Roeser and colleagues (2001) conducted a cross-cultural study of self-esteem comparing adolescents from the United States and the Netherlands. In their article they had a section labeled "Covariates" in which they note, "A measure of social desirability was included in each analysis because prior research has shown that self-report measures of mental health often elicit socially desirable response patterns". (That is, social desirability was a kind of nuisance variable they were concerned would disturb their pattern of results.) They then reported a series of analyses of covariance. Here is one of their results: "ANCOVA results for the CBCL [a measure of emotional/behavioral problems] . . . revealed only one significant effect: After covarying out the effect of social desirability . . . , American adolescents reported significantly more externalizing problems than their Dutch peers, F(1, 136) = 9.80, p < .01, Eta squared = .07". Explain what is being done, using this result as an example, to someone who is familiar with ordinary analysis of variance and with partial correlation but who is not familiar with the analysis of covariance.