Welcome to the fourth installment of the Psychology thread series, enjoy!

I will begin by discussing the Hawthorn effect. This is relatively simple to understand, what the Hawthorne effect essentially is that people have the tendency to change in performance just by being in an experiment itself. This demonstrates our need for a control group as we can account for a good deal of the Hawthorne just by measuring the numerical value reflecting how much the control group is affected and subtracting it from the experimental, or some mathematical procedure of the sort. The significance, nonetheless, is that the effect is known and can be accounted to a certain extent.

Within experiments it is essential to define the operational definitions well, as there are many variables that could take effect, so it is good for replication of the experiment for verification by other experimenters at some other date that the definitions are as precise as needed. For example, if you were conducting a study on the effects of alcohol on the blood during various parts of the day, it is necessary to specify in what form the alcohol is, even the brand name or other label by with the same substance can be obtained is necessary. This makes sure that in the future others use the same exact methods in order to verify that the results were not fraudulent or consisting of other types of bias.

As discussed to a smaller extent earlier, the taking of the placebo pill is known to have effects on people. This is termed the placebo effect. By having a control that takes the inert substance, the psychological and physical effects of the placebo can be determined and the results of the experiment can be accounted for in ways that do not allow the placebo effect to alter the results.

Sometimes scientists use counterbalancing procedures to have accurate results. Counterbalancing is when the scientists have the subjects be their own control by having them function as the control and the experiment in succession. I will explain an example to make it clearer. If a scientist wanted to measure the adrenaline levels in someone reading a warning out loud and just reading random letters, with the reading a warning being the experiment and the random words being the control, then they can have the same person do both in succession and see the results and form conclusions. This is a really good technique in the fact that each personal difference is accounted for with most of the variables that will vary from person to person being removed. Especially in the case of adrenaline, depending on the experiences that someone has been through the adrenaline increase might be different. Even naturally the levels might vary from person to person just based on physiology, this technique accounts for that and allows for more accurate results. However, a problem arises when you think about it a little further. What happens if the first control test or the experiment test affects in some manner the results of the experiment test or the control test respectively? This is called order bias and is a scenario that might occur within this type of experiment. In order to most effectively get rid of it while still employing the same method is to randomly assign the people to complete the control or the experiment group first before proceeding on to the next one. This is usually done 50 / 50, and is our best current method.

Now we get into some actual statistics that is relevant to Psychology, the mathematics of correlation. Correlation deals with the ASSOCIATION between two variables. It does not, however deal with causation. Remember: CORRELATION DOES NOT IMPLY CAUSATION! Just because two things are correlated does not mean that one influences or is a cause of the other. This will become more clear as I progress with telling you what exactly correlation is and supply relevant examples. Correlation is defined to be in two forms, positive correlation and negative correlation. Positive correlation is when the increase of one variable is associated with the increase of another variable. Negative correlation is when the increase of one variable is associated with the decrease of another variable. That is two say, two variables are positively correlated when both goes up (within geometry you may refer to this as positive slope when looking at a graph) and two variables are negatively correlated when one goes up as the other goes down (in geometry this is known as negative slope if you were to look at a graph of the variables). I make the connections with geometry to make it easiest to understand and tie many things that you have learned in the past together as Psychology is pretty much a mix of biology, statistics, and certain historical aspects, so being able to tie information learned from these and related subjects is essential.

In many cases human behavior is not able to be easily determined through an experiment. This is especially true when it comes to behavior. Since behavior is spontaneous, it is not easily done under a lab setting and is often done after the relationship is viewed as a likely one. In this case the experimenter will study the different possible variables and conduct all of the other parts of the research process, this we call an ex facto study. The purpose of the ex facto study is to search for a relationship for the variables through researching the variables without actually conducting the experiment. I keep repeating the paragraph within my head, but cannot say that I am coherently writing down what I am thinking, so here is a website that can help with the definition of this type of study. Website

Surveys can be used to show associations. Since the experiment is not actually being conducted, there is no way for the scientists / experimenters to be sure that one variable is causing a change in the other variable, but by looking at the answers to the survey, the experimenter can check for an association. Within association studies it is important to note that there is no independent and dependent variable, which is simple to understand as no variables are being proved to affect the other, this leads to a statistical explanation with graphs. Within normal representation of data, graphs are used in which the x axes normally depicts the independent variable while the y axes reflects the changes. (For example, if X was time and Y was number of bubbles forming from a solution, then the amount of time is what is causing the amount of bubbles that are forming). However, within an association, since there are no independent nor dependent variables, then it correct to put either variable on the x axes and the other on the y. The key problem with surveys is that external variables cannot be controlled for as each person is different so different groups will not be accurate as there is an assortment of different variables, and the person themselves cannot be a control without actually experimenting. Since there are a lot of differences, the researcher will not be able to conclude which one of the differences is the cause of change within the variables.

Often times the survey is used because of convenience, because it can be mailed to many people. However, this opens the gateway to a whole parade of confounding variables depending on where, when, how, and why the people respond to the survey. If the survey is on whether they would be willing to try out a new sun screen type of product, then they could be more likely to pick yes, if they are answering outside in the sun or they could be less likely if they rarely leave the house and are currently inside. This because of all the differences all that can be garnered is the idea that a change in one is associated with a particular change in the other without being able to relate cause at all. Another downside to surveying is the fact that the surveys can rarely represent the population for which the survey is supposedly forming conclusions on. This is because even if the random sample (as discussed before in the previous thread) is used there are going to be many people who are picked, but just not choose to respond to the survey. This is called low response rate and is the reason why caution must be exerted in believing what the results of a survey are. Seeing as people that feel more strongly about a subject might respond whereas those who do not may not means that the survey may be biased, so do not always believe all of those online internet surveys!

Control is also not present within naturalistic observational studies. What these are, are researchers observing the subjects without taking part. This observation prevents the presence of the researcher to cause effects (which occurs quite often) and create a realistic idea on the behavior of different participants. This differs from field experiments in the fact that naturalistic experiments do not attempt to remove any sort of confounding variable, but rather keep everything the way it is.

I will end here for today, but I hope you enjoyed the fourth installment of the Psychology Series !