Abstract : According to the classic Galam model of opinion dynamics, each agent participates at each update of an opinion interaction. While the scheme gives everyone the same chance to in- fluence others, in reality, social activity and influence vary considerably from one agent to another. To account for such a feature, we introduce a new individual attribute – ‘‘active- ness’’ – which makes some agents more inclined than others at engaging in local discus- sions. To enhance the corresponding effect, opinion updates are shifted from all-out agent interaction cycles to few agent interaction cycles. Using dynamic analysis and simulations the resulting model is found to exhibit a ‘‘Minority Counteroffensive’’ phenomenon, which under some initial conditions makes the minority to win the opinion competition despite a threshold tipping point at fifty percent. The associated probabilistic phenomenon persists in the case ‘‘activeness’’ is held equal for all agents. The effect of ‘‘opinion leaders’’ is also investigated. Indeed, a leader is an inflexible agent, i.e., an agent who does not change opin- ion. The results reveal that two opinion leaders with moderate social influence may have a stronger effect than one opinion leader with a strong social influence. The model may shed a new light to the understanding of opinion formation and public voting.