biases are human tendencies that lead us to follow a particular quasi

8/3/2019 Biases Are Human Tendencies That Lead Us to Follow a Particular Quasi

1/22

Biases are human tendencies that lead us to follow a particular quasi-

logical path, or form a certain perspective based on predetermined

mental notions and beliefs. When investors act on a bias, they do not

explore the full issue and can be ignorant to evidence that contradictstheir initial opinions. Avoiding cognitive biases allows investors to

reach impartial decision based solely on available data

Astatisticis biased if it is calculated in such a way that it issystematically different from thepopulation parameterof interest. Thefollowing lists some types of, or aspects of, bias which should not beconsidered mutually exclusive:

Selection bias, where individuals or groups are more likely to takepart in aresearchproject than others, resulting inbiased samples.This can also be termedBerksonian bias.

[1]

Spectrum biasarises from evaluating diagnostic tests on biasedpatient samples, leading to an overestimate of thesensitivity andspecificityof the test.

Thebias of an estimatoris the difference between an estimator'sexpectation and the true value of the parameter being estimated.

Omitted-variable biasis the bias that appears in estimates ofparameters in a regression analysis when the assumedspecification is incorrect, in that it omits an independent variablethat should be in the model.

Instatistical hypothesis testing, a test is said to be unbiased whenthe probability of rejecting the null hypothesis is less than or equalto the significance level when the null hypothesis is true, and the

probability of rejecting the null hypothesis is greater than or equalto the significance level when the alternative hypothesis is true,

Detection bias is where a phenomenon is more likely to beobserved and/or reported for a particular set of study subjects. Forinstance, thesyndemicinvolvingobesityanddiabetesmay meandoctors are more likely to look for diabetes in obese patients thanin less overweight patients, leading to an inflation in diabetesamong obese patients because of skewed detection efforts.

Funding biasmay lead to selection of outcomes, test samples, ortest procedures that favor a study's financial sponsor.
http://en.wikipedia.org/wiki/Statistichttp://en.wikipedia.org/wiki/Statistichttp://en.wikipedia.org/wiki/Statistichttp://en.wikipedia.org/wiki/Statistical_parameterhttp://en.wikipedia.org/wiki/Statistical_parameterhttp://en.wikipedia.org/wiki/Statistical_parameterhttp://en.wikipedia.org/wiki/Selection_biashttp://en.wikipedia.org/wiki/Selection_biashttp://en.wikipedia.org/wiki/Researchhttp://en.wikipedia.org/wiki/Researchhttp://en.wikipedia.org/wiki/Researchhttp://en.wikipedia.org/wiki/Biased_samplehttp://en.wikipedia.org/wiki/Biased_samplehttp://en.wikipedia.org/wiki/Biased_samplehttp://en.wikipedia.org/wiki/Bias_(statistics)#cite_note-0http://en.wikipedia.org/wiki/Bias_(statistics)#cite_note-0http://en.wikipedia.org/wiki/Bias_(statistics)#cite_note-0http://en.wikipedia.org/wiki/Spectrum_biashttp://en.wikipedia.org/wiki/Spectrum_biashttp://en.wikipedia.org/wiki/Sensitivity_and_specificityhttp://en.wikipedia.org/wiki/Sensitivity_and_specificityhttp://en.wikipedia.org/wiki/Sensitivity_and_specificityhttp://en.wikipedia.org/wiki/Sensitivity_and_specificityhttp://en.wikipedia.org/wiki/Bias_of_an_estimatorhttp://en.wikipedia.org/wiki/Bias_of_an_estimatorhttp://en.wikipedia.org/wiki/Bias_of_an_estimatorhttp://en.wikipedia.org/wiki/Omitted-variable_biashttp://en.wikipedia.org/wiki/Omitted-variable_biashttp://en.wikipedia.org/wiki/Statistical_hypothesis_testinghttp://en.wikipedia.org/wiki/Statistical_hypothesis_testinghttp://en.wikipedia.org/wiki/Statistical_hypothesis_testinghttp://en.wikipedia.org/wiki/Syndemichttp://en.wikipedia.org/wiki/Syndemichttp://en.wikipedia.org/wiki/Syndemichttp://en.wikipedia.org/wiki/Obesityhttp://en.wikipedia.org/wiki/Obesityhttp://en.wikipedia.org/wiki/Obesityhttp://en.wikipedia.org/wiki/Diabetes_mellitushttp://en.wikipedia.org/wiki/Diabetes_mellitushttp://en.wikipedia.org/wiki/Funding_biashttp://en.wikipedia.org/wiki/Funding_biashttp://en.wikipedia.org/wiki/Funding_biashttp://en.wikipedia.org/wiki/Diabetes_mellitushttp://en.wikipedia.org/wiki/Obesityhttp://en.wikipedia.org/wiki/Syndemichttp://en.wikipedia.org/wiki/Statistical_hypothesis_testinghttp://en.wikipedia.org/wiki/Omitted-variable_biashttp://en.wikipedia.org/wiki/Bias_of_an_estimatorhttp://en.wikipedia.org/wiki/Sensitivity_and_specificityhttp://en.wikipedia.org/wiki/Sensitivity_and_specificityhttp://en.wikipedia.org/wiki/Spectrum_biashttp://en.wikipedia.org/wiki/Bias_(statistics)#cite_note-0http://en.wikipedia.org/wiki/Biased_samplehttp://en.wikipedia.org/wiki/Researchhttp://en.wikipedia.org/wiki/Selection_biashttp://en.wikipedia.org/wiki/Statistical_parameterhttp://en.wikipedia.org/wiki/Statistic


2/22

Reporting bias involves a skew in the availability of data, such thatobservations of a certain kind may be more likely to be reportedand consequently used in research.

Data-snooping biascomes from the misuse of data miningtechniques.

Instatistics,sampling bias is when a sample is collected in sucha way that some members of the intendedpopulationare lesslikely to be included than others. It results in a biased sample, anon-random sample[1]of apopulation(or non-human factors) inwhich all individuals, or instances, were not equally likely tohave been selected.[2]If this is not accounted for, results can be

erroneously attributed to the phenomenon under study ratherthan to the method ofsampling.

Medical sources sometimes refer to sampling biasas ascertainment bias.[3][4]Ascertainment bias has basically thesame definition,[5][6]but is still sometimes classified as aseparate type of bias.[5]

Distinction from selection bias

Sampling bias is mostly classified as a subtype ofselectionbias,[7]sometimes specifically termed sample selection bias,[8][9]butsome classify it as a separate type of bias.

[10]A distinction, albeit not

universally accepted, of sampling bias is that it underminestheexternal validityof a test (the ability of its results to begeneralized to the rest of the population), while selection bias mainlyaddressesinternal validityfor differences or similarities found in thesample at hand. In this sense, errors occurring in the process of

gathering the sample or cohort cause sampling bias, while errors inany process thereafter cause selection bias.

However, selection bias and sampling bias are often usedsynonymously.

[11]

[edit]Types of sampling bias

Selection from a specific area. For example, a survey of highschool students to measure teenage use of illegal drugs will be a

biased sample because it does not include home-schooled students
http://en.wikipedia.org/wiki/Data-snooping_biashttp://en.wikipedia.org/wiki/Data-snooping_biashttp://en.wikipedia.org/wiki/Statisticshttp://en.wikipedia.org/wiki/Statisticshttp://en.wikipedia.org/wiki/Statisticshttp://en.wikipedia.org/wiki/Statistical_populationhttp://en.wikipedia.org/wiki/Statistical_populationhttp://en.wikipedia.org/wiki/Statistical_populationhttp://en.wikipedia.org/wiki/Random_samplehttp://en.wikipedia.org/wiki/Random_samplehttp://en.wikipedia.org/wiki/Random_samplehttp://en.wikipedia.org/wiki/Statistical_populationhttp://en.wikipedia.org/wiki/Statistical_populationhttp://en.wikipedia.org/wiki/Statistical_populationhttp://en.wikipedia.org/wiki/Ascertainment_bias#cite_note-1http://en.wikipedia.org/wiki/Ascertainment_bias#cite_note-1http://en.wikipedia.org/wiki/Ascertainment_bias#cite_note-1http://en.wikipedia.org/wiki/Sampling_(statistics)http://en.wikipedia.org/wiki/Sampling_(statistics)http://en.wikipedia.org/wiki/Sampling_(statistics)http://en.wikipedia.org/wiki/Ascertainment_bias#cite_note-2http://en.wikipedia.org/wiki/Ascertainment_bias#cite_note-2http://en.wikipedia.org/wiki/Ascertainment_bias#cite_note-2http://en.wikipedia.org/wiki/Ascertainment_bias#cite_note-saem-4http://en.wikipedia.org/wiki/Ascertainment_bias#cite_note-saem-4http://en.wikipedia.org/wiki/Ascertainment_bias#cite_note-saem-4http://en.wikipedia.org/wiki/Ascertainment_bias#cite_note-saem-4http://en.wikipedia.org/wiki/Ascertainment_bias#cite_note-saem-4http://en.wikipedia.org/wiki/Ascertainment_bias#cite_note-saem-4http://en.wikipedia.org/wiki/Selection_biashttp://en.wikipedia.org/wiki/Selection_biashttp://en.wikipedia.org/wiki/Selection_biashttp://en.wikipedia.org/wiki/Ascertainment_bias#cite_note-6http://en.wikipedia.org/wiki/Ascertainment_bias#cite_note-6http://en.wikipedia.org/wiki/Ascertainment_bias#cite_note-6http://en.wikipedia.org/wiki/Ascertainment_bias#cite_note-7http://en.wikipedia.org/wiki/Ascertainment_bias#cite_note-7http://en.wikipedia.org/wiki/Ascertainment_bias#cite_note-7http://en.wikipedia.org/wiki/Ascertainment_bias#cite_note-9http://en.wikipedia.org/wiki/Ascertainment_bias#cite_note-9http://en.wikipedia.org/wiki/Ascertainment_bias#cite_note-9http://en.wikipedia.org/wiki/External_validityhttp://en.wikipedia.org/wiki/External_validityhttp://en.wikipedia.org/wiki/External_validityhttp://en.wikipedia.org/wiki/Internal_validityhttp://en.wikipedia.org/wiki/Internal_validityhttp://en.wikipedia.org/wiki/Internal_validityhttp://en.wikipedia.org/wiki/Ascertainment_bias#cite_note-10http://en.wikipedia.org/wiki/Ascertainment_bias#cite_note-10http://en.wikipedia.org/wiki/Ascertainment_bias#cite_note-10http://en.wikipedia.org/w/index.php?title=Sampling_bias&action=edit&section=2http://en.wikipedia.org/w/index.php?title=Sampling_bias&action=edit&section=2http://en.wikipedia.org/wiki/Ascertainment_bias#cite_note-10http://en.wikipedia.org/wiki/Internal_validityhttp://en.wikipedia.org/wiki/External_validityhttp://en.wikipedia.org/wiki/Ascertainment_bias#cite_note-9http://en.wikipedia.org/wiki/Ascertainment_bias#cite_note-7http://en.wikipedia.org/wiki/Ascertainment_bias#cite_note-7http://en.wikipedia.org/wiki/Ascertainment_bias#cite_note-6http://en.wikipedia.org/wiki/Selection_biashttp://en.wikipedia.org/wiki/Selection_biashttp://en.wikipedia.org/wiki/Ascertainment_bias#cite_note-saem-4http://en.wikipedia.org/wiki/Ascertainment_bias#cite_note-saem-4http://en.wikipedia.org/wiki/Ascertainment_bias#cite_note-saem-4http://en.wikipedia.org/wiki/Ascertainment_bias#cite_note-2http://en.wikipedia.org/wiki/Ascertainment_bias#cite_note-2http://en.wikipedia.org/wiki/Sampling_(statistics)http://en.wikipedia.org/wiki/Ascertainment_bias#cite_note-1http://en.wikipedia.org/wiki/Statistical_populationhttp://en.wikipedia.org/wiki/Random_samplehttp://en.wikipedia.org/wiki/Random_samplehttp://en.wikipedia.org/wiki/Statistical_populationhttp://en.wikipedia.org/wiki/Statisticshttp://en.wikipedia.org/wiki/Data-snooping_bias


3/22

or dropouts. A sample is also biased if certain members areunderrepresented or overrepresented relative to others in thepopulation. For example, a "man on the street" interview whichselects people who walk by a certain location is going to have anoverrepresentation of healthy individuals who are more likely to beout of the home than individuals with a chronic illness. This maybe an extreme form of biased sampling, because certain membersof the population are totally excluded from the sample (that is, theyhave zero probability of being selected).

