http://columsblog.wordpress.com/2012/03/25/making-the-case-for-more-field-experiments/#comment-84
http://kmusial.wordpress.com/2012/03/25/why-is-hydration-important/#comment-43
And for now… good bye, farewell, adieu, adios, au revior…
Claiming that psychological research can be defined as “biased”, is hardly much of a stress on the imagination. Experiments or studies we come across during our degree (or beyond), will always have some element to them that reduces its reliability, validity, generalisability, or ethical approval. In this blog I will outline a few examples of bias in research, some well-known and some not so much, what problems they cause and how they can be rectified.
The most common bias we usually come across is down to the experimenter. For example, when conducting an interview, the researcher could already have preconceived ideas or beliefs about the subject they are talking about. As such, their analysis could be biased as a different researcher may not code a response the same (Freud, 1909). A way to prevent this could be to use inter-rater reliability, whereby two or more individuals code behaviours and then see if their results match (Milgram, 1963). Another method that could be implemented is a double blind procedure, whereby neither the participant nor the researcher know what condition they are taking part in/testing, so the results can’t be altered to correspond with the hypotheses of the study (Lustman et al, 2000).
It has also been observed that over the years, that a heterosexual bias can occur within studies. Indeed, in the 1960’s, homosexuality was considered a mental illness. The reports written in those times considered homosexual people to be hereditarily insane (Moreau, 1887; Szasz, 1974). From studies I have participated in the past, I believe some researchers have managed to neglect the idea that their participants could be attracted to people other than the opposite sex, and how this could have an effect on what they were attempting to measure e.g. trustworthiness. However since then, a plethora of studies have helped change people’s preconceived ideas about this; mainly that it is not so “unnatural” as previously thought (Ford & Beach, 1951). I believe these possible confounding variables could be prevented if homosexual people are taken into consideration when designing a study, so that it does not automatically make the assumption their participants will be straight.
Bias can also be present in the samples taken for experiments. For example, participants that are all males (Milgram, 1963) could mean that the sample in not generalisable to a wider population, and so reduces the usefulness of the study (although, Milgram was attempting to show the inhumane acts done by Nazi Germans was nothing down to the fact they are more obedient than other cultures… and all the sergeants were men). This can obviously be rectified by having an even amount of males and females in each condition you are testing for, as well as using a matched pairs design so that one group is not so different from another on what you are trying to test (Bandura, 1961)
Also, Wilson and Brekke (1994) introduced the idea of mental contamination, that they defined as “the process whereby a person has an unwanted judgment, emotion, or behavior because of mental processing that is unconscious or uncontrollable”. An example of this is what we have been talking about in cognitive psychology; an attentional bottleneck whereby we will gather information from our environment that is important or useful to us. This could be rectified by clearly explaining the procedure to participants, so that they know a bit more about what is happening and so will not get bored and just give up trying.
In conclusion, bias can be present in many different forms and can cause many problems. However, if there are more hard set rules and consideration in studies, there is no reason why it can’t be rectified. What do you guys think? What’s the best way to combat these problems? Thanks for reading.
I’m not going to lie, I really struggled for a blog topic this week. I seem to have depleted my “interest” in research methods and statistics, and so went trowling through all the TA’s links to see what other people have been writing about. I was quite impressed at how much people are managing to link to stats, from specific types of therapies used in clinical settings, to the crazy cat lady off The Simpsons and whether she needs help (personally, I think she does). So, I have settled on a topic I am rather passionate about, and attempting to link it to some sort of research methods that are applied in psychology.
I decided on exams, and the conditions you have to be in when taking one. I may be slightly biased in the fact that, well, I hate them. My results in them tend to be the same of other assignments we are set, but only like 475843567438965 times more stressful. This blog will therefore address the questions: are they really worthwhile? Are they really the best way of measuring one’s understanding, or are they just unneeded stress-inducing bane’s of people’s existence? I will explore this by looking at the benefits and costs of exams, and their overall usefulness.
Firstly, I’ll attempt to analyse how useful exams are in the same way a psychology study is. The most obvious point seems to be that of ecological validity; how applicable are exam conditions in the outside world? Can they predict how efficient someone will work once they have passed the exam? While it is important to know and understand components of your job for yourself, have you ever heard of a single individual working on a majorly important project for their employers? Of course not. People confer with each other at all different stages e.g. from research and planning to presenting a final design. Imagine the absurdity of employers forcing their employees to work alone under silent conditions and being unable to work together with others. This showcases the very problem with a laboratory experiments; while a high degree of control prevents any confounding variables, it also means it is very unlike real world occurrences.
