Interpreting sign of the beta cofficient in binary regression

I have questionnaire
Three independent variables and one binary dependent variable(use of x)

A1 (construct)
itemA11(temA11 is positively related to use of x) Likert 1-5
itemA12(itemA12 is positively related to use of x) Likert 1-5
(not exactly the question but gives the meaning)

Hypothesis is "A1 is positively related to use of X".

Similarly, another construct

B1 (construct)
itemB11(lack of itemB11 is negatively related to use of x)
itemB12(lack of itemB12 is negatively related to use of x)

Hypothesis is "lack of B1 is negatively related to use of X".

Now I understand that negative coefficient of A1 means it is inversely related to the dependent variable and if it increases, dependent variable decreases and similar on the increase.

But I am a little bit confused about the negative sign in beta coefficient of B1 as it is already kind of negative statement(lack of B1).
So, can I safely conclude that - sign means when "lack of b1" increases dependent variable decreases and when "lack of B1" decreases dependent increases?
And Can I accept that hypothesis that "lack of B1 is negatively related to use of X", if value is significant and sign is negative?

It makes sense, but kind of double negatives was causing the problem.

And in the final model for a construct, can I put just put B1 (rather than "lack of B1") and say it is negatively related?

Bit confused would be grateful for the help.



Not a robit
Your write-up is difficult to interpret. It would probably be better in this case not to provide so much context language and stick to Xs and Ys.

So, Y = X. Without completely following you, a negative beta in a logistic regression model represents a decrease in the log odds of the outcome when increasing the independent variable by a unit. So for a generic example, eating more bananas (X) associated with lower odds of hypertension (Y). If you need more clarification, try rewriting the question and I would be happy to try and help.

P.S., I don't recall the rules of treating an ordinal/interval (Likert-style variable) as continuous, but I know some people like to get all technical on the appropriateness of this (e.g., are intervals even, is their a neutral value, etc.).