3. Evaluation Methods

To ground the design rationale in practice, the prototype of the NAO for PwD will be evaluated in a formative evaluation with the PwD (subject) and their caregivers. The evaluation will investigate the process of how interaction with the NAO can alleviate potential symptoms caused by early-stage dementia in PwD.  Three methods of evaluation will be applied to have reliable and accurate results, as Bethel. et al (2020)[1] suggests. The study will focus on the prototype's potential effects, which are based on the desired value of autonomy as a part of self-direction. Before the study, possible confounding variables need to be examined such as the subject's attitude towards robots and the baseline relatedness and autonomy.

Study design claims

The study will investigate the claims on the following questions:

1. Does the design increase the sense of autonomy in PwD?

2. Does the design positively affect PwD's affective state? Do subjects like the system?

3. Is the design dependable; is the design accessible enough for PwD to rely on it? Does it feel natural?

For a sample size as small as 20 participants, it is most adequate to apply a within-subjects design (they require fewer participants)[1]. That means there is an approach where every subject is experiencing all of the conditions examined. A within-subjects design might be prone to confounds such as pre-existing notions in the environment. That is why the attitude towards robots and the pre-study sense of affect and autonomy should be examined and evaluated as such. Another confounder variable to look into is the study location and environment. The evaluation method will be self-assessment, which can only be included in the study when it is validated.

/

References

(1) Bethel, C.L., Henkel, Z., Baugus, K. (2020). Conducting Studies in Human-Robot Interaction. In: Jost, C., et al. Human-Robot Interaction. Springer Series on Bio- and Neurosystems, vol 12. Springer, Cham. https://doi.org/10.1007/978-3-030-42307-0_4