Design Considerations: Minimizing the chances of confounding variables

Confounding variables are the one which are varied between experiments and which affects the measured values of dependent variable. While testing an interface with the different people, there might be case that we could ask participants to enter different messages for different interfaces. This has effect on how long it takes to enter a message, regardless of the interface. This message length is the confounding variable. Others might be complexity of entering certain text compared to other words. The simplest solution is to make sure that same message is entered on all interfaces. Thus, even if the message takes different times, over the whole experiment, the effect of different messages would be evenly spread out across all interfaces. This is called counter balancing.

The aim is to vary independent variable in a known manner, to measure dependent variable(s) and to minimize the effect of confounds on outcome of study

How was this performed during the tests session?

To reduce the confounding between the different feedbacks, a series of questions were prepared. Although the questions were random, appropriate questionnaire was not difficult due to factors like:

• Time allocation constraints for students during the test sessions: usually we only had a chance to test the platform on a short time period, as the testing was usually conducted after school hours, which usually lasts for 2-2.5 hours per sessions on average. Most of these times students spend their time in activities like being comfortable with the platform, learning about the information given in the information box, enjoying the stories provided and discussing with the peers about it.

• Language barrier: Some of the participants from the lower form were not very comfortable with the English. Due to this reason, most of them just use simple way of communication to say that everything is good and they don’t have anything to say. Also due to the observer basic skills in Swahili, participants’ feedbacks were based on some other students’ interpretations (which again depend on their English deviation). This might loose out the possibility of the participants real comment and evaluation from feedback.

• Writing over was difficult part rather than comfortable talking: As it was a known that students appear in the test session after their classes in the schools, and they have just been out of possible stressful classes throughout the day, it was not sensible to ask them to fill up the questionnaire. Although to give out a simple evaluation, a simple fillout-form was prepared and handed over to students during one of the sessions. It was observed that most of them had a feeling of an official writing and felt difficult to write as most of them thought it would be recorded and given off to the teachers for their evaluation. Even after consenting that it is not the case and the feedback will not affect any of the educational activities, they emphasized on oral feedback to the questionnaire. Possible cause were being tired after a school day, being afraid of the writing being recorded, thoughts about their language(grammar) not being quality enough.