
Advanced Statistics Advanced Probability and Statistics Math Forum 
 LinkBack  Thread Tools  Display Modes 
August 5th, 2016, 08:50 PM  #1 
Newbie Joined: Mar 2015 From: New Zealand Posts: 6 Thanks: 0  sample size determinant
Hi everyone, I'm trying to determine a sample size. I am listening to calls to a Contact Centre from customers with Internet Banking or a banking app. I'll be listening for 3 different areas. area 1 gets 3.3 million calls annually area 2 gets 1.3 million calls annually area 3 gets 800k calls annually. 40% of customers who call, have internet banking or banking app. Once I've listened to the calls I'll categorize what the call was about. Is there a model or algorithm that can help me determine how many calls I should listen to in each area to get, say a 95% confidence interval, for using my sample as a representation of the whole population? Thanks 
August 6th, 2016, 03:31 PM  #2 
Global Moderator Joined: May 2007 Posts: 6,728 Thanks: 689 
You need an accuracy requirement for your result. 95% confidence is about 2 standard deviations from the mean. You need to specify how far from the mean is acceptable.

August 7th, 2016, 09:05 PM  #3 
Newbie Joined: Mar 2015 From: New Zealand Posts: 6 Thanks: 0  No mean
What would the mean be tho? If I was taking the mean age of each person who responded to a survey then I could provide one but I'm looking to get a good sample representation for call types for which they'd be no mean. What do people do in cases of nonparametric data? Thanks 
August 8th, 2016, 01:07 PM  #4 
Global Moderator Joined: May 2007 Posts: 6,728 Thanks: 689 
I am somewhat confused. What defines call types? What are you trying to represent?

August 8th, 2016, 09:59 PM  #5 
Newbie Joined: Mar 2015 From: New Zealand Posts: 6 Thanks: 0 
Hey Mathman, Call types are defined by people who call. For example say you get 10000 calls a year 3000 might be about Credit Cards, 2000 about loans, 4000 about accounts and 1000 about general inquiries. It depends on why the customer is calling, only way to know is to listen to the call and record the information. We get about 4.4 million calls a year and I want to listen to a sample of those so I can make judgments on future changes. e.g. Credit card calls are now going to be 50% cheaper and they cost $4 per call, how much would we save. So $2*3000 = $6000. That's a simplistic and unrealistic overview but essentially what I'm trying to do; get a sample size that represents the population. Thanks Scott 
August 9th, 2016, 01:59 PM  #6 
Global Moderator Joined: May 2007 Posts: 6,728 Thanks: 689 
If I understand you correctly, the mean you are trying to estimate is the number of calls of a given type? Using your credit card numbers, $\displaystyle \sigma = \sqrt{npq}=\sqrt{10000\times .3\times .7}=45.8$. Your 95% confidence interval is $\displaystyle 3000\pm 91.6$. 

Tags 
determinant, sample, size 
Thread Tools  
Display Modes  

Similar Threads  
Thread  Thread Starter  Forum  Replies  Last Post 
sample size  Engr Faisal hbaib  Advanced Statistics  0  November 23rd, 2013 06:25 PM 
Chi Square Sample Size  Aqil  Advanced Statistics  0  March 21st, 2012 01:28 AM 
Sample Size and Std Dev  jskrzy  Algebra  1  November 16th, 2011 04:17 PM 
Sample Size  waza086  Algebra  0  March 21st, 2010 02:54 AM 
Sample Size  symmetry  Advanced Statistics  0  March 30th, 2007 04:09 PM 