Also, even with this high level of control in exam settings, I believe it is no where near enough to allow and compensate for confounding variables. There is no way reliability can be high – exact conditions for every exam can’t be achieved. For example, exams aim to measure the level of understanding an individual knows about a certain subject. However in the majority of cases, this exam is usually administered only once at a specific point in time. What if the individual is unwell? Or too stressed? Having a generally bad day? What if the weather conditions on the day are too hot, or too cold? There is no way of ensuring that every person who takes an exam is at their optimum level of health or state of mind, to show how much they really know in an exam. This is usually the main reason longitudinal studies are conducted within psychology; they allow changes in behaviour to be observed over time by collecting many data points (Watt et al, 2003). Surely this would show progression of actual knowledge over time, which would also encourage individuals doing A levels etc to not leave learning everything until the last few weeks of the course.
In the example of A levels, the idea of condensing a years worth of knowledge into one exam can also be severely daunting, especially to those who need the top grades to get into university. This is commonly known as test anxiety (Liebert & Morris, 1967), and often interferes with the learning needed to pass the exam, as well as lowering test performance. Also, Johansson et al (1978) highlighted how repetitive and isolated working can increase stress levels… sounds a lot like exam settings to me. So basically the whole experience of exams, from attempting to learn all the material needed, revising, and actually taking the exam seems to induce stress to an individual in some way.
On the other hand, what if we were to scrap exams? Is coursework and a variety of assessments the right way to go about assessing someone’s knowledge or competency? Group work may seem a more beneficial and valid way of assessment, but I myself can think of problems with this. When living back at home I applied to work at Argos, in which I had to take part in group interviews with other prospective employees. The groups ranged from three to six people in each one, and those with more seemed to put less effort in and were unsure of what they were doing. This is known as the Ringelmann effect (Ringelmann, 1913) and will result in a loss of motivation; there is a diffusion of responsibility and people in larger groups will feel they do not need to try as hard. This is seen all the time in schools where some students will take a back seat approach to group work, allowing others to do all the tasks. This would obviously be very detrimental to the validity of assessing knowledge outside of an exam setting.
In conclusion, exams seem to only stress people out, and do not seem the ideal way in which to gauge a person’s knowledge. However, there seems no well-defined or straight-forward way in which to otherwise assess individuals, without blatant problems. While education seems to be a top agenda for the government, I believe more research is needed to come up with new ways in which to test people fairly.
Case studies can be described as an in depth investigation into a particular individual. The methods usually used within them are observation or interviews, and originated from a clinical and idiographic perspective, such as looking at a patients history. This blog will attempt to look into exactly how case studies are conducted, and how useful they can really be within the area of psychology, giving points for and against using them.
Firstly; how are they conducted? There are several well known examples within psychology, the most prominent and memorable perhaps being Freud’s (1909) study on Little Hans and his phobia with horses. Data in this case study was collected through interviews – Hans’s father, who was a supporter of Freud, would talk to the boy and send his findings on to Freud, and Freud in turn would offer advice on how to deal with what he believed to be the Oedipus complex. In the end, Hans’s irrational thoughts and phobias were removed and so he benefited from the experience. However, the methodology raises several points to consider. Surely Hans’s father was biased in the fact he supported Freud, and therefore been narrow minded when analysing the data he had collected from his son? Could Freud (1909) really convince other psychologists if he only met Hans once before, instead leaving the data collection to an amateur? Objectivity seems to have gone out of the window on both points mentioned.
Another study worth looking at is Thigpen and Cleckley (1954), who conducted a case study on a woman with multiple personality disorder. This study used a large amount of procedures including interviews, psychometric tests (IQ and memory), and projective tests (Rorscahch and figure drawing). “Eve”, the participant, was interviewed extensively for 14 months; creating a long and detailed case history. This obviously provided a lot of data, both qualitative and quantitative. Eve’s relatives were also interviewed to increase the level of understanding in certain situations, that the researchers themselves may not have thought of. They also hired several independent experts to run other tests such as EEG to reduce the chance of experimenter bias.
So, from looking at these two studies, what can we infer? Some people argue the point that case studies on an individual are too specific to be generalisable to a wider population; who is to say that Hans or Eve were a “typical” human being? Psychology aims to help society and the people within it. To use so many resources on a single participant could be seen by some as a waste of time – resources are limited and should be saved to benefit more people. While this can be true, such as Eve’s extensive testing, I believe that if help can be given, then it should. Think about it from an ethical point of view – some argue that if one person in an experiment is or could be harmed, then it should not take place (Milgram, 1963). So, by this logic, surely conducting a study to benefit an individual should be okay? In my two examples and many others I have read, the case study has assisted them in their road to recovery, very much due to the large amount of rich data obtained. The researchers are not attempting to “average” a behaviour, they are tailoring it to the actual individual in an attempt to help them.
