Influence of frequency on contigency table

#1
If there are binomial observations, and binomial conditions, what is the influence of frequency and how to determine the "correct" frequency.

For example, say there is a engine light that is either on or off. And the cars either had the oil changed or they didn't by a mechanic who could tell from experience and/or other inspections if the oil change was needed.

oil change status
performed not performed
signal alarm correct alarm false alarm
no alarm missing alarm correct no alarm

We have records of when the alarms went off, and also what maintenance was performed. To make the contingency table, a sample rate is needed. One rate is per hour of operation, and a second rate could be off/on/off cycles. The number of alarms and the number of maintenance actions are fixed in the historic data. What is the effect of using per hour or per off/on/off cycle as the basis for counting to fill out the contingency table?

P.S. I don't know how to present a table in this post. The 2x2 contingency table rows are signals, alarm and no-alarm. The columns are "oil change status" and the columns are performed and not-performed. The alarm/performed entry is "correct alarm", the alarm/not-performed entry is "false alarm", the no-alarm/performed entry is "missing-alarm", and the no-alarm/not-performed entry is "correct no alarm". Those are the entries to be counted.
 
Last edited:

hlsmith

Less is more. Stay pure. Stay poor.
#2
If there are binomial observations, and binomial conditions, what is the influence of frequency and how to determine the "correct" frequency.
So what exactly is your question? Is it the frequency of the potential exposure (rare event) on your probability of event?