【正文】 te to variation, causing wrong decisions to be made 測(cè)量誤差的每一項(xiàng)都可能對(duì)變差造成影響而使我們做出錯(cuò)誤的決策 Categories Of Measurement Error Which Affect Location 影響位置的測(cè)量誤差 Accuracy/ Bias Linearity Stability Categories Of Measurement Error Which Affect Spread 影響分布的測(cè)量誤差 Repeatability Reproducibility Precision Resolution/Discrimination 分辨率 Can change be detected? 能偵測(cè)到改變嗎？ Resolution? Accuracy/Bias? Linearity? Stability? Precision (RR)? OK OK OK OK Resolution Definitions 分辨率定義 ? Resolution/Discrimination 分辨率 – Capability to detect the smallest tolerable changes 可以偵測(cè)最小變化的能力 ? Inadequate Measurement Units 不充分的度量單位 – Measurement units too large to detect variation present 度量單位過大而不能偵測(cè)到變化 ? Guideline: “10 Bucket Rule” 1/10原則 – Increments in the measurement system should be onetenth the product specification or process variation 測(cè)量系統(tǒng)必須精確到產(chǎn)品范圍或過程變差的 1/10 Same process output being measured 1 2 3 4 5 Better Discrimination 1 2 3 4 5 Poor Discrimination 1 Resolution/Discrimination 分辨率 Resolution 分辨率 On Hold plaints per hour 每小時(shí)的投訴 Complaint Number Transfers 50 Disputes 210 Information 143 Other 12 Total 415 What is the customer’s biggest plaint? On Hold plaints per hour Complaint Number Transfers 50 Set up or Maintenance Disputes 70 Service Received Disputes 60 Billing Amount Disputes 80 Update Account Information 115 Request Information 28 Other 12 ? Total 415 What is the customer’s biggest plaint? Resolution Accuracy/Bias 準(zhǔn)確性 /偏離 Measurements are “shifted” from “true” value 測(cè)量值偏離真值 Resolution? Accuracy/Bias? Linearity? Stability? Precision (RR)? OK OK OK OK Difference between the observed average value of measurements and the master value 測(cè)量平均值與基準(zhǔn)值之間的差異 Master Value (Reference Standard) Average Value Master value is an accepted, traceable reference standard 基準(zhǔn)值是公認(rèn)的標(biāo)準(zhǔn)值 Accuracy/Bias 準(zhǔn)確性 /偏離 x x x x x x x x x x x x x x x x x x Less accurate More accurate Accuracy/Bias 準(zhǔn)確性 /偏離 Linearity 線性 Measurement is not “true” and/or consistent across the range of the “gage” 測(cè)量系統(tǒng)在測(cè)量范圍內(nèi)與儀器范圍的不一致性 Resolution? Accuracy/Bias? Linearity? Stability? Precision (RR)? OK OK OK OK Linearity Full Range of Gage Reference Value No Bias Observed Average Value Bias Linearity Attribute Example Survey scoring: __ Super Outstanding! 10 __ Outstanding! 9 __ Incredible 8 __ Excellent 7 __ Great 6 __ Very Good 5 __ Good 4 __ OK 3 __ Fair 2 __ Poor 1 Is this a fair scale? Stability 穩(wěn)定性 Measurement drifts 測(cè)量系統(tǒng)偏移 Resolution? Accuracy/Bias? Linearity? Stability? Precision (RR)? OK OK OK OK Stability 穩(wěn)定性 ? Measurements remain constant and predictable over time 測(cè)量系統(tǒng)隨時(shí)間保持一致性與可 預(yù)見性 – For both mean and standard deviation 含均值與標(biāo)準(zhǔn)偏差 ? Evaluated using control charts 可用控制圖來檢查 Time 2 Time 1 Master Value (Reference Standard) Precision 精確性 Repeatability and Reproducibility 重復(fù)性與再現(xiàn)性 Resolution? Accuracy/Bias? Linearity? Stability? Precision (RR)? OK OK OK OK Precision 精確性 ?2total = ?2product/process + ?2repeatability + ?2reproducibility Good Precision Poor Precision Master Value A B Also known as Gage RR Repeatability 重復(fù)性 Variation that occurs when repeated measurements are made of the same item under absolutely identical conditions – Same: ? Operator ? Setup ? Units ? Environmental conditions 重復(fù)性：同一個(gè)人使用同樣的設(shè)備、同樣的儀器在同樣的條件 下測(cè)量同一個(gè)樣品的差異 Repeatability 重復(fù)性 Master Value mean mean Good Repeatability Bad Repeatability Master Value Reproducibility再現(xiàn)性 Variation that occurs when different operators make the measurements under absolutely identical conditions –Same: ? Setups ? Test units ? Environmental conditions ? Locations ? Companies 再現(xiàn)性 ： 不同的人使用同樣的儀器，在同樣的條件下測(cè)量同 一個(gè)樣本之間的差異 Reproducibility 再現(xiàn)性 評(píng)價(jià)者 A 評(píng)價(jià)者 B 評(píng)價(jià)者 C 評(píng)價(jià)者 C 評(píng)價(jià)者 A 評(píng)價(jià)者 B Good Reproducibility Bad Reproducibility Master Value A B C Master Value A B C RR 重復(fù)性與再現(xiàn)性 The Big Picture: Linking Them All Together ?2Total = ?2RR + ?2Process output ?2Total = ?2Repeat + ?2Reproducibility + ?2Process output ?2Total = ?2Repeat+?2Oper+?2Oper ? Process output + ?2Process output Measurement Error Matching Exercise Match the measurement element to the picture that best describes it 1 2 3 4 5 A. Acc