Which statistical test to use for binary data?

Posted earlier, but I think I didnt explain well enough. So, reposting in a different thread. Apologize if this is treated as a duplicate. Really needing some help

I have an experiment that has different levels of vibration measured in volts as my independent variable and a binary value of HIGH (1) or LOW(0) as my dependent variable. At a specific threshold, the pulse changes to HIGH.

The data will look like this:

Input Output
Voltage Voltage
0 0
0.1 0
0.2 0
0.3 0
0.4 0
0.5 1 <-- Threshold set at 0.5
0.6 1
0.7 1
3.3 1
I have two similar sets of data as above with different thresholds set at 2 Volts and 3 Volts.

What statistical test could i use with such binary data for my dependent variable? I am trying to find out how reliable my data is with these tests. But most importantly, statistical analysis is part of the grade, but I dont know much beyond Mean, median, mode and SD. Excuse my ignorance. Just looking for help.

When I see the forum and research, Chi square and ANOVA sound greek and latin.


Omega Contributor
Yeah your content and description is a little confusing. Can you post the full datasets? Is the whole issue that the binary labels can disagree with the threshold? Otherwise I don't get the purpose?

As mentioned above, data sets have an expected outcome and an actual outcome. There are only two places where the expected outcome doesnt match the actual outcome and they are marked in red. Any thing else required, please let me know and i can explain.

Hope to get some help..


Ambassador to the humans
I'm still not clear what your question actually is. "What test can I do" isn't really a question. What is a question about your data that you want answered? Is it what threshold should be used? If so it kind of seems like you're just looking for a simple heuristic. Are you looking to test if the thresholds are different for the datasets? That might be something we could talk about but we'd want to know more about the collection methods and such. Basically statistics isn't magic. You need to have to want to know something about your data and some sort of randomness/variation makes it hard to give a perfect answer about that.
Thanks for your patience.
I am trying to understand how confident can I can be that the binary data (Dependent variable) is accurate. As an example, 32 out of the 33 readings were correct (Meaning.. my actual outcome was the same as the expected outcome), but 1 wasn't in data Set 1. Would calculating a confidence interval work in this case?

The thresholds were set (in the potentiometer) differently on purpose for the three datasets. So, the HIGH voltage should be showing up in the output once the threshold is hit.
Last edited: