# Comparison of coefficients in two logistic regressions

##### New Member
Hi,
I carried out a logistic model on malnutrition status on a whole underfive children and I found that my variable poverty is significant. But, when I rerun models for each sex of children, I noticed that the poverty variable is not significant in any of the two models. How can I explain this ?

#### hlsmith

##### Less is more. Stay pure. Stay poor.
Was sex in the first model? Please write out your models so we can be sure of what you are describing. You can also post model output.

##### New Member
Here are the models:

xi: svy: logit ta_2 sexenft agenft agenftsq i.imc i.nivvie i.jumellitr i.tailmerc i.anemie i.persassi ptz ptzsq intervpr i.diarrhee i.instrmer i.ethnier i.religion i.anje2 i.grosnair, or

xi: svy: logit ta_2 agenft agenftsq i.imc i.nivvie i.jumellitr i.tailmerc i.anemie i.persassi ptz ptzsq intervpr i.diarrhee i.instrmer i.ethnier i.religion i.anje2 i.grosnair if sexenft==1, or

xi: svy: logit ta_2 agenft agenftsq i.imc i.nivvie i.jumellitr i.tailmerc i.anemie i.persassi ptz ptzsq intervpr i.diarrhee i.instrmer i.ethnier i.religion i.anje2 i.grosnair if sexenft==2, or

The variable sexenft represents the sex of the child and the ta_2 the malnutrition status in two items. As I said, in the first model concerning all children, poverty variable (nivvie) is significant, but in the last two ones (for respectively males and females) , it is not significant. Could it be a matter of low number of cases and how can I explain this?

#### Karabiner

##### TS Contributor
Please describe the numer of cases for each sex, and the coefficients, standard errors, odds ratios and p-values for poverty in all 3 models.

With kind regards

Karabiner

##### New Member
Please find bellow the results of the three models. The first one has been carried on both sexes. The following two concern each sex. My independent variable of interest is nivvie (poverty). I hope that what you are looking for.

Survey: Logistic regression
Number of strata = 26 Number of obs = 2569
Number of PSUs = 538 Population size = 2645.7052
Design df = 512
F( 23, 490) = 11.98
Prob > F = 0.0000

Linearized
ta_2 Odds Ratio Std. Err. t P>t [95% Conf. Interval]

_Isexenft_1 1.302401 .118093 2.91 0.004 1.089884 1.556355
agenft 1.022242 .0031073 7.24 0.000 1.016156 1.028365
_Ianje2_0 1.646547 .4987195 1.65 0.100 .9081221 2.985409
_Ipersassi_2 1.410396 .1923972 2.52 0.012 1.078823 1.843875
_Itailmerc_1 2.500761 .4179843 5.48 0.000 1.800789 3.472814
_Itailmerc_2 1.846668 .2331651 4.86 0.000 1.440984 2.366565
_Iethnier_1 1.829222 .4494636 2.46 0.014 1.128806 2.964239
_Iethnier_3 1.520993 .3200396 1.99 0.047 1.005999 2.299626
agebirth .9853814 .0078892 -1.84 0.066 .9700035 1.001003
_Iimc_1 2.3806 .5315126 3.88 0.000 1.535287 3.691331
_Iimc_2 1.412376 .2867215 1.70 0.090 .9478535 2.10455
intervpr .9940734 .003109 -1.90 0.058 .9879841 1.0002
instrmer .6030546 .0847828 -3.60 0.000 .4575132 .7948948
_Inivvie_1 1.585762 .3234038 2.26 0.024 1.06226 2.367256
_Inivvie_2 1.516249 .298936 2.11 0.035 1.029329 2.233504
_Igrosnair_2 1.324713 .143638 2.59 0.010 1.070552 1.639215
_Igrosnair_3 1.446154 .2222215 2.40 0.017 1.069315 1.955795
_Ijumellitr_1 4.003614 1.157877 4.80 0.000 2.268269 7.066589
_Ianemie_1 2.065912 .3832185 3.91 0.000 1.434971 2.974271
_Ianemie_2 1.563194 .2390432 2.92 0.004 1.157548 2.110993
_Ianemie_4 1.188793 .237346 0.87 0.387 .8030801 1.759762
_Ireligion_1 1.306835 .1529353 2.29 0.023 1.038415 1.644639
_Ireligion_3 1.663168 .2855364 2.96 0.003 1.187008 2.330336
_cons .0195026 .0104757 -7.33 0.000 .0067888 .0560264

Survey: Logistic regression
Number of strata = 26 Number of obs = 1317
Number of PSUs = 475 Population size = 1350.4941
Design df = 449
F( 22, 428) = 7.12
Prob > F = 0.0000
Linearized
ta_2 Odds Ratio Std. Err. t P>t [95% Conf. Interval]
agenft 1.015942 .0041866 3.84 0.000 1.007748 1.024204
_Ianje2_0 2.419378 1.029046 2.08 0.038 1.048768 5.581206
_Ipersassi_2 1.372574 .2432072 1.79 0.075 .9689543 1.944323
_Itailmerc_1 2.400365 .5159462 4.07 0.000 1.573336 3.662126
_Itailmerc_2 2.100481 .3345685 4.66 0.000 1.53593 2.872541
_Iethnier_1 2.582524 .9391103 2.61 0.009 1.263795 5.277307
_Iethnier_3 1.907515 .589694 2.09 0.037 1.038996 3.502045
agebirth .9797989 .0103091 -1.94 0.053 .9597468 1.00027
_Iimc_1 2.129003 .6228981 2.58 0.010 1.198011 3.783482
_Iimc_2 1.392131 .3370991 1.37 0.173 .8649838 2.240538
intervpr .9949808 .0037705 -1.33 0.185 .9875983 1.002418
instrmer .5431931 .1029086 -3.22 0.001 .3743322 .788227
_Inivvie_1 1.430976 .3880431 1.32 0.187 .8398174 2.438258
_Inivvie_2 1.358264 .3575763 1.16 0.245 .8096386 2.278647
_Igrosnair_2 1.193513 .1683143 1.25 0.210 .9046129 1.574678
_Igrosnair_3 1.241296 .2938327 0.91 0.362 .7795416 1.976568
_Ijumellitr_1 4.260104 1.628328 3.79 0.000 2.009965 9.029256
_Ianemie_1 1.974866 .4820662 2.79 0.006 1.222355 3.190643
_Ianemie_2 1.430805 .2900668 1.77 0.078 .9606171 2.131134
_Ianemie_4 1.067733 .3212577 0.22 0.828 .5911017 1.928692
_Ireligion_1 1.431257 .2265805 2.26 0.024 1.048578 1.953595
_Ireligion_3 1.983774 .4546943 2.99 0.003 1.26434 3.112581
_cons .0215881 .0159665 -5.19 0.000 .0050462 .092355

