# Resampling an analasis

4 messages
Open this post in threaded view
|

## Resampling an analasis

 This post was updated on . CONTENTS DELETED The author has deleted this message.
Open this post in threaded view
|

## Re: Resampling an analasis

 I think re-sampling won't gain anything that you can't have directly. "Noise" is almost always some version of the variance; given variance, you have the standard deviation; dividing Mean by SD gives you a z-score. Potential conceptual problems arise if variances vary -- It could matter if decide to pool the four variances. For some data, it would be right to compute each z based on its own sample; other data, not. If variances differ by a lot, you may have to decide the meaning of a higher mean in a stage which yields a smaller z-score because the variance is huge. (If means increase directly with increased variances, is the need for (log?) transformation implied?) -- Rich Ulrich From: SPSSX(r) Discussion <[hidden email]> on behalf of Juleke2 <[hidden email]> Sent: Wednesday, December 13, 2017 6:17:01 PM To: [hidden email] Subject: Resampling an analasis   Dear all, I need som help with bootstrapping or resampling in SPSS. For an analysis I'm calculating the signal to noise ratios (SNRs) of a measurement in 4 different disease stages. The signal and noise are calculated from 100 tests with this measurement. The mean signal and mean noise for each stage are then divided, resulting in a signal-to-noise ratio for each stage. We want to compare these ratio's between the stages, only, because there is no distribution of the ratio (it's only one ratio for each stage calculated from the signal and noise data (100 tests) which do have a distribution, it is not possible to define the (significant) differences between the ratios of the different stages. However, if I can somehow resample the signal and noise and the calculation of the ratio for each stage, a distribution of the ratio can be formed. Does anyone perhaps know how to this kind of bootstrapping or resampling? Many thanks in advance! -- Sent from: http://spssx-discussion.1045642.n5.nabble.com/ ===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD ===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD