologist On 15 Mar, 2010, at 10:21 AM, Ginkel, Joost van wrote: > I've carried out a logistic regression with gender as the dependent > variable once. Even though gender can't be a dependent variable in the > strict sense (except in some rare examples, such as the one Ryan > gave), > there may be reasons to use it in a logistic regression as a dependent > variable. For example, suppose you want to know whether men and women > differ on a large number of background variables such as income, age, > education etc. One could carry out a large number of t-tests for the > continuous variables and chi-square tests for the categorical ones, to > compare men and women on each of these variables. However, in this way > you don't correct for the relations among the background variables. > Moreover, since you don't perform one overall test, you don't correct > for capitalization of chance either. Thus, it may be argued that a > logistic regression is most appropriate here, even though gender can't > be a dependent variable in the strict sense. Besides, how bad is it > that > gender can't really be a dependent variable? The statistical technique > itself doesn't care about the content of the variables, it just > computes > associations and corrects these associations for the associations > among > the background variables. The bottom line is that I think it would not > necessarily be a problem to do a logistic regression using gender as a > dependent variable. > > Best regards, > > Joost van Ginkel > > > Joost R. Van Ginkel, PhD > Leiden University > Faculty of Social and Behavioural Sciences > Data Theory Group > PO Box 9555 > 2300 RB Leiden > The Netherlands > Tel: +31-(0)71-527 3620 > Fax: +31-(0)71-527 1721 > > > -----Original Message----- > From: SPSSX(r) Discussion [mailto:[hidden email]] On > Behalf Of > rblack > Sent: 15 March 2010 14:41 > To: [hidden email] > Subject: Re: Using gender as an outcome variable in logistic > regression > > Joesy, > > I suppose there could be situations where using gender as the > dependent > variable in a binary logistic regression analysis may be appropriate > (i.e. > using a set of variables to predict the gender of a child). For a more > precise response, you'll need to give more details regarding your > research question(s). > > Ryan > > > Java Joe wrote: >> >> Hi, >> >> I was wondering if it's acceptable to use gender as an outcome >> variable in a logistic regression since it can be treated as a >> dichotomous variable? >> >> Thanks very much for your patience and help. I am thankful for both. >> >> Joesy >> >> > > -- > View this message in context: > http://old.nabble.com/Using-gender-as-an-outcome-variable-in- > logistic-re > gression-tp27903356p27904425.html > Sent from the SPSSX Discussion mailing list archive at 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 > > ********************************************************************** > This email and any files transmitted with it are confidential and > intended solely for the use of the individual or entity to whom they > are addressed. If you have received this email in error please notify > the system manager. > ********************************************************************** > > ===================== > 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 |
Re: Stacked percentages in bar chart?
|
|
In reply to this post by Robert L
Are the values under A and B percentages? They don’t add to 100. Row 1 adds to 100 but Row 2 adds to 101. Have you tried to chart this data directly from an activated table? From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Robert Lundqvist I have got this material where one way to represent the data is an ordinary cross table with values expressed in row-wise percentages. The goal is to get a graph with stacked bar segments and where the height of each bar is 100%. There must be a clever way to achieve this, but so far, I haven't found it. The table which is the result from a CTABLE sequence: (%) A B C Total 1 11 23 66 100% 2 23 24 43 100% Put horizontally, the graph would look like the following: 1 ___|_____|____________| 2 _____|______|_________| 0-----------------------------------100% This should be a straightforward thing, shouldn't it? Suggestions are more than welcome. Robert |
Re: Using gender as an outcome variable in logistic regression
|
|
In reply to this post by magda abdel hamid
Magda,
I'm not quite sure what you want to do exactly but here are some possibilities. 1) if you want to read the file and save it in different order, you can physically re-arrange the columns around in the data window before saving 2) you can use the Keep subcommand of the Save command to specify an order (see the docmentation). 3) if you are writing the file back out as a text file, you can specify an order on the Write command (again, see the documentation). There are probably other methods also; others may comment. Gene Maguin >>I am asking if there is a way to read some variables from a file, then write them in a different order. I mean re arrange the order of the variables in the file in different order than it is existed. regards ahmed ===================== 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 Ryan
Ryan,
Thanks for the help. I have a limited data set which has some information concerning technology and science courses and wanted to see how female/male students fare.