Survey: Logistic regression
Number of strata = 26 Number of obs = 1252
Number of PSUs = 471 Population size = 1295.2111
Design df = 445
F( 22, 424) = 5.23
Prob > F = 0.0000
Linearized
ta_2 Odds Ratio Std. Err. t P>t [95% Conf. Interval]
agenft 1.029439 .0044429 6.72 0.000 1.020744 1.038208
_Ianje2_0 1.113405 .4772801 0.25 0.802 .479486 2.585416
_Ipersassi_2 1.476941 .2961445 1.94 0.052 .9959129 2.190306
_Itailmerc_1 2.602066 .5821472 4.27 0.000 1.676347 4.03899
_Itailmerc_2 1.575727 .275256 2.60 0.010 1.117849 2.221154
_Iethnier_1 1.135574 .4461761 0.32 0.746 .5246379 2.457941
_Iethnier_3 1.092881 .3546218 0.27 0.784 .577589 2.067888
agebirth .9923263 .0124229 -0.62 0.539 .9682094 1.017044
_Iimc_1 2.554923 .8637124 2.77 0.006 1.314732 4.964987
_Iimc_2 1.356324 .3989269 1.04 0.301 .7608904 2.417712
intervpr .9920849 .0051281 -1.54 0.125 .9820577 1.002214
instrmer .6949417 .1425154 -1.77 0.077 .4644211 1.039884
_Inivvie_1 1.967428 .7056009 1.89 0.060 .9722776 3.981139
_Inivvie_2 1.899081 .6526607 1.87 0.063 .9665222 3.73143
_Igrosnair_2 1.493199 .245599 2.44 0.015 1.080766 2.063022
_Igrosnair_3 1.749503 .4115544 2.38 0.018 1.101872 2.777781
_Ijumellitr_1 3.827345 1.38622 3.71 0.000 1.878285 7.798905
_Ianemie_1 2.318173 .6503387 3.00 0.003 1.33567 4.023395
_Ianemie_2 1.794669 .3917804 2.68 0.008 1.16858 2.756198
_Ianemie_4 1.455358 .4373485 1.25 0.212 .8062656 2.627008
_Ireligion_1 1.160847 .1992671 0.87 0.385 .8284418 1.626626
_Ireligion_3 1.345974 .3457285 1.16 0.248 .8124559 2.22984
_cons .022462 .0181951 -4.69 0.000 .0045716 .1103656

#### Karabiner

##### TS Contributor
Please describe coefficients, standard errors, odds ratios and p-values for poverty in all 3 models.

Such a bunch of 360 coefficients of the "1.3459782" type are hard to read.

With kind regards

Karabiner

Last edited:

##### New Member
Find bellow what you requested:

Both sexes:
Number of obs = 2569
Odds Ratio Std. Err. T P>t
Poverty_1 1.585762 .3234038 2.26 0.024
Poverty_2 1.516249 .298936 2.11 0.035

Male:
Number of obs = 1317
Odds Ratio Std. Err. T P>t
Poverty_1 1.430976 .3880431 1.32 0.187
Poverty_2 1.358264 .3575763 1.16 0.245

Female:
Number of obs = 1252
Odds Ratio Std. Err. t P>t
Poverty_1 1.967428 .7056009 1.89 0.060
Poverty_2 1.899081 .6526607 1.87 0.063

I draw your attention on the fact that the independent variable poverty is coded on three items.

##### New Member
Here is the information asked. Hope that it will do. I have put spaces between items with hyphens to make it more readable.

Both sexes:
Number of obs = 2569
-----------------Odds Ratio----Std. Err. -------T----------P>t
Poverty_1-------1.585762------0.3234038-----2.26------0.024
Poverty_2-------1.516249------0.298936------2.11-------0.035

Male:
Number of obs = 1317
----------------Odds Ratio----Std. Err. ------T------P>t
Poverty_1------1.430976-------0.3880431---1.32---0.187
Poverty_2------1.35826--------0.3575763----1.16---0.245

Female:
Number of obs = 1252
-----------------Odds Ratio-------Std. Err. -------t---------P>t
Poverty_1----1.967428--------0.7056009-------1.89-------0.060
Poverty_2----1.899081--------0.6526607-------1.87-------0.063

##### New Member
Dear,

I would be very pleased to receive any details on the way to estimate effect size of independant variable after a logistic regression. If you have a detailed procedure on your own example, it would be interesting to share it.

Regards

#### obh

##### Well-Known Member
Hi Traore, To your last question, you may say the coefficients are measures of the effect sizes of the IVs.
But due to intercorrelations among the IVs, you may get different effect sizes for different models, so part of the effect size may be taken or given from/to other IV