# Which of these descriptions match one-way ANOVA, factorial ANOVA, logistic regression (from a tutorial)

I've tried about 10 combinations of this, and none of them are correct. This is from an online tutorial hosted by a university. This is not graded. The interactive graphic associated with this only says whether your entire selection of choices is correct or not. (So I don't know what's wrong.)

For 1-6, match with A-G. A-G can only be used once.

(I reordered the items in graphic so they align with my answers, I didn't use G.
I have 1: A, 2: B, 3: C, 4: D, 5: E, 6: F.)

QUESTIONS

1. Three-level factor moderated by two-level factor, continuous target.
2. Three-level factor, interval target.
3. Continuous independent variable, binary target.
4. Two independent variables which are interval, continuous target.
5. There is curvilinear relationship between two continuous variables.
6. There is an interaction effect between two two-level factors, continuous target.

A. Factorial ANOVA, with post-hocs.
B. One-way ANOVA, no post-hocs.
C. Logistic regression.
D. Linear regression.
E. None of the choices.
F. Factorial ANOVA, without post-hocs.
G. One-way ANOVA, with post-hocs.

• Welcome to this site! Please add the self-study tag, and tell us if you understand all models placed on the right panel. – chl Oct 27 '20 at 17:44
• I thought I knew what all the models on the right were for... I added the self-study tag. – JohnT Oct 27 '20 at 17:50
• Then, start by isolating: (1) the type of the response variable, (2) the type and number of predictors, and ask yourself (3) how those predictors are involved in the design (i.e, are subjects exposed to different levels of a single variable or to a combination of levels from multiple variables), and (4) are we interested in a global (omnibus) test or in specific comparisons (between levels of one or more predictors), before seeing the data or afterwards. – chl Oct 27 '20 at 17:54
• John, update your questions with the above propositions, aand delete your previous comments, this will help cleaning up the comment thread (comments are not for extended discussion anyway), and people will be able to answer or comment directly by referring to those answers. – chl Oct 27 '20 at 18:48
• Thanks! Let's then hope that someone will come in handy and not add another vote to close, after all your efforts to clarify your question. – chl Oct 27 '20 at 20:37

The Independent variable is continuous, and the target is binary. Logistic Regression. Correct?
Two interval predictors, one interval target. Linear regression. Correct?

Yes, I agree with you here.

Advertising platform is a factor. Intent is an interval target. One factor= one-way ANOVA. Correct?

Given you cannot use linear regression again by the conditions of the puzzle this is correct too.

Two factors, gender and level. Continuous target (salary). Factorial ANOVA. Is there an interaction between the two factors? Factorial ANOVA. (The R output directly shows if the interaction is significant - no post-hocs required.) You run a single command in R. Correct? anova(salary ~ gender * level, data=MyData)

Yes, no post hoc testing will be needed since the interaction is estimated separately.

Consumer confidence and gender are factors. Spending is the continuous target. Two factors = factorial ANOVA. Correct?

Since they are asking about how the effect is moderated by gender they will be thinking you need post hoc testing. I am not convinced this is the best approach but of the options left this is clearly the best.

Curvilinear relationship. None of the choices. Correct?

This is the one I disagree with depending on what they mean by curvilinear relationship. It is perfectly possible to do this with a linear regression by transforming the predictor. However None of the choices is the only one available.

• This says the answers I gave that you refer to are correct. There are more questions. I have moved on because I spent to much time trying options, and I have other things to do. I still want to know what the systems wants. I'll check back here periodically. – JohnT Oct 28 '20 at 20:41