business statistics contents the one sample test for
play

Business Statistics CONTENTS The one-sample -test for Hypotheses - PowerPoint PPT Presentation

: THE ONE-SAMPLE -TEST AND SPSS Business Statistics CONTENTS The one-sample -test for Hypotheses and SPSS Old exam question Further study THE ONE-SAMPLE -TEST FOR When to use the one-sample -test? It


  1. ๐œˆ : THE ONE-SAMPLE ๐‘ข -TEST AND SPSS Business Statistics

  2. CONTENTS The one-sample ๐‘ข -test for ๐œˆ Hypotheses and SPSS Old exam question Further study

  3. THE ONE-SAMPLE ๐‘ข -TEST FOR ๐œˆ When to use the one-sample ๐‘จ -test? โ–ช It relies on the fact that the distribution of เดค ๐‘Œ is normal with ๐œ ๐‘Œ ๐‘Œ = ๐œˆ ๐‘Œ and ๐œ เดค ๐‘œ , so on the CLT ๐œˆ เดค ๐‘Œ = โ–ช So it can be used to test a hypothesis of the mean only โ–ช not a hypothesis on ๐œ , median, etc. โ–ช It works when the population ๐‘Œ is normal โ–ช or when ๐‘Œ is symmetric and ๐‘œ โ‰ฅ 15 โ–ช or when ๐‘œ โ‰ฅ 30 โ–ช It requires knowledge of ๐œ ๐‘Œ Rule of thumb: โ–ช so knowing ๐‘ก ๐‘Œ will not work use of CLT for mean justified when ๐‘œ โ‰ฅ 30 โ–ช that is a problem!

  4. THE ONE-SAMPLE ๐‘ข -TEST FOR ๐œˆ So, what to do if you donโ€™t know ๐œ ๐‘Œ ? 2 from the โ–ช Luckily, we can estimate the population variance ๐œ ๐‘Œ 2 sample variance ๐‘ก ๐‘Œ เดค เดค เดค เดค ๐‘Œโˆ’๐œˆ เดฅ ๐‘Œโˆ’๐œˆ เดฅ ๐‘Œโˆ’๐œˆ ๐‘Œ ๐‘Œโˆ’๐œˆ ๐‘Œ โ–ช calculate the test statistic ๐‘ก ๐‘Œ / ๐‘œ instead of ๐œ ๐‘Œ / ๐‘œ ? ๐‘Œ ๐‘Œ = = ๐‘ก เดฅ ๐œ เดฅ ๐‘Œ ๐‘Œ Sure, we can ... เดค ๐‘Œโˆ’๐œˆ เดฅ โ–ช but instead of ๐‘Œ ~๐‘‚ 0,1 , we have ๐‘Œ ๐œ เดฅ เดค ๐‘Œ โˆ’ ๐œˆ เดค ๐‘Œ ~๐‘ข ๐‘œโˆ’1 ๐‘ก เดค ๐‘Œ โ–ช where ๐‘ข df is the ๐‘ข -distribution with df degrees of freedom

  5. THE ONE-SAMPLE ๐‘ข -TEST FOR ๐œˆ Example: โ–ช sample of ๐‘œ = 100 body heights โ–ช sample mean าง ๐‘ฆ = 179.1 cm 2 = 212.4 cm 2 โ–ช sample variance ๐‘ก ๐‘Œ We think (or hope, or fear) that ๐œˆ ๐‘Œ < 181 cm Use the five-step procedure โ–ช Steps 1-2: as before ( ๐ผ 0 : ๐œˆ โ‰ฅ 181 ) โ–ช Steps 3-5: see below

  6. THE ONE-SAMPLE ๐‘ข -TEST FOR ๐œˆ โ–ช Step 3 เดค ๐‘Œโˆ’181 โ–ช under least extreme version of ๐ผ 0 : 212.4/100 ~๐‘ข 99 โ–ช no further assumptions required (because ๐‘œ โ‰ฅ 30 ) โ–ช Step 4 โ–ช value of test statistic ๐‘ข calc = โˆ’1.3037 โ–ช critical value of ๐‘ข from the table is ๐‘ข crit,lower,0.05,df=99 = โˆ’ 1.660 โˆ’1.660 0 โ–ช Step 5 โ–ช ๐‘ข calc โˆ‰ ๐‘† crit , so do not reject ๐ผ 0 โ–ช conclude that there is no evidence for ๐œˆ < 181

