A confounder is a variable that influences the relationship between the experimental variables. To gain clear vision that influence must be held constant or neutralized.
Variables can be of several kinds. Most well known are the experimental variables, consisting of one or more independent variables and one or more dependent variables, and non-experimental variables.
A confounder is a variable that covariates with the experimental variables. A confounder is always disturbing the relationship of interest. Therefore it always needs to be controlled for.
The best way to deal with a confounder is statistical. In an ANOVA it can be treated as a covariate - this changes the name to ANCOVA - or as a control variable in regression analysis, which changes the name to hierarchical regression.