On the other hand, there are several disadvantages to case studies. One factor could be that the study is retrospective. This means that the individual is basing their answers from memory, which may not be entirely accurate for several reasons such as leading questions from the researchers (Loftus and Palmer, 1974). This obviously reduces the validity and reliability of the results obtained, damaging the study’s usefulness as well as using the data for the individuals recovery. In Thigpen and Cleckley’s (1954) study, people also argued that they could have been easily misled by Eve; they got to know her on a personal basis which would only be natural after studying her for over a year. This could be all the difference between an actual personality change in the subject being studies, and the researchers actual perception of them; objectivity is reduced.
In conclusion, I believe that case studies are well worth doing. They benefit an individual by looking at them on so many levels, to understand and help to cope with abnormal behaviour. However, the procedures need to be done with care to remain objective. One way of doing this could be to set up a study with a hypothesis, and then use a double blind measure whereby the researcher conducting it is unaware of what you are looking for. By using more well defined rules such as this, I believe they can become even more useful as a method within psychology, to prove their worth and use of resources.
References…
Freud (1909) http://www.psychblog.co.uk/where-to-start/developmental-freud-1909
Thigpen & Cleckley (1954) http://www.holah.karoo.net/thigpenstudy.htm
Milgram (1963) http://www.nickoh.com/emacs_files/psychology/ss_dir/milgram1963.html
Loftus & Palmer (1974) http://www.holah.co.uk/summary/loftus/
Case study overview http://www.simplypsychology.org/case-study.html
Experimental research is the main building block within psychology; allowing us to investigate in specific areas, and attempting to reject our null hypotheses. These studies aim to further our understanding of humans and the environments we live in, striving to benefit society on a individual and global level.
However, to make these investigations worthwhile and useful in the outside world, they need to be grounded in well defined rules and procedures to retain reliability and validity. While this is relevant to concepts such as control, sampling, reducing bias etc, this blog will attempt to highlight the importance of how to measure variables; ones that lack clarity and cannot always be directly observed e.g. distress, aggression, self-esteem.
Therefore, theories establish mechanisms and elements that are only assumed to exist, in order to describe and explain behaviours we observe. Variables mainly looked at in psychology are in fact made from these theories and speculations, and according named hypothetical constructs. By observing behaviour by this unified and organised approach, predictions of behaviour can be generated.
These constructs are then operationalized; a procedure figuring out how to define and measure them. The most well known case of this in everyday life are IQ tests. Intended to measure intelligence (a hypothetical construct), whereby it influences behaviours exerted in the real world that can be observed and measured (Sternberg, 1988).
However, a clear problem with operationalization is that it can be reductionist; too simplistic when defining a variable. For example, when diagnosing depression, this can be done in terms of behavioural symptoms e.g. lack of a social life, low self-esteem. But with a lack of physiological measures e.g. urine samples (Johansson et al, 1978); this could be all the difference between a short period of depression and major depressive disorder (MDD). The behaviour only represents a part of the total hypothetical construct.
Another example not often thought of in this context could be the concept of love. Attraction is an important part of all social animals’ lives, not only for the social interaction gained, but also for judgement on who should be lovers or friends. Castellow et al (1990) even established the importance of attraction in a court room setting, whereby defendants were found less guilty if perceived to be more attractive. On the other hand, should we as psychologists even attempt to operationalize such a construct? Surely such an innate and complex state of mind can’t be boiled down to one or two observable behaviours?
In conclusion, when variables in studies are hypothetical constructs, they must be properly operationalized for defining and measurement. This does not always mean creating your own – instead establishing concurrent validity with previous research that have examined related variables to your own. Also, I hope I have emphasised enough that such variables should not be taken at face value, from simplistic and narrow minded views. If psychology has taught us anything, it’s that humans are complex, and by extension, their wildly varying behaviours and attitudes.
References for studies…
Operationalization – http://www.experiment-resources.com/operationalization.html
Sternberg (1988) – http://www.indiana.edu/~intell/sternberg.shtml
Johansson et al (1978) – http://psychology4a.com/stress.htm
Castellow et al (1990) – http://core.ecu.edu/psyc/wuenschk/Articles/JSB&P1990/JSB&P1990.htm
For the final blog of the year I have decided to write about another study, and analyse it’s relevance and usefulness, by looking at the practices and methods they undertook and how this affected the reliability and validity. I have chosen a study with a physiological background, for one due to personal preference, and for another reason because I feel it demonstrates several points on research methods only briefly mentioned by lecturers.
