Cell
Phone Data Collection
About US
OGI stands for Oregon Graduate Institute (www.ogi.edu). We are located outside of Portland Oregon. We offer Masters and Ph.D. degrees in the sciences. Last year, we merged with the Oregon Health and Science University(www.ohsu.edu), which is the largest medical school in Oregon. OHSU was ranked number 2 in the nation by US News and World report for primary care education. http://www.ohsu.edu/news/033001usnews.html
Companies such as United Airlines and Goldman Sachs are replacing their touchtone systems by systems that will let you talk to them. However, if you try calling them with an unusual accent or say from a cell phone, or from a noisy street, their performance can become very bad, resulting in the computer misunderstanding what you are saying. To make these systems more robust, we need to collect actual speech from people with all sorts of different accents calling from different noise environments. For this project, we are focusing on people calling from their cell phone.
We want your organization to recruit participants (who are english native speakers) from your geographic region of the US. We need between 223 to 670 different callers from your region.
Participants
For the callers, the procedure is simple; they just call our toll-free phone number from their cell phone and a computer will answer and give them further instructions. They will be asked to read some words, numbers, or phrases from the prompt sheet. The system will also ask a few questions where they will need to answer the question directly. For the individual caller, this whole process should take around 10 minutes. Click here to try one of our sample prompt sheets. It is active so you can call it to test it out.
Our toll-free phone number is available 24 hours a day so participants can call whenever it is most convenient for them or when they have free minutes: daytime, nights, or weekends.
Distribution Requirements of
Participants
We are looking for the following distribution / characteristics of your participants.
Note: Each person can participate only once; but the same phone can be used for multiple participant sessions.
1. Gender: Approximately 50% female & 50% male.
2. Age: At least 20% in the 16-30 range, 20% in the 31-45 range, 15% in the 46-60 range. The other 45% can be any age from 10 to 99.
3. Calling Environment: This is the most important characteristic: approximately equal distributions between the following 4 environments. The final requirement is that each caller must call from one of these four environments:
|
Environment |
Description |
|
home or office |
The participant must call from home, the office or a quiet room inside, with no background noise or conversations. |
|
public place |
The participant must call from a public place or establishment where there are background conversations and noise. |
|
street |
The participant must call while he/she is a stationary pedestrian standing near a busy street or road where background traffic noise may be heard. |
|
car, bus, train |
The participant must call while he/she is a passenger in a moving car, train, bus, or another form of transportation which will allow background traffic and other noise to be heard. |
The first three environments are easy and anyone you find will be able to call from these environments. The car, bus, train environment will be a little more difficult finding participants, but again, anyone can be a passenger in a car, bus or train.
All of the distribution requirements are independent of each other. So, for instance it doesn't matter how many males in the 16-30 category do the home/office noise environment condition.
Recruitment Strategy
To help you control for the noise environments the prompt sheets have been color coded according to environment. The table below shows the color coordination.
| home or office | street | public place | car, bus, train |
| Yellow | Blue | Orange | Purple |
We suggest that you do not distribute all of your prompt sheets at once. Instead, as your first stage, distribute about one-third to half of the prompt sheets taking color coding into account, then wait about a week to give people a chance to do the session. After the week, check on our project statistic website at http://www.cslu.ogi.edu/collection, there you can monitor how your organization is progressing and whether you are meeting the distribution requirements based on completed sessions. Your second pass at distributing the prompt sheets can now focus on whichever areas have low numbers. You can even break it down into three passes at distribution, the last two really narrowing the focus of your recruitment on lacking areas.
Note: Some people will agree to participate, but can forget or lose the prompt sheet. That is why we suggest to distribute in stages. Also, more prompt sheets are available upon request.
Contact Information
If any part of this document is unclear, or if there are any problems, or if there are questions about the requirements, please do not hesitate to e-mail project@ece.ogi.edu or you can call Seth or Alena at (503) 748-7492.
As mentioned above, we have an active example of a prompt sheet. You can see the sample prompt sheet here, please feel free to try it.
For more information about our research center please visit our web site www.cslu.ogi.edu.