Finally, the new version, "JMP 19," has been released!
Of course, JMP 19 adds a variety of new features. The main new features and selling points are introduced in another blog post (original English).
So, over the next few articles I will be introducing some features that are not widely promoted but that I think will be convenient and useful for users.
In this first installment, we will cover the new "multiple response" feature, which is useful for consumer surveys and satisfaction studies.JMP 19 now supports filtering and creating summary tables using "multiple response" columns.
Multiple Response and JMP Settings
Multiple response is a measure or property in JMP that corresponds to "multiple choice" in surveys and other similar situations (where respondents select all or some of the answers that apply).
The figure below shows a JMP data table summarizing the responses to a survey for a corporate training program. The same training program was held 11 times, and a total of 452 participants responded.
The column "Reasons for taking the course" is indeed a multiple response column, but it contains answers to the following questions:
----------------------------------------
Q. What is your purpose in participating in this training? Please select all that apply.
□ Necessary for business
□ Supervisor's instructions
□ Interest in the training content
□ Improve your own skills
□ Other ( )
----------------------------------------
For example, in the 33rd row, the course number "06-033" has the following values in the "Reason for taking the course" column, separated by commas: "Necessary for work, Interest in the training content, Improvement of my own skills." This means that multiple answers were given for these three items.
For JMP to recognize such input data as multiple responses, you must specify the measurement level of the column as Multiple Response .
How to set up multiple responses
Select the appropriate column in the data table, then select Columns > Column Info to:
- Change the scale to "Multiple Response"
- Set "Multiple Responses" in the Column Properties and specify the delimiter (in this example, a comma (,)).
Depending on the input format, for example, "Necessary for work; Interest in the training content; Improvement of own skills"In some cases, values are separated by a semicolon (;). In this case, specify (;) as the separator.
*The delimiter defaults to a comma (,), so you do not need to set the Multiple Response column property in this example.
Using Multiple Response Columns in the Data Filter
One of the features of JMP is that it allows you to interactively narrow down analysis conditions using the "Data Filter (Local Data Filter)," and starting with JMP 19, you can now specify multiple response options as filter conditions.
(Until JMP 18, you could not specify a Multiple Response column as a filter condition.)
The graph below, created using Graph Builder, shows the average level of satisfaction with training for each session (1st to 10th). Perhaps because the training staff are constantly improving the content of the training, satisfaction tends to increase with each session.
全回答者（452名）に対するグラフGraph for all respondents (452 people)
However, motivation for training may differ depending on the reason for taking the course, so we will consider filtering by the reason for taking the course.
In Graph Builder, use the "Local Data Filter" to select the multiple response column "Reason for taking the course" and narrow down the participants who answered "Interest in the training content" and "I want to improve my own skills." These items can be interpreted as participants who are participating in the training with a relatively positive attitude, such as being interested or wanting to improve their skills.
The number of people who answered yes was 123, which is the total number of people who answered at least one of "interest in the training content" or "improving my own skills."
Looking at the satisfaction levels of this group, there are no clear trends depending on the number of sessions. Even with repeated improvements, it is possible that the content is not resonating sufficiently with participants who have positive motivations, and it would be effective to explore the reasons for this through other questions.
Use multiple response columns in Tabulate
In JMP 19, the Tabulate platform also supports multiple response columns and allows you to calculate summary statistics (N, mean, etc.) for each response.
(Until JMP 18, multiple response columns could not be included in the analysis.)
The figure below summarizes the level of satisfaction with the training by "reason for taking the course" using the "Table Creation" tool.
This table shows that the average value (6.96) of those who answered "Interest in the training content" and "Improvement of my own skills" was lower than the average values for "Necessary for work" and "Supervisor's instructions."
Expanding the scope of survey use
In recent years, anyone can easily create, conduct, and compile surveys using tools such as Google Forms and Microsoft Forms.
However, if you want to use the survey to gain insights into improving your company's products or developing new products, you will need to gain detailed insight into the survey results.
In such cases, the functions introduced here are useful for using JMP to extract attributes that cause dissatisfaction with a product or to focus analysis on specific attributes.
by Naohiro Masukawa (JMP Japan)
Naohiro Masukawa - JMP User Community
