ASM 275 / ASM275 LAB 2 - ACCURACY, PRECISION, AND ERROR

ASM 275 / ASM275 LAB 2 - ACCURACY, PRECISION, AND ERROR O °' Look at the placement of the darts with respect to the green bullseye (the true value). Are the darts accurate and precise’ [4] E*plananaw TheanmverisC boMhacourateamd precise. Answered - Correct! No attempts IeR . — , O °' Look at the placement of the darts with respect to the bultseye (the true value]. Are the darts accurate and precise* [4] The answer is A because even though the darts are all settered with respect to the bulfseye they do surround the target. So they are not precise but on average they do approximate the measurement. O °' Look at the placement of the darts with respect to the bullseye {the true value). Are the darts accurate and precise* [4] A the darts are accurate but not precise The answer is B, precise but not accurate. Amsweced-Cor«xt! No attempts IeR O °' Look at the placement of the darts with respect to the bultseye {the true value). Are the darts accurate and precise* [4] ' ' A the darts are accurate but not precise I' s{•iann•ion The answer is D, neither precise nor accurate. Amsweced-Correct! No attempts IeR Fière is a dartbaard shek ng the bulles fthe ta rget or true value) and the Iccatioro where the ciarts •ære throvm fyelloîv dote). The bottom in-lage shows a non-nal distributiori where thie horizontal axis represents thie rrieasurerrierrt value on the dartboard and the vertical arJs représente ho•z common those values are. The vertiœ I li ne la beled "Refererœ value" is directly beneath the buIlseye. This is our TRUE value th+at ue are irying to aczorately ar+d precisely +îæasure. TÏne horizontal line A Is the differenœ between the TRUE value ar+d thie MEASURED value and li ne B represents ho•z spread out the darts are. In This diag+am, Ifnes A and B repræent •Aat? [5] Reference value Value TI-ie a ns+ver is A The diflérence beb+even the true a nd rriaasureci value is accoray. In this eample accuracy Is poor. §6 Based on the a bove, which diagram represents a more precise measurement set? [5] Reference value Value A Reference value B Value B B The answer is A, precision i Efers to how clustered the diffei ent measurement attempts are. Nore clustered = less spread out. Answered - Correct! No attempts left Assigned as O • There are many ways to reduce inter-observer and intra-observer error. One way is to have the same observer take measurements at the same time of day with the same caliper What does this accomplish? Explanation The answer is B. Answered - Co rrect! No attempts left Assigned as O • What if two observers used the same measurement instrument to measure diPerent specimens. This would have what aPect? Explanation The answer is A. Using the same measurement tool would reduœ inter-observer error. O • In the next unit we spend œnsiderable time discussing sexual dimorphism, the fact that men tend to be larger than women. Does this apply to the human skull as well? This is a basic hypothesis: Males have significantly larger skulls than females. [8] To test this, you measure a sampie of male skulls and a sample of female skulls and compare the mean values to see which is larger. However, the male skulls included in your sample are from unusually small individuals. Your study suffers from which type of error? Consider the definitions above before answering Exptanaôon The answer is B. The sample selected is not appropriate and that decision will bias the results, this is systematic error that is impossible to correct for. Answered - Correct! No attempts left O °l • During the same research project you later find out that the calipers you used to measure the females skulls were not calibrated correctly. Your study suPers from: [8] Hint Incorrect. The measurement tool was defective for only part of the sample. Furthermore, you only used these calipers for females, this is systematic error and you have to redo the study. A lot of money has been wasted. The answer is B. Answered - Sorry, that is not correct. No attempts left O °’2 How could you avoid introducing the kind of error caused by faulty ? Select all that apply. I-1.i It ipI.° z n s'.•.'e rd: You can select more than one option _ A Use the same equipment for all measurements Hint Incorrect. All three answers are ways of minimizing the chances of introducing error into your data set. Answered - Sorry, you didn’t select ALL the correct answers. No attempts left What type of error would you be testing by comparing measurements recorded a couple weeks apart by the same person? _ A Inter-observer error Explanation The answer is B. Answered - Correct! No attempts left O ••4 What type of error would you be testing by comparing measurements taken by you and then a second researcher whose data you wanted to use in your study? _ B Intra-observer error Explanation The answer is A. this would reduce inter-observer error. Answered - Correct! No attempts left Based on your understanding of error and how it is measured, what COULD Observation 1 and Observation 2 represent in this equation. Select all that apply. Hint: there are two correct responses. I- ipI.° z ns'.•.'en : You can select more than one option C Observation 1 and 2 could reflect the measurement of different specimens by the same observer D Observation 1 and 2 could reflect the measurement different specimens by different observers Explanation The answer is A and B. The two observations could either be two measurements taken by the same observer days a part, or they could represent measurements of the same specimen by two different observers. Answered - Correct! No attempts left You are combining femur head data with another researcher to evaluate the sexual dimorphism of the human femur. You both measured a test case to see if your Please useehe Aria omn bobli H e ate ' 2Scm, them = 29 cm. this va'lue [10] What is the 96 error rounded to the nearest .1? I; umer”c Ars'.’,'e•. The answer is 7.496 Answered - Correct! No attempts left Is your answer to the previous question within the acceptable range of error, as defined previously in this lab? Explanation The answer is B. The generally accepted error rate is about 5°ti. The value reported here is T.4°4, which is above the standard. You should not combine measurements. Answered - Correct! No attempts left Despite what you learned in this class, you proceed with the data collection as planned. You collect femur data from a sample of males and another researcher collects femur data from a sample of females. You find that there are no differences between the means of the male and females, contrary to expectations. based on what you know about the inter-observer error pattern, what sort of error is affecting the results? Explanation The answer is & This is an example of systemic error. What little data we have to compare suggests your measurements are minimizing (male} values while their measurements are maximizing (female) values, thus biasing your results and reducing the true differences between males and females. Answered - Correct! No attempts left Based on your understa nding of different types of error (systematic and random), which is worse? That is, which type of error is likely to have a greater negative effect on your analysis if not corrected? Although the rate of error should always be calculated, in general random e rror introduces e rrors that are more _ B serious. Explanation The a nswer is A. Random error will reduce precision and increase variability of your sample but systemic error is more serious and could totally invalidate your study. Answered - Correct! No attempts left O °2• You are interested in male/female diPerences in overall height (represented by femur length) and your hypothesis is that males are taller than females, on average. However, after collecting your data the results seem to suggest otherwise. You begin to wonder if the samples you used were biased (unusually short men paired with unusually tall women), your calipers were poorly calibrated, but only for the men, or some other explanation. After all, decades of previous research indicate that males are on average taller than women. Fortunately you did not just collect data that specifically related to the hypothesis being tested. You also collected metadata. Select three metadata variables from this list that would help reduce error or help you identify error that may have crept into your sample. Remember you are trying to identify both inter- and intra-observer error as well as the effects of both random error and systematic error. Think about what each of these is when making your decisions about which variables below constitute metadata. I-1.i It ip I.° z ns'.•.'e rd: You can select more than one option _ B Femu r length _ D Caliper serial number Hint X Incorrect. Metadata tell you about the quality of your data but are not part of the primary analysis. Date/time of measurement, serial num ber, and date of last calibration are all examples of metadata because they don’t relate to your specific hypothesis — do males have larger skulls than females. The answer is A, D, and E. Answered - Sorry, you didn’t select ALL the correct answers. No attempts left