【正文】 20232023 35 Measurement Systems Analysis (MSA) Attribute 謝謝觀看 /歡迎下載 BY FAITH I MEAN A VISION OF GOOD ONE CHERISHES AND THE ENTHUSIASM THAT PUSHES ONE TO SEEK ITS FULFILLMENT REGARDLESS OF OBSTACLES. BY FAITH I BY FAITH 。, Breakthrough Value Services174。 20232023 33 Measurement Systems Analysis (MSA) Attribute References Kappa Cohen J., A Coefficient of Agreement for Nominal Scales Educational and Psychological Measurement Vol. 20, pp 37– 46, 1960 Fleiss j, l., “The Measure of Interrater Agreement” Statistical Methods for Rates and Proportions 2nd Edition, John Wiley and Sons, pp. 212–304 Kendall’s Coefficient of Concordance ? Sheshkin D. J., “Kendall’s Coefficient of Concordance” Handbook of Parametric and Nonparametric Statistical Procedures, CRC Press, pp. 641–651, 1997 ? Daniel W. W., Applied Nonparametric Statistics Houghton Mifflin, pp. 326334, 1978 34 Measurement Systems Analysis (MSA) Attribute The following are trademarks and service marks of Six Sigma Academy International, LLC: Breakthrough Lean174。 20232023 31 Measurement Systems Analysis (MSA) Attribute Statistical Report Copyright 169。 20232023 29 Measurement Systems Analysis (MSA) Attribute Scoring Example ? 100% is target for all scores 100% indicates training required ? % Appraiser score = repeatability ? Screen % Effectiveness Score = reproducibility ? % Score vs. Attribute individual error against a known population ? Screen % Effective vs. Attribute Total error against a known population % % % % % % SCREEN % EFFECTIVE SCORE % SCREEN % EFFECTIVE SCORE vs. ATTRIBUTE % % APPRAISER SCORE % SCORE VS. ATTRIBUTE Copyright 169。 for use only in pliance with SSA license. Attribute MSA Excel Method ? Allows for RR analysis within and between appraisers 可以評(píng)價(jià)者內(nèi)和評(píng)價(jià)者間的重復(fù)性和再現(xiàn)性研究 ? Test for effectiveness against standard 檢驗(yàn)和標(biāo)準(zhǔn)值比較的有效性 ? Limited to nominal data at two levels 僅限于 2水平的 nominal 數(shù)據(jù) Copyright 169。 20232023 26 Measurement Systems Analysis (MSA) Attribute Attribute MSA – Supplemental Files Copyright 169。 20232023 Six Sigma Academy International, LLC All rights reserved。你的決定取決于你愿意承擔(dān)的風(fēng)險(xiǎn)。 P值表示純粹由于偶然而發(fā)生聯(lián)系的可能性。 Kappa Statistics Response Kappa SE Kappa Z P(vs 0) 1 2 3 4 5 Overall Kendall39。使用 15的標(biāo)準(zhǔn)， 1表示差， 5表示極好， Proposal Judge 1 Judge 2 Judge 3 1 4 3 3 2 4 2 4 3 5 4 5 4 3 4 4 5 3 2 1 6 2 3 2 Would you like to see more proposal to proposal variation or judge to judge? 你愿意看到更多的提議間的波動(dòng)呢還是判斷間的波動(dòng) ? Let’s use KCC to assess the degree of rater association. 17 Measurement Systems Analysis (MSA) Attribute Kendall’s Analysis In MINITAB 1. Put data into MINITAB, each judges trial in a separate column 2. To analyze, go to Stat ? Quality Tools ? Attribute Agreement Analysis 1 2 18 Measurement Systems Analysis (MSA) Attribute Kendall’s Analysis In MINITAB (Cont’d) 3. Select columns that contain the values 4. Enter Judges (Appraisers) and quantity 5. Check the box to show that the categories are ordered (this will trigger the Kendall’s calculation) 6. Select “Results” button 7. Click last option to display the results 6 5 4 3 7 Hit OK twice to execute 8 8 19 Measurement Systems Analysis (MSA) Attribute Attribute Agreement Analysis for Judge 1, Judge 2, Judge 3 Between Appraisers Assessment Agreement Inspected Matched Percent 95 % CI 6 0 (, ) Matched: All Appraisers39。 30/70 ratio is acceptable – Beyond this level, single disagreements can have large leverage on the Kappa / 計(jì)算 Kappa，為了最大化信賴性，最好 good/bad的部件各占 50%， 30/70也是可以接受的 – 超過(guò)這個(gè)限度，單個(gè)的不一致對(duì) Kappa有很大影響 ? Execution 實(shí)施 Parts should be rated in random order independently (no parisons) 部件應(yīng)該以隨機(jī)順序評(píng)價(jià) Study should be blind 研究應(yīng)該是盲測(cè) Categories need to be mutually exclusive and exhaustive 分類既不重復(fù)也不遺漏 Rating time should be similar to that “normally” used 評(píng)價(jià)時(shí)間也和正常使用相同 14 Measurement Systems Analysis (MSA) Attribute Guidelines For Kappa Studies (Cont’d) ? Analysis 分析 Review the repeatability portion first (Within Appraiser Kappa), if an Appraiser cannot agree with themselves, ignore parisons to other Appraisers and go understand why (see Improvement below) 先檢查重復(fù)性的部分 (評(píng)價(jià)者內(nèi)的 Kappa)，如果評(píng)價(jià)者自己不一致，忽略和其他評(píng)價(jià)者的比較，而要找出原因 (看下面的改進(jìn) ) For Appraisers that have acceptable repeatability, review the reproducibility portion (Between Appraiser) 如果重復(fù)性可接收，檢查再現(xiàn)性部分 (評(píng)價(jià)者之間 ) If a “Gold Standard” is available (ratings of the samples known by some other means as being “correct”), pare each Appraiser to them for “calibration” / 如果 ”標(biāo)準(zhǔn) ”已知 (已知的正確的等級(jí) )，通過(guò)比較每個(gè)評(píng)價(jià)者和已知值來(lái)校正 ? Use the field in MINITAB, “Known Standa