A(n) - is an extreme data point that can often inaccurately alter an effect size. - occurs when all possible scores on a variable are not present, which can often inaccurately decrease an effect size. A(n) - is when a set of data is not well represented by a straight line and can result in a zero effect size. All three of these issues can decrease - validity. curvilinear relationshipthird-variable probleminternalspurious associationoutlierrestriction of rangestatisticalconstruct