Errors and Statistics INTRODUCTION "RE Inon of he ans obtin a el a ero he tm value a pot bythe coms sp ne rede eployed-The dere of confidence hat he analyst may esoy in is ea rcs eh tnow ge oth acura nd pesisin of the mated wie ms wel aan furans a enor wich ay be noder. Quatiave analysis not simpy acne of aaa ce soy oats ane Jacinto nd the claiming a! the value cid mtu. Nr et sovilg ef he cheisty imate he possibiis of inererenes from ober 1 eee conde at wel as dial Usrbuon of values. Te main purpose ofthis chapter se ean oth term employed and o oui te Statistical methods hat may be api Fecvraere? ppb {essa af vekeninton may b deeds cosorinse einen tan tee oF mos peokaboeale The blue ears measure ofthe accuracy ofthe measremen Inpro cori of meson J preasin® OK Precision may be dfne as the concordance of series of measurements ofthe sme quantity. The ned Messe or tive mean deviation isa measure of prosison. In general, prevsion expresses the reproduiiiy of measure: rcin vans acconpfes accuracy, but ahigh degree of precision does not imply accuracy. To explain thie following problem i usa, Problem 2A substance was known t contain 49.06 0.02 per cent of a given consiuent A. The ens bined by Ivo observes sing th same sbstance and he same general echnigue were ‘bverer (1): 49.01; 4921 4908 ‘Obverer 2): 49.40, 4944 49.42 Solution: For observer (1) 1+ 89.08 _ 4519 96 96Relative mean deviation = £0.09 0.11+6.02),_100, Relative mean err Relative mean devstin » (0924 0.62+0.00) 100 3 19.42 Te analyses of observe (1) were therefore accurate and proce; thos observer (2) were unsaly precise, but ss acurate than those of cbr) ‘Treerors which fet an experimental esull maybe conveninl vided ato those ofthe determine nd the indeterminate Kind. 1. Determi oF Constant Errors: These ae eors whi an be determined. The mos. important ofthese are tbe moied oF whose magne (a) Operational and Personal Erors These ave di o factors fr which the Indvidlanlayst x responsible and are nol connected with the metho er procedure ‘Taeerors are mostly physical in are tl occur wh sound anal cal echique isnt flloned. Examples are : mechanical los of materi in vatous steps of an ysis, underashing or over washing of prespiates, ignition of precipitates at icone temperatures, allowing hygroscopic ‘materials to absocb moisture before or after weighing Personal errs may arse fom the constitutional inability of an individual to make crsin observations this kind is, some persons rately. An example for sable to judge colour changes sarpy in vial izations, which may result in asligh overtepping ofthe end-point (0) Insrumental and Reagent Errore: These as from the faculty constuction of balances, the use of uncalibrated or improperly calibrated welts, gradustod slaseware, and other instruments, the tack of reagents upon sassvars and the use of reagents conaining impurits. Errors of Method: Tese arse fom incore sampling nd from incompleteness of reaction, Tn gravimetric analysis erors may arse owing to appreciable solubiliy of presipiates, 2o- ‘precipitation and post-prespitaion. In tivimevie analysis erors may eecur owing fo failure of Feactions to peseed 1 completion, oceurrence of side and indused reactions and a difference between the obseyed end-point and sioichiometie endpoint of a earion, Aadiive and Proportional Erors The sbsolle valve of an ative errors independent of the ‘amount ofthe eonstiven present in the determination, Some examples of aditive error in weight ofa crucible in which a precipita is ined and erors in weight, “The absolute value of proportional eror deponds upen the amount ofthe constituent, An example ‘of proportional errs impuriy in primary standard, which leeds oan incorrect valve fr the normality of fa sandard solution, 2 Indeterminate or Accidental Errors : These errors manifest themselves by the slight varietions that oveur in sucessive measurements made by the same observer with the greats. ca o @ ‘detcal conditions as possible. They are due to causes over which the analyst has no contol, and which,ici an therfore These eros cannot be pred seme lint raves chown in 9 enh above rep itis evden th “iy very lage errs ae alkey to oecer Gp. Seat ewors ocr wth eestor frequency than large oor nd (Gah Te emors on eve andr sie occur with equal probably / es z | PRE pit ie aioe ‘einai fen bin to nny of on tre athe het Naa | Fal ati oe ps ami moro Me [MINIMISATION OF ERRORS ~~ ‘Determinate or constant eror’ean be reduced by one of the following appropriate methods 1 Craton ofapparats eed ppicaian of corrections AN intent (weighs, asks, ures. ee ae caida eapepine a8 rebel neaenens