1- unit root tests are an important tool for econometric analysis. However, univariateunit root tests are known to lack power for samples of small or medium size.

In early approaches to panel unit root testing, the often unrealistic assumption of cross-sectional independence is made. For instance, the tests proposed in Levin, Lin and Chu(2002) and Im, Pesaran and Shin (2003), denoted respectively as LLC and IPS, assume cross-sectional independence, but allow for heterogeneity of the form of individual deterministiceffects (constant and/or linear time trend) and heterogenous serial correlation structure ofthe error terms. Both methods test the same null hypothesis of non-stationarity, but differ interms of the considered alternative and hence, in the way information is pooled. Levin, Lin andChu (2002) study balanced panels withNcross-sectional units andTtime series observations.

2- the following artical can you help:-

thanx a lot Ramiro and Maiami

Consider the p-values. If the p-values are greater than 0.05, then the variable is NOT stationary, hence your parameter estimates will be unreliable. In such a case, you may need to transform the variable before using it in any regression analysis.