The purpose of our study is to compare our results of minimum dietary diversity for women during 2 seasons: spring and summer.
We want to check if the number of women who achieve dietary diversity is different between spring and summer and if the number of women who do not achieve minimum dietary diversity is different between spring and summer.
We want to create a table like this:
Minimum dietary diversity | spring | summer | p-value |
No | ......% | .....% |
|
Yes | ......% | .....% |
We used a proc logistic to take into account the confounding variables:
proc logistic data = results2;
Class Land_acces (param=ref ref='No')
region (param=ref ref='1-South')
v_urban_rural (param=ref ref='rural')
season (param=ref ref='spring')
model MDD_W (event='Yes')=
Land_acces
region
season;
v_urban_rural
Land_acces * season
region * season
v_urban_rural * season
format season saison. ;
run;
This proc logistic gave us just one p-value for both seasons!!
How can I obtain the p-value for women who achieve minimum dietary diversity and women who do not while taking into account the confounding variables?
I don't understand. You asked about season, which is not a dependent variable in the model you show.
LSMEANS always gives you the mean value of the DEPENDENT variable, split by one (or more) of the independent variables.
Contrasts are never needed when you are comparing a level of a variable to another level of the same variable.
I answered this the other time you asked.
https://communities.sas.com/t5/SAS-Statistical-Procedures/Proc-logistic-p-value/m-p/450253
Yes, I know and it was really helpful Thank you. But last time I asked about the independent variables and this time I wanted to do the same thing for the dependent variable. For the other question I used contrast instead of LSMEANS to obtain my results and it worked but not for the dependant variable in both seasons!!
I used this Code :
proc logistic data = results2;
Class
Land_acces (param=ref ref='Non')
region (param=ref ref='1-Sud')
v_urban_rural (param=ref ref='rural')
season (param=ref ref='spring');
model MDD_W(event='Yes')=
Land_acces
region
v_urban_rural
Land_acces * season
region * season
v_urban_rural * season
season/ expb;
contrast 'region = 1-Sud season summer vs spring' region 0 0 season 1 / estimate = exp ;
contrast 'region = 2-GAE season summer vs spring' region 1 0 season 1 / estimate = exp;
contrast 'region = 3-GAO season summer vs spring' region 0 1 season 1 / estimate = exp;
contrast 'v_urban_rural = rural season summer vs spring' v_urban_rural 0 season 1 / estimate = exp;
contrast 'v_urban_rural = urban season summer vs spring' v_urban_rural 1 season 1 / estimate = exp;
contrast 'Land_acces = Non season summer vs spring' Land_acces 0 season 1 / estimate = exp;
contrast 'Land_acces = Wi season summer vs spring' Land_acces 1 season 1 / estimate = exp;
contrast 'summer v spring' season 0 / estimate = exp;
contrast 'summer v spring' season 1 / estimate = exp;
format season saison.;
TITLE'COMP MDD_W PRINTEMPS ÉTÉ';
run;
I tried to add the code in red to answer my question but it didn't seem to work !!
Will it work with LSMEAN for the dependent variable ???
I don't understand. You asked about season, which is not a dependent variable in the model you show.
LSMEANS always gives you the mean value of the DEPENDENT variable, split by one (or more) of the independent variables.
Contrasts are never needed when you are comparing a level of a variable to another level of the same variable.
Thank you so much for all your answers. I just need to confirm one last thing if you don't mind...
As I said in my post: https://communities.sas.com/t5/Base-SAS-Programming/Proc-logistic-p-value/td-p/452359
I want to create a table like this: So I thought I need 2 p-values to compare people from the 2 rows: women who achieved dietary diversity during both seasons and women who didn't during both seasons.
Minimum dietary diversity | spring | summer | p-value |
No | 76.79% | 79.03% | ....... |
Yes | 23.21% | 20.97% | ...... |
By using LSMEANS I got only one p-value for the variable season (spring/summer) so my question is what can I do to create the table I want?
this is the code I used :
proc logistic data = results2;
Class
terres_acces (param=ref ref='Non')
region (param=ref ref='1-Sud')
v_urban_rural (param=ref ref='rural')
season (param=ref ref='spring');
model MDD_W_5(event='Oui')=
terres_acces
region
season
v_urban_rural
terres_acces * season
region * season
v_urban_rural * season/ expb;
lsmeans terres_acces * season
region * season
v_urban_rural * season / lines;
format season saison.;
run;
@sebai wrote:
I want to create a table like this: So I thought I need 2 p-values to compare people from the 2 rows: women who achieved dietary diversity during both seasons and women who didn't during both seasons.
Minimum dietary diversity
spring
summer
p-value
No
76.79%
79.03%
.......
Yes
23.21%
20.97%
......
Again, this isn't making sense. What model term is: "women who achieved dietary diversity during both seasons and women who didn't during both seasons". What term in the model tells us if the women achieved dietary diversity during both seasons or not??? Specifically, what numbers in the table shown do you want to compare?
I want to compare:
Minimum dietary diversity | spring | summer | p-value |
No | 76.79% | 79.03% | ……….. |
And then do another comparison:
Minimum dietary diversity | spring | summer | p-value |
Yes | 23.21% | 20.97% |
|
The minimum dietary diversity (MDD) is another variable that I created in my programme:
MDD =
MDD_W_1_grains_root_tuber +
MDD_W_2_pulses +
MDD_W_3_nuts_seeds +
MDD_W_4_dairy +
MDD_W_5_meat_fish +
MDD_W_6_eggs +
MDD_W_7_leafy_veg +
MDD_W_8_vit_a_veg_fruit +
MDD_W_9_other_veg +
MDD_W_10_other_fruit;
IF MDD < 5 THEN MDD_W_5 = "No";
IF MDD GE 5 THEN MDD_W_5 = "Yes";
@sebai wrote:
I want to compare:
- Minimum dietary diversity=NO in spring and Minimum dietary diversity=NO in summer (compare 76% and 79%)
Minimum dietary diversity
spring
summer
p-value
No
76.79%
79.03%
………..
And then do another comparison:
- Between Minimum dietary diversity=Yes in spring and Minimum dietary diversity=Yes in summer
Minimum dietary diversity
spring
summer
p-value
Yes
23.21%
20.97%
No, you don't want to do this. If you know the values for "Yes", you also know the values for "No". This is only a single comparison.
Since the logistic regression output shows season with a p-value of 0.2984, the difference between 23.21% in spring and 20.97% in summer is not a statistically significant difference.
Are you ready for the spotlight? We're accepting content ideas for SAS Innovate 2025 to be held May 6-9 in Orlando, FL. The call is open until September 25. Read more here about why you should contribute and what is in it for you!
Learn how use the CAT functions in SAS to join values from multiple variables into a single value.
Find more tutorials on the SAS Users YouTube channel.