Joesy
---------- Forwarded message ---------- From: rblack <[hidden email]> Date: Mon, Mar 15, 2010 at 3:40 AM Subject: Re: Using gender as an outcome variable in logistic regression To: [hidden email] Joesy, I suppose there could be situations where using gender as the dependent variable in a binary logistic regression analysis may be appropriate (i.e. using a set of variables to predict the gender of a child). For a more precise response, you'll need to give more details regarding your research question(s). Ryan Java Joe wrote: > > Hi, > > I was wondering if it's acceptable to use gender as an outcome variable in > a > logistic regression since it can be treated as a dichotomous variable? > > Thanks very much for your patience and help. I am thankful for both. > > Joesy > > View this message in context: http://old.nabble.com/Using-gender-as-an-outcome-variable-in-logistic-regression-tp27903356p27904425.html Sent from the SPSSX Discussion mailing list archive at 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 Joost van Ginkel
Thanks for the thoughtful advice! It's reassuring to know treating gender as an outcome variable is possible and not problematic. ~joesy
On Mon, Mar 15, 2010 at 4:21 AM, Ginkel, Joost van <[hidden email]> wrote: I've carried out a logistic regression with gender as the dependent |
|
|
In reply to this post by Java Joe
Thanks, Ian. This is good to know, particularly for locating empirical research employing the gender/DV approach. ~joesy
Taxonomists and evolutionary biologists often do this sort of thing. Many species of animals with 2 sexes can't be distinguished except by dissection or genetic testing. Think of something like larval insects, where no sex organs or external sex characteristics are shown in this immature stage. A suite of morphometric measures and a PCA or DFA is used to predict -- usually very accurately -- the sex of individuals.
It's certainly a useful predictive tool, even if sex isn't actually a dependent variable in the classic definition.
regards, Ian
Ian D. Martin, Ph.D. Aquatic Ecologist
|
Re: Using gender as an outcome variable in logistic regression
|
Administrator
|
In reply to this post by Maguin, Eugene
To add to what Gene said, /KEEP can be used to specify the order of the variables when opening a file too. Read the fine manual to see examples -- look for "KEEP (subcommand)". And if you have v16 or later, look up SORT VARIABLES while you're at it.
--
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/). |
Re: Using gender as an outcome variable in logistic regression
|
|
In reply to this post by Java Joe
GIven your interests in understanding the effect of a student's sex on grades, it would seem more sensible to use this dichotomous variable as an independent variable predicting grades, which you could do using multinomial or ordinal logistic regression, with other student characteristics added to the list of predictors if you have the data. David Greenberg, Sociology Department, New York University.
----- Original Message ----- From: Java Joe <[hidden email]> Date: Monday, March 15, 2010 3:21 pm Subject: Re: Using gender as an outcome variable in logistic regression To: [hidden email] > Ryan, > > Thanks for the help. I have a limited data set which has some information > concerning technology and science courses and wanted to see how female/male > students fare. > > Joesy > > ---------- Forwarded message ---------- > From: rblack <[hidden email]> > Date: Mon, Mar 15, 2010 at 3:40 AM > Subject: Re: Using gender as an outcome variable in logistic regression > To: [hidden email] > > > Joesy, > > I suppose there could be situations where using gender as the dependent > variable in a binary logistic regression analysis may be appropriate (i.e. > using a set of variables to predict the gender of a child). For a more > precise response, you'll need to give more details regarding your research > question(s). > > Ryan > > > Java Joe wrote: > > > > Hi, > > > > I was wondering if it's acceptable to use gender as an outcome > variable in > > a > > logistic regression since it can be treated as a dichotomous variable? > > > > Thanks very much for your patience and help. I am thankful for both. > > > > Joesy > > > > > > -- > View this message in context: > http://old.nabble.com/Using-gender-as-an-outcome-variable-in-logistic-regression-tp27903356p27904425.html > Sent from the SPSSX Discussion mailing list archive at 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 |
|
|
In reply to this post by Java Joe
Thank you for the suggestion. I'm hampered by my limited data set which includes only course numbers/titles (no grades, GPA, or exam scores), the number of courses taken, and demographics. So I thought maybe using the course numbers (recoded to reflect level of difficulty or course progression) might predict gender (in this case, females). ~joesy
-----Original Message-----
GIven your interests in understanding the effect of a student's sex on grades, it would seem more sensible to use this dichotomous variable as an independent variable predicting grades, which you could do using multinomial or ordinal logistic regression, with other student characteristics added to the list of predictors if you have the data. David Greenberg, Sociology Department, New York University.
