After many years of working with, researching, and reading survey research about pregnant adolescents, the conclusion was reached that samples of pregnant adolescents provide self-report data of questionable psychometric value. First, they display very labile moods. Second, they may experience cognitive-developmental changes due to the experience of being pregnant (Blinn, 1988). Third, they tend to provide inaccurate information about areas such as family or parental socioeconomic status, and tend toward social desirability when it comes to self-report data on sensitive information such as substance use, sexual behavior, and delinquent behavior. An adolescent who is one month pregnant is quite different in her thoughts, attitudes, and expectations from one who is further along in her pregnancy, even by a few weeks. A pregnant adolescent's responses at one time may be very different one hour, day, week or month later. To report which trimester of pregnancy an adolescent is in does not recognize changes that occur on a more frequent basis. The purpose of the present pilot study was to illustrate the lability of pregnant adolescents' moods which may impact on both their decision-making processes and the psychometric properties of the data they provide.
Researchers in the area of adolescent pregnancy have several unique problems, and researchers who want to do short-term longitudinal work and address changes over the course of the pregnancy have an even more difficult task. Gaining access to samples of pregnant adolescents can be difficult. Samples of pregnant adolescents are generally heterogeneous in terms of gestational progress, life situation, and age of the adolescents. Recognizing these sampling issues as unavoidable in this area of research, it is imperative that researchers compensate by reporting more detail about the gestational variation in the samples, at least the months of the subjects' pregnancies. They also need to more critically address the issues of the validity and reliability of their data.
To confirm the conclusion that there was a lack of specificity in the gestational and psychometric data provided on survey research with pregnant adolescents, an informal content analysis was conducted using 30 research articles published in refereed journals from 1977 to the present dealing with the measurement of psychosocial variables in pregnant adolescents. All of the studies were selected because they used survey methodologies. Two methodological patterns were revealed. First, for understandable and practical reasons, all of the research involved convenience samples with adolescents in various stages of pregnancy. However, little specific information was provided about gestational progress other than by trimesters. Common phrases used to describe the pregnancy status of the adolescent samples were: "All of them were past the first trimester." "All were in the second trimester or beyond." "All were in the end of their first trimester or in the second trimester when they entered the study." "All were at least four weeks from delivery." None of the authors reported the mean and standard deviation of the pregnancy status of the sample or recognized that there may be variations due to finer distinctions in pregnancy status.
Second, none of the researchers reported the test-retest or split-half reliabilities of the instruments used based on the pregnant samples being studied. None addressed validity issues by including measures of social desirability. Those authors who included information on the reliability of the survey instruments they used reported reliability coefficients previously determined with nonpregnant samples. The instruments used were all well known for their application with nonpregnant samples and included the Piers-Harris Self-Concept Questionnaire (Piers & Harris, 1983). Rosenberg Self-Esteem Questionnaire (Rosenberg, 1965), Offer Self-Image Questionnaire for Adolescents (Offer, Ostrov, & Howard, 1982), State-Trait Anxiety Inventory (Spielberger, Gorsuch, & Lushene, 1970), Beck Depression Inventory (Beck et al. …