Why social sciences are not scientific enough -- Can social science approaches find the law of gravitation? -- How to construct predictive models: simplicity and nonabsurdity -- Example of model building: electoral volatility -- Physicists multiply, social scientists add---even when it does not add up -- All hypotheses are not created equal -- Why most numbers published in social sciences are dead on arrival -- Forbidden areas and anchor points -- Geometric means and lognormal distributions -- Example of interlocking models : party sizes and cabinet duration -- Beyond constraint-based models : communication channels and growth rates -- Why we should shift to symmetric regression -- All indices are not created equal -- From descriptive to predictive approaches -- Recommendations for better regression -- Converting from descriptive analysis to predictive models -- Are electoral studies a rosetta stone for parts of social sciences? -- Beyond regression : the need for predictive models.