Before we begin answering the question, let's breakdown the basic components of the question.

First what type of research design is being employed? We are presented with a quasi-experimental design, where the independent variables are not assigned by the researcher (as the random assignment is not mentioned) but are a pre-existing state of the subjects. In this case the subjects either practice yoga Nidra or regular seated meditation for 45 minutes versus those that did not practice either. Since all the measurements are pre-existing, and no other controls or random assignments are mentioned, this must be an observational study.

Secondly, we are asked about the generalizability (known also as external validity) of the research, i.e. are we able to make the conclusion that the results of the research apply to all adults? Given we have identified this as an observational research design, there is no randomization, and the population is poorly defined in the question, the generalizability to all adults is very low. We do not know or are not able to negate pre-existing relationships exist among the subjects, the circumstances of recruitment (did all the yoga subjects com from the same clinic?), among other covariates, so it is impossible to say, with any confidence, that the positive results of yoga/meditation would be seen across all adults, or even outside the group of subjects.

Lastly, what could we change so we could have a completely randomized design? Well, the key here is randomized. We need to randomly choose our subjects from across all adults and then randomly assign each adult to either the yoga/meditation condition or none. In fact, I would prefer three conditions; yoga, meditation, or neither, as this would allow me to make inferences about yoga and meditation separately. Using the random selection and assignment we also effectively control for other covariates (age, gender, lifestyle, culture, among many others), as we assume that if we did randomization correctly that each groups' subjects will have demographics roughly equivalent to each other, thereby ensuring the effects of these covariates are equal across all groups, essentially removing their effect from the research.

Bringing this all together we can provide the answers to the three parts of the question.

a) This was an observational study as no conditions were assigned to the subjects by the researcher. The practice of yoga/meditation or neither was a pre-existing condition among the subjects included in the study.

b) No the researcher cannot make that conclusion. The study did not include random selection across the whole adult population, thus the study findings are not generalizable across the whole adult population.

c) Adding a random selection of the subjects, along with a random assignment to the yoga/meditation or neither conditions by the researcher would make this a completely randomized design, where ensuring the subjects are randomly sampled from the whole adult population would allow for generalization of the findings to the whole adult population.