Help in regression interpretation, including interaction terms I m doing a quantitative analysis for my dissertation
and I m having issues with the interpretation of my results (I have little knowledge of SPSS). 
I m doing a panel data analysis with multiple regression.
What, I want to understand is when I should keep or remove a control variable? 
My R2 and adjusted R2 are optimal when I keep my set of variables.
But I have a variable (political stability) that, when I add my explanatory variable (FDI) becomes insignificant and another variable (inflation rate) which logically should be negative becomes positive. I have trouble understanding it: what does it mean?
And also how to interpret interaction terms?
For instance I m studying the effect of FDI on economic growth: 
both openness of the economy and FDI are significant and positive on FDI, but the association of FDI and openness comes with a negative coefficient. 
How can I explain this?
 A: "What I want to understand is when I should keep or remove a control variable."
If your objective is to construct a theoretically sound model, then you shouldn't put too much weight in the R^2 metric. You should include all variables that are justified by theory. If political stability is theoretically related to your independent variable (which you seem to indicate is economic growth), then you should include it.
"I have a variable (political stability) that when I add my explanatory variable (FDI) becomes insignificant and another variable (inflation rate) which logically should be negative becomes positive."
It's not immediately clear why the relationship between inflation and growth should be negative. While the matter is up for debate, according to the AD-AS model in macroeconomics, there is a positive relationship between inflation and growth (at least in the short run).
"And also how to interpret interaction terms."
This really depends on how the openness variable is coded. If it's a binary variable that is 0 for open and 1 for closed, then the negative coefficient would make sense because when economy is closed, then var=1 -> growth= beta1*FDI*var -> growth=beta1*FDI
We would expectd beta1 to be negative for a close economy.