Self-selectionbias, which is possible whenever the group of peoplebeing studied has any form of control over whether to participate.Participants' decision to participate may be correlated with traitsthat affect the study, making the participants a non-representativesample. For example, people who have strong opinions orsubstantial knowledge may be more willing to spend timeanswering a survey than those who do not. Another exampleisonline and phone-in polls, which are biased samples because therespondents are self-selected. Those individuals who are highlymotivated to respond, typically individuals who have strongopinions, are overrepresented, and individuals that are indifferent

or apathetic are less likely to respond. This often leads to apolarization of responses with extreme perspectives being given adisproportionate weight in the summary. As a result, these types ofpolls are regarded as unscientific.

Pre-screening of trial participants, or advertising for volunteerswithin particular groups. For example a study to "prove" thatsmoking does not affect fitness might recruit at the local fitnesscenter, but advertise for smokers during the advanced aerobics

class, and for non-smokers during the weight loss sessions. Exclusion bias results from exclusion of particular groups from the

sample, e.g. exclusion of subjects who have recentlymigratedintothe study area (this may occur when newcomers are not availablein a register used to identify the source population). Excludingsubjects who move out of the study area during follow-up is ratherequivalent of dropout or nonresponse, aselection biasin that itrather affects the internal validity of the study.
http://en.wikipedia.org/wiki/Self-selectionhttp://en.wikipedia.org/wiki/Self-selectionhttp://en.wikipedia.org/wiki/Online_and_phone-in_pollshttp://en.wikipedia.org/wiki/Online_and_phone-in_pollshttp://en.wikipedia.org/wiki/Online_and_phone-in_pollshttp://en.wikipedia.org/wiki/Human_migrationhttp://en.wikipedia.org/wiki/Human_migrationhttp://en.wikipedia.org/wiki/Human_migrationhttp://en.wikipedia.org/wiki/Selection_biashttp://en.wikipedia.org/wiki/Selection_biashttp://en.wikipedia.org/wiki/Selection_biashttp://en.wikipedia.org/wiki/Selection_biashttp://en.wikipedia.org/wiki/Human_migrationhttp://en.wikipedia.org/wiki/Online_and_phone-in_pollshttp://en.wikipedia.org/wiki/Self-selection


4/22

Healthy user bias, when the study population is likely healthierthan the general population, e.g. workers (i.e. someone in ill-healthis unlikely to have a job as manual laborer).

Overmatching, matching for an apparent confounder that actuallyis a result of the exposure. The control group becomes more similarto the cases in regard to exposure than the general population.

[edit]Symptom-based sampling

The study of medical conditions begins with anecdotal reports. Bytheir nature, such reports only include those referred for diagnosis andtreatment. A child who can't function in school is more likely to bediagnosed withdyslexiathan a child who struggles but passes. A

child examined for one condition is more likely to be tested for anddiagnosed with other conditions, skewingcomorbiditystatistics. Ascertain diagnoses become associated with behavior problemsormental retardation, parents try to prevent their children from beingstigmatized with those diagnoses, introducing further bias. Studiescarefully selected from whole populations are showing that manyconditions are much more common and usually much milder thanformerly believed.

[edit]Truncate selection in pedigree studies

Simple pedigree example of sampling bias

Geneticists are limited in how they can obtain data from humanpopulations. As an example, consider a human characteristic. We areinterested in deciding if the characteristic is inherited as asimpleMendeliantrait. Following the laws ofMendelian inheritance, if the

parents in a family do not have the characteristic, but carry the allele
http://en.wikipedia.org/wiki/Healthy_user_biashttp://en.wikipedia.org/wiki/Healthy_user_biashttp://en.wikipedia.org/wiki/Overmatchinghttp://en.wikipedia.org/wiki/Overmatchinghttp://en.wikipedia.org/w/index.php?title=Sampling_bias&action=edit&section=3http://en.wikipedia.org/wiki/Dyslexiahttp://en.wikipedia.org/wiki/Dyslexiahttp://en.wikipedia.org/wiki/Dyslexiahttp://en.wikipedia.org/wiki/Comorbidityhttp://en.wikipedia.org/wiki/Comorbidityhttp://en.wikipedia.org/wiki/Comorbidityhttp://en.wikipedia.org/wiki/Mental_retardationhttp://en.wikipedia.org/wiki/Mental_retardationhttp://en.wikipedia.org/wiki/Mental_retardationhttp://en.wikipedia.org/w/index.php?title=Sampling_bias&action=edit&section=4http://en.wikipedia.org/wiki/Autosomal_recessivehttp://en.wikipedia.org/wiki/Autosomal_recessivehttp://en.wikipedia.org/wiki/Autosomal_recessivehttp://en.wikipedia.org/wiki/Autosomal_recessivehttp://en.wikipedia.org/wiki/Mendelian_inheritancehttp://en.wikipedia.org/wiki/Mendelian_inheritancehttp://en.wikipedia.org/wiki/Mendelian_inheritancehttp://en.wikipedia.org/wiki/File:Ascertainment_bias.pnghttp://en.wikipedia.org/wiki/File:Ascertainment_bias.pnghttp://en.wikipedia.org/wiki/File:Ascertainment_bias.pnghttp://en.wikipedia.org/wiki/File:Ascertainment_bias.pnghttp://en.wikipedia.org/wiki/Mendelian_inheritancehttp://en.wikipedia.org/wiki/Autosomal_recessivehttp://en.wikipedia.org/wiki/Autosomal_recessivehttp://en.wikipedia.org/w/index.php?title=Sampling_bias&action=edit&section=4http://en.wikipedia.org/wiki/Mental_retardationhttp://en.wikipedia.org/wiki/Comorbidityhttp://en.wikipedia.org/wiki/Dyslexiahttp://en.wikipedia.org/w/index.php?title=Sampling_bias&action=edit&section=3http://en.wikipedia.org/wiki/Overmatchinghttp://en.wikipedia.org/wiki/Healthy_user_bias


5/22

for it, they are carriers (e.g. a non-expressiveheterozygote). In thiscase their children will each have a 25% chance of showing thecharacteristic. The problem arises because we can't tell which familieshave both parents as carriers (heterozygous) unless they have a childwho exhibits the characteristic. The description follows the textbookby Sutton.[12]

The figure shows the pedigrees of all the possible families with twochildren when the parents are carriers (Aa).

Nontruncate selection. In a perfect world we should be able todiscover all such families with a gene including those who aresimply carriers. In this situation the analysis would be free fromascertainment bias and the pedigrees would be under "nontruncateselection" In practice, most studies identify, and include, familiesin a study based upon them having affected individuals.

Truncate selection. When afflicted individuals have an equalchance of being included in a study this is called truncate selection,signifying the inadvertent exclusion (truncation) of families whoare carriers for a gene. Because selection is performed on theindividual level, families with two or more affected children wouldhave a higher probability of becoming included in the study.

Complete truncate selection is a special case whereeachfamily with an affected child has an equal chance of beingselected for the study.

The probabilities of each of the families being selected is given in thefigure, with the sample frequency of affected children also given. Inthis simple case, the researcher will look for a frequency

of

4

7 or

5

8 for the characteristic, depending on the type of truncateselection used.

[edit]The caveman effect

An example of selection basis is called the "caveman effect." Much ofour understanding ofprehistoricpeoples comes from caves, suchascave paintingsmade nearly 40,000 years ago. If there had beencontemporary paintings on trees, animal skins or hillsides, they wouldhave been washed away long ago. Similarly, evidence of fire

pits,middens,burial sites, etc. are most likely to remain intact to the
http://en.wikipedia.org/wiki/Heterozygotehttp://en.wikipedia.org/wiki/Heterozygotehttp://en.wikipedia.org/wiki/Heterozygotehttp://en.wikipedia.org/wiki/Ascertainment_bias#cite_note-11http://en.wikipedia.org/wiki/Ascertainment_bias#cite_note-11http://en.wikipedia.org/wiki/Ascertainment_bias#cite_note-11http://en.wikipedia.org/w/index.php?title=Sampling_bias&action=edit&section=5http://en.wikipedia.org/wiki/Prehistoryhttp://en.wikipedia.org/wiki/Prehistoryhttp://en.wikipedia.org/wiki/Prehistoryhttp://en.wikipedia.org/wiki/Cave_paintinghttp://en.wikipedia.org/wiki/Cave_paintinghttp://en.wikipedia.org/wiki/Cave_paintinghttp://en.wikipedia.org/wiki/Middenhttp://en.wikipedia.org/wiki/Middenhttp://en.wikipedia.org/wiki/Middenhttp://en.wikipedia.org/wiki/Ceremonial_burialhttp://en.wikipedia.org/wiki/Ceremonial_burialhttp://en.wikipedia.org/wiki/Ceremonial_burialhttp://en.wikipedia.org/wiki/Ceremonial_burialhttp://en.wikipedia.org/wiki/Middenhttp://en.wikipedia.org/wiki/Cave_paintinghttp://en.wikipedia.org/wiki/Prehistoryhttp://en.wikipedia.org/w/index.php?title=Sampling_bias&action=edit&section=5http://en.wikipedia.org/wiki/Ascertainment_bias#cite_note-11http://en.wikipedia.org/wiki/Heterozygote


6/22

modern era in caves. Prehistoric people are associated with cavesbecause that is where the data still exists, not necessarily becausemost of them lived in caves for most of their lives.

[edit]Problems caused by sampling biasA biased sample causes problems because anystatisticcomputedfrom that sample has the potential to be consistently erroneous. Thebias can lead to an over- or underrepresentation of thecorrespondingparameterin the population. Almost every sample inpractice is biased because it is practically impossible to ensure aperfectly random sample. If the degree of underrepresentation issmall, the sample can be treated as a reasonable approximation to a

random sample. Also, if the group that is underrepresented does notdiffer markedly from the other groups in the quantity being measured,then a random sample can still be a reasonable approximation.

The wordbiasin common usage has a strong negative wordconnotation, and implies a deliberate intent to mislead orotherscientific fraud. In statistical usage, bias merely represents amathematical property, no matter if it is deliberate or eitherunconscious or due to imperfections in the instruments used forobservation. While some individuals might deliberately use a biasedsample to produce misleading results, more often, a biased sample isjust a reflection of the difficulty in obtaining a truly representativesample.

Some samples use a biased statistical design which neverthelessallows the estimation of parameters. The U.S.National Center forHealth Statisticsfor example, deliberately oversamples from minority

populations in many of its nationwide surveys in order to gainsufficient precision for estimates within these groups.[13]

Thesesurveys require the use of sample weights (see below) to produceproper estimates across all racial and ethnic groups. Provided thatcertain conditions are met (chiefly that the sample is drawn randomlyfrom the entire sample) these samples permit accurate estimation ofpopulation parameters.