----- Original Message ----- From: Java Joe <[hidden email]> Date: Monday, March 15, 2010 3:21 pm Subject: Re: Using gender as an outcome variable in logistic regression To: [hidden email]
> Ryan, > > Thanks for the help. I have a limited data set which has some information > concerning technology and science courses and wanted to see how female/male > students fare. > > Joesy > > ---------- Forwarded message ---------- > From: rblack <[hidden email]> > Date: Mon, Mar 15, 2010 at 3:40 AM > Subject: Re: Using gender as an outcome variable in logistic regression > To: [hidden email] > > > Joesy, > > I suppose there could be situations where using gender as the dependent > variable in a binary logistic regression analysis may be appropriate (i.e. > using a set of variables to predict the gender of a child). For a more > precise response, you'll need to give more details regarding your research > question(s). > > Ryan > > > Java Joe wrote: > > > > Hi, > > > > I was wondering if it's acceptable to use gender as an outcome > variable in > > a > > logistic regression since it can be treated as a dichotomous variable? > > > > Thanks very much for your patience and help. I am thankful for both. > > > > Joesy > > > > > > -- > View this message in context: > Sent from the SPSSX Discussion mailing list archive at 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 |
Re: Using gender as an outcome variable in logistic regression
|
|
Just to add a few philosophical points to the discussion, remember those
urn problems that you had way back when in your probability classes. Those balls were either black or white and but when you draw one randomly from an urn, it has a probability distribution. Similarly, men and women have a probability distribution when drawn randomly from an urn. So if you draw a person randomly from the urn labelled "does not like to ask for directions when lost" the probability that that person is male is approximately 98%. Since gender can have a probability distribution, it can be modeled using tools like logistic regression. Now if you think in terms of cause and effect, not wanting to ask for directions does not cause you to turn into a male, but that is a problem with thinking of independent variables as causes and depdendent variables as effects. If you wanted to you could use the urn model to reverse the time arrow. Draw one ball and throw it away. Then draw ten balls without replacement and note their colors. The data from these ten balls can be used to make inferences about the first ball, even though they occurred in time after the first ball. So go ahead and model gender (or race or any other immutable characteristic) as a dependent variable. It doesn't lead to a logical contradiction as long as you discard the idea of independent variables "causing" a dependent outcome. -- Steve Simon, Standard Disclaimer Sign up for The Monthly Mean, the newsletter that dares to call itself "average" at www.pmean.com/news ===================== 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 Java Joe
Joesy,
I agree with others that it is permissable to run a logistic regression analysis for similar purposes to running a discriminant function analysis. Your research design/question(s) are still a bit nebulous to me, so forgive me if my comments are off base. I just don't know enough about the courses at this school to state one way or the other if course numbers are a reasonable proxy to level of difficulty or course progression. And what do you plan on doing with the other demographic variables? Do you plan on treating those as independent variables? If yes, which ones and why? Simply adding demographic variables as independent variables may not be the correct approach. I would imagine that year in school would play an important role (i.e. freshman, sophmore, junior or senior). A freshman taking college algebra is quite different than a senior taking college algebra, for instance. I also wonder if there is some sort of nesting that one might need to consider, if you even have such information to account for nesting. Unfortunately, SPSS does not currently offer a procedure to handle mutilevel logistic regression. Anyway, these are just some of the questions/issues I'd consider if I were you. HTH, Ryan
|
|
|
In reply to this post by Steve Simon, P.Mean Consulting
Thank you very much for your explanation which helps my perspective on using gender as a DV. My male friends especially appreciate your funny (and true) cause-and-effect example that "not wanting to ask for directions does not cause you to turn into a male". ;) ~joesy
On Tue, Mar 16, 2010 at 4:43 AM, Steve Simon, P.Mean Consulting <[hidden email]> wrote: Just to add a few philosophical points to the discussion, remember those urn problems that you had way back when in your probability classes. Those balls were either black or white and but when you draw one randomly from an urn, it has a probability distribution. Similarly, men and women have a probability distribution when drawn randomly from an urn. So if you draw a person randomly from the urn labelled "does not like to ask for directions when lost" the probability that that person is male is approximately 98%. Since gender can have a probability distribution, it can be modeled using tools like logistic regression. |
«
Return to SPSSX Discussion
|
1 view|%1 views
| Free forum by Nabble | Edit this page |