search this site.

130501P - BIAS IN RESEARCH: EPISTEMOLOGICAL AND PRACTICAL CONSIDERATIONS

Print Friendly and PDFPrint Friendly



Presentation at the Annual Research Day of the University of Qasim, Buraydah on May 1, 2013 by Professor Omar HasanKasule Sr. MB ChB (MUK), MPH (Harvard), DrPH (Harvard) Faculty of Medicine King Fahad Medical City Riyadh Saudi Arabia. EM: omarkasule@yahoo.com


ABSTRACT

The paper starts by introducing the Qur’anic concept of istiqamatas the basis for valid and unbiased search for the truth. Humans by their nature can fall prey to bias away from objectivity by hiwa al nafs, prior philosophical assumptions due to dhann, and rejection of hujjiyat al burhan. Epistemological bias arises in biased formulation of research hypotheses and interpretation of results because of prior philosophical biases some of which are sometimes unconscious. The researcher must acknowledge these biases and develop systematic approaches to identifying and eliminating or curing them. Using the refutationist null hypothesis alongside the alternative hypothesis protects the researcher from the effect of dogmatic assumptions. The Bayesian approach in statistics enables a researcher to acknowledge and incorporate prior beliefs to reach posterior conclusions. Practically 4 forms of bias occur during study design (misclassification bias, selection bias, confounding bias, and sampling bias) and in several ways in the analysis stage the most important being misspecification bias. Misclassification is inaccurate assignment of exposure or disease status. Selection bias arises when subjects included in the study differ in a systematic way from those not included. Confounding bias arises when the disease-exposure relationship is disturbed by an extraneous factor called the confounding variable. Sampling bias includes sampling errors (wrong sampling frame, wrong sampling unit) and non-sampling errors (non-response error, coverage error, and measurement error). Mis-specification bias arises when a wrong statistical analytic model is used for example using parametric methods for non-parametric data.

 

Key terms: epistemological bias, null hypothesis, misclassification bias, selection bias, confounding bias, sampling bias.


TRUTH, haqq, and OBJECTIVITY, istiqamat

·         Truth, haqq[1], and objectivity, istiqamat[2], are the default position in a normal person and are a bed-rock of valid un-biased research.
·         Truth is at 3 levels: ‘ain al yaqiin[3], haqq al yaqiin[4], ‘ilm al yaqiin. Truth at the human level is relative, nisbiyat al haqiiqat, because of limited knowledge.
·         The unknown is 2 levels: ghaibmutlaq (e.g five known only by Allah[5]) and ghaibnisbi (accessed through empirical research)

CAUSES OF BIAS
·         Diseases of the heart, amradh al qalb[6], that can color and distort objective observation and interpretation resulting in bias manifesting as (a) personal whims, hiwa al naf[7](b) conjecture not based on evidence, dhann[8],(c) laziness and negligence, ghaflat[9] (d) human error, khata’u[10](e) forgetfulness, nisyaan[11](prevented by recording, kitaabat[12], and witnessing, ishhaad[13])
·         Distorted world view can lead to bias. King Nimrod’s concept of life and death based on his thinking he was a god. The objective world view is the ruuyatkawniyatquraniyat.
·         False knowledge that manifests as: usturat[14], khurafat[15], kadhb[16], lahw[17], and wahm[18], blind following, taqlid[19].
·         Failure to base knowledge on evidence, hujjiyat al burhaan: burhan[20], daliil[21], bayyinat[22], tathabbut[23], andsidq[24], and hujjat[25].

BIASES IN HYPOTHESIS FORMULATION
·         The scientific method consists of formulating and testing hypotheses. Bias can start at the hypothesis formulation stage. The researcher must acknowledge these biases and develop systematic approaches to identifying and eliminating or curing them.
·         Using the refutationist null hypothesis alongside the alternative hypothesis protects the researcher from the effect of dogmatic assumptions.
·         The Bayesian approach in statistics enables a researcher to acknowledge and incorporate prior beliefs to reach posterior conclusions.

 

PROBLEMS OF THE EMPIRICAL METHODOLOGY 1

·         The empirical methodology is Qur’anic. The Qur’an calls upon humans to observe the universe, al nadhar fi al anfaaqwa al anfusafaaq&anfus[26].Ibrahim used empirical evidence to calm his heart, liyatumainnatqalbi[27]
·         The empirical methodology is innately good but the manner and context of its use lead to the following 4 problems: biases due to a priori assumptions, limitations of observation by human senses, limitations of human intellect, and lack of an integrating paradigm
·         A priori assertions or non-assertions, (assertions by default) bias the selection of fields/issues of investigation, formulation of hypotheses, selection of hypotheses for testing, reporting of data, interpretation of data, and use of information. These assumptions arise from the western world view that is basically materialistic and lacks an integrating paradigm like tauhid. There is no morally/ethically neutral, hiyaadakhlaqi because of these assumptions.
·         Limitations of empirical observations: Empirical knowledge is relativistic and probabilistic. It depends on human senses that are limited and can be deceived. There are 3 sources of knowledge: wahy, kawn, & ‘aql. Kawn is superior in quantity buy wahy is superior in quality. ‘Aql is a tool used by both sources.

