# Measuring originality of a person on a sentence completion task

I conducted an experiment where I had 20 people complete 100 sentences with only one word. They had to come up with their completion independently of others and could give any answer they thought was suitable.

Here's a sample question with responses from 20 people:

Yesterday I bought a:

1. House
3. Sandwich
4. Car
5. Ticket
6. Sandwich

...

1. Car
2. Sandwich

I now need to measure, how creative every participant was, meaning I want to rate each participant on a certain scale from completely uncreative to absolutely creative (let's say 0 is uncreative, sticks to popular answers and 100 is highly creative with unique answers to every question).

Crucially important: some sentences were left unanswered by some people, meaning total number of answers for a question was smaller.

Thank you.

## migrated from stats.stackexchange.comJun 6 at 11:50

This question came from our site for people interested in statistics, machine learning, data analysis, data mining, and data visualization.

• @JakeWestfall, thank you! – Y. Lapin Jun 3 at 1:34
• Questions: (a) Are participants told this is a test of creativity? (b) Is creativity to be judged entirely on the basis of picking unpopular answers? In that case somebody could score as creative by picking some unusual animal, seldom-mentioned city/town, etc. for every question: platypus; iguana; Cairo IL, Moenchengladbach, Germany etc. // Then you could rank the frequency of answers to each question and give high 'creativity' points to least frequent answers. 'Unanswered' might always be scored least creative. // For "Yesterday I bought a:", is 'congressman' creative or just cynical? – BruceET Jun 3 at 2:55
• @BruceEt, (a) They are told to finish a sentence differently of others. (b) Yes, least used words mean higher creativity and we don’t judge for hidden meanings or cynicism – Y. Lapin Jun 3 at 5:21

(a) They are told to finish a sentence differently of others.

Are they given any other instructions? Do the answers have to "make sense" somehow? What's to stop someone from answering "yesterday I bought a stomach" for instance? That would certainly be unique. You do say they had to have answers they thought were "suitable" but, still, that's a huge range! Maybe not "stomach" but what about "heart lung machine"?

Second, I think you have to precisely define what you mean by "creative". You say (also in a comment) that "least used words mean higher creativity" but I am not so sure that I agree.

But, if you want the least used word and you want to account for the "no answers" then one way is:

1. All no answers get a 0 score on creativity.
2. Call the number of answers (excluding those with no answers) $$N$$
3. For each answer, call the number of people who gave that answer $$Q_i$$ where i ranges from 1 to the number of different answers.
4. The creativity score of each answer is $$N/Q_i$$

The key is step 4. There are certainly other formulas you could use. Which one to choose depends on what you want the distribution of creativity to look like and whether you thing "only person with this answer" is twice as creative as "one of two people with this answer".

But, again, I'd hesitate to call this creativity.