When medical regimes are administered to patients, there can be several reasons as to why they do or do not adhere to it. Bulpitt et al (1988) attempted to find these reasons, reaching the conclusion that negative side effects outweigh the benefits, especially when treating an asymptomatic problem (no physical indications showing illnesses e.g. hypertension, where blood pressure is too high). It is important to be able to measure adherence levels, however there are no standardised methods for this. As mentioned earlier, self report always runs the risk of social desirability bias, especially in this case where doctors may refuse to help anymore if patients are not taking their medication. To overcome this, a physiological measure is needed, which Lustman et al (2000) attempted to do by assessing the efficacy of an anti-depressant known as Fluoxetine for treating depression. Therefore, this study makes the assumption that treating depression can lead to increased adherence.
Lustman et al’s (2000) methodology involved the use of a randomised, controlled double blind procedure. This method has several advantages such as the use of random assignment, which reduces any experimenter bias. Also, a double blind test means that neither participants nor researchers know who belongs in each condition (the experimental or control) until all the data has been recorded. This greatly reduces any bias that could be present and removes any demand characteristics.
The participants included 60 patients with either type 1 or type 2 diabetes, and have been diagnosed with depression. Type 1 diabetes is where the body no longer creates insulin to allow glucose into cells, which is used for cells to respire (broken down by Glycolysis creating pyruvate, which is used in a process called the Link Reaction to create acetylcoenzyme A, which is used in the Krebs cycle, which creates electrons to be used in the Electron Transfer Chain, which yields much ATP for energy for the body… thought I may as well use my A level biology knowledge for something!). On the other hand, type 2 diabetes is where the cells do not correctly respond to the insulin produced in the pancreas, which can be due to a number of reasons e.g. too much food, resistance to insulin, and too much can be toxic for cells so they will shut down receptor sites.
The patients/participants were therefore in either the Fluoxetine or placebo group. For those who don’t know, a placebo is where a harmless and inactive medication is administered that will not do anything; people taking it believe it contains something active that will help them and their illness. This seems like a major flaw in the ethics of Lustman’s (2000) study; people diagnosed with depression are not receiving the medication they have been described, which could surely cause harm to them!
Patients were assessed for their depression through psychometric measures, and their adherence to the medical regime was measured through their GHb levels, indicating glycemic control. GHb is glycohaemoglobin (I hope I’m not scaring you all with these unusual words, I just love biology…) which shows the amount of glucose in the blood; it should be lower if they are taking their medication. This use of physiological measures increases the level of objectivity within the results. For one, patients/participants cannot lie about their adherence, and there is no chance of experimenter bias affecting the results.
In the findings, the patients given Fluoxetine reported lower levels of depression, as well as lower (nearer to normal) levels of GHb, indicating improved adherence. This would seem to demonstrate that measuring GHb for those with diabetes is a good indicator to levels of adherence to a medical regime. While this is beneficial to those who fit into that category, it cannot and should not be assumed that we can generalise these findings to other illnesses; but this study may well be able to spark future research into this domain, and a more unified approach to measuring adherence to medical regimes can be found.
References for studies…
Lustman et al (2000) http://www.ncbi.nlm.nih.gov/pubmed/10834419
Bulpitt et al (1988) http://www.slideshare.net/psychexchange.co.uk/psychexchangecouk-shared-resource-2825001
And also, if I completely failed at describing the different types of diabetes, this is a good link that describes diabetes http://www.medicalnewstoday.com/articles/7504.php
In previous blogs, I have focused mainly on the design and analysis of psychological investigations, and the different levels of data you can use. However, I have hardly discussed the people taking part in these studies; the participants. It is the intent of psychology to help and understand the human condition, and this often requires the use of willing and human participants. The researcher has a responsibility – that no physical or psychological harm will be subjected to these people, whether that be through the process of taking part, nor the impact of the findings.
Therefore, all research wishing to be conducted requires ethical approval from a committee. These committee’s are found at every single establishment undertaking research, following the guidelines set by the British Psychological Society (BPS). The rest of this blog will list, describe, and explain these different features within these guidelines by the BPS, as well as the implications of abiding by them.
