TA Comments

Comment 1: http://aglinskas.wordpress.com/2012/03/25/of-correlation-and-causality-a-push-for-a-more-pragmatical-science/#comment-33 (link to study in first comment: http://aglinskas.wordpress.com/2012/03/25/of-correlation-and-causality-a-push-for-a-more-pragmatical-science/#comment-34 )

Comment 2: http://scarlettrose23.wordpress.com/2012/03/25/final-blog/#comment-118

Comment 3: http://suedonym344.wordpress.com/2012/03/25/mind-reading-it-isnt-that-scary/#comment-37

Comment 4: http://kiwifruit8.wordpress.com/2012/03/25/music-therapy/#comment-54

Archival Research

For this week’s blog, I have decided to write about the seldom mentioned archival research methods. An archive is basically a collection of documents, usually extending some period into the past. When I hear the word archive, I always think of a dingy cellar in some grand building like the British museum or the Louvre, with an old librarian type person opening a huge drawer with some historically important artefacts in it. But the reality is that in today’s modern age, lots of archiving is done online, whether stored on the internet or on some electronic tape or something. This makes it incredibly easy for people to access. And if it can be accessed, you can be pretty sure someone is going to want to use the data for a study. This seems like a very good thing, as you can use very rich historical data, which you would not have been able to access before. I will discuss some of the advantages, and also the disadvantages to archival research.

This study into trends in certain groups used archival research. This allowed them to look back into the past, and access a huge amount of information. The sheer amount of available information is in itself a huge positive. Not only does it provide huge amounts of support, but it also offers a very comprehensive picture to be built. This Study addressed some important questions into archival research. They highlight the fact that different people have different ways of archiving. This means that any data you find may be missing some information, or just be different to what you found in another individuals archive. Whilst this may cause some problems (for example in writing a flipping reference list), they are very keen to emphasise the fact that archival research allows researchers to go down completely new research routes, which is hugely advantageous and beneficial to the continuation of science. This Document lists some further advantages, such as the fact that the data has already been collected, you are just analysing it. This can help to lower costs and overall effort. It also mentions that the area of interest may be difficult or impossible to study, such as life span development.

This Document, whilst mentioning a number of advantages, also puts forward a few disadvantages with using the archival research method. It points out that the experimenter has no control over how the data was collected, and this can lead to a number of problems, such as introducing unwanted bias if only a certain group of people was used. It also explores the idea of improper archiving, such as if a date was missing. This would put you in a very tricky situation. You could either guess the date and use the data which may introduce unwanted bias, or discard it, which is missing out on the rich data. There are also issues with extrapolation and application of findings. Because by their very nature, archives are very heavily weighted towards the past, you have to be extremely careful when applying results to current society. For example (and this may hyperbolic or convoluted, but it illustrates the point) let’s say you were looking at life expectancies. If you were to use archive data from many years ago, such as this, with its fantastic graph, from let’s say 1850 you may conclude that we will only live for an average of about 38 years. This is hugely wrong compared to today’s life expectancies.

Overall, archives are hugely important for ensuring the gathering and collection of huge amounts of data. This provides an extremely vast array of things to study, and the richness of the data makes it a wonder they aren’t used more. But it is very important that care is taken when applying results to now, and dealing with missing data (such as a date).

Homework for TA

Comment 1: https://tallesttales.wordpress.com/2012/03/10/prospective-studies/#comment-38

Comment 2: http://aglinskas.wordpress.com/2012/03/11/is-it-possible-to-operationalize-every-variable/#comment-24

Comment 3: http://scarlettrose23.wordpress.com/2012/03/10/sociometric-tests/#comment-104

Comment 4: http://psychblogld.wordpress.com/2012/03/10/approaching-significance/#comment-53

Prospective Studies

Following on from last week’s blog on retrospective studies, I thought it made sense to regale you this week with tales of Prospective studies. Retrospective studies are essentially studies that look into the past, on events that have already happened. Prospective studies are thus the opposite; they start out with a group, and then follow them over a set period of time. I think it is all too easy to try and compare these two types of studies, and it happens quite a lot. Just because they are two different ways of making observations, people seem to think that you pick to do one or the other. But I think that this is not the case. In essence, they are both very unique, and should be used for testing different things. This means that to try and compare them in terms of which one you should use is unfair and unscientific, as it’s not a case of ‘oh, I’ll choose retrospective because it’s cheaper’. To get the full story on retrospective studies, read my last blog. Now! Up to date? Good. Here, I will discuss what prospective studies are, and some positives and negatives.

