I ran a logistic regression (using R) but for one of my subgroups (by Race), got a warning that the Algorithm does not converge.
When I look at my data, for 800 participants, 798 are one outcome, and 2 are the other for my independent variable.
Here is what the output looks like.
Call:
svyglm(formula = ExposureLevel ~ Chemical1 + Age + Gender + TimeofYearWinterorSummer + SmokingCotinine + PRSlevels,
design = subsetasian,
family = quasibinomial)
Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) 174.785286 9.496548 18.405 < 2e-16 ***
Chemical1 -0.009771 0.462753 -0.021 0.983
Age -10.968026 0.342395 -32.033 < 2e-16 ***
GenderFemale 97.004972 2.501516 38.778 < 2e-16 ***
TimeofYearWinterorSummerWinterSpring -12.444436 0.828608 -15.018 < 2e-16 ***
SmokingCotinineSecondHand 22.924403 1.634055 14.029 < 2e-16 ***
SmokingCotinineSmoker 172.356472 5.313671 32.436 < 2e-16 ***
PRSlevel -76.706172 2.592161 -29.592 < 2e-16 ***
---
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
What conclusions if any can I make from these results? Suddenly all of my variables besides the chemical have a significant pvalue.
I know that Age is NOT significiantly associated with the independent variable in any of my other log regressions. But here it says it is.
