The study was an experiment.  An experiment manipulates an independent variable and measures its effect on a dependent variable.  In this case, the intervention (or no intervention) was the IV and the number of steps was the DV.  In experiments, participants are randomly allocated to conditions.  In addition, because an IV is manipulated, a cause-and-effect relationship can be established.

The sample was an opportunity sample taken from a girls' school. The sample was made up of girls of roughly the same age (16) who were in a physical education class. The girls were randomly allocated to conditions. An opportunity sample means that the participants will share certain characteristics - in this case, they were all from the same school which means that they are most likely to have the characteristics of the local community with regard to socioeconomic status and culture.

Unstructured interviews could be used in order to get a better sense of the girls' attitudes toward exercise and how they perceive their own exercise habits.  It could also be done as a focus group, which may prompt the girls to respond with answers that they would not come up within a one-on-one interview.  Finally, this could be done as a case study which would be longitudinal in nature.  The results of the study saw no change in the level of self-efficacy but this could be due to the fact that it was only a 12-week experiment.  A case study would also be more holistic in its approach and attempt to look at other variables that may affect the girls' exercise habits.

The girls would have to give informed consent to the study - but they may not be told the full aim of the study in order to avoid expectancy effect - that is, changing one's behaviour in order to assist the researcher's goal. The researchers did not deceive them by giving them false information, but deceived by omission, leaving out important information.  It is not clear whether the girls knew that they could drop out of the experiment at any time.  This may be especially problematic for the girls that were simply in their regular PE class. Their data would have to be kept anonymous and they would have to be debriefed at the end of the study.  The debriefing would not only share the findings but explain their importance to the girls.  They would also be reminded of their rights as participants. The study should lead to no undue stress or harm.  As the control group would have had the PE class anyway, they are not really "denied" anything in this study in a way that would unduly hurt their health or well-being.

When considering the generalizability of an experimental study, we should consider three things: population validity, the size of the sample, and whether the study is ecologically valid.

The sample was made up of American teenagers that were going to an all-girls school.  There are several concerns about this sample.  First, the sample is American, which means that the sample would be taken from an individualistic culture and therefore could not necessarily be generalized to a collectivistic culture.  In addition, males are excluded from the sample, meaning that we could not generalize to that population.  Finally, going to an "all girls" school implies that this is a private school.  This means that it is highly likely that these girls are middle and upper class and do not represent poor or working-class girls.  Therefore, the bias of the sample may mean that generalizability is limited to similar populations.

There were about 90 girls altogether - with roughly half of the control group and half in the treatment group.  This may be a large enough sample to generalize the findings to the school population from which it was drawn, but it would not be large enough to generalize to a much broader population. They would have to replicate this study with more girls, and preferably with boys as well, in order to be able to generalize this study with more confidence to a larger population.

Finally, the study has high ecological validity. Although there are some controls in the study (for example, not allowing the participants to see their own progress and the presence of a control group), the study is otherwise carried out under naturalistic conditions.  A higher level of ecological validity means that we are able to generalize the findings with more confidence, although the internal validity of the study may have, to some extent, been compromised.