PROBLEMS OF THE EMPIRICAL METHODOLOGY 2

·         Lack of balance, tawazun: The way science and technology are used today shows lack of balance which leads to transgression. The Qur’an told the story of the people of thamud who were technologically advanced but lacked spiritual and social balance and the end-result was evil-doing[28].
·         Lack of purpose, abath[29]: Technology seems to have become an automaton with its own dynamism that is sometimes not related to any understandable human purpose. This is what the Qur’an described as building structures for amusement only[30] with no underlying purpose.
·         Lack of an integrating paradigm, tauhid: Knowing the parts and not the whole. Tauhid is needed as an integrating paradigm to understand the whole.

MISCLASSIFICATION BIAS 1
·         Misclassification is inaccurate assignment of exposure or disease status. Random or non-differential misclassification of disease biases the effect measure towards the null and underestimates the effect measure but does not introduce bias. Non-random or differential misclassification is a systematic error that biases the effect measures away from the null exaggerating or underestimating the effect measure. Positive association may become negative and negative associations association may become positive.

MISCLASSIFICATION BIAS 2
·         Misclassification bias is classified as information bias, detection bias, and proto-pathic bias. Information bias is systematic incorrect measurement on response due to questionnaire defects, observer errors, respondent errors, instrument errors, diagnostic errors, and exposure mis-specification. Detection bias arises when disease or exposure are sought more vigorously in one comparison more than the other group. Protopathic bias arises when early signs of disease cause a change in behaviour with regard to the risk factor.
·         Misclassification bias can be prevented by using double-blind techniques to decrease observer and respondent bias. Treatment of misclassification bias is by the probabilistic approach or measurement of inter-rater variation.

EXAMPLES OF MISCLASSIFICATION BIAS IN THE LITERATURE

SELECTION BIAS 1
·         Selection bias arises when subjects included in the study differ in a systematic way from those not included. It is due to biological factors, disease ascertainment procedures, or data collection procedures.
·         Selection bias due to biological factors includes the Neyman fallacy and susceptibility bias. The Neyman fallacy arises when the risk factor is related to prognosis (survival) thus biasing prevalence studies. Susceptibility bias arises when susceptibility to disease is indirectly related to the risk factor.
·         Selection bias due to disease ascertainment procedures includes publicity, exposure, diagnostic, detection, referral, self-selection, and Berkson biases.

SELECTION BIAS 2
·         The Hawthorne self selection bias is also called the healthy worker effect since sick people are not employed or are dismissed.
·         The Berkson fallacy arises due to differential admission of some cases to hospital in proportions such that studies based on the hospital give a wrong picture of disease-exposure relations in the community.
·         Selection bias during data collection is represented by non-response bias and follow-up bias.
·         Prevention of selection bias is by avoiding its causes that were mentioned above.  There is no treatment for selection bias once it has occurred. There are no easy methods for adjustment for the effect of selection bias once it has occurred.

EXAMPLES OF SELECTION BIAS IN THE LITERATURE

CONFOUNDING BIAS 1
·         Confounding is mixing up of effects. Confounding bias arises when the disease-exposure relationship is disturbed by an extraneous factor called the confounding variable. The confounding variable is not actually involved in the exposure-disease relationship. It is however predictive of disease but is unequally distributed between exposure groups.
·         Being related both to the disease and the risk factor, the confounding variable could lead to a spurious apparent relation between disease and exposure if it is a factor in the selection of subjects into the study.
·         A confounder must fulfil the following criteria: relation to both disease and exposure, not being part of the causal pathway, being a true risk factor for the disease, being associated to the exposure in the source population, and being not affected by either disease or exposure.

CONFOUNDING BIAS 1
·         Prevention of confounding at the design stage by eliminating the effect of the confounding factor can be achieved using 4 strategies: pair-matching, stratification, randomisation, and restriction.
·         Confounding can be treated at the analysis stage by various adjustment methods (both non-multivariate and multi-variate). Non-multivariate treatment of confounding employs standardization and stratified Mantel-Haenszel analysis. Multivariate treatment of confounding employs multivariate adjustment procedures: multiple linear regression, linear discriminant function, and multiple logistic regression.
·         Care must be taken to deal only with true confounders. Adjusting for non-confounders reduces the precision of the study.

EXAMPLES OF CONFOUNDNING BIAS IN THE LITERATURE


OTHER CAUSES OF BIAS: mis-specification and survey error
·         Mis-specification bias arises when a wrong statistical model is used.
·         Use of parametric methods for non-parametric data biases the findings.
·         Total survey error is the sum of the sampling error and three non-sampling errors (measurement error, non-response error, and coverage error). Sampling errors are easier to estimate than non-sampling errors. Sampling error decreases with increasing sample size. Non-sampling errors may be systematic like non-coverage of the whole sample or they may be non-systematic. Non-systematic errors cause severe bias.
·         Sampling bias, positive or negative,  arises when results from the sample are consistently wrong (biased) away from the true population parameter.
·         The sources of bias are: incomplete or inappropriate sampling frame, use of a wrong sampling unit, non-response bias, measurement bias, coverage bias, and sampling bias.

EXAMPLES OF MISCLASSIFICATION BIAS
·         Misclassification of birth defects as ‘other neonatal conditions’ on birth certificates in West Virginia[1]

REFERENCES