Stats don’t have to be intimidating
Strategies to grow confidence with graphing and statistics at any level
Statistics can be intimidating. But they don’t have to be.
DataClassroom supports students (and teachers!) in doing more with data and statistics at any level.
Former high school bio teacher and DataClassroom co-founder, Dr. Aaron Reedy and Dr. Elizabeth Schultheis, the co-founder of Data Nuggets show you how you can strengthen graphing and statistics in your biology class no matter what level you are teaching at. From descriptive statistics like mean and median to inferential statistical tests like t-test and chi-square, Reedy and Schultheis share concrete strategies for improving the statistical thinking of your students when they approach a graph.
Above is the full recording of the workshop by Dr. Aaron Reedy, teacher and co-creator of DataClassroom and Dr. Elizabeth Schultheis, co-creator of Data Nuggets that was presented live on July 27, 2021
Resources
The Digital Data Nugget for Which would a woodlouse prefer? *We didn’t get to this example in the webinar.
Questions
Here is some Q&A from the chat that we didn’t get to answer in the workshop.
Can students upload their data and use the DataClassroom to get their p-value, standard deviation, etc?
Yes, students can do all these things in DataClassroom! Here are links for learning to work with their own data, for getting a p-value with different statistical tests, and for calculating descriptive stats like mean, median, standard deviation, standard error, or confidence intervals.
Do you have worksheet for kids to fill out while they are working with a dataset in DataClassoom?
We have ready to teach lessons that have questions for the students to answer that are built into the description of the dataset. If you would like to attach a worksheet, answer sheet, or any other material to a dataset for students to use, that can be achieved by the attachments feature in DataClassroom. This lets you attach any Google Doc, Google Slideshow, or PDF to a file. You can then share this custom made dataset or activity with your students.
What is the difference between Data Nuggets and DataClassroom?
Data Nuggets are free classroom activities created by scientists at Michigan State University, co-designed by scientists and teachers, designed to bring contemporary research and authentic data into the classroom. Data Nuggets include a connection to the scientist behind the data and the true story of their research. Each pencil and paper based activity gives students practice working with “messy data” and interpreting quantitative information.
DataClassroom is a web-application designed by a former high school teacher and scientist to be used by beginners for graphing and statistics in grades 6-12 and beyond. The app is meant to grow with students so that younger grades will start with graphing and more advanced students will move into descriptive (mean, median, error bars) and inferential statistics (P-values and hypothesis testing). DataClassroom has a Resource Library of free Ready-To-Teach lesson plans and datasets that include Digital Data Nuggets. Data Classroom works on any computer or tablet with an internet connection.
When I choose frequency histogram graph (instead of dot scattered plot), there is no option to choose descriptive statistics. Is there a way to do so?
Yes! Remember that in order to use descriptive statistics in DataClassroom you need to be plotting numeric (rather than categorical) data. The option for descriptive statistics will appear on the control panel on the right whenever you have made a histogram with numeric data. It will not appear if you make a frequency plot with categorical data (which does not show variation around a mean).
Do you, as a statistician, worry about the students’ learning about statistical fishing or worry about applying normal stats to non-normal data?
Great question! This is a real concern for older high school or college level students working with inferential statistics and interpreting P-values. Statistical fishing, P-hacking, or whatever you want to call it is probably one of the most serious problems in professional science given the pressures of the publish or perish reality that early career scientists are facing. For advanced students I think it is probably best to teach about those things and how to spot them. Careful experimental design with a hypothesis created before the experiment is conducted can go a long way to guard against these abuses of P-value interpretation. I also think that running multiple simulations of statistical samples is a great way to teach how inferential statistics can reliably lead to false conclusions a some predictable frequency. DataClassroom will soon (fall 2021) be adding a simulation tool for this very purpose.
I worry less about students applying normal (parametric) stats to non-normal data. The reason I feel this way is that I think students can get bogged down in the assumptions of different statistical tests when they don’t make a difference for so many datasets. I feel like an in-depth exploration of these assumptions works better when students already have a little experience with statistics as an applied tool. Many of the statistical tests commonly used in intro course are very robust to violations of normality, and especially so when sample sizes are large. Having said that, DataClassroom is about to release a package for non-parametric statistical tests in our forthcoming college level web-app. The current version of DataClassroom can also test a dataset for normality when a histogram has been created.