Trying to design a Multiple Regression model to predict vehicle repair
times (Dependent) based on range of vehicle variables (no. screws,
bolts, length of adhesive, cable etc.). Some of variables take on wide
range of values (eg. Screws) but others only take on values 1 or 0 (i.e.
either they feature in repair or not) - or sometimes 0, 1, 2.
Query - I thought it might be appropriate to use Dummy Variables to
model these??
Query - how do dummy variables like these affect required sample size? -
i.e. could I get away with a smaller sample size requirement (i.e.
number of vehicle repairs!) given that these dummy variables only take
on limited values (0, 1, 2).
Any advice much appreciated.
Regards,
Murray
Murray Doyle
Statistician
Direct Tel: (0)1635 293124
Main Switchboard Tel: (0)1635 868855
Fax: +44(0)1635 871346
Email : [log in to unmask]
www.thatcham.org
Think before you print - please consider the effect on the environment before printing this email!!
This e-mail has been sent in the strictest confidence to the addressee(s). If you are not the intended recipient of this e-mail you must not disclose or distribute this communication in any form and you are kindly requested to contact the Sender immediately. The information, images, documents and views expressed in this e-mail are personal to the Sender and do not expressly represent official positions and policies of The Motor Insurance Repair Research Centre (Thatcham). No authority exists on behalf of Thatcham to make any agreements, representations or other commitment by the means of e-mail.
Company name: The Motor Insurance Repair Research Centre
Place of registration: England and Wales
Registered number: 967763
Registered address: Colthrop Way, Thatcham, Berkshire, RG19 4NR
Legal form of the company: Private Company Limited by Guarantee
|