The first guideline I will look at is informed consent. This is where participants must agree to take part, either verbally or in writing. “Informed” means that they fully understand what the study will involve, and any risks that the procedure contains. However, should they be fully informed? If so, they may display a well known problem in the analysis of data – demand characteristics. This is where they will attempt to please the researcher by doing what they think is expected of them, which is otherwise known as the “please you” effect. On the other hand, some people may go against the wishes of the experimenter, by doing the opposite of the purpose of the study… appropriately named the “screw you” effect. Either of these effects would certainly have a detrimental effect on the validity of the results, as they would be nowhere near a true score and thus the findings would be useless. For example, a study by Janis and Feshbeck (1953) into fear arousal used questionnaires within their procedure. They administered one a week after delivering a lecture igniting fear arousal, to see if a change in health behaviour occurred. However, it would not take a genius within the participants to figure out what the researchers were looking for, and some may have lied about their health practices to either conform or rebel to the hypotheses. Another problem with “informed” consent would have to be the use of children within studies; surely they wouldn’t know or understand all the implications of what they were taking part in? Due to this, researchers usually get parents to do it on their behalf. Yet, I do not believe this alleviates the possibility of stress to the young child; they may not understand what is happening and being interrogated by a stranger could be very frightening to them (Bandura, Ross & Ross, 1961).
The next guideline is you are not allowed to deceive participants. This includes the fact they cannot be misled into what the study is looking at or the way in which they are being measured. As stated above, this can lead to demand characteristics, but can also present the problem of the participants not taking the study seriously or not being very valid. Milgram (1963) deceived his participants by telling them they were administering electric shocks to other participants, whereas there was no shock and the other participant was a confederate within the study. Although, if he hadn’t, the participants would have certainly acted differently; the study would not have been measuring obedience and the breakthrough on learning more about human nature by Milgram’s (1963) study would never have existed.
Researchers must also reduce the chance of participants being exposed to stress. This includes embarrassing, frightening or offending them; the risk of harm being no greater than you would from everyday lifestyles. For example, a study by Johansson et al (1978) aimed to measure the stress levels of Swedish sawmill workers. The methodology they undertook and their implications however, could have in itself caused more stress; they took consecutive urine samples throughout the day which could have caused more stress. And if they had too much stress, this information could have reached the ears of their boss at work, and could have lost their job as a result. Another guideline that links to reducing the risk of stress is the right to withdraw, meaning participants must be aware of their right to withdraw at any time within the study without penalty; even after the study has finished they can ask for their data to not be used.
Confidentiality is the next one. This means that participants should not be able to be identified in any research material that is published. This can be achieved through assigning participant’s numbers or changes/abbreviations of names, for example those who went to the qualitative data seminar this week would have seen the conversation we were annotating and analysing at used different names as well as censor university names that the interviewee went to, so no one could identify the girl.
Lastly, debriefing participants is a crucial and highly beneficial process within a study, done at the end when the testing is over, whereby researchers summarise the nature of the study and provide an opportunity for the participants to ask any questions or issues they may have. If this is not done, the study could have negative long term effects on those that took part, due to their unawareness of what was taking place. For example, Piliavin et al (1969) looked into diffusion of responsibility and the idea of a cost-benefit analysis, by covertly observing people on a subway in New York. Not only does this ignore the idea of informed consent, but also means that no one was able to be debriefed. As the participants witnessed an individual collapse, this could have had detrimental effects on their wellbeing as they may not have helped the individual. This also relates back to the idea of not putting the participants at risk due to the study.
In conclusion, while ethics are useful in protecting participants in many ways, they can often be refuted for many reasons. They can have serious detrimental impacts on validity, both internal and external, causing the study to lack any usefulness. A waste of money and resources, in other words. It all comes down to the well known phrase “do the ends justify the means?”, whereby is risking some things worth the benefit to others in the end, if it furthers knowledge to help others? It would seem that some of the most influential studies e.g. Milgram (1963) did not abide by these guidelines, yet furthered the area of psychology because of it. It would seem that a balance between the two should be made, if that is possible. Thoughts?
References for studies…
Janis and Feshbeck (1953) http://changingminds.org/techniques/general/articles/fear_persuasion.htm
Bandura, Ross & Ross (1961) http://www.holah.karoo.net/bandurastudy.htm
Milgram (1963) http://www.nickoh.com/emacs_files/psychology/ss_dir/milgram1963.html
Johansson et al (1978) http://psychology4a.com/stress.htm
Piliavin et al (1969) http://www.helium.com/items/329831-piliavin-rodin-and-piliavin-investigating-good-samaritanism-behavior
jameezio 