It looks like you want a multiple regression or similar model. If you just had a categorical predictor (representing two groups) then the t test from the regression is identical to an independent t test (without corrections such as Welch-Satterthwaite if that's relevant). So a t test is just a special case of multiple regression.

I note your response is ordinal so you might also consider ordinal logistic regression.

For the moderator analysis I would not group the moderator scores (as that loses information and hence statistical power). You can add an interaction to the model (a moderator analysis is just a term for an interaction model involving a continuous predictor) by created a product term (multiplying the indicator for text/video by the continuous visual literacy score). Add the product term to the model with both the main effects (learning method and vis literacy) in the model. This gives you a test of the interaction/moderator effect.

For ease of interpretation it may help to code the learning method as 0/1 or -0.5/0.5 and to centre the visual literacy effect (bu subtracting its mean so it has a mean of zero). If you code learning method 0/1 then the intercept is the mean of the group coded 0 and the learning method effect the difference between the groups. For -0.5/0.5 coding the intercept is the grand mean and the and the learning method effect is still the difference between the groups. (This assumes no covariates added or any covariates are also centred). Interpreting an interaction model with an uncensored variable is not straight forward.

Thank you for your help Thom, I really appreciate it!

https://www.researchgate.net/post/Moderation-for-the-independent-sample-t-test

Check https://libguides.library.kent.edu/spss/independentttest

Also check https://statistics.laerd.com/spss-tutorials/dichotomous-moderator-analysis-using-spss-statistics.php