There have been very few dietary assessments that have been developed and tested specifically for Latino adolescents, who have very different eating habits than White adults. This study developed a modified picture sort food frequency questionnaire (P-S FFQ) for use with this population and examined its validity by comparing mean values of nutrient intakes from the questionnaire to three, 24-hour recalls and its test-retest reliability. Validity correlation coefficients ranged from 0.206 for saturated fat to 0.459 for carbohydrates, with 3 statistically significant values (P<.05). Following log transformation, reliability correlation coefficients ranged from .478 for carbohydrates to .720 for cholesterol, and all values were statistically significant (P<.01). The tool shows promise for assessing the diets of minority adolescents but may require modifications, such as adapting it to a computer based program, in order to be most effective.