Fare ee: THscomiss crying pare trina, hear een we epic Condos a enleyed i the aca a er sande" empress Uoueh people aeons aa iil er deo ne aan eee Cimmtnal cin oa 4 uly of «sand whan ich Fe arto oanton ws tlc tive sale Te wih fhe constituent in the unknown can then be calculated from the elation, Result forthe standard _ weight of constituent in tandaré Result forthe unknown z 5 where “¥" isthe weight of the constituent in the unknown,) Standar Ton : A know? artctnt of the constituent being determined is added to the sample, a aon eve ating fe amount asaeenes mati Ea ee ey of known concen the mati be messed. The sa nc erane or pra ae) fein saad ad ts Sf AOS TY Of a opine eoeeatn aos: Ti sos ge LAY once an hen be cerned yang th sme quanityof eral sonar aoa hrs the rat obtained fll on the concentration sae Fenn oncom wah very sm anon of alist Ampito Me ns he appara abl 1 och ete ihe Sal 00 this moy be ber cb inch a ay at vey sk rots wo of mo moles of ome ofr measurable mati Dae knoe anoint of cement bing ens amin oS Lot in nth ple aed he leet sean 8 pe frm Cu 8 ‘sacri swede oer, dterined. Thera ofthese Set Ja xtICANT FIGURES (NUMBERS) AND COMPUTATIONAL RULES” ‘ Kae re iy ie eases ah rie of Se SStoaamcenay esas soc ban tel GAS ug te a sie 126 0 1 el manne arte) pepe os 23 we cla milan 21545 Ts eae he we is es an Fi Segmumere tims‘ARNACEUTICALANALYSIS wit give only one une=tain snares orn any data a wil 96 0017 Roe 1: Retain & many signin ewes 14.20. mt should be written 2s 20.6 ‘osween 20.59 ml and 20.61 ml vtrigue, ad one 10 the Tat igure Forexanpe a volume, whichis known Y ster woud inde a ae reiea toting off guns he comet numberof Ta rennet lowing Figur (wich as een rejected) ESN rea satan inthe wii of 126.12 + 018+ 02678 Baie ediionorsubecton, bre shald be cach mune! eS caceteteher Tins me eeibn ties Se be wrincnas obi Ans 40022100 seer cate on dso, similar rounding off the goes ae HT 8 aera pemipiotnn of 12% 3788, the wloes sed Bix eletoue” 112x230 x 058 = 149 ‘STATISTICAL TREATMENT OF DATA ‘Before applying ssistical weatment 1 observations Ae ated in tate, Some portant tes ae spe, population, average, range, absolue ror clave eon mean deviation, relative mean deviation, standard deviation and variance Semple 7 Samples generally refer to aamall numberof observations The sample values ae the ‘Sbservations the dats, the readings taken frm a population. Population; Population is commonly known as universe oa very large poo of data, Manufoc “abaich of capsules fa population inter of capsules. Twenty capsules taken for determination of uniformity isa sample TArerage 11s commonly known as mean of aihmetic average Tis denoted by z (x bar). Average ivesa menor of he conte of dstbutin. I gies an average ofall dings, For example, the average of readings of 12, 6,7, 5 and 8157. Range i represcns the difference betwoon the largest and smallest readings ina series of ‘measurements ori a dita set. This is denoted by ‘R’. Th range is based on only two values (.., smallest tnd Jages) It s more variable than standard deviation SOLUTE ERROR ‘The absolute eso of determination isthe difference betwoon the observed or measured value and the rue oF most probable value ofthe qvanity measured. Ii generally a measure ofthe Yacuracy’ of the RELATIVE ERROR “The cave errs the absolute eor divided byte or most probable value Tis usually expres {in tems of percentage or paris per housand nner pwoen 20.5 ml an snl bt nota 20.60 msn th ould be writen a5 128.12 + 6402+ 027 aly a many sigiiea FiBUES aS role. or readings its necesary to understand the eof (MEAN (AVERAGE) DEVIATION #\, wis eauned by dtemining te ride] mean ofthe ets ten caluating te deviation of och nda essremen fom he mesh a nly diviig the some ofthe deviations (egress of sign) by the number of measurements. It is denoted by "d", rsaspoxs ANDSTATISICS A Indiv sal raging 3 = Somat ELATIVE MEAN DEVIATION: pe tis ihe mean devistion divided bythe mean, This may be cxpresein ers of percentage orn Pa pertssand Relative msan deviation = J staNpARD DEVIATION gneen te values and mean those values. This sls called asthe roo mean sau Tre consider a eles of observations arranged in ase the mean i given by + ee sués measured mont efficiently by the standard deviation (Senoted Py °S') ‘The spread of | VARIANCE Tie add Bovine te acs Amore pears PRS AERO nvr ate oeicen of variation (CY), is Bye ee (Relative andar deviation)» PROBLEMS Example 1. Analyses of iron oF 708,716,114, 107, 7.14, 7.18 and 7.11, Caleuat forthe vals, ve the flfowing values forthe ion content: 7.08, 7.21, 7.12 can, standard deviation and coefficient of variation