  7. THE ONE-SAMPLE ๐‘ข -TEST FOR ๐œˆ Summing up: โ–ช ๐œ ๐‘Œ known: โ–ช use ๐‘จ -table/perform ๐‘จ -test โ–ช ๐œ ๐‘Œ estimated by ๐‘ก ๐‘Œ : โ–ช use ๐‘ข -table/perform ๐‘ข -test with df = ๐‘œ โˆ’ 1 degrees of freedom โ–ช In both cases: โ–ช use two-tailed critical value ( ๐›ฝ/2 ) to construct a confidence interval โ–ช use two-tailed critical value ( ๐›ฝ/2 ) to construct a critical region for a two-sided hypothesis test โ–ช use one-tailed critical value ( ๐›ฝ ) to construct a critical region for a one-sided hypothesis test

  8. EXERCISE 1 We test a hypothesis ๐ผ 0 : ๐œˆ = 310 with a sample size ๐‘œ = 50 and ๐œ unknown. Which statements are correct? A. We can use the ๐‘จ -test. B. We can use the ๐‘ข -test. C. We canโ€™t perform this test. D. The sample is normally distributed. E. The population is normally distributed. F. The sampling distribution of the mean is normal.

  9. HYPOTHESES AND SPSS Using SPSS for the one-sample ๐‘จ -test โ–ช This is not possible! โ–ช SPSS (realistically) assumes you donโ€™t know ๐œ ๐‘Œ โ–ช SPSS will always use ๐‘ก ๐‘Œ to estimate ๐œ ๐‘Œ and then do a one-sample ๐‘ข -test

  10. HYPOTHESES AND SPSS ๐‘Œ =life Using SPSS for the one-sample ๐‘ข -test expectancy โ–ช Claim: the average life expectancy is 68 year ๐ผ 0 : ๐œˆ ๐‘Œ = 68 Under ๐ผ 0 : ๐‘ˆ~๐‘ข 152 ๐‘ž -value ๐‘ข calc

  11. HYPOTHESES AND SPSS But look carefully: โ–ช it is a two-sided test (SPSS says: two-tailed) ๐ผ 0 : ๐œˆ ๐‘Œ = 68 ๐‘ž -value

  12. HYPOTHESES AND SPSS Doing a one-sided test with SPSS โ–ช How to do a one-sided ๐‘ข -test in SPSS? โ–ช we could do it by hand with the intermediate results โ–ช we could do it by hand with the two-sided results โ–ช you must be able to do both โ–ช Example: โ–ช null hypothesis ๐ผ 0 : ๐œˆ โ‰ฅ 68 โ–ช sample with ๐‘œ = 153 yields าง ๐‘ฆ = 64.515 and ๐‘ก = 12.7937 64.515 โˆ’ 68 153 โˆ’ 1

  13. HYPOTHESES AND SPSS 68 + 68 โˆ’ 64.515 Two-sided โ–ช ๐‘„ เดค ๐‘Œ โ‰ค 64.515 + ๐‘„ เดค ๐‘Œ โ‰ฅ 71.485 = ๐‘„ แ‰€ ๐‘ข โ‰ค 64.515โˆ’68 71.485โˆ’68 12.7937/ 153 = ๐‘„แˆบ 12.7937/ 153 + ๐‘„ ๐‘ข โ‰ฅ แ‰ ๐‘ข โ‰ค แˆป โˆ’3.369 + ๐‘„ ๐‘ข โ‰ฅ 3.369 โ–ช use SPSS for ๐‘ข df=152 : ๐‘ž = 0.001

