EXAMPLE 7.25
Length of audio files on an iPod. Table 7.5 presents data on the length (in seconds) of audio files found on an iPod. There was a total of 10,003 audio files, and 50 files were randomly selected using the “shuffle songs’’ command.41 We would like to give a confidence interval for the average audio file length μ for this iPod.
240 | 316 | 259 | 46 | 871 | 411 | 1366 |
233 | 520 | 239 | 259 | 535 | 213 | 492 |
315 | 696 | 181 | 357 | 130 | 373 | 245 |
305 | 188 | 398 | 140 | 252 | 331 | 47 |
309 | 245 | 69 | 293 | 160 | 245 | 184 |
326 | 612 | 474 | 171 | 498 | 484 | 271 |
207 | 169 | 171 | 180 | 269 | 297 | 266 |
1847 |
471
A Normal quantile plot of the audio data from Table 7.5 (Figure 7.20) shows that the distribution is skewed to the right. Because there are no extreme outliers, the sample mean of the 50 observations will nonetheless have an approximately Normal sampling distribution. The t procedures could be used for approximate inference. For more exact inference, we will transform the data so that the distribution is more nearly Normal. Figure 7.21 is a Normal quantile plot of the natural logarithms of the time measurements. The transformed data are very close to Normal, so t procedures will give quite exact results.