Threats to validity

When establishing validity in reference to research, we need to ask “is the research measuring what it intended to measure?” We must be sure that the research is measuring what is intended. This is very important, as we have to consider that measuring such things as intelligence, depression, esteem and so on, can not be measured or observed directly as they are hypothetical, and we have to develop other methods of measuring.

In a research study as a whole, validity has two crucial parts which are internal and external validity. The aim of many research studies is to state “this is what happened and this is what it means”, if there are any doubts here, then there is a threat to validity. In research we often search for a cause and effect between variables, for example we study that if we eat less we loose weight, providing that there is no other explanation for effect, the research is valid, if we have doubts and there may be other explanations for the effect, this is a threat to the internal validity of the research. Extraneous (additional) variables in a research are often a threat to internal validity, and they are often numerous in studies. They become a threat when they influence the variables being investigated, and are referred to as confounding variables, and can sometimes offer alternative explanations to findings. Therefore, it is important for researchers to identify confounding variables. Confounding variables can be categorised into three groups; (1)Environmental variables, when environment between treatment conditions are noticeably different, for example one treatment in a warm room and other in a cold room, or different times of the day and so on, this category can be a confounding variable for all research designs; (2)Assignment bias, if research has different groups of people for each treatment, then each group may have different characteristics, for example one group could be younger than the other or that there are more males in one group and so on; (3) the third category is concerned with confounding variables to a research design which studies a group of participants over a period of time and can be; history, when events outside of the study has an impact on the research; maturation, when participants change physiogically or psychologically, this is especially a concern when studying children and the elderly; instrumentation, when instrument of measurement changes with time, this is often the researcher who can become more or less efficient, or equipment can become more advanced and so on; testing effects, such as practice and fatigue and finally regression, this is a mathematical phenomenon where extreme scores regress towards the mean.

To establish external validity, research findings should be able to be generalised outside the study, if not this could be considered a threat to the external validity of the research. Most common threats to external validity can be categorised into three groups; (1)generalise across participants or subjects, are participants representative of the general public, more often they are not, research is often subject to students and volunteers, and these are not representative of the general public, research suggests that students and volunteers have higher IQ than the general public, more educated and so on; (2)generalise across features of a study, threats can include novelty effect, suggesting that actual study can cause arousal and excitement that may incur different results and reactions than real world. Other threats in this category include experimenter characteristics that can affect results, such as age, class, friendliness and so on; (3) generalise across features of the measures, threats include sensitisation, means of measurement, where one form of measurement produces results to another form of measurement and so on.

There are confounding variables that are a threat to both internal and external validity. Special attention must be given as the threat can be twice as damaging. Experimenter bias is one, where the researcher predicts the outcome and influences settings and procedures to accomplish the outcome. For example, may encourage one group of participants where experimenter expects results to be high, and discourage other group where results are expected low. Another threat to both is demand characteristics where researchers have suggested what the findings will be and somehow tried to influence the outcome. I believe both these are detrimental and should be avoided at all costs, and as future researchers should be identified in existing and potential research.

Establishing internal and external validity is crucial for a successful and respected research, if both criterions are met, and then findings are influential. Researchers aim to produce and maintain validity, and it is why before research is carried out confounding variables are identified and dealt with. It is often a delicate matter of balance and elimination. Sometimes dealing with one confounding variable can increase the effects of another and sometimes introduce another. Often is the case, by increasing internal validity of a research study decreases external validity, for example if studying the effects of colour on mood, where all other extraneous variables are controlled such as time, temperature, age, gender and so on are similar, then external validity is threatened, because findings would be hard to generalise to the general public and real world. Consider the opposite, where researchers aim to maximise external validity by providing a natural/real world environment allows for possible confounding variables to enter research and threaten internal validity. This often happens in research, it is near impossible to not have any confounding variables in a research, but what must be decided is which is most important for that research, internal or external validity, often is the case that one must have priority over the other. We must decide which is important for relevant research.