This Link explains prospective studies well. They basically start out with a group who share a characteristic and follows them over time to determine the relationship between the characteristic and a condition. An example would be smoking. It is very hard to retrospectively study those who have died as a result of smoking. Prospectively, you are able to progress with them, and see the effects it has. This highlights one of the advantages, as it can measure things that you would not have otherwise been able to. This Link highlights the fact that as you are not relying on memory, you are able to produce true incident rates and results. It also, however, lists many problems. For example, they are usually very expensive and take huge amounts of time, as you have gathered potentially years worth of data, then need to analyse it all. They can also require huge amounts of people, and are particularly prone to attrition, where people leave the study midway through.

This Study looked at effects of diet on mental health. They followed a set of adolescents and found that poor diet is related to poor mental health and vice versa. This shows the usefulness of prospective studies, as they are able to look at a difference between people (diet), and see the effect that this has over time. We know that diet has a big effect on health, and so to experimentally alter people’s diet to see the outcome would be unethical. So prospective studies allow you to test things that otherwise you wouldn’t have been able to. I think that one of the most important uses of prospective studies Is this; the ability to provide valuable information about the course of a disease. This allows for more accurate prognoses and treatment. This Study looked at a particular disease and gathered information that has led to doctors being able to more accurately diagnose and treat it. This is useful for both the doctor and the patient, as the patient gets the most effective treatment and realistic idea of course, and huge amounts of money aren’t wasted in the fruitless endeavour of curing. These positives can be equally applied to the field of psychology, for example without prospective studies, much that is currently known about the course of schizophrenia would be unknown.   This Study again shows the usefulness of prospective studies in being able to isolate causes, in this specific case identifying disease-related genes. But also makes a point of the fact that they can be long and costly.

One of the main issues I feel there is with these type of studies, is that for important things like diseases, the researcher is just sitting back to see what happens, much like how nature cameramen don’t help the limping gazelle before it is ripped to shreds by bloodthirsty lions. Can this be ethically acceptable? Surely the researchers should be doing everything to help those people. I understand that the results will eventually help many people in the future, but you are effectively sacrificing those people who are studied.

Whilst having a major thorny ethical issue, I think that the advantages of prospective studies make them an invaluable research tool. Rather than being used instead of retrospective studies, they examine different things, and different natures. Surely the most effective method of research would be to use both!

Homework for TA

Comment 1:

http://aglinskas.wordpress.com/2012/02/19/file-drawer-in-a-nutshel/#comment-20

Comment 2:

http://psychblogld.wordpress.com/2012/02/18/science-for-scientists/#comment-40

Comment 3:

http://psud63.wordpress.com/2012/02/19/explanation-of-a-hypothesis/#comment-49

Comment 4:

http://whatalodeofblog.wordpress.com/2012/02/17/unfortunate-results-of-unpredictable-research-some-times-we-cannot-protect-every-one/#comment-37

Retrospective Study

                Hi people, I thought for my latest ‘blog’ I would discuss retrospective studies. Like most things in psychology, I spent most of my A level trying to find bad points about them, so naturally I will list a few of the major ones. But in the object of fairness, I will also mention some good points, and how they can be a very useful tool.

Retrospective studies are essentially exactly what they say on the tin. They are studies that gain data by looking to the past. This means that they can quite often take the form of a questionnaires and surveys, in which the participant fills it out to the best of their knowledge. This Study looked at trauma in school children, suggesting it may be linked to teacher abuse. The study brilliantly highlights to very big positives and negatives of the retrospective approach. Firstly it relies on memory, and I don’t know about other people but I’m only 20 and have a terrible memory, I can barely remember what I was doing last week. This calls into question the results that may have been recalled from a very long time ago. The positive, however, is it would be highly unethical to conduct a study in which you got teachers to abuse children then measure the effects. It would also be unethical to identify abusing teachers then simply let them continue and measure what happens. Whilst what has happened to victims of abuse is terrible, it is obvious that this type of study can access information that would otherwise have been unobtainable, and can greatly help in future interventions.