[edit] Statistical corrections for a biased sample
http://en.wikipedia.org/w/index.php?title=Sampling_bias&action=edit&section=6http://en.wikipedia.org/wiki/Statistichttp://en.wikipedia.org/wiki/Statistichttp://en.wikipedia.org/wiki/Statistichttp://en.wikipedia.org/wiki/Parameterhttp://en.wikipedia.org/wiki/Parameterhttp://en.wikipedia.org/wiki/Parameterhttp://en.wikipedia.org/wiki/Biashttp://en.wikipedia.org/wiki/Biashttp://en.wikipedia.org/wiki/Biashttp://en.wikipedia.org/wiki/Scientific_fraudhttp://en.wikipedia.org/wiki/Scientific_fraudhttp://en.wikipedia.org/wiki/Scientific_fraudhttp://en.wikipedia.org/wiki/National_Center_for_Health_Statisticshttp://en.wikipedia.org/wiki/National_Center_for_Health_Statisticshttp://en.wikipedia.org/wiki/National_Center_for_Health_Statisticshttp://en.wikipedia.org/wiki/National_Center_for_Health_Statisticshttp://en.wikipedia.org/wiki/Ascertainment_bias#cite_note-12http://en.wikipedia.org/wiki/Ascertainment_bias#cite_note-12http://en.wikipedia.org/wiki/Ascertainment_bias#cite_note-12http://en.wikipedia.org/w/index.php?title=Sampling_bias&action=edit&section=7http://en.wikipedia.org/w/index.php?title=Sampling_bias&action=edit&section=7http://en.wikipedia.org/w/index.php?title=Sampling_bias&action=edit&section=7http://en.wikipedia.org/wiki/Ascertainment_bias#cite_note-12http://en.wikipedia.org/wiki/National_Center_for_Health_Statisticshttp://en.wikipedia.org/wiki/National_Center_for_Health_Statisticshttp://en.wikipedia.org/wiki/Scientific_fraudhttp://en.wikipedia.org/wiki/Biashttp://en.wikipedia.org/wiki/Parameterhttp://en.wikipedia.org/wiki/Statistichttp://en.wikipedia.org/w/index.php?title=Sampling_bias&action=edit&section=6


7/22

If entire segments of the population are excluded from a sample, thenthere are no adjustments that can produce estimates that arerepresentative of the entire population. But if some groups areunderrepresented and the degree of underrepresentation can bequantified, then sample weights can correct the bias.

For example, a hypothetical population might include 10 million menand 10 million women. Suppose that a biased sample of 100 patientsincluded 20 men and 80 women. A researcher could correct for thisimbalance by attaching a weight of 2.5 for each male and 0.625 foreach female. This would adjust any estimates to achieve the sameexpected value as a sample that included exactly 50 men and 50

women, unless men and women differed in their likelihood of takingpart in the survey.

Instatistics,bias (or bias function) of anestimatoris thedifference between this estimator'sexpected valueand the truevalue of the parameter being estimated. An estimator or decisionrule with zero bias is called unbiased. Otherwise the estimatoris said to be biased.

In ordinary English, the term bias ispejorative. In statistics,there are problems for which it may be good to use an estimatorwith a small, but nonzero, bias. In some cases, an estimator witha small bias may have lessermean squared erroror bemedian-unbiased (rather thanmean-unbiased, the standard unbiasednessproperty). The property ofmedian-unbiasedness is invariantunder transformations, while the property ofmean-unbiasednessmay be lost under nonlinear transformations

Information bias

Definition

Also referred to as observational bias and misclassification.ADictionary of Epidemiology, sponsored by theInternationalEpidemiological Association, defines this as the following:

1. A flaw in measuring exposure, covariate, or outcome variables

that results in different quality (accuracy) of information between
http://en.wikipedia.org/wiki/Statisticshttp://en.wikipedia.org/wiki/Statisticshttp://en.wikipedia.org/wiki/Statisticshttp://en.wikipedia.org/wiki/Estimatorhttp://en.wikipedia.org/wiki/Estimatorhttp://en.wikipedia.org/wiki/Estimatorhttp://en.wikipedia.org/wiki/Expected_valuehttp://en.wikipedia.org/wiki/Expected_valuehttp://en.wikipedia.org/wiki/Expected_valuehttp://en.wikipedia.org/wiki/Pejorativehttp://en.wikipedia.org/wiki/Pejorativehttp://en.wikipedia.org/wiki/Pejorativehttp://en.wikipedia.org/wiki/Mean_squared_errorhttp://en.wikipedia.org/wiki/Mean_squared_errorhttp://en.wikipedia.org/wiki/Mean_squared_errorhttp://en.wikipedia.org/wiki/Medianhttp://en.wikipedia.org/wiki/Medianhttp://en.wikipedia.org/wiki/Medianhttp://en.wikipedia.org/wiki/Expected_valuehttp://en.wikipedia.org/wiki/Expected_valuehttp://en.wikipedia.org/wiki/Expected_valuehttp://en.wikipedia.org/wiki/International_Epidemiological_Associationhttp://en.wikipedia.org/wiki/International_Epidemiological_Associationhttp://en.wikipedia.org/wiki/International_Epidemiological_Associationhttp://en.wikipedia.org/wiki/International_Epidemiological_Associationhttp://en.wikipedia.org/wiki/International_Epidemiological_Associationhttp://en.wikipedia.org/wiki/International_Epidemiological_Associationhttp://en.wikipedia.org/wiki/Expected_valuehttp://en.wikipedia.org/wiki/Medianhttp://en.wikipedia.org/wiki/Mean_squared_errorhttp://en.wikipedia.org/wiki/Pejorativehttp://en.wikipedia.org/wiki/Expected_valuehttp://en.wikipedia.org/wiki/Estimatorhttp://en.wikipedia.org/wiki/Statistics


8/22

comparison groups. The occurrence of information biases may not beindependent of the occurrence ofselection biases.

2. Bias in an estimate arising from measurement errors.[1]

Information bias, essentially, refers to bias arising from measurementerror.[2]

[edit]Misclassification

Misclassification thus refers to measurement error. There are twotypes of misclassification in epidemiological research: nondifferential misclassification and differentialmisclassification.

[edit]Non differential misclassification

Non differential misclassification is when all classes, groups, orcategories of a variable (whether exposure, outcome, or covariate)have the same error rate or probability of being misclassified for allstudy subjects.

[1]The traditional assumption has been that, in the case

of binary or dichotomous variables, this would result inan underestimate of the hypothesized relationship between exposureand outcome. This has more recently been challenged however in thatresults of individual studies represent a single estimate and not the

average of repeated measurements and thus can be farther (or nearer)from the null value (i.e. zero) than the true value.[3]

[edit]Differential misclassification

Differential misclassification occurs when the error rate or probabilityof being misclassified differs across groups of study subjects.[1]Forexample, the accuracy of blood pressure measurement may be lowerfor heavier than for lighter study subjects, or a study of elderly

persons may find that reports from elderly persons with dementia areless reliable than those without dementia. The effect(s) of suchmisclassification can vary from an overestimation to anunderestimation of the true value.

[4]Statisticians have developed

methods to adjust for this type of bias, which may assist somewhat incompensating for this problem when known and when it isquantifiable.[5]

Lead time bias

From Wikipedia, the free encyclopedia
http://en.wikipedia.org/wiki/Selection_biashttp://en.wikipedia.org/wiki/Selection_biashttp://en.wikipedia.org/wiki/Selection_biashttp://en.wikipedia.org/wiki/Information_bias_(epidemiology)#cite_note-A_dictionary_of_epidemiology-0http://en.wikipedia.org/wiki/Information_bias_(epidemiology)#cite_note-A_dictionary_of_epidemiology-0http://en.wikipedia.org/wiki/Information_bias_(epidemiology)#cite_note-1http://en.wikipedia.org/wiki/Information_bias_(epidemiology)#cite_note-1http://en.wikipedia.org/wiki/Information_bias_(epidemiology)#cite_note-1http://en.wikipedia.org/w/index.php?title=Information_bias_(epidemiology)&action=edit&section=2http://en.wikipedia.org/w/index.php?title=Information_bias_(epidemiology)&action=edit&section=3http://en.wikipedia.org/wiki/Information_bias_(epidemiology)#cite_note-A_dictionary_of_epidemiology-0http://en.wikipedia.org/wiki/Information_bias_(epidemiology)#cite_note-A_dictionary_of_epidemiology-0http://en.wikipedia.org/wiki/Information_bias_(epidemiology)#cite_note-A_dictionary_of_epidemiology-0http://en.wikipedia.org/wiki/Information_bias_(epidemiology)#cite_note-2http://en.wikipedia.org/wiki/Information_bias_(epidemiology)#cite_note-2http://en.wikipedia.org/wiki/Information_bias_(epidemiology)#cite_note-2http://en.wikipedia.org/w/index.php?title=Information_bias_(epidemiology)&action=edit&section=4http://en.wikipedia.org/wiki/Information_bias_(epidemiology)#cite_note-A_dictionary_of_epidemiology-0http://en.wikipedia.org/wiki/Information_bias_(epidemiology)#cite_note-A_dictionary_of_epidemiology-0http://en.wikipedia.org/wiki/Information_bias_(epidemiology)#cite_note-A_dictionary_of_epidemiology-0http://en.wikipedia.org/wiki/Information_bias_(epidemiology)#cite_note-3http://en.wikipedia.org/wiki/Information_bias_(epidemiology)#cite_note-3http://en.wikipedia.org/wiki/Information_bias_(epidemiology)#cite_note-3http://en.wikipedia.org/wiki/Information_bias_(epidemiology)#cite_note-4http://en.wikipedia.org/wiki/Information_bias_(epidemiology)#cite_note-4http://en.wikipedia.org/wiki/Information_bias_(epidemiology)#cite_note-4http://en.wikipedia.org/wiki/Information_bias_(epidemiology)#cite_note-4http://en.wikipedia.org/wiki/Information_bias_(epidemiology)#cite_note-3http://en.wikipedia.org/wiki/Information_bias_(epidemiology)#cite_note-A_dictionary_of_epidemiology-0http://en.wikipedia.org/w/index.php?title=Information_bias_(epidemiology)&action=edit&section=4http://en.wikipedia.org/wiki/Information_bias_(epidemiology)#cite_note-2http://en.wikipedia.org/wiki/Information_bias_(epidemiology)#cite_note-A_dictionary_of_epidemiology-0http://en.wikipedia.org/w/index.php?title=Information_bias_(epidemiology)&action=edit&section=3http://en.wikipedia.org/w/index.php?title=Information_bias_(epidemiology)&action=edit&section=2http://en.wikipedia.org/wiki/Information_bias_(epidemiology)#cite_note-1http://en.wikipedia.org/wiki/Information_bias_(epidemiology)#cite_note-A_dictionary_of_epidemiology-0http://en.wikipedia.org/wiki/Selection_bias


9/22

Lead time is the length of time between the detection of a disease(usually based on new, experimental criteria) and its usual clinicalpresentation and diagnosis (based on traditional criteria).

Lead time bias is the bias that occurs when two tests for a disease arecompared, and one test (the new, experimental one) diagnoses thedisease earlier, but there is no effect on the outcome of the diseaseitmay appear that the test prolonged survival, when in fact it onlyresulted in earlier diagnosis when compared to traditional methods. Itis an important factor when evaluating the effectiveness of a specifictest.[1]

Lead time bias occurs when testing increases perceived survival timewithout affecting the course of the disease.

[edit]Relationship between screening and survival

Main article:Screening (medicine)

By screening, the intention is to diagnose a disease earlier than itwould be without screening. Without screening, the disease may bediscovered later oncesymptomsappear.

