Five or Seven point likert scale is more preferred

One option is to do a set of 3x3 chi-squares, and if a result is significant, then examine the appropriate percentages in the cells to determine the nature of the relationship. By "appropriate" percentages, I mean that you should calculate the percentages within the independent variable, so that they add to 100% in that direction.

You could try the ordinal regression approach. I guess that most people would be reluctant to try out the ordinary (OLS) regression in your three-ordinal-categories-case :-)

I partially agree with Christer Thrane. However, you can apply 2 approaches: (1) you can convert your 3 categories independent variable into 2 categories, if possible, like a dummy variable. Then, apply Binary Logit Model. (2) Using 3 categories case, you can apply Ordered Logit Model. Then, you can compare both models. Using STATA, you can do this very easily.

For example, 1 = No cleanliness, 2 = Moderately clean, 3 = Very clean. Now, this can be converted into a dummy variable, e.g. 1 = Cleanliness and 0 = No cleanliness. The same can be done for privacy and comfort variables, if possible. Then, apply a Binary Logit model.

As I have seen above, many of the experts shared their knowledge. That is very good for interchanging knowledge.

Yes, you can use an ordinal logistic regression which is called ordinal regression is used to predict an ordinal dependent variable given one or more independent variables.

More, these links and others will give you more information.

https://www.youtube.com/watch?v=rSCdwZD1DuM.

https://statistics.laerd.com/spss-tutorials/ordinal-regression-using-spss-statistics.php

Ordinal logistic regression (ordinal regression) is applicable when one aims to predict an ordinal dependent variable given one or more independent variables. Independent variable(s) may be categorical or continuous. Data used should meet the following assumptions:

(1). The dependent variable should be measured at the ordinal level;

(2). One or more independent variables should be continuous, ordinal or categorical; and

(3). No multicollinearity (situation where two or more independent variables are highly correlated with each other).

A Likert scale is a close-ended, forced-choice scale used in a questionnaire that provides a series of answers that go from one extreme to another. For example, a scale might have five choices that start at one end with "strongly agree" and end at the other with "strongly disagree," with less extreme choices in the middle three points.

Now, depends on your dependent variable(s) you need to identify both dependent and independent variable before you can talk of ordinary regression, but in another way if your dependent variables are more than one then try logistic regression.