According to the CDC, the majority of female adolescents will have had at least one sexual partner by the time they reach the age of 18. Once reserved for high school students, sex ed is now being introduced in junior high and even elementary schools. It is undeniable that our teenagers are becoming sexually active earlier and inevitably, are vulnerable to sexually transmitted infections and HIV.
While it is impossible to alter the choices of sexually active teens once they have been made, it is a mandate for those who interact with them to equip teens with the proper resources to make smart choices and offer suggestions to modify risky behavior. The aim of this particular study was two-fold. First, the authors studied the possibility and practicality of reducing the HIV risk among adolescent girls through the utilization of a small group, community-based setting. Secondly, the effectiveness and usefulness of such an intervention was determined and further analyzed using a controlled design. In the examination of statistical conclusion validity pertaining to this study, one threat that is imperative for the reader to consider lies in the low statistical power of the study.
With 129 girls recruited, less than half (62) attended either the HIV or control intervention groups with only 48 completing the three month follow up session. Polit and Beck states that studies with low statistical power may not succeed in establishing a relationship between the variables. As a pilot study, it is hardly dismissive and worth noting that, at best, a relationship may be present or could be present between the two variables, however, the low statistical power causes me to hesitate to assign an irrefutable relationship between the intervention group and reduced risk behavior as demonstrated by the behavioral changes (decreased substance abuse, increased knowledge on HIV prevention, fewer engagement in risky sexual behavior). The authors acknowledged this threat to the statistical conclusion validity but also noted that outcomes were “encouraging”.
Attrition presents a threat to both statistical conclusion and internal validity. In the aforementioned statistics, it is obvious to deduce that attrition was a direct result of the unavailability of many girls, which consequently resulted in low statistical power. The attrition rate in this study can be considered random as it was blamed on work schedules and inability to form contact and did not alter extraneous characteristics of those remaining in the study. The attrition rate along with selection bias that may have occurred make it difficult for me to determine a resolute causal inference between the intervention and the outcome.
The outcome could have also been influenced by the girls’ desire to please the researchers, “give them what they’re looking for”, or motivated by the monetary reimbursement. Again, as a pilot study in which the efficacy and feasibility of such interventions is also studied, this study can act as a catalyst to strengthen future studies of similar nature. Perhaps the most exigent validation tool in the notion of design validity, construct validity observes whether the cause in a study, when operationalized, represent what the investigator(s) say they are representing. In the discussion of confounding of constructs in which a treatment is “more than” what it is conceived to be, one potential source of confounding may that of participant expectancies. In providing a group-based setting in which discussions were facilitated, we cannot eliminate the possibility of the participants gradually viewing the “treatment” as more of a support group or safe haven for discussion of sensitive topics and for answers to urban legends or myths in relation to sexual behavior and HIV.
Following this, participant expectancies are altered and may present a threat to construct validity. In this case, altered participant expectancies can be considered an insignificant threat due to the fact that the altered participant expectancies may actually enhance or contribute to the outcome. In assessing whether an inadequate pre-experimental explication of the effect is a threat to the construct validity, I believe this study pointed towards a reasonable and plausible outcome for its treatment. While there are many opportunities to further strengthen the study’s statistical conclusion and internal validity, the effect of the treatment was appropriate and did not present a .