Even if in both cases a person will die at the same time, because thedisease was diagnosed early with screening, the survival time sincediagnosis is longer with screening. No additional life has been gained(and indeed, there may be added anxiety as the patient must live withknowledge of the disease for longer). For example, most people withthe genetic disorderHuntington's diseaseare diagnosed whensymptoms appear around age 50, and they die around age 65. The

typical patient therefore lives about 15 years after diagnosis. With agenetic test, it is possible to diagnose this disorder at birth. If this
http://en.wikipedia.org/wiki/Lead_time_bias#cite_note-0http://en.wikipedia.org/wiki/Lead_time_bias#cite_note-0http://en.wikipedia.org/wiki/Lead_time_bias#cite_note-0http://en.wikipedia.org/w/index.php?title=Lead_time_bias&action=edit&section=1http://en.wikipedia.org/wiki/Screening_(medicine)http://en.wikipedia.org/wiki/Screening_(medicine)http://en.wikipedia.org/wiki/Screening_(medicine)http://en.wikipedia.org/wiki/Symptomhttp://en.wikipedia.org/wiki/Symptomhttp://en.wikipedia.org/wiki/Symptomhttp://en.wikipedia.org/wiki/Huntington%27s_diseasehttp://en.wikipedia.org/wiki/Huntington%27s_diseasehttp://en.wikipedia.org/wiki/Huntington%27s_diseasehttp://en.wikipedia.org/wiki/File:Lead_time_bias.svghttp://en.wikipedia.org/wiki/File:Lead_time_bias.svghttp://en.wikipedia.org/wiki/File:Lead_time_bias.svghttp://en.wikipedia.org/wiki/File:Lead_time_bias.svghttp://en.wikipedia.org/wiki/Huntington%27s_diseasehttp://en.wikipedia.org/wiki/Symptomhttp://en.wikipedia.org/wiki/Screening_(medicine)http://en.wikipedia.org/w/index.php?title=Lead_time_bias&action=edit&section=1http://en.wikipedia.org/wiki/Lead_time_bias#cite_note-0


10/22

newborn baby dies around age 65, he or she will have "survived" 65years after diagnosis, without having actually lived any longer thanthe people who were diagnosed late in life.

Looking at raw statistics, screening will appear to increase survivaltime (this gain is called lead time). If we do not think about whatsurvival time actually means in this context, we might attributesuccess to a screening test that does nothing but advance diagnosis.

Lead time bias can affect interpretation of thefive-year survival rate.[2

Observer-expectancy effect

The observer-expectancy effect (also called the experimenter-expectancy effect, observer effect, or experimenter effect) is aform ofreactivity, in which aresearcher'scognitive biascausesthem to unconsciously influence the participants of anexperiment. It is a significant threat to a study'sinternal validity,and is therefore typicallycontrolledusing adouble-blindexperimental design.

An example of the observer-expectancy effect is demonstratedin musicbackmasking, in which hidden verbal messages aresaid to be audible when a recording is played backwards. Somepeople expect to hear hidden messages when reversing songs,and therefore hear the messages, but to others it sounds likenothing more than random sounds. Often when a song is playedbackwards, a listener will fail to notice the "hidden" lyrics untilthey are explicitly pointed out,after which they are obvious.Other prominent examples includefacilitated

communicationanddowsing.Omitted-variable bias


Instatistics,omitted-variable bias (OVB) occurs when a model iscreated which incorrectly leaves out one or more important causalfactors. The 'bias' is created when the model compensates for themissing factor by over- or under-estimating one of the other factors.
http://en.wikipedia.org/wiki/Five-year_survival_ratehttp://en.wikipedia.org/wiki/Five-year_survival_ratehttp://en.wikipedia.org/wiki/Lead_time_bias#cite_note-isbn1-4160-4002-1-1http://en.wikipedia.org/wiki/Lead_time_bias#cite_note-isbn1-4160-4002-1-1http://en.wikipedia.org/wiki/Lead_time_bias#cite_note-isbn1-4160-4002-1-1http://en.wikipedia.org/wiki/Reactivity_(psychology)http://en.wikipedia.org/wiki/Reactivity_(psychology)http://en.wikipedia.org/wiki/Reactivity_(psychology)http://en.wikipedia.org/wiki/Researcherhttp://en.wikipedia.org/wiki/Researcherhttp://en.wikipedia.org/wiki/Researcherhttp://en.wikipedia.org/wiki/Cognitive_biashttp://en.wikipedia.org/wiki/Cognitive_biashttp://en.wikipedia.org/wiki/Cognitive_biashttp://en.wikipedia.org/wiki/Internal_validityhttp://en.wikipedia.org/wiki/Internal_validityhttp://en.wikipedia.org/wiki/Internal_validityhttp://en.wikipedia.org/wiki/Experimental_controlhttp://en.wikipedia.org/wiki/Experimental_controlhttp://en.wikipedia.org/wiki/Experimental_controlhttp://en.wikipedia.org/wiki/Double-blind_experimenthttp://en.wikipedia.org/wiki/Double-blind_experimenthttp://en.wikipedia.org/wiki/Double-blind_experimenthttp://en.wikipedia.org/wiki/Double-blind_experimenthttp://en.wikipedia.org/wiki/Backmaskinghttp://en.wikipedia.org/wiki/Backmaskinghttp://en.wikipedia.org/wiki/Backmaskinghttp://en.wikipedia.org/wiki/Confirmation_biashttp://en.wikipedia.org/wiki/Confirmation_biashttp://en.wikipedia.org/wiki/Confirmation_biashttp://en.wikipedia.org/wiki/Facilitated_communicationhttp://en.wikipedia.org/wiki/Facilitated_communicationhttp://en.wikipedia.org/wiki/Facilitated_communicationhttp://en.wikipedia.org/wiki/Facilitated_communicationhttp://en.wikipedia.org/wiki/Dowsinghttp://en.wikipedia.org/wiki/Dowsinghttp://en.wikipedia.org/wiki/Dowsinghttp://en.wikipedia.org/wiki/Statisticshttp://en.wikipedia.org/wiki/Statisticshttp://en.wikipedia.org/wiki/Statisticshttp://en.wikipedia.org/wiki/Statisticshttp://en.wikipedia.org/wiki/Dowsinghttp://en.wikipedia.org/wiki/Facilitated_communicationhttp://en.wikipedia.org/wiki/Facilitated_communicationhttp://en.wikipedia.org/wiki/Confirmation_biashttp://en.wikipedia.org/wiki/Backmaskinghttp://en.wikipedia.org/wiki/Double-blind_experimenthttp://en.wikipedia.org/wiki/Double-blind_experimenthttp://en.wikipedia.org/wiki/Experimental_controlhttp://en.wikipedia.org/wiki/Internal_validityhttp://en.wikipedia.org/wiki/Cognitive_biashttp://en.wikipedia.org/wiki/Researcherhttp://en.wikipedia.org/wiki/Reactivity_(psychology)http://en.wikipedia.org/wiki/Lead_time_bias#cite_note-isbn1-4160-4002-1-1http://en.wikipedia.org/wiki/Five-year_survival_rate


11/22

More specifically, OVB is thebiasthat appears in the estimatesofparametersin aregression analysis, when theassumedspecificationis incorrect, in that it omits an independentvariable (possibly non-delineated) that should be in the model.

Recall bias


Inpsychology,recall bias is a type ofsystematic biaswhich occurswhen the way asurveyrespondent answers a question is affected notjust by the correct answer, but also by the respondent'smemory.[1][2]This can affect the results of the survey. As ahypothetical example, suppose that a survey in 2005 asked

respondents whether they believed thatO.J. Simpsonhad killed hiswife, 10 years after the criminal trial. Respondents who believed himinnocent might be more likely to have forgotten about the case, andtherefore to state no opinion, than respondents who thought himguilty. If this is the case, then the survey would find a higher-than-accurate proportion of people who believed that Simpson did kill hiswife.

Relatedly but distinctly, the term might also be used to describe aninstance where a survey respondent intentionally responds incorrectlyto a question about their personal history which results inresponsebias. As a hypothetical example, suppose that a researcher conducts asurvey among women of group A, asking whether they have had anabortion, and the same survey among women of group B. If theresults are different between the two groups, it might be that womenof one group are less likely to have had an abortion, orit mightsimply be that women of one group who have had abortions are lesslikely to admitto it. If the latter is the case, then this would skew thesurvey results; this is a kind ofresponse bias. (It is also possible thatboth are the case: women of one group are less likely to have hadabortions, andwomen of one group who have had abortions are lesslikely to admit to it. This would still affect the survey statistics.)

Response bias

http://en.wikipedia.org/wiki/Estimator_biashttp://en.wikipedia.org/wiki/Estimator_biashttp://en.wikipedia.org/wiki/Estimator_biashttp://en.wikipedia.org/wiki/Parameterhttp://en.wikipedia.org/wiki/Parameterhttp://en.wikipedia.org/wiki/Parameterhttp://en.wikipedia.org/wiki/Regression_analysishttp://en.wikipedia.org/wiki/Regression_analysishttp://en.wikipedia.org/wiki/Regression_analysishttp://en.wikipedia.org/wiki/Specification_(regression)http://en.wikipedia.org/wiki/Specification_(regression)http://en.wikipedia.org/wiki/Specification_(regression)http://en.wikipedia.org/wiki/Psychologyhttp://en.wikipedia.org/wiki/Psychologyhttp://en.wikipedia.org/wiki/Psychologyhttp://en.wikipedia.org/wiki/Systematic_biashttp://en.wikipedia.org/wiki/Systematic_biashttp://en.wikipedia.org/wiki/Systematic_biashttp://en.wikipedia.org/wiki/Statistical_surveyhttp://en.wikipedia.org/wiki/Statistical_surveyhttp://en.wikipedia.org/wiki/Statistical_surveyhttp://en.wikipedia.org/wiki/Recall_bias#cite_note-0http://en.wikipedia.org/wiki/Recall_bias#cite_note-0http://en.wikipedia.org/wiki/Recall_bias#cite_note-0http://en.wikipedia.org/wiki/O.J._Simpsonhttp://en.wikipedia.org/wiki/O.J._Simpsonhttp://en.wikipedia.org/wiki/O.J._Simpsonhttp://en.wikipedia.org/wiki/Response_biashttp://en.wikipedia.org/wiki/Response_biashttp://en.wikipedia.org/wiki/Response_biashttp://en.wikipedia.org/wiki/Response_biashttp://en.wikipedia.org/wiki/Response_biashttp://en.wikipedia.org/wiki/Response_biashttp://en.wikipedia.org/wiki/Response_biashttp://en.wikipedia.org/wiki/Response_biashttp://en.wikipedia.org/wiki/Response_biashttp://en.wikipedia.org/wiki/Response_biashttp://en.wikipedia.org/wiki/O.J._Simpsonhttp://en.wikipedia.org/wiki/Recall_bias#cite_note-0http://en.wikipedia.org/wiki/Recall_bias#cite_note-0http://en.wikipedia.org/wiki/Statistical_surveyhttp://en.wikipedia.org/wiki/Systematic_biashttp://en.wikipedia.org/wiki/Psychologyhttp://en.wikipedia.org/wiki/Specification_(regression)http://en.wikipedia.org/wiki/Regression_analysishttp://en.wikipedia.org/wiki/Parameterhttp://en.wikipedia.org/wiki/Estimator_bias


12/22

Response bias is a type ofcognitive biaswhich can affect the resultsof astatistical surveyif respondents answer questions in the way theythink the questioner wants them to answer rather than according totheir truebeliefs. This may occur if the questioner is obviouslyangling for a particular answer (as inpush polling) or if therespondent wishes to please the questioner by answering what appearsto be the "morallyright" answer. An example of the latter might be ifa woman surveys a man on his attitudes todomestic violence, orsomeone who obviously cares about theenvironmentasks people howmuch they value awilderness area.

