STAT 202 Lab: Human response times
Hello, dear friend, you can consult us at any time if you have any questions, add WeChat: daixieit
STAT 202
Lab: Human response times
Prerequisites:
You’ll create a personal dataset then analyze it. You must submit your personal dataset (CSV file) with this lab! You should also save your session to make sure you don’t lose it.
Objectives:
You will generate and use your own data set from Prof. Dietz’s website. We will compare our response times against standard distributions such as Normal, Exponential, and Gamma. This lab is to stimulate understanding of Quantile-Quantile plots for Normal-Quantile plots by using simulated data plotted
against our collected data.
Work instructions:
For this lab: You may consult with a colleague as you work. Each student will type up their own answers using their own unique data set to submit on Canvas. Writeup should be a pdf file. (This should not be the Mac
proprietary “pages” format.) The write-up directions and checklist can be found on the last page.
YOU MUST SUBMIT YOUR OWN .csv DATA FILE
OR YOUR SUBMISSION WILL NOT BE ACCEPTED FOR GRADING!!!
1. Gotohttps://www.donnadietz.com/clickme
Make sure the page is purple and you see 12 identical images. If not, please reload.
We will do this as a class, but if you missed class, you can do this yourself.
Be ready to time yourself for four minutes, then click “Start” .
One of the pictures will change.
As soon as you see the picture change, click on it to make it change back again.
This will repeat every 1 to 4 seconds, so you won't be able to fully predict when it will happen. After 4 minutes, you should have at least 60 data points.
Then, just stop, leaving the picture unchanged.
To grab your data, just scroll down in your browser and copy the text which is now below. Data > Load > From Paste (If using StatCrunch)
Paste your data into text box provided, and leave the checkbox on to use first line as column names. Double check that you are using "whitespace" as the delimiter. Sometimes this is automatic.
2. Export your data. Data > Export (all columns should be highlighted) using comma to export
to a csv file. This will be stored where you store downloads for your browser. It will be “Untitled.csv” by default but you can change the name to some more descriptive in the StatCrunch window.
3. Create ahistogram for your response times. Graph > Histogram and choose “responseTime” .
Choose bin size of 100, starting at zero. (That gives you bins of 0.1 second, which seem to look nice.)
4. Create a scatterplot to see if your response time has anything to do with the wait time.
Graph > Scatter Plot where X=waitTime andy=responseTime. Overlay polynomial of order 1.
(Note: A polynomial of order one is just a best-fit line.) Does your response time relate to the wait time in a meaningful way? Do the same for X=trial andy=responseTime. Did you get better or worse overtime?
5. If you have any response times over 1500 (which is over 1.5 seconds) you may remove them now. If your dataset becomes too small, only remove times over 2000.
Click on the outliers either in the histogram or in the scatterplot, and those data lines will turn pink. Edit > Rows > Delete then Compute! Then, on each graph: Options > Refresh.
6. On histogram, Options > Edit. Display options - overlay a Normal curve and see what that looks like on your data set. Again, on this histogram, go back and this time overlay a Gamma curve then an
Exponential. In all three cases, let the software figure out the parameters, and write down what those
parameters are to use in a future step, simulation! (Note: They show up in the title of your plot!) Note you will need to include these three graphics in your writeup.
7. We will simulate two normal distributions to help us understand Quantile-Quantile plots. Scroll down to the bottom of your screen to figure out how many trials you have. Data > Simulate > Normal. Rows=size of
your data set. Columns=1. Mean=0. Std Dev=1. Repeat this again for the mean and standard deviation you recorded above that StatCrunch gave you in your histogram overlay plot. Name the second one
“myNormal” so you know it’s the one based off of the parameters for your collected data.
8. We will create two more simulations as follows, then make several comparison plots.
1. Data > Simulate > Exponential. Rows=size of your data set. Columns=1. Mean = (See step 6.)
2. Data > Simulate > Gamma. Rows=size of your data set. Columns=1. Set alpha and beta to agree with step 6.
9. Sort all simulated columns and your response time. Be sure to ONLY sort one column at a time! Data > Sort. Click "Select Columns" (but only select one column each time!).
Sort By: (Select the same column again). It should already say "Ascend". Keep it that way.
Click "Create New Column" or it will copy over the column you already have.
Then hit "Compute".
10. Graph > Scatter Plot. Choose x=Sort(Normal1) andy=Sort(myNormal)
Fit a polynomial of order 1. (This is just another name for a line of best fit!)
This one should work perfectly. The points should lie very close to the line. Now let’s see about our data!
11. We’ll make three graphics here. Let’s see if any of them behave the way the previous graphic did. Graph > Scatter Plot. Choose x=Sort(responseTime) for all three. Include best fit lines on these plots.
1. y=Sort(myNormal)
2. y=Sort(Exponential1)
3. y=Sort(Gamma1)
Theoretically, the Exponential or Gamma distribution is a closer fit than the normal, but it might not be! This is truly an experiment! The result will vary for different students! Use only your results to answer this question in your own writeup! Our data sets are also quite small, addingto the issues.
Double Check Before Submission:
* Each student will have different data. Each student should
submit a pdf file with the writeup and additionally, submit your.csv file with your original uncleaned data in it!
The writeup should contain:
* Your histogram before data cleaning was done, for your response times
* Your scatterplot before data cleaning was done, for wait time versus response times, with best-fit line
* Your scatterplot before data cleaning was done, for trial versus response times.
* Your comments for both scatterplots as to whether you think there is a trend. (Hint: Probably not!)
* Tell me whether or not you cleaned your data and why or why not. (Hint: Most people have to clean it!)
* Include the three graphics where you overlaid standard curve families.
* Include the scatterplot you made of the two sorted normal simulations. It should be right on the line of best-fit. Is it? Why or why not? If it looks like stars in the night sky, you did NOT SORT it!!!
* Include each of the three remaining scatterplots with best-fit lines which compare your sorted data
against the sorted simulated distributions. These are done on the cleaned data. If any plots look like the nighttime sky, something is NOT SORTED correctly!!!
* Choose one of the three last plots, by choosing the one where the points lie most closely along the line. This is the distribution that most closely fits your data!
* Write in complete sentences.
* Explain what you did, so you could pickup the document a year from now, without the instructions, and understand what you did and why.
* Use a proper heading with name, date, course, assignment title, etc.
2023-11-06