I assume that when you say "five point scale" that you have a single 5-point Likert item.

In this is the case,

• The Cochran-Armitage test is an ideal test.
• Ordinal regression is also ideal, though more difficult in some software.
• The Mann-Whitney test is commonly used in this case, although there is some controversy about using it for an ordinal dependent variable with few ties.

In any case, I would recommend reporting an effect size statistic. A significant p-value by itself may not tell you very much.

• r ---- the one appropriate for Mann-Whitney ---- is the most common
• epsilon-squared, Freeman's theta, and Kendall's tau-b are also options.

Some information on interpreting these statistics is here: http://rcompanion.org/handbook/F_04.html

Also consider expressing the results graphically. Histograms of counts for each group would be very effective.

It depends how your variable is distributed. If highly skewed or non-normal, then it would be better to use a non-parametric test, most likely Mann-Whitney since you only have two comparison groups. If your variable is relatively normal, then you can simply use a standard t-test.

Thank you so much for all your responses.