You would probably do better to do a logistic regression with gender as the DV. Good luck with your work. Best wishes, David Booth

I do not see the difficulty: regardless of the type of variables, if they can be ordered it is good to calculate the relative correlation indices (Sperman, Pearson, etc.).

The rest comes by itself.

The question mentions a nominal variable with five levels, so Spearman or Pearson correlation won't work for this. "Scale variable" isn't well defined, but in SPSS-speak it means "numeric variable". ... It's not quite correlation, but probably a reasonable approach to this problem would be to use anova or Kruskal-Wallis, and then use eta-squared or epsilon-squared ---- there are forms that are appropriate for either anova or K-W ----. eta-squared for anova is the same as r-squared, and the usual form of epsilon-squared for K-W is similar. You can take the square root to get an r-like value.

When one variable is dichotomous and the other is quantitative, you can compute rpb, the point-biserial correlation.

• https://en.wikipedia.org/wiki/Point-biserial_correlation_coefficient

For your 5-level nominal variable (shelter type), I suggest you look at the thorough answer by Silverfish in this discussion:

• https://stats.stackexchange.com/questions/119835/correlation-between-a-nominal-iv-and-a-continuous-dv-variable

Notice especially this excerpt, and replace "topic of conversation" with "type of shelter":

If your independent variable is nominal then it doesn't make sense to talk about what happens "as x increases". In your case, "Topic of conversation" doesn't have a numerical value that can go up and down. So you can't correlate "Topic of conversation" with "Duration of conversation". But as @ttnphns wrote in the comments, there are measures of strength of association you can use that are somewhat analogous.

HTH.

Thank you, Dr. Candiafor your valuable comment. By the way, do you have any references concerning this? Gioacchino de Candia

Md Atikur Rahman I have already indicated them to you; you just have to apply them (if you are capable).

I did it already. But researchers suggest to put references in my research paper because in general correlations between nominal and scale data are not common to show. That’s why I request the references earlier. Gioacchino de Candia

Md Atikur Rahman , if you have a nominal variable with more than two categories, it makes no sense to try to use correlation (Pearson, Spearman, Kendall). Gioacchino de Candia had an if in their suggestion: "if they can be ordered..." ... That being said, if you are performing a correlation between a nominal variable with two levels and a continuous variable, I'm not sure this requires a citation. It's just correlation. Maybe just a citation that covers the correlation test you are using.

Thank you so much. This is really understandable. Salvatore S. Mangiafico

Point-biserial correlation coefficient?