Today we are very happy to have an external speaker, Professor Sigmond of Uni Passau, come

and visit us and she will talk about controlled experiments as you can see.

She has interestingly enough a diploma in both psychology and computer science which

of course makes her the perfect computer science researcher, certainly when it comes to empirical

methods or as we call it confirmatory research.

And so she's right now in Passau as well as a professor in software engineering, computer

science and she will talk about controlled experiments, I think I will simply hand over

to her.

So thank you very much for being here and we are looking forward to your talk.

So also it's not as much as I am talking, but like we are talking so I don't like to

talk too much, that's why I have many questions for you and also when you have questions anytime

just raise your hand.

So it's designed to be more like a discussion.

So, controlled experiments, I have some examples of what can be controlled experiments.

So like when you are watching stars in the sky and you want to detect planets, it's also

kind of an experiment because you are observing something and you want to observe the outcome

of something.

Or when you have a new walk drive in your spaceship, you want to evaluate that, it's

also kind of an experiment.

Or when you have two teaching methods you want to compare, you can also form two groups

and then you also have a controlled experiment.

Okay, so what are experiments in general?

They are defined as a systematic research study, so you do some research and you are

systematic about that.

You have factors in that study, you intentionally vary those factors and you try to hold everything

else constant during that time.

And then you observe the result of the systematic variation.

So when you compare two teaching methods, you observe the result of the teaching methods

or how students learn based on this new teaching method.

And what I want to do here is focus on human participants, so always when you have humans

in your experiment, but all the stuff we are talking about here also applies for non-human

studies.

So when you want to evaluate the performance of computer system or your new walk drive

and also all the methods here apply there as well.

So if you are interested in that stuff a bit more about historical stuff, you can read

something about that.

It's only available in German.

I guess there are translations, but I'm not sure about that.

Okay, so when you design an experiment, you have these stages you go through.

So first you define what you want to measure, what you want to observe.

You define your hypothesis that you want to evaluate.

You define the variables that you measure and the variables that you observe.

And based on the objective definition, you can then select an experimental design.

So how you arrange the participants into groups, what exactly you are measuring and observing.

And then when you have planned your design, you can execute the study.

Hopefully nothing goes wrong, but there will always be something wrong.

Then you get the data.

You have to analyze the data.

And then also very importantly, you have to interpret the data.

So what do the results mean for the hypothesis that you defined?