One major problem for retrospective studies is selection bias. This is essentially error caused by choosing a non-representative sample. This Study shows the effects that by excluding family’s friends and neighbours from a study into child abuse, interpretations of determinants change for black children, but not white. This shows that selection bias has had a double-fold negative result, not only will it have a drastic effect on how those children would be dealt with, but it has also introduced a racial component that simply does not exist. One type of selection bias is sampling bias, and This Document explains very well what it is, which basically is a consistent error arising due to a non-random sampling method. A sample could be biased if a certain group was over- or under-represented in the sample. This is very problematic when you take into account the nature of retrospective studies, and the fact that many take the form of questionnaires and surveys. For example, if you are collecting retrospective data on health, you would introduce a sampling bias if you stood outside a gym handing out questionnaires.

This interesting article lists a number of additional problems, such as an over-reliance on memory. However it also introduces the idea that all you are gaining is correlations, making causality impossible (read the article for more information). It’s not all doom and gloom for retrospective studies though, as This Study shows. It lists a number of strengths, such as being relatively inexpensive, especially questionnaires. They also take a relatively short amount of time to gather data, meaning large volumes can be collected to gain a more comprehensive view of what you are interested in. Retrospective studies are used very often within the field of medicine, and the study indicates why, as they can be used to study very rare occurrences, and diseases which have a long period between exposure and disease.

Overall, there may be a wide range of negatives surrounding retrospective studies, but they remain an essential research tool for a variety of reasons, as mentioned. This requires very strict and diligent attention must be paid when using this method, to ensure that any source of bias or error is not introduced. Interpretation of results must also be handled carefully, as memory often makes fools of us all. As long as these strict guidelines are adhered to, retrospective studies remain a very effective tool in the garage of life.

Homework for TA

Comment 1: http://aglinskas.wordpress.com/2011/12/09/must-the-need-for-documented-ethics-procedures-hinder-research-progress/#comment-14

Comment 2: http://tinastakeon.wordpress.com/2012/02/05/should-psychology-be-written-for-the-layman-or-should-science-be-exclusively-for-scientists/#comment-42

Comment 3:http://psychblogld.wordpress.com/2012/02/04/research-ethics/#comment-30

Comment 4: http://vickygoodwin.wordpress.com/2012/02/05/6th-feb-quantitative-vs-qualitative-data/#comment-26

Observation: Friend or Foe

There are many instances in psychology, and the wider world, where precisely controlled laboratory studies are either not necessary, or simply cannot be undertaken. This is where observations and observational studies step in. This Fantastic website ably describes observational studies, explaining how the researcher essentially just observes natural behaviours, rather than manipulating variables and then looking at the response. It also mentions that there may be certain issues with ethics that prevent a researcher from manipulating variables and carrying out a lab study, such as when looking at effects of smoking on cancer for example. This means that observational studies can be vital in looking at relationships between variables that simply could not be studied otherwise. But there are a number of problems. As the website mentions, there is little to no control over variables or groups, which as we know could cause a range of problems. Bias is a particularly big problem with observations and it can come in a number of forms, such as confirmation bias. This means that whilst these studies can be greatly useful, care is needed in both conducting and reviewing them.

This study looked at the smoking behaviours of people who had recently survived cancer. If trying to study this using a lab experiment, it would necessitate either giving smokers cancer, or making people with cancer smoke. For obvious reasons this is not only highly unethical, but impractical. The study, however is aiming to create successful smoking cessation interventions. This is obviously a very positive aim, and could go on to save countless lives.  By using an observational study, the researcher has been able to access this important information without breaking any ethical guidelines. This would allow the researcher to fulfil the study aims and create the intervention. This is just one area where observational studies are invaluable and necessary in order for important research to be conducted.