This occurs most often in the wording of the question. Response bias

is present when a question contains a leading opinion. For example,saying "Given that at the age of 18 people are old enough to fight anddie for your country, don't you think they should be able to drinkalcohol as well?" yields a response bias. It is better to say "Do youthink 18-year-olds should be able to drink alcohol?"

It also occurs in situations of voluntary response, such as phone-inpolls, where the people who care enough to call are not necessarily astatistically representative sample of the actual population.

Non-response biasis not the opposite of "response bias" and is not atype of cognitive bias: it occurs in astatistical surveyif those whorespond to the survey differ in the outcome variable (for example,evaluation of the need for financial aid) from those who do notrespond. Often, the differences, which may include race, gender orsocioeconomic status, are reported and/or accounted for throughstatistical modelling in any publication of the results.

Selection bias


Selection bias is astatistical biasin which there is an error inchoosing the individuals or groups to take part in ascientificstudy.[1]It is sometimes referred to as the selection effect. The term"selection bias" most often refers to the distortion ofastatisticalanalysis, resulting from the method of collecting samples.If the selection bias is not taken into account then any conclusions

drawn may be wrong.
http://en.wikipedia.org/wiki/Cognitive_biashttp://en.wikipedia.org/wiki/Cognitive_biashttp://en.wikipedia.org/wiki/Cognitive_biashttp://en.wikipedia.org/wiki/Statistical_surveyhttp://en.wikipedia.org/wiki/Statistical_surveyhttp://en.wikipedia.org/wiki/Statistical_surveyhttp://en.wikipedia.org/wiki/Beliefhttp://en.wikipedia.org/wiki/Beliefhttp://en.wikipedia.org/wiki/Beliefhttp://en.wikipedia.org/wiki/Push_pollinghttp://en.wikipedia.org/wiki/Push_pollinghttp://en.wikipedia.org/wiki/Morallyhttp://en.wikipedia.org/wiki/Morallyhttp://en.wikipedia.org/wiki/Domestic_violencehttp://en.wikipedia.org/wiki/Domestic_violencehttp://en.wikipedia.org/wiki/Domestic_violencehttp://en.wikipedia.org/wiki/Natural_environmenthttp://en.wikipedia.org/wiki/Natural_environmenthttp://en.wikipedia.org/wiki/Natural_environmenthttp://en.wikipedia.org/wiki/Wilderness_areahttp://en.wikipedia.org/wiki/Wilderness_areahttp://en.wikipedia.org/wiki/Wilderness_areahttp://en.wikipedia.org/wiki/Non-response_biashttp://en.wikipedia.org/wiki/Non-response_biashttp://en.wikipedia.org/wiki/Statistical_surveyhttp://en.wikipedia.org/wiki/Statistical_surveyhttp://en.wikipedia.org/wiki/Statistical_surveyhttp://en.wikipedia.org/wiki/Statistical_biashttp://en.wikipedia.org/wiki/Statistical_biashttp://en.wikipedia.org/wiki/Statistical_biashttp://en.wikipedia.org/wiki/Scientific_methodhttp://en.wikipedia.org/wiki/Scientific_methodhttp://en.wikipedia.org/wiki/Scientific_methodhttp://en.wikipedia.org/wiki/Selection_bias#cite_note-0http://en.wikipedia.org/wiki/Selection_bias#cite_note-0http://en.wikipedia.org/wiki/Selection_bias#cite_note-0http://en.wikipedia.org/wiki/Statisticalhttp://en.wikipedia.org/wiki/Statisticalhttp://en.wikipedia.org/wiki/Statisticalhttp://en.wikipedia.org/wiki/Statisticalhttp://en.wikipedia.org/wiki/Selection_bias#cite_note-0http://en.wikipedia.org/wiki/Scientific_methodhttp://en.wikipedia.org/wiki/Scientific_methodhttp://en.wikipedia.org/wiki/Statistical_biashttp://en.wikipedia.org/wiki/Statistical_surveyhttp://en.wikipedia.org/wiki/Non-response_biashttp://en.wikipedia.org/wiki/Wilderness_areahttp://en.wikipedia.org/wiki/Natural_environmenthttp://en.wikipedia.org/wiki/Domestic_violencehttp://en.wikipedia.org/wiki/Morallyhttp://en.wikipedia.org/wiki/Push_pollinghttp://en.wikipedia.org/wiki/Beliefhttp://en.wikipedia.org/wiki/Statistical_surveyhttp://en.wikipedia.org/wiki/Cognitive_bias


13/22

Contents

[hide]

1 Typeso 1.1 Sampling biaso 1.2 Time intervalo 1.3 Exposureo 1.4 Datao 1.5 Studieso 1.6 Attrition

2 Avoidance

3 Related issues

4 See also

5 Notes

[edit]Types

There are many types of possible selection bias, including:

[edit]Sampling bias

Sampling biasis systematic error due to a non-random sampleof apopulation,[2]causing some members of the population to be lesslikely to be included than others, resulting in abiased sample, definedas astatistical sampleof apopulation(or non-human factors) in whichall participants are not equally balanced or objectivelyrepresented.

[3]It is mostly classified as a subtype of selection

bias,[4]

sometimes specifically termed sample selection bias,[5][6]

butsome classify it as a separate type of bias.

[7]

A distinction, albeit not universally accepted, of sampling bias is thatit undermines theexternal validityof a test (the ability of its results tobe generalized to the rest of the population), while selection biasmainly addressesinternal validityfor differences or similarities foundin the sample at hand. In this sense, errors occurring in the process ofgathering the sample or cohort cause sampling bias, while errors inany process thereafter cause selection bias.

Examples of sampling bias includeself-selection, pre-screening oftrial participants, discounting trial subjects/tests that did not run to
http://en.wikipedia.org/wiki/Selection_biashttp://en.wikipedia.org/wiki/Selection_biashttp://en.wikipedia.org/wiki/Selection_bias#Typeshttp://en.wikipedia.org/wiki/Selection_bias#Typeshttp://en.wikipedia.org/wiki/Selection_bias#Sampling_biashttp://en.wikipedia.org/wiki/Selection_bias#Sampling_biashttp://en.wikipedia.org/wiki/Selection_bias#Time_intervalhttp://en.wikipedia.org/wiki/Selection_bias#Time_intervalhttp://en.wikipedia.org/wiki/Selection_bias#Exposurehttp://en.wikipedia.org/wiki/Selection_bias#Exposurehttp://en.wikipedia.org/wiki/Selection_bias#Datahttp://en.wikipedia.org/wiki/Selection_bias#Datahttp://en.wikipedia.org/wiki/Selection_bias#Studieshttp://en.wikipedia.org/wiki/Selection_bias#Studieshttp://en.wikipedia.org/wiki/Selection_bias#Attritionhttp://en.wikipedia.org/wiki/Selection_bias#Attritionhttp://en.wikipedia.org/wiki/Selection_bias#Avoidancehttp://en.wikipedia.org/wiki/Selection_bias#Avoidancehttp://en.wikipedia.org/wiki/Selection_bias#Related_issueshttp://en.wikipedia.org/wiki/Selection_bias#Related_issueshttp://en.wikipedia.org/wiki/Selection_bias#See_alsohttp://en.wikipedia.org/wiki/Selection_bias#See_alsohttp://en.wikipedia.org/wiki/Selection_bias#Noteshttp://en.wikipedia.org/wiki/Selection_bias#Noteshttp://en.wikipedia.org/w/index.php?title=Selection_bias&action=edit&section=1http://en.wikipedia.org/w/index.php?title=Selection_bias&action=edit&section=2http://en.wikipedia.org/wiki/Sampling_biashttp://en.wikipedia.org/wiki/Sampling_biashttp://en.wikipedia.org/wiki/Random_samplehttp://en.wikipedia.org/wiki/Random_samplehttp://en.wikipedia.org/wiki/Selection_bias#cite_note-1http://en.wikipedia.org/wiki/Selection_bias#cite_note-1http://en.wikipedia.org/wiki/Selection_bias#cite_note-1http://en.wikipedia.org/wiki/Biased_samplehttp://en.wikipedia.org/wiki/Biased_samplehttp://en.wikipedia.org/wiki/Statistical_samplehttp://en.wikipedia.org/wiki/Statistical_samplehttp://en.wikipedia.org/wiki/Statistical_samplehttp://en.wikipedia.org/wiki/Statistical_populationhttp://en.wikipedia.org/wiki/Statistical_populationhttp://en.wikipedia.org/wiki/Statistical_populationhttp://en.wikipedia.org/wiki/Selection_bias#cite_note-2http://en.wikipedia.org/wiki/Selection_bias#cite_note-2http://en.wikipedia.org/wiki/Selection_bias#cite_note-2http://en.wikipedia.org/wiki/Selection_bias#cite_note-3http://en.wikipedia.org/wiki/Selection_bias#cite_note-3http://en.wikipedia.org/wiki/Selection_bias#cite_note-3http://en.wikipedia.org/wiki/Selection_bias#cite_note-4http://en.wikipedia.org/wiki/Selection_bias#cite_note-4http://en.wikipedia.org/wiki/Selection_bias#cite_note-4http://en.wikipedia.org/wiki/Selection_bias#cite_note-6http://en.wikipedia.org/wiki/Selection_bias#cite_note-6http://en.wikipedia.org/wiki/Selection_bias#cite_note-6http://en.wikipedia.org/wiki/External_validityhttp://en.wikipedia.org/wiki/External_validityhttp://en.wikipedia.org/wiki/External_validityhttp://en.wikipedia.org/wiki/Internal_validityhttp://en.wikipedia.org/wiki/Internal_validityhttp://en.wikipedia.org/wiki/Internal_validityhttp://en.wikipedia.org/wiki/Self-selectionhttp://en.wikipedia.org/wiki/Self-selectionhttp://en.wikipedia.org/wiki/Self-selectionhttp://en.wikipedia.org/wiki/Self-selectionhttp://en.wikipedia.org/wiki/Internal_validityhttp://en.wikipedia.org/wiki/External_validityhttp://en.wikipedia.org/wiki/Selection_bias#cite_note-6http://en.wikipedia.org/wiki/Selection_bias#cite_note-4http://en.wikipedia.org/wiki/Selection_bias#cite_note-4http://en.wikipedia.org/wiki/Selection_bias#cite_note-3http://en.wikipedia.org/wiki/Selection_bias#cite_note-2http://en.wikipedia.org/wiki/Statistical_populationhttp://en.wikipedia.org/wiki/Statistical_samplehttp://en.wikipedia.org/wiki/Biased_samplehttp://en.wikipedia.org/wiki/Selection_bias#cite_note-1http://en.wikipedia.org/wiki/Random_samplehttp://en.wikipedia.org/wiki/Sampling_biashttp://en.wikipedia.org/w/index.php?title=Selection_bias&action=edit&section=2http://en.wikipedia.org/w/index.php?title=Selection_bias&action=edit&section=1http://en.wikipedia.org/wiki/Selection_bias#Noteshttp://en.wikipedia.org/wiki/Selection_bias#See_alsohttp://en.wikipedia.org/wiki/Selection_bias#Related_issueshttp://en.wikipedia.org/wiki/Selection_bias#Avoidancehttp://en.wikipedia.org/wiki/Selection_bias#Attritionhttp://en.wikipedia.org/wiki/Selection_bias#Studieshttp://en.wikipedia.org/wiki/Selection_bias#Datahttp://en.wikipedia.org/wiki/Selection_bias#Exposurehttp://en.wikipedia.org/wiki/Selection_bias#Time_intervalhttp://en.wikipedia.org/wiki/Selection_bias#Sampling_biashttp://en.wikipedia.org/wiki/Selection_bias#Typeshttp://en.wikipedia.org/wiki/Selection_bias


14/22

completion and migration bias by excluding subjects who haverecently moved into or out of the study area.

[edit]Time interval

Early termination of a trial at a time when its results support adesired conclusion.

