This might be useful "Some researchers treat variables measured with Likert scales (e.g., with labels such as 1 = strongly disagree, 2 = disagree, 3 = neutral, 4 = agree, and 5 = strongly agree) as interval variables. However, treating Likert scale responses as interval data carries the assumption that the differences between points on the scale are all equal. That is to say, using the 5-point Likert scale as an interval scale assumes that the difference between strongly agree and agree is the same relative difference as between neutral and agree. This is often not a safe assumption to make, so Likert scale responses are usually better off treated as ordinal." ( https://www.statisticssolutions.com/data-levels-and-measurement/).

I've found this useful for selecting the statistical test, maybe it has what you need?

https://stats.idre.ucla.edu/other/mult-pkg/whatstat/

I am unfamiliar with the instruments you are using, such as how many options on the Likert scales, how the options are anchored, how many questions, whether the instruments have internal scales and if so how they were constructed and validated. These are all material issues. As you know, Likert data are ordinal but under well-justified circumstances can be treated as continuous, the most common way being to take the sum or mean of two or more ordinal variables to create an approximately continuous variable -- even here, the choice of methodology can be important.

My questions lie behind yours. Is there a particular reason why you want to give up what might be potentially valuable information by transforming your IVs and your DV into dichotomous data? Is there good measurement science behind doing so or is the basis some form of common sense face "validity?" Instead, I would recommend a design that maximizes degrees of freedom. Much depends on the validity of the instruments you are using as one IV and the DV and also on the precision and reliability which you can apply your classification scheme to create the binary "victim" IV. Is this a clean classification or does it admit of shades of grey in the real world of human behavior? If it is the latter, you might be better off to capture this nuance for analysis.

In sum, it could make sense to re-examine your design before settling on a statistic. I hope this is helpful. One final thought: these questions and their answers are pretty fundamental. Do you not have a measurement scientist on you committee or as an available advisor to guide you in these matters? Your university should make available to you someone having a Ph.D. in methodology, design, and statistics. It would also be helpful if they had training in instrument development and the basic techniques of construct validation, item analysis, etc. I have advised my own doctoral students in these matters but doing so requires much more information about your research question, etc.

for your research before determining the scale it is better to have clear direction and purpose of the research to ensure the variables and for the variables themselves must be clear indicators and dimensions to make a statement so that we can use the Likert scale method. The dependent variable is divided into 2 levels, avoidance and approach. You should treat this as two separate levels or 2 DV.