Design Research in one Day
Hunches
Experiments
1 Introduction
How can we do Design Research in One Day?? Let us put together a set of hunches, and find out, today, if there is any truth in them.
2
Important
Have you got money? Open your wallet and show me how much. Guys and Gals. Will this work in these UPI days? And well-dressed Gals don’t have pockets in their clothing, Uncle…
Al right, then ask “How much did you spend today?” Again, Guys and Gals.
And then a two-independent-sample test for means. And Permutation and all that stuff.
3
Important
Are vegetarians more kanjoos? Or it is the meat-eaters?
So Swiggy/Zomato/Dining Out bills. For both sets of people, and look at the tip amounts. And then ? Which test, peasants?
4
Important
So? Grades of course, for a good sample from all three groups of people..and then?
ANOVA of course.
5
Important
The Anywhere Door makes Doraemon the greatest children’s show on earth. But Dragon Tales also has that Dragon Stone thingy…“I wish, I wish..”
And that flute musical in Chhota Bheem…Uff!
So? Get Opinion scores from people. Scale of 10. And then? Oh, peasants! Think !!
6
Important
So? Get Opinion scores from people. Scale of 10. What next, peasants?
7
Important
What are we looking at, data-wise? A proportion, which if more than 50% would justify our hunch. So we do an MBTI https://www.16personalities.com/free-personality-test on some unsuspecting sample of people, and try to generalize that result to the population.
8
Important
Again, a survey of a sample. Opinions, yes or no. A Proportion for the sample, and an extension to the population. A proportion test.
9
Important
Let’s say Art people vs Design people. Ask people if they have a BF/GF and since when…And then a prop.test to see if there are differences in proportion.
Make sure you have equal/large nos of all genders in both groups (Art/Design).
10
Important
Even if some of them happen to be Girl-Dads.
So how do we find out? Ask for a “yes/no” opinion from a sample of students. And then a prop.test to see if the proportion is significantly greater than 50%.
11
Important
So, a survey of people from different colleges on campus, as to what pronouns they use, if any ( He/She/Him/Her, or They/Them). And then a chi-square test to see if there are differences in proportions.
12
Important
So, a survey of people from different colleges on campus, as what their opinion is on tattoos and attractiveness ( Yes / No). Survey Males and Females, Students and Faculty. Check the number of respondents in each sub-group!! And then a chi-square test to see if there are differences in proportions.
13 How to Sample a Population?
Here are some simple steps to sample a population (of people)
- Ensure you know how many samples you need ( n = ? )
- Ensure that the population is accessible to you: are the specific category of people numerous and available to you?
- Method:
- Walk up to any person whom you have determined to be part of your target population
- Explain in brief (< 80 seconds) what you are doing, and why you want their help, and in what way
- Clarify on any data-privacy concerns they may have. There is no name column in your data.
- If they agree to participate:
- Give them a coin to toss, and you call it
- If you call correctly, they are part of your sample. If not, not.
- Record their responses in your data form. Show them the final recorded response if asked.
- Either way, pay them.
14 References
- Facing the Abyss: How to Probe Unknown Data. https://shancarter.github.io/ucb-dataviz-fall-2013/classes/facing-the-abyss/