A trial may be terminated early at an extreme value (oftenforethicalreasons), but the extreme value is likely to be reached bythe variable with the largestvariance, even if all variables have asimilarmean.

[edit]Exposure

Susceptibility bias Clinical susceptibility bias, when one disease predisposes for a

second disease, and the treatment for the first diseaseerroneously appears to predispose to the second disease. Forexample,postmenopausal syndromegives a higher likelihood ofalso developingendometrial cancer, so estrogens given for thepostmenopausal syndrome may receive a higher than actualblame for causing endometrial cancer.

[8]

Protopathic bias, when a treatment for the first symptoms of adisease or other outcome appear to cause the outcome. It is apotential bias when there is a lag time from the first symptomsand start of treatment before actual diagnosis.[8]It can bemitigated bylagging, that is, exclusion of exposures thatoccurred in a certain time period before diagnosis.

[9]

Indication bias, a potential mix up between cause and effectwhen exposure is dependent on indication, e.g. a treatment is

given to people in high risk of acquiring a disease, potentiallycausing a preponderance of treated people among thoseacquiring the disease. This may cause an erroneous appearanceof the treatment being a cause of the disease.[10]

[edit]Data

Partitioning data with knowledge of the contents of the partitions,and then analyzing them with tests designed for blindly chosen

partitions.
http://en.wikipedia.org/w/index.php?title=Selection_bias&action=edit&section=3http://en.wikipedia.org/wiki/Ethicalhttp://en.wikipedia.org/wiki/Ethicalhttp://en.wikipedia.org/wiki/Ethicalhttp://en.wikipedia.org/wiki/Variancehttp://en.wikipedia.org/wiki/Variancehttp://en.wikipedia.org/wiki/Variancehttp://en.wikipedia.org/wiki/Meanhttp://en.wikipedia.org/wiki/Meanhttp://en.wikipedia.org/wiki/Meanhttp://en.wikipedia.org/w/index.php?title=Selection_bias&action=edit&section=4http://en.wikipedia.org/w/index.php?title=Postmenopausal_syndrome&action=edit&redlink=1http://en.wikipedia.org/w/index.php?title=Postmenopausal_syndrome&action=edit&redlink=1http://en.wikipedia.org/w/index.php?title=Postmenopausal_syndrome&action=edit&redlink=1http://en.wikipedia.org/wiki/Endometrial_cancerhttp://en.wikipedia.org/wiki/Endometrial_cancerhttp://en.wikipedia.org/wiki/Endometrial_cancerhttp://en.wikipedia.org/wiki/Selection_bias#cite_note-fenstein-7http://en.wikipedia.org/wiki/Selection_bias#cite_note-fenstein-7http://en.wikipedia.org/wiki/Selection_bias#cite_note-fenstein-7http://en.wikipedia.org/wiki/Selection_bias#cite_note-fenstein-7http://en.wikipedia.org/wiki/Selection_bias#cite_note-fenstein-7http://en.wikipedia.org/wiki/Selection_bias#cite_note-fenstein-7http://en.wikipedia.org/wiki/Lagging_(epidemiology)http://en.wikipedia.org/wiki/Lagging_(epidemiology)http://en.wikipedia.org/wiki/Lagging_(epidemiology)http://en.wikipedia.org/wiki/Selection_bias#cite_note-8http://en.wikipedia.org/wiki/Selection_bias#cite_note-8http://en.wikipedia.org/wiki/Selection_bias#cite_note-8http://en.wikipedia.org/wiki/Selection_bias#cite_note-9http://en.wikipedia.org/wiki/Selection_bias#cite_note-9http://en.wikipedia.org/wiki/Selection_bias#cite_note-9http://en.wikipedia.org/w/index.php?title=Selection_bias&action=edit&section=5http://en.wikipedia.org/w/index.php?title=Selection_bias&action=edit&section=5http://en.wikipedia.org/wiki/Selection_bias#cite_note-9http://en.wikipedia.org/wiki/Selection_bias#cite_note-8http://en.wikipedia.org/wiki/Lagging_(epidemiology)http://en.wikipedia.org/wiki/Selection_bias#cite_note-fenstein-7http://en.wikipedia.org/wiki/Selection_bias#cite_note-fenstein-7http://en.wikipedia.org/wiki/Endometrial_cancerhttp://en.wikipedia.org/w/index.php?title=Postmenopausal_syndrome&action=edit&redlink=1http://en.wikipedia.org/w/index.php?title=Selection_bias&action=edit&section=4http://en.wikipedia.org/wiki/Meanhttp://en.wikipedia.org/wiki/Variancehttp://en.wikipedia.org/wiki/Ethicalhttp://en.wikipedia.org/w/index.php?title=Selection_bias&action=edit&section=3


15/22

Rejection of "bad" data on arbitrary grounds, instead of accordingto previously stated or generally agreed criteria.

Rejection of "outliers" on statistical grounds that fail to take intoaccount important information that could be derived from "wild"observations.[11]

[edit]Studies

Selection of which studies to include in ameta-analysis(seealsocombinatorial meta-analysis).

Performing repeated experiments and reporting only the mostfavorable results, perhaps relabelling lab records of otherexperiments as "calibration tests", "instrumentation errors" or

"preliminary surveys". Presenting the most significant result of adata dredgeas if it were

a single experiment (which is logically the same as the previousitem, but is seen as much less dishonest).

[edit]Attrition

Attrition bias is a kind of selection bias caused by attrition (loss ofparticipants),

[12]discounting trial subjects/tests that did not run to

completion. It includesdropout

,nonresponse

(lowerresponserate), withdrawal andprotocol deviators. It gives biased results whereit is unequal in regard to exposure and/or outcome. For example, in atest of a dieting program, the researcher may simply reject everyonewho drops out of the trial, but most of those who drop out are thosefor whom it was not working. Different loss of subjects inintervention and comparison group may change the characteristics ofthese groups and outcomes irrespective of the studiedintervention.[12]

[edit]AvoidanceIn the general case, selection biases cannot be overcome withstatistical analysis of existing data alone, thoughHeckmancorrectionmay be used in special cases. An informal assessment ofthe degree of selection bias can be made by examining correlationsbetweenexogenous(background) variables and a treatment indicator.However, inregressionmodels, it is correlationbetween unobserveddeterminants of the outcome

and unobserveddeterminants of selection into the sample which bias
http://en.wikipedia.org/wiki/Selection_bias#cite_note-10http://en.wikipedia.org/wiki/Selection_bias#cite_note-10http://en.wikipedia.org/wiki/Selection_bias#cite_note-10http://en.wikipedia.org/w/index.php?title=Selection_bias&action=edit&section=6http://en.wikipedia.org/wiki/Meta-analysishttp://en.wikipedia.org/wiki/Meta-analysishttp://en.wikipedia.org/wiki/Meta-analysishttp://en.wikipedia.org/wiki/Combinatorial_meta-analysishttp://en.wikipedia.org/wiki/Combinatorial_meta-analysishttp://en.wikipedia.org/wiki/Data_dredginghttp://en.wikipedia.org/wiki/Data_dredginghttp://en.wikipedia.org/wiki/Data_dredginghttp://en.wikipedia.org/w/index.php?title=Selection_bias&action=edit&section=7http://en.wikipedia.org/wiki/Selection_bias#cite_note-Juni-11http://en.wikipedia.org/wiki/Selection_bias#cite_note-Juni-11http://en.wikipedia.org/wiki/Selection_bias#cite_note-Juni-11http://en.wikipedia.org/wiki/Response_ratehttp://en.wikipedia.org/wiki/Response_ratehttp://en.wikipedia.org/wiki/Response_ratehttp://en.wikipedia.org/wiki/Health_interventionhttp://en.wikipedia.org/wiki/Health_interventionhttp://en.wikipedia.org/wiki/Selection_bias#cite_note-Juni-11http://en.wikipedia.org/wiki/Selection_bias#cite_note-Juni-11http://en.wikipedia.org/wiki/Selection_bias#cite_note-Juni-11http://en.wikipedia.org/w/index.php?title=Selection_bias&action=edit&section=8http://en.wikipedia.org/wiki/Heckman_correctionhttp://en.wikipedia.org/wiki/Heckman_correctionhttp://en.wikipedia.org/wiki/Heckman_correctionhttp://en.wikipedia.org/wiki/Heckman_correctionhttp://en.wikipedia.org/wiki/Exogenoushttp://en.wikipedia.org/wiki/Exogenoushttp://en.wikipedia.org/wiki/Exogenoushttp://en.wikipedia.org/wiki/Regression_analysishttp://en.wikipedia.org/wiki/Regression_analysishttp://en.wikipedia.org/wiki/Regression_analysishttp://en.wikipedia.org/wiki/Regression_analysishttp://en.wikipedia.org/wiki/Exogenoushttp://en.wikipedia.org/wiki/Heckman_correctionhttp://en.wikipedia.org/wiki/Heckman_correctionhttp://en.wikipedia.org/w/index.php?title=Selection_bias&action=edit&section=8http://en.wikipedia.org/wiki/Selection_bias#cite_note-Juni-11http://en.wikipedia.org/wiki/Health_interventionhttp://en.wikipedia.org/wiki/Response_ratehttp://en.wikipedia.org/wiki/Response_ratehttp://en.wikipedia.org/wiki/Selection_bias#cite_note-Juni-11http://en.wikipedia.org/w/index.php?title=Selection_bias&action=edit&section=7http://en.wikipedia.org/wiki/Data_dredginghttp://en.wikipedia.org/wiki/Combinatorial_meta-analysishttp://en.wikipedia.org/wiki/Meta-analysishttp://en.wikipedia.org/w/index.php?title=Selection_bias&action=edit&section=6http://en.wikipedia.org/wiki/Selection_bias#cite_note-10


16/22

estimates, and this correlation between unobservables cannot bedirectly assessed by the observed determinants of treatment.[13]

[edit]Related issues

Selection bias is closely related to:

publication biasorreporting bias, the distortion produced incommunity perception ormeta-analysesby not publishinguninteresting (usually negative) results, or results which go againstthe experimenter's prejudices, a sponsor's interests, or communityexpectations.

confirmation bias, the distortion produced by experiments that aredesigned to seek confirmatory evidence instead of trying todisprove the hypothesis.

exclusion bias, results from applying different criteria to cases andcontrols in regards to participation eligibility for a study/differentvariables serving as basis for exclusion.

