Hello Silas,
It appears that you have several questions in your posting, so I will attempt to sort them out and take a stab at offering suggestions for each.
First, you mention that you have your DV measured at both the continuous level (interval or ratio), and also at the ordinal level. If you develop a regression model using the continuous DV, you should not add this same variable (measured at the ordinal level) to your model as an IV. On the other hand, if you prefer to use as your DV the version measured at the ordinal level, you should then instead develop an Ordinal Logistic regression model (which would present you with a very different set of coefficient interpretation questions and issues than you now have).
Second, it seems that you have both some p-value output problems and also some coefficient interpretation difficulties with your existing model (the continuous DV model).
I agree with the other answers that you have received; if you are getting 'periods' in place of p-values for some of your IV coefficients, then there is some problem with your data input for that variable. As some have suggested, perhaps missing some values for some cases? Also, I think that you might be formatting at least one of your qualitative IV incorrectly ( although I do not know if this is the cause of the 'period' problem). For example, I would not code the Housing IV as ordinal (1,2,3), but would instead code it as two separate dummy variables. This is categorical data and using codes of 1 2 3 only confuses the model (as Temporary housing is not twice the value of Permanent housing).
When one has several sets of dummy variables in the model, interpreting those coefficients is certainly possible, but definitely complicated to sort out and explain easily. However, I would caution against the temptation to instead create several separate regression models. If you did, it would adversely affect your DF and your pooled variance.
Regarding the interpretation of the significant dummy coefficients, I suggest you read (or reread) sections of textbooks on such interpretations.
Assuming the DV is continuous, and assuming the IV of Gender is coded as Male = 1, Female = 0,, if the p-value for the Male coefficient is significant--the Male coefficient represents the difference in the DV for Males and Females. This basic interpretation principle can be extended to all your significant dummy coefficients, although it can get confusing and complicated if you have a large number of these in your model (but it is doable).
You can also create interactive terms for your dummies, but then interpretations get even more complicated (although again, doable).
I hope that I didn't insult your intelligence with my rather basic comments. It was not my intention to do so.
Good luck with your model.
------------------------------
Gretchen Donahue
------------------------------