coding and analyzing events based on dates of the year

eugene077

Hello, I'm relatively new to JMP and have a question on how to code and analyze data for patterns that primarily involve dates of the year.

 

Specifically, I'd like to know if students take more sick days right before or right after a holiday. This is a 3-year program, 10-20 students per class, so ideally I'd like to look at patterns by class and by academic year. An example of the raw data is below.

 

The variables of interest would be specific dates (e.g. Dec 31 and Jan 2 that repeat every year), dates that change from year to year (weekday before and after Thanksgiving, weekday before after Memorial day), as well as range of dates (e.g. flu season).

 

1. How should I set up the columns to be analyzable (should I have a column for each weekday for 3 years - that's hundreds of columns, or a column for each class/year combination)? 

I have fair understanding of chi square, etc, but I'm confused as to what are the variables and observations here. Does each student get a row, or does each workday of the year get a row??

 

Thank you so much!

Eugene

 

raw data example ...

 

ACADEMIC Year	CLASS	Sick days taken
2018 -2019 academic year	"Class of 2019"	Sep 3-5, Nov 15, 16,  Jan 2, Jan 2 (2 students out on same day), Jan 7, June 20 …
	"Class of 2020"	July 3, Oct 15, Oct 16, Oct 30, Oct 30 (2 students out), Dec 20…
	"Class of 2021"	Oct 24, July 15, Dec 24, Dec 24, Jan 20....
2017-18 academic year	"Class of 2018"	Oct 15, Oct 16...
	"Class of 2019"	...
	"Class of 2020" ...	...

 

 

txnelson

I would have a tendency to layout the data in the following format:

I have also attached the table I created.

You do not need to put in a row for every day.  JMP knows what days are, and will be able to handle only having to have the days when students are absent.  

Concerning your thought of having a separate row for each student, it may not be necessary.  The table I have attached counts the number of students missing.  If you want to do comparisons between specific students, having a separate row for each absent student, where you have a student identifier may be what you want to do.

With the data laid out the way I am suggesting, it is a simple task to graph the number of students missing, by the number of days away from the closest holiday.  You will see if there is a trend of more missing students the closer to a holiday.

datedata.jmp