[edit] Systematic error


(Redirected fromSystematic bias)Systematic errors arebiasesinmeasurementwhich lead to thesituation where themeanof many separate measurements differssignificantly from the actual value of the measured attribute. Allmeasurements are prone to systematic errors, often of severaldifferent types. Sources of systematic error may be imperfectcalibration of measurement instruments (zero error), changes intheenvironmentwhich interfere with the measurement process and

sometimes imperfect methods ofobservationcan be either zero erroror percentage error. For example, consider an experimenter taking areading of the time period of a pendulum swinging past a fiducialmark: If his stop-watch or timer starts with 1 second on the clock thenall of his results will be off by 1 second (zero error). If theexperimenter repeats this experiment twenty times (starting at 1second each time), then there will be apercentage errorin thecalculated average of his results; the final result will be slightly larger

than the true period.Distancemeasured byradarwill be
http://en.wikipedia.org/wiki/Selection_bias#cite_note-12http://en.wikipedia.org/wiki/Selection_bias#cite_note-12http://en.wikipedia.org/wiki/Selection_bias#cite_note-12http://en.wikipedia.org/w/index.php?title=Selection_bias&action=edit&section=9http://en.wikipedia.org/wiki/Publication_biashttp://en.wikipedia.org/wiki/Publication_biashttp://en.wikipedia.org/wiki/Reporting_biashttp://en.wikipedia.org/wiki/Reporting_biashttp://en.wikipedia.org/wiki/Reporting_biashttp://en.wikipedia.org/wiki/Meta-analysishttp://en.wikipedia.org/wiki/Meta-analysishttp://en.wikipedia.org/wiki/Meta-analysishttp://en.wikipedia.org/wiki/Confirmation_biashttp://en.wikipedia.org/wiki/Confirmation_biashttp://en.wikipedia.org/w/index.php?title=Selection_bias&action=edit&section=10http://en.wikipedia.org/w/index.php?title=Systematic_bias&redirect=nohttp://en.wikipedia.org/w/index.php?title=Systematic_bias&redirect=nohttp://en.wikipedia.org/wiki/Biashttp://en.wikipedia.org/wiki/Biashttp://en.wikipedia.org/wiki/Biashttp://en.wikipedia.org/wiki/Measurementhttp://en.wikipedia.org/wiki/Measurementhttp://en.wikipedia.org/wiki/Measurementhttp://en.wikipedia.org/wiki/Meanhttp://en.wikipedia.org/wiki/Meanhttp://en.wikipedia.org/wiki/Meanhttp://en.wikipedia.org/wiki/Surroundingshttp://en.wikipedia.org/wiki/Surroundingshttp://en.wikipedia.org/wiki/Surroundingshttp://en.wikipedia.org/wiki/Observationhttp://en.wikipedia.org/wiki/Observationhttp://en.wikipedia.org/wiki/Observationhttp://en.wikipedia.org/wiki/Percentage_errorhttp://en.wikipedia.org/wiki/Percentage_errorhttp://en.wikipedia.org/wiki/Percentage_errorhttp://en.wikipedia.org/wiki/Distancehttp://en.wikipedia.org/wiki/Distancehttp://en.wikipedia.org/wiki/Distancehttp://en.wikipedia.org/wiki/Radarhttp://en.wikipedia.org/wiki/Radarhttp://en.wikipedia.org/wiki/Radarhttp://en.wikipedia.org/wiki/Radarhttp://en.wikipedia.org/wiki/Distancehttp://en.wikipedia.org/wiki/Percentage_errorhttp://en.wikipedia.org/wiki/Observationhttp://en.wikipedia.org/wiki/Surroundingshttp://en.wikipedia.org/wiki/Meanhttp://en.wikipedia.org/wiki/Measurementhttp://en.wikipedia.org/wiki/Biashttp://en.wikipedia.org/w/index.php?title=Systematic_bias&redirect=nohttp://en.wikipedia.org/w/index.php?title=Selection_bias&action=edit&section=10http://en.wikipedia.org/wiki/Confirmation_biashttp://en.wikipedia.org/wiki/Meta-analysishttp://en.wikipedia.org/wiki/Reporting_biashttp://en.wikipedia.org/wiki/Publication_biashttp://en.wikipedia.org/w/index.php?title=Selection_bias&action=edit&section=9http://en.wikipedia.org/wiki/Selection_bias#cite_note-12


17/22

systematically overestimated if the slight slowing down of the wavesin air is not accounted for. Incorrect zeroing of an instrument leadingto a zero error is an example of systematic error in instrumentation.

Systematic errors may also be present in the result ofanestimatebased on a mathematical model or physical law. Forinstance, the estimatedoscillation frequencyof apendulumwill besystematically in error if slight movement of the support is notaccounted for.

Systematic errors can be either constant, or be related (e.g.proportional or a percentage) to the actual value of the measuredquantity, or even to the value of a different quantity (the reading of

arulercan be affected by environment temperature). When they areconstant, they are simply due to incorrect zeroing of the instrument.When they are not constant, they can change sign. For instance, if athermometer is affected by a proportional systematic error equal to2% of the actual temperature, and the actual temperature is 200, 0,or 100, the measured temperature will be 204 (systematic error =+4), 0 (null systematic error) or 102 (systematic error = 2),respectively. Thus, the temperature will be overestimated when it will

be above zero, and underestimated when it will be below zero.Constant systematic errors are very difficult to deal with, becausetheir effects are only observable if they can be removed. Such errorscannot be removed by repeating measurements or averaging largenumbers of results. A common method to remove systematic error isthroughcalibrationof the measurement instrument.

In astatisticalcontext, the term systematic errorusually arises wherethe sizes and directions of possible errors are unknown.

Contents

[hide]

1 Drift

2 Systematic versus random error

3 See also

4 References
http://en.wikipedia.org/wiki/Computational_mechanicshttp://en.wikipedia.org/wiki/Computational_mechanicshttp://en.wikipedia.org/wiki/Computational_mechanicshttp://en.wikipedia.org/wiki/Oscillation_frequencyhttp://en.wikipedia.org/wiki/Oscillation_frequencyhttp://en.wikipedia.org/wiki/Oscillation_frequencyhttp://en.wikipedia.org/wiki/Pendulumhttp://en.wikipedia.org/wiki/Pendulumhttp://en.wikipedia.org/wiki/Pendulumhttp://en.wikipedia.org/wiki/Rulerhttp://en.wikipedia.org/wiki/Rulerhttp://en.wikipedia.org/wiki/Rulerhttp://en.wikipedia.org/wiki/Calibrationhttp://en.wikipedia.org/wiki/Calibrationhttp://en.wikipedia.org/wiki/Calibrationhttp://en.wikipedia.org/wiki/Statisticshttp://en.wikipedia.org/wiki/Statisticshttp://en.wikipedia.org/wiki/Statisticshttp://en.wikipedia.org/wiki/Systematic_biashttp://en.wikipedia.org/wiki/Systematic_biashttp://en.wikipedia.org/wiki/Systematic_bias#Drifthttp://en.wikipedia.org/wiki/Systematic_bias#Drifthttp://en.wikipedia.org/wiki/Systematic_bias#Systematic_versus_random_errorhttp://en.wikipedia.org/wiki/Systematic_bias#Systematic_versus_random_errorhttp://en.wikipedia.org/wiki/Systematic_bias#See_alsohttp://en.wikipedia.org/wiki/Systematic_bias#See_alsohttp://en.wikipedia.org/wiki/Systematic_bias#Referenceshttp://en.wikipedia.org/wiki/Systematic_bias#Referenceshttp://en.wikipedia.org/wiki/Systematic_bias#Referenceshttp://en.wikipedia.org/wiki/Systematic_bias#See_alsohttp://en.wikipedia.org/wiki/Systematic_bias#Systematic_versus_random_errorhttp://en.wikipedia.org/wiki/Systematic_bias#Drifthttp://en.wikipedia.org/wiki/Systematic_biashttp://en.wikipedia.org/wiki/Statisticshttp://en.wikipedia.org/wiki/Calibrationhttp://en.wikipedia.org/wiki/Rulerhttp://en.wikipedia.org/wiki/Pendulumhttp://en.wikipedia.org/wiki/Oscillation_frequencyhttp://en.wikipedia.org/wiki/Computational_mechanics


18/22

[edit]Drift

Systematic errors which change during an experiment (drift) areeasier to detect. Measurements show trends with time rather than

varying randomly about amean.Drift is evident if a measurement of a constant quantity is repeatedseveral times and the measurements drift one way during theexperiment, for example if each measurement is higher than theprevious measurement which could perhaps occur if an instrumentbecomes warmer during the experiment. If the measured quantity isvariable, it is possible to detect a drift by checking the zero readingduring the experiment as well as at the start of the experiment (indeed,

the zero reading is a measurement of a constant quantity). If the zeroreading is consistently above or below zero, a systematic error ispresent. If this cannot be eliminated, for instance by resetting theinstrument immediately before the experiment, it needs to be allowedfor by subtracting its (possibly time-varying) value from the readings,and by taking it into account in assessing the accuracy of themeasurement.

If no pattern in a series of repeated measurements is evident, thepresence of fixed systematic errors can only be found if themeasurements are checked, either by measuring a known quantity orby comparing the readings with readings made using a differentapparatus, known to be more accurate. For example, suppose thetiming of a pendulum using an accuratestopwatchseveral times givesreadings randomly distributed about the mean. A systematic error ispresent if the stopwatch is checked against the 'speaking clock' of thetelephone system and found to be running slow or fast. Clearly, the

pendulum timings need to be corrected according to how fast or slowthe stopwatch was found to be running. Measuring instruments suchasammetersandvoltmetersneed to be checked periodically againstknown standards.

Systematic errors can also be detected by measuring already knownquantities. For example, aspectrometerfitted with adiffractiongratingmay be checked by using it to measure thewavelength of the

D-lines of thesodiumelectromagnetic spectrumwhich are at 600nmand 589.6 nm. The measurements may be used to determine the
http://en.wikipedia.org/wiki/Systematic_bias#Referenceshttp://en.wikipedia.org/w/index.php?title=Systematic_error&action=edit&section=1http://en.wikipedia.org/wiki/Drifthttp://en.wikipedia.org/wiki/Meanhttp://en.wikipedia.org/wiki/Meanhttp://en.wikipedia.org/wiki/Meanhttp://en.wikipedia.org/wiki/Stopwatchhttp://en.wikipedia.org/wiki/Stopwatchhttp://en.wikipedia.org/wiki/Speaking_clockhttp://en.wikipedia.org/wiki/Speaking_clockhttp://en.wikipedia.org/wiki/Speaking_clockhttp://en.wikipedia.org/wiki/Ammeterhttp://en.wikipedia.org/wiki/Ammeterhttp://en.wikipedia.org/wiki/Ammeterhttp://en.wikipedia.org/wiki/Voltmeterhttp://en.wikipedia.org/wiki/Voltmeterhttp://en.wikipedia.org/wiki/Voltmeterhttp://en.wikipedia.org/wiki/Spectrometerhttp://en.wikipedia.org/wiki/Spectrometerhttp://en.wikipedia.org/wiki/Spectrometerhttp://en.wikipedia.org/wiki/Diffraction_gratinghttp://en.wikipedia.org/wiki/Diffraction_gratinghttp://en.wikipedia.org/wiki/Diffraction_gratinghttp://en.wikipedia.org/wiki/Diffraction_gratinghttp://en.wikipedia.org/wiki/Wavelengthhttp://en.wikipedia.org/wiki/Sodiumhttp://en.wikipedia.org/wiki/Sodiumhttp://en.wikipedia.org/wiki/Electromagnetic_spectrumhttp://en.wikipedia.org/wiki/Electromagnetic_spectrumhttp://en.wikipedia.org/wiki/Electromagnetic_spectrumhttp://en.wikipedia.org/wiki/Electromagnetic_spectrumhttp://en.wikipedia.org/wiki/Sodiumhttp://en.wikipedia.org/wiki/Wavelengthhttp://en.wikipedia.org/wiki/Diffraction_gratinghttp://en.wikipedia.org/wiki/Diffraction_gratinghttp://en.wikipedia.org/wiki/Spectrometerhttp://en.wikipedia.org/wiki/Voltmeterhttp://en.wikipedia.org/wiki/Ammeterhttp://en.wikipedia.org/wiki/Speaking_clockhttp://en.wikipedia.org/wiki/Stopwatchhttp://en.wikipedia.org/wiki/Meanhttp://en.wikipedia.org/wiki/Drifthttp://en.wikipedia.org/w/index.php?title=Systematic_error&action=edit&section=1


19/22

number of lines per millimetre of the diffraction grating, which canthen be used to measure the wavelength of any other spectral line.

[edit]Systematic versus random error

Measurement errors can be divided into two components:randomerrorand systematic error.[1]Random error is always present in ameasurement. It is caused by inherently unpredictable fluctuations inthe readings of a measurement apparatus or in the experimenter'sinterpretation of the instrumental reading. Random errors show up asdifferent results for ostensibly the same repeated measurement.Systematic error cannot be discovered this way because it alwayspushes the results in the same direction. If the cause of a systematic

error can be identified, then it can usually be eliminated.

