Bias in Conducting Research: Guidelines for Young Researchers Regarding Gender Differences

Every scientific discipline is determined by the object of measurement and the selection of appropriate methods of data collection and statistical analysis. Faulty methodology can lead to incorrect information in the results, without the researcher being aware of this. Taking incorrect knowledge as correct into account while conducting further research has far-reaching negative consequences. One of these errors present, to some degree, in every single research is bias. It is a particularly dangerous one, because it usually goes undetected by the researcher. But if you are aware of its threat there are ways to avoid it.  In research, it occurs when systematic error is introduced into sampling or testing by selecting or encouraging one outcome or answer over others. It comes in numerous ways and forms. The rest of this post will focus on causes of bias in the field of gender studies.

Bias is defined as any tendency which prevents unprejudiced conside ration of a question. Gender bias occurs because researchers’ stereotypes and prejudices about gender become implicitly, unknowingly, but systematically implemented in the research process. It could be defined as a systematically erroneous gender dependent approach related to a social construct, which incorrectly regards women and men as similar/different (Ritchie, 2009).

This type of cognitive bias can occur at any phase of research, including planning, data collection and analysis. The following section sheds light into the typical research stages and the respective threats that occur in each of them, as well as providing advice on how to deal with the different flaws or threats to thorough research practice.


  • Hypothesis and conceptual frameworks have to be based on reality, not on assumptions about gender characteristics, roles, and cultural values. If you want to be more objective while conducting research in order to minimise the impact of your own maybe unfounded beliefs (don’t feel bad, we all have them) you have to acknowledge your own bias; e.g. prejudices and stereotypes. It means you have to engage in introspection, and grasp the preconceived notions about what men and women are like, and what they should ideally be like. Afterwards, you should try to accept them as something that may or may not be true or right; just don’t let them lead your research project and be open to new solutions. In essence, keep an open mind concerning your assumptions and question them at every opportunity.
  • Furthermore, you should avoid trying to understand the problem from a single perspective. When you make a decision about your topic of the interest you are responsible for reviewing literature relevant to the research question. That means going beyond your own point of view and reading-up thoroughly the literature on different perspectives which may propose different answers to your question. You are free to choose any theory you consider relevant, and the same theory will probably guide you in formulation of the hypothesis.
  • Another problem is androcentrism. It is “the practice of giving overriding importance to men human beings or to the masculine point of view on the world, its culture and its history”. It is a by-product of male majority in science (today thankfully less than before, but still existing). Therefore, an androcentric view holds men and masculine characteristic as the norm; a standard for comparing all people (Ruiz-Cantero et al., 2007). It is clear that it can lead to prejudice or discrimination based on sex (in this case against women), also known as sexism. I think it should be clear to all why it is undesired.  That means it is useful to have the perspective of female scientist on the research problem framework. This can be accomplished by building a gender balanced research team, or having it peer reviewed by a female scientist. Besides, if you want to have a representative sample of a larger population and results you can generalise to both men and women make sure you choose a gender heterogeneous sample.


  • Develop a gender sensitive methodology. Gender sensitive methodology takes into account gender differences e.g. intelligence tests that measures only special intelligence (in which man are, on average, better) are not gender sensitive because they put men at an advantage and may lead to incorrect conclusion that men are more intelligent than women. Psychometric Item Response Theory offers techniques for indicating which items cause bias. One of them is Differential Item Functioning. It gives insight into items on which examinees with equal level of knowledge or skill, have different probabilities of success and failure, depending on a group which they belong to such as gender (Ritchie, 2009).
  • Standardized protocols for data collection, which include training of study personnel can minimise inter-observer variability (Milas, 2005). The training should, among other things, include educating study personnel about gender biases. Another solution is to use double-blind procedure, an experimental practice where the researcher doesn’t know the critical aspects of the experiment, and therefore his lack of expectations dissolve biases and ensure this doesn’t confound the results of the study.

Data analysis

  • Some variables function differently for men and women. Check whether measures of central tendency of men and women differ at a level with is statistically significant.  Thereby you will be sure if you can generalize results across gender.
  • It is not enough to conclude that men and women differ in a statistically significant way.  Very small differences may have no practical significance. The statistical significance is usually calculated as p-value; the probability that a difference is caused by chance. The p-value depends on the effect size as well as on the sample size. So even the smallest differences are statistically significant if the number of participants is large enough. Use effect size measures that complement your statistical procedure in question. Cohen’s D is an effect size measure equivalent to a Z-score. It tells us how relevant the difference between the means of two groups is. In other words, it tells us how significant the overlap among the distributions of the different groups is.


  • Make sure to conclude only what your research results indicate; don’t be seduced by your personal, prior believes and possible expectations.
  • Also, keep in mind that no research is perfect or optimal. Be aware of potential flaws in your research and possible omissions you have made.
  • Use gender sensitive language. American Psychological Associations guidelines suggest the following. Write clearly and concisely. Use sex only when it is thought on biological differences and gender when you talk about men and women as parts of social groups. Avoid the terms male and female (that implicate only biological differences), and replace them with men/boy or women/girl. To fight androcentrism eliminate the generic use of ‘man’ for ‘man’, substitute ‘person’/’people’, ‘individual(s)’, ‘human(s)’, ‘human being(s)’; for ‘mankind’, substitute ‘humankind’, ‘humanity’, ‘the human race’; for ‘manhood’, substitute ‘adulthood’, ‘maturity’; and delete unnecessary references to generic ‘man’.  Try to eliminate sexual stereotyping of roles by using the same term (which avoids the generic ‘man’) for both women and men (e.g., ‘department chair’ or ‘chairperson’), or by using the corresponding verb (e.g., ‘to chair’), not calling attention to irrelevancies (e.g., ‘lady lawyer’, ‘male nurse’) (American Psychological Association, 1994).

In short, this article guides you through the research process by identifying specific moments and the ways in which gender bias occurs, as well as recommends ways in which the impact of gender bias on your research results and interpretation can be reduced.

In conclusion, I offer you a checklist which you can use to make sure that you have done as much as you can to get gender bias free, and therefore more accurate, research results.


  1. Acknowledge your own gender bias
  2. Review  literature relevant to the research question
  3. Have the perspective of female and male scientist on the research problem framework
  4. Develop a gender sensitive methodology
  5. Standardise  protocols for data collection and/or use double-blind procedure
  6. Check whether measures of central tendency of men and women differ at a level which is statistically significant
  7. Use effect size measures that complement your statistical procedure
  8. Make sure to conclude only what your research results indicate
  9. Be aware of potential flaws in your research and possible omissions you have made
  10. Use gender sensitive language

Good luck with your research!


American Psychological Association. (1994). Publication manual of the American Psychological Association (4th ed.). Washington, DC: American Psychological Association.

Coe, R., (2002). It’s the Effect Size, Stupid. What effect size is and why it is important. Presentation to the Annual Conference of the British Educational Research Association, England 2002. Retrieved from

Milas, G. (2005). Survey methods in social investigation. Jastrebarsko: Naklada Slap

Ritchie, T. D. (2009). Gender bias in research. In J. O’Brien, J. Fields, & E. Shapiro (Eds.), Encyclopedia of gender and society (vol. 2, pp. 713-715). Thousand Oaks, CA: Sage Publications.

Ruiz Cantero, M.T. et al. (2007). A framework to analyse gender bias in epidemiological research. Journal of epidemiology and community health, 61(2), 46-53.

Nina Jelić

Nina Jelić

Nina Jelić is a graduate psychology student at Faculty of Humanities and Social Sciences, University of Zagreb, Croatia. Her fields of interest are social and clinical psychology.

More Posts