[video]
Script
I came across a great blog on Multidimensional Scaling and decided that I’d like to experiment with a couple visualizations myself with now that this platform is available JMP under Consumer Research.
Multidimensional Scaling or MDS is a technique that is used to create a visual representation of the pattern of proximities among a set of objects.
For example, given a matrix of distances between cities like the Flight Distances sample data in JMP, MDS can be used to generate a map of the cities in two dimensions.
Multidimensional Scaling is frequently used in consumer research where researchers have measures of perceptions about brands, tastes, or other product attributes. And it’s is applicable to many other areas.
I’m going to demo this in Jmp using some data referenced in the blog and study below.
In 1962, a psychological experiment was performed on 30 students.
Each student was asked to look at pictures of a woman making faces and rate on a 9-point scale the facial expression dissimilarities, in terms of emotion and content.
The facial expressions are shown here:
1. Grief at death of mother
2. Savoring a Coke
3. Very pleasant surprise
4. Maternal love, baby in arms
5. Physical exhaustion
6. Something wrong with plane
7. Anger at seeing dog beaten
8. Pulling hard on seat of chair
9. Unexpectedly meets old boyfriend
10. Revulsion
11. Extreme pain
12. Knows plane will crash
13. Light sleep
In JMP 13, I have this data showing pairwise dissimilarities between 13 different expressions made by the woman, and we want to use JMP’s MDS platform to try to understand how these emotions are perceived.
Since I want to visualize the Facial Reactions next to all of the resulting plot points, I’ll select all the rows, right click and toggle on Label
Now I need to select any Column I want to add to that label by right clicking in the Column Panel then selecting Lanel/Unlabel.
Now up to the Analyze menu then down under Consumer Research, then select Multidimensional Scaling.
The MDS platform launches.
I’ll leave the Data Format to Distance Matrix.
I decide I want to see the MDS results in two dimensions, so I select 2 in the drop down, which is the default, so I’ll just leave that alone.
I’ll add all of the 13 Continuous columns to Y by dragging them over.
And that’s it, select Ok.
And here is our MDS plot.
The display below now tells me a lot of information about how the students perceived the woman’s emotional faces. The points that are closer together mean that those facial expressions portrayed more similar emotions.
Some things are predictable, like the patterns in Sad and Disgust, while for others we might learn something interesting. Like Keci points out in her blog linked below, I too would NOT have associated meeting an old boyfriend with maternal love
Try it for yourself now using the Flight Distances sample data in JMP or with the linked data below. Thanks for joing me here.