Systemic bias is the inherent tendency of a process to favor particularoutcomes. The term is aneologismthat generally refers to humansystems; the analogous problem in non-human systems (such asmeasurement instruments or mathematical modelsused to estimatephysical quantities) is often calledsystematic bias, and leads

tosystematic errorin measurements or estimates.[citation needed]

Contents

[hide]

1 Bias in human institutions

2 Examples

3 Systemic versus systematic bias

4 See also5 References

6 Further reading

[edit]Bias in human institutions

For example, one might refer to the systemic, systematic, orinstitutional bias of a particular institution in devaluing contributionsby women, men or ethnic minorities. For example, a poetry

competition that was consistently won by white women could be
http://en.wikipedia.org/w/index.php?title=Systematic_error&action=edit&section=2http://en.wikipedia.org/wiki/Random_errorhttp://en.wikipedia.org/wiki/Random_errorhttp://en.wikipedia.org/wiki/Random_errorhttp://en.wikipedia.org/wiki/Random_errorhttp://en.wikipedia.org/wiki/Systematic_bias#cite_note-Taylor-0http://en.wikipedia.org/wiki/Systematic_bias#cite_note-Taylor-0http://en.wikipedia.org/wiki/Systematic_bias#cite_note-Taylor-0http://en.wikipedia.org/wiki/Neologismhttp://en.wikipedia.org/wiki/Neologismhttp://en.wikipedia.org/wiki/Neologismhttp://en.wikipedia.org/wiki/Computational_mechanicshttp://en.wikipedia.org/wiki/Computational_mechanicshttp://en.wikipedia.org/wiki/Systematic_biashttp://en.wikipedia.org/wiki/Systematic_biashttp://en.wikipedia.org/wiki/Systematic_biashttp://en.wikipedia.org/wiki/Systematic_errorhttp://en.wikipedia.org/wiki/Systematic_errorhttp://en.wikipedia.org/wiki/Systematic_errorhttp://en.wikipedia.org/wiki/Wikipedia:Citation_neededhttp://en.wikipedia.org/wiki/Wikipedia:Citation_neededhttp://en.wikipedia.org/wiki/Wikipedia:Citation_neededhttp://en.wikipedia.org/wiki/Systemic_biashttp://en.wikipedia.org/wiki/Systemic_biashttp://en.wikipedia.org/wiki/Systemic_bias#Bias_in_human_institutionshttp://en.wikipedia.org/wiki/Systemic_bias#Bias_in_human_institutionshttp://en.wikipedia.org/wiki/Systemic_bias#Exampleshttp://en.wikipedia.org/wiki/Systemic_bias#Exampleshttp://en.wikipedia.org/wiki/Systemic_bias#Systemic_versus_systematic_biashttp://en.wikipedia.org/wiki/Systemic_bias#Systemic_versus_systematic_biashttp://en.wikipedia.org/wiki/Systemic_bias#See_alsohttp://en.wikipedia.org/wiki/Systemic_bias#See_alsohttp://en.wikipedia.org/wiki/Systemic_bias#Referenceshttp://en.wikipedia.org/wiki/Systemic_bias#Referenceshttp://en.wikipedia.org/wiki/Systemic_bias#Further_readinghttp://en.wikipedia.org/wiki/Systemic_bias#Further_readinghttp://en.wikipedia.org/w/index.php?title=Systemic_bias&action=edit&section=1http://en.wikipedia.org/w/index.php?title=Systemic_bias&action=edit&section=1http://en.wikipedia.org/wiki/Systemic_bias#Further_readinghttp://en.wikipedia.org/wiki/Systemic_bias#Referenceshttp://en.wikipedia.org/wiki/Systemic_bias#See_alsohttp://en.wikipedia.org/wiki/Systemic_bias#Systemic_versus_systematic_biashttp://en.wikipedia.org/wiki/Systemic_bias#Exampleshttp://en.wikipedia.org/wiki/Systemic_bias#Bias_in_human_institutionshttp://en.wikipedia.org/wiki/Systemic_biashttp://en.wikipedia.org/wiki/Wikipedia:Citation_neededhttp://en.wikipedia.org/wiki/Systematic_errorhttp://en.wikipedia.org/wiki/Systematic_biashttp://en.wikipedia.org/wiki/Computational_mechanicshttp://en.wikipedia.org/wiki/Neologismhttp://en.wikipedia.org/wiki/Systematic_bias#cite_note-Taylor-0http://en.wikipedia.org/wiki/Random_errorhttp://en.wikipedia.org/wiki/Random_errorhttp://en.wikipedia.org/w/index.php?title=Systematic_error&action=edit&section=2


20/22

subject to suspicion of a bias if there were no inherent reason thatwhite women would consistently be the best poets. Such a bias couldbe deliberate on the part of the judges or entirely unconscious. [citationneeded]

For example, the poetry contest might be judged by a pool drawnfrom its own previous winners, reasoning that prize-winning poets arethe best to judge a poetry contest. However, it might be that inaddition to choosing for poetic skill, they are also inclined to choosepeople with whom they have values in common, either about poetryor about other matters, resulting in a continuous stream ofprizewinning white female poets. In this case, the bias could arise

from either conscious or unconscious defenseofgenderandracialinterests or simply from their sharedpoint ofview. In either case, it results in a biased representation of the realitythey are describing in terms of quality of poets and poetry.

[citation needed]

Becausecognitive biasis inherent in the experiences, loyalties, andrelationships of people in their daily lives, it cannot be eliminated byeducation or training, butawarenessof biases can be enhanced,allowing for the adoption of compensating correction mechanisms.

For example, the theory behindaffirmative actionin theUnitedStatesis precisely to counter biases in matters of gender, race,andethnicity, by opening up institutional participation to people witha wider range of backgrounds, and hence presumably a wider range ofpoints of view. InIndiathe system ofscheduled castes and tribeswasintended to address systemic bias within thecastesystem. Similar toaffirmative action, it mandates the hiring of persons within certaindesignated groups. However, in both instances (as well as numerous

others), many people claim that a reverse systemic bias now exists[1]

.[edit]Examples

Financial Weekreported May 5, 2008 (emphasisadded):

But we travel in a world with a systemic bias to optimism thattypically chooses to avoid the topic of the impending bursting ofinvestment bubbles. Collectively, this is done for career or businessreasons. As discussed many times in the investment business,

pessimism or realism in the face of probable trouble is just plain badfor business and bad for careers. What I am only slowly realizing,
http://en.wikipedia.org/wiki/Wikipedia:Citation_neededhttp://en.wikipedia.org/wiki/Wikipedia:Citation_neededhttp://en.wikipedia.org/wiki/Wikipedia:Citation_neededhttp://en.wikipedia.org/wiki/Wikipedia:Citation_neededhttp://en.wikipedia.org/wiki/Genderhttp://en.wikipedia.org/wiki/Genderhttp://en.wikipedia.org/wiki/Genderhttp://en.wikipedia.org/wiki/Race_(classification_of_human_beings)http://en.wikipedia.org/wiki/Race_(classification_of_human_beings)http://en.wikipedia.org/wiki/Race_(classification_of_human_beings)http://en.wikipedia.org/wiki/Perspective_(cognitive)http://en.wikipedia.org/wiki/Perspective_(cognitive)http://en.wikipedia.org/wiki/Perspective_(cognitive)http://en.wikipedia.org/wiki/Perspective_(cognitive)http://en.wikipedia.org/wiki/Wikipedia:Citation_neededhttp://en.wikipedia.org/wiki/Wikipedia:Citation_neededhttp://en.wikipedia.org/wiki/Wikipedia:Citation_neededhttp://en.wikipedia.org/wiki/Cognitive_biashttp://en.wikipedia.org/wiki/Cognitive_biashttp://en.wikipedia.org/wiki/Cognitive_biashttp://en.wikipedia.org/wiki/Awarenesshttp://en.wikipedia.org/wiki/Awarenesshttp://en.wikipedia.org/wiki/Awarenesshttp://en.wikipedia.org/wiki/Affirmative_actionhttp://en.wikipedia.org/wiki/Affirmative_actionhttp://en.wikipedia.org/wiki/Affirmative_actionhttp://en.wikipedia.org/wiki/United_Stateshttp://en.wikipedia.org/wiki/United_Stateshttp://en.wikipedia.org/wiki/United_Stateshttp://en.wikipedia.org/wiki/United_Stateshttp://en.wikipedia.org/wiki/Ethnicityhttp://en.wikipedia.org/wiki/Ethnicityhttp://en.wikipedia.org/wiki/Ethnicityhttp://en.wikipedia.org/wiki/Indiahttp://en.wikipedia.org/wiki/Indiahttp://en.wikipedia.org/wiki/Scheduled_castes_and_tribeshttp://en.wikipedia.org/wiki/Scheduled_castes_and_tribeshttp://en.wikipedia.org/wiki/Scheduled_castes_and_tribeshttp://en.wikipedia.org/wiki/Castehttp://en.wikipedia.org/wiki/Castehttp://en.wikipedia.org/wiki/Castehttp://en.wikipedia.org/wiki/Systemic_bias#cite_note-0http://en.wikipedia.org/wiki/Systemic_bias#cite_note-0http://en.wikipedia.org/w/index.php?title=Systemic_bias&action=edit&section=2http://en.wikipedia.org/wiki/Crain_Communications_Inc.http://en.wikipedia.org/wiki/Crain_Communications_Inc.http://en.wikipedia.org/wiki/Emphasishttp://en.wikipedia.org/wiki/Emphasishttp://en.wikipedia.org/wiki/Emphasishttp://en.wikipedia.org/wiki/Emphasishttp://en.wikipedia.org/wiki/Crain_Communications_Inc.http://en.wikipedia.org/w/index.php?title=Systemic_bias&action=edit&section=2http://en.wikipedia.org/wiki/Systemic_bias#cite_note-0http://en.wikipedia.org/wiki/Castehttp://en.wikipedia.org/wiki/Scheduled_castes_and_tribeshttp://en.wikipedia.org/wiki/Indiahttp://en.wikipedia.org/wiki/Ethnicityhttp://en.wikipedia.org/wiki/United_Stateshttp://en.wikipedia.org/wiki/United_Stateshttp://en.wikipedia.org/wiki/Affirmative_actionhttp://en.wikipedia.org/wiki/Awarenesshttp://en.wikipedia.org/wiki/Cognitive_biashttp://en.wikipedia.org/wiki/Wikipedia:Citation_neededhttp://en.wikipedia.org/wiki/Perspective_(cognitive)http://en.wikipedia.org/wiki/Perspective_(cognitive)http://en.wikipedia.org/wiki/Race_(classification_of_human_beings)http://en.wikipedia.org/wiki/Genderhttp://en.wikipedia.org/wiki/Wikipedia:Citation_neededhttp://en.wikipedia.org/wiki/Wikipedia:Citation_needed


21/22

though, is how similar the career risk appears to be for the Fed. Itdoesn't want to move against bubbles because Congress and businessdo not like it and show their dislike in unmistakable terms. EvenFederal reserve chairmen get bullied and have their faces slapped ifthey stick to their guns, which will, not surprisingly, be rare sinceeveryone values his career or does not want to be replaced laMr.Volcker. So, be as optimistic as possible, be nice to everyone, baileveryone out and hope for the best. If all goes well, after all, you willhave a lot of grateful bailees who will happily hire you for $300,000 apop.

[2]

[edit]Systemic versus systematic biasThere is some contention over the choice of the word systemic asopposed to systematic.

[citation needed]

"Systemic bias" and the older, more common expression "systematicbias" are often used to refer to the same thing; some users seek todraw a distinction between them, suggesting that systemicbiasismost

biases are human tendencies that lead us to follow a particular quasi

Documents