  14. HYPOTHESES AND SPSS One-sided (right-sided) and าง ๐‘ฆ < ๐œˆ 0 64.515โˆ’68 โ–ช ๐‘„ เดค ๐‘Œ โ‰ค 64.515 = ๐‘„ ๐‘ข โ‰ค 12.7937/ 153 = ๐‘„ ๐‘ข โ‰ค โˆ’3.369 โ–ช this is exactly half of the reported two-sided ๐‘ž -value โ–ช ๐‘žโˆ’value = 0.0005 So, to move from the SPSS-reported two-sided ๐‘ž -value to right-sided ๐‘ž -value, in case of าง ๐‘ฆ < ๐œˆ 0 : โ–ช divide by 2 ๐‘ž right = ๐‘ž two /2

  15. HYPOTHESES AND SPSS One-sided (left-sided) and าง ๐‘ฆ < ๐œˆ 64.515โˆ’68 โ–ช ๐‘„ เดค ๐‘Œ โ‰ฅ 64.515 = ๐‘„ ๐‘ข โ‰ฅ 12.7937/ 153 = ๐‘„ ๐‘ข โ‰ฅ โˆ’3.369 โ–ช this is exactly 1 โˆ’ ๐‘„ ๐‘ข โ‰ค โˆ’3.369 โ–ช ๐‘žโˆ’value = 0.9995 So, to move from the SPSS-reported two-sided ๐‘ž -value to left- sided ๐‘ž -value, in case of าง ๐‘ฆ > ๐œˆ 0 : โ–ช divide by 2 and subtract the result from 1 ๐‘ž left = 1 โˆ’ ๐‘ž two /2

  16. าง HYPOTHESES AND SPSS Another example โ–ช Suppose โ–ช ๐ผ 0 : ๐œˆ = 3 โ–ช ๐›ฝ = 0.05 โ–ช ๐‘ฆ = 2.84 ๐œˆ=3 เดค โ–ช ๐‘žโˆ’value = 2 ร— ๐‘„ ๐‘Œ โ‰ค 2.84 = 0.0456

  17. HYPOTHESES AND SPSS Both าง ๐‘ฆ = 2.84 and าง ๐‘ฆ = 3.16 give the same two-sided ๐‘ž - value! โ–ช so which is true when ๐ผ 0 : ๐œˆ โ‰ฅ 3 and which when ๐ผ 0 : ๐œˆ โ‰ค 3 ?

  18. าง าง HYPOTHESES AND SPSS ๐‘ฐ ๐Ÿ : ๐‚ โ‰ค ๐Ÿ’ ๐‘ฐ ๐Ÿ : ๐‚ โ‰ฅ ๐Ÿ’ (right-sided) (left-sided) ๐‘ž SPSS ๐‘ž SPSS ๐‘ฆ = 2.84 ๐œˆ=3 เดค ๐œˆ=3 เดค ๐‘„ ๐‘Œ โ‰ฅ 2.84 = 1 โˆ’ ๐‘„ ๐‘Œ โ‰ค 2.84 = 2 2 ( าง ๐‘ฆ < ๐œˆ 0 ) ๐‘ž SPSS ๐‘ž SPSS ๐œˆ=3 เดค ๐œˆ=3 เดค ๐‘ฆ = 3.16 ๐‘„ ๐‘Œ โ‰ฅ 3.16 = ๐‘„ ๐‘Œ โ‰ค 3.16 = 1 โˆ’ 2 2 ( าง ๐‘ฆ > ๐œˆ 0 )

  19. EXERCISE 2 Suppose we test a hypothesis on the mean ๐ผ 0 : ๐œˆ โ‰ฅ 310 with significance level ๐›ฝ = 0.05 . We sample data, and perform the test and find a sample mean าง ๐‘ฆ = 307 and a two- sided ๐‘ž -value 0.08 . What do we conclude?

  20. EXERCISE 3 Suppose we have the following result: A. reject ๐ผ 0 : ๐œˆ โ‰ค 6 . B. ๐‘žโˆ’value < ๐›ฝ . C. probably ๐œˆ > 6 . What is a correct sequence? Aโ†’Bโ†’C , Cโ†’Bโ†’A , Bโ†’Aโ†’C , etc?

  21. OLD EXAM QUESTION 26 March 2015, Q3b

  22. FURTHER STUDY Doane & Seward 5/E 9.1-9.5 Tutorial exercises week 2 one-sided test in SPSS

Recommend


More recommend