1. ## Relevant statistical test

Hello,

I would like to ask for a help with what relevant statistical test to use.

I am doing research on internet broadband connection and its specific technologies (ADSL, cable, wireless).

I have data on quality and prices for couple hundred people over time (2006-2009), showing how many they have paid for their broadband connection per month, what technology they had and what internet speed they had.

I have created few speed groups (0-512kbps, 513kbps-1mbps, ..., 18mbps-200mbps) and for each group (and each year) and I calculated average speed of the three technologies. So I know that products with speed 1-2mbps were cheaper (and by how much) from ADSL compared to cable and wireless.

Now I would like to test, whether the differences in means (average prices) between technologies in each speed group are due to random effects or not. I would like to avoid doing 8 separate independent sample t-tests, one for each group, because then I would go from 0,05 to 8*0,05 => 0,4 level of confidence. I would like to test (in each year), whether the different average prices between technologies that I see in all speed groups were due to randomness or are statistically significant.

What test to use for this? I am using SPSS.... In general I would like to avoid doing 8*4 (8 speed groups and 4 years) independent sample t-tests - that is a lot to present and it seems like that has to be a better way to do this...
I will be grateful for any tips on how to handle the problem...
Thank you!!

ps: a graphical representation of the problem is here: http://is.muni.cz/www/99107/graf.JPG Graph shows ADSL, Cable ("kabel") and wireless ("WiFi") connections and their average price in each speed group. I have more groups, but unless the technology had more than 1% of its customers in certain speed group, I did not report the price. Couple speed groups were left empty (they get filled in later years). And for example in year 2006, speed group 0-512, the differences in average prices are quite small (prices are reported in local currency) so I would like to test where they are due to random effects (chance) or if it is statistically significant.

#### Posting Permissions

• You may not post new threads
• You may not post replies
• You may not post attachments
• You may not edit your posts