As mentioned before, the very nature of observational studies makes them very prone to bias and other problems. There is typically very little control in these studies, meaning that many confounding variables could be manipulating what you are trying to measure. For example in the smoking cancer study, entire patient histories would need to be known before any inferences could be made about cancer and smoking cessation. One type of bias found commonly in observational studies is observation bias. This occurs when researchers tend to look only where they expect to find the most positive significant results. This is also known as the ‘street light effect’. David Freeman explains this in his blog, containing an extract from a very interesting article of his. I particularly liked the analogy of the drunk man, who when questioned by a police officer, was found to be looking for his wallet in a completely different place from where he lost it simply because there was more light there. It does raise a serious point though, about how researchers will introduce their own bias when conducting observations. They will enter the study with preconceived notions of where and what to look for, which means important behaviour may go unnoticed.  This links in heavily with another form of bias that researchers can be guilty of; Confirmation bias. This is a tendency for people to focus only on evidence that confirms their hypothesis or question. This study looks at multiple sources, and this book looks at a social perspective, and both find that hypotheses frequently get tested in a very one-sided way in order to get the best confirmatory evidence. This can be either a conscious or subconscious thing. The preconceived notions of researchers may have a bias towards the confirmatory without them even knowing about it, or it may be more sinister, with greedy researchers wanting more funding. This book brilliantly explains how researchers tend to look for the consequences of their hypothesis being true, rather than false, which may sound sensible, but could miss valuable evidence, increasing the likelihood of a type 1 (maybe 2) error.

Overall, observational studies hold a very valuable and integral part of psychological research, as they enable the testing of areas that would otherwise be impossible to test (practically or ethically). But the numerous sources of bias and problems leads to great care needed when conducting, and also scrutiny and vigilance when reading, interpreting and utilising results of these studies.

Blog: https://tallesttales.wordpress.com/2011/12/09/meta-analysis/

Comment 1: http://destroypenguins.wordpress.com/2011/10/27/is-there-a-need-to-learn-stats-when-we-have-spss/#comment-40

Comment 2: http://nirapsy.wordpress.com/2011/11/25/once-upon-a-time/#comment-28

Comment 3: http://cppap.wordpress.com/2011/11/25/qualitative-research-isnt-as-scientific-as-quantitative-methods/#comment-24

Comment 4: http://groundblogday20.wordpress.com/2011/11/25/the-milgram-experiment-the-epitome-of-good-ethics/#comment-24

Meta-Analysis

Hello wonderful people, I decided that this week I would discuss meta-analyses as I think they are a very useful tool to use in research methods. They basically involve collating results from many different studies, in order to look at the bigger picture and gain better knowledge. Whilst being useful though, there are also many pitfalls and areas that require specific attention to ensure that meta-analyses are accurate.

This study is a nice example of a meta-analysis, and it looked at the results of 12 different psychological intervention studies into paediatric oncology. After all of the 12 studies results were examined, it was found that psychological interventions play an important role in a number of ways, e.g. reducing stress. But why conduct a meta-analysis? Surely if each of the 12 studies showed significant results then a meta-analysis is unnecessary. Well this is the good thing about meta-analyses, they allow you to look at a very broad range of literature, so you aren’t just taking one study as fact; you are actually looking at that study in relation to other studies. This describes excellently how a meta-analysis can be used to improve the power of small or inconclusive studies. This means that you can use more of the research available to you, as before you may have only focussed on the results of the bigger studies. So this ability to access a greater range of research enables you gain a more comprehensive understanding of the topic area. Another page also highlights a number of the advantages of meta-analyses; you are able to generalise the results to a larger population. Rather than just looking at the population studied in one piece of research, you can look at the effects that many different studies had into many different populations. So there are many positives to meta-analyses (I’m not going to list them all because I’m sure you know them/ know where to find them, and there aren’t enough hours in the day), however there are some serious issues that need to be addressed.

The two pages linked above contain a number of disadvantages. For example, if one of the original studies was conducted poorly, or had bad statistics, your meta-analysis has no way to improve this, and so you have just used some research that may be bad. One of the biggest issues with meta-analyses is of course Publication Bias. This study explains how publication bias is basically the tendency for people to only publish studies that show significant results. Being the brilliant scientific minds that we are, we of course know that not significant results have the ability to tell us just as much about the study area than significant results, however these are far more likely to not get published. This means that any meta-analysis conducted may be missing a huge portion of unpublished studies that have the potential to produce wildly different results to the ones presented.

The area of publication bias poses a huge threat to meta-analyses and highlights the point that whilst meta-analyses are important, very strict care must be taken to ensure that results are accurate, and even then, a great deal of care must be taken in interpreting the results. And this study suggests that maybe people conducting meta-analyses are not following the proper procedures, which is having a huge effect on the results found. This shows that for meta-analysis to be considered good, they need to be undertaken with extreme care, however when they are, they are very useful indeed.