Quantitative Evaluations

From CS1635 Spring 2014
Jump to: navigation, search

slides

Contents

Readings

Charlie Koch 12:53:31 2/23/2014

Today's reading covered how to run an experiment. There are a ton of variables and threats to validity within a single test. Only through careful consideration and planning can a test be conducted successfully without major bias or skewing. The independent variable is the condition you are aiming to test. The dependent variables are the variables you observe during the test. Along with the independent variable, you must also consider all of the control variables, which should remain stable throughout the test. Some of the control variables can be randomized as random variables. Using a good mix of control and random variables can give strong, generalized results. The last type of variable is confounded variables, which as the name suggests, can complicate test results. These are variables that may not be obvious or immediately observable, but do have an impact on the subjects of the test. Along with the variables, there are also several things that can threaten the validity of a test. History and maturation of the subjects can skew their results, both of which are also tied to the selection of the subjects. The testing itself can also condition the subjects to react differently. Statistical regression is a factor that can also negatively affect a test result. Overall, running an experiment requires a lot of planning and consideration to do successfully.

Brian Kelly 9:11:25 2/24/2014

The required reading for this week dealt with something that was relatively unrelated to computer science in general but can be applied to things like our current group project. By approaching our interviews with potential customers like an experiment, we may be able to ascertain more useful information than if we were just improvising. We could establish dependent, independent and control variables in our "experiments" to learn more about our user's experience. The independent variables may be things like the user's age or gender. The dependent variables could be the ease of interaction with the application. The rest should be control variables that we do not want to change: similar environment, same instructions, etc... Overall though, I am not sure how necessary it is for us to perform a full, scientific psychological experiment. It may be a bit of overkill to go to all of that trouble.

David Grayson 22:24:20 2/24/2014

“How to Do Experiments” is a description of just that…how to do experiments. The article seems like a review of many things I have learned many times when learning to do experiments. Your independent variables are the ones that you manipulate during the experiment (think of this variable as the x-axis variables). The dependent variables, or y-axis variables, is the variable that you typically expect to change based on the particular independent variable. To ensure that independent variable changing is the cause of the dependent variable changing, experimenters must try to control the setting in which the experiment is conducted through control variables. However, when controlling variables it is best to keep in mind the target population so you do not control variables that sacrifice external validity and reduce the generalizability of your research. One way to take account for the variables that cannot be controlled due to external validity concerns, researchers must try to randomize the sample of test subjects based on the target population. The researchers should also randomly assign test subjects to control groups. Aside from external validity concerns there are internal validity concerns such as confounding variables. Confounding variables include things like maturation, or when participants improve with practice. Assuming random selection, participants dropping out of studies also impacts internal validity if there could be a reason why they dropped out related to the research. This article contains valuable threats to validity to look out for when designing experiments and is an excellent reminder of the features of a good experiment.

Steven Bauer 17:52:27 2/25/2014

Todays reading is "How to Do Experiments", the article starts off by defining the key components of an experiment. These components are the Independent, Dependent, Control, and Random variables. It is important for any researcher to know these definitions because without having clearly defined variables it might not be clear to others and even ourselves what we are testing and what assumptions we are making. For an experiment to be valid we need to have an independent variable that we are going to manipulate and a dependent variable that we are going to measure. Because the real world is not black and white it is difficult if not impossible to conduct a perfect experiment. This is because the real world has so many variables and we try to make the conditions of the experiment controlled as to minimize changes other than what we want. There are many threats to internal validity, these include history, maturation, selection, mortality, testing, statistical regression, and interactions with selection. Due to the sheer number of these I am only going to summarize one of them. Selection is when the experiments participants are self selected. The problem is that the people who selected themselves for your experiment are clearly interested in it in some way and this might skew your results. The example they gave was comparing college students who sign up for a study at the beginning of a term compared to ones that were forced to do it at the end. It is possible that the ones at the end are very uninterested compared to the early sign ups. The next part of the article summarizes the "Experimental method". It delves into picking the variables and the dangers that accompany it. One example was that when we are picking which variables go into each category, the more specific we make our test, the better our internal validity is but the worse we are able to generalize our results. In conclusion the Experimental Method allows us to make statements about when a circumstance is manipulated that it causes a change.

Longhao Li 14:17:56 2/26/2014

The topic of this article is very interesting and very useful while designing user interface. The method to do experiment gives developers the way to test their design of interface. The author firstly introduces the most important component of experiment, which are the variables. There are two types of variables: independent variable, which will not be effect by the participants, dependent variables, which is the behavior that chose to measure. The dependent variable always followed hypothesis, since the measurement are trying proof the hypothesis. Then he talked about control variable. It is the variables that are controlled in the experiment. Controlling variable will help us to find if the single variable affect the result of experiment. But we cannot control too much since if we did that, we may narrow our target group, which may not reflect the general case that the experiment host set at the beginning. Also the author talked about the random variable. It is the opposite of control variable. Once we control one independent variable, we need randomly selection so that we can make sure the generalizability of the results and also reduce the bias. This is what random variable gave us. Also we cannot fully rely on truly random since it may lead to difference on the amount of treatment in different group. So it is a good way to do random with restriction, like random select people into group, we need to control the number in different group to make balance. While we want to do an experiment we need to be careful with the confound variable, which we confound during the experiment. It may lead to worthless of the experiment since the result will not reflect the change of independent variable. Meanwhile, the author also talked about the threats to internal validity: history, maturation, selection, mortality, testing and also statistical regression may influence the experiment result. In general, when design an experiment, people need to be careful about these things. Good design experiment can help people to get the right result easier.

James Devine 19:45:12 2/26/2014

This reading is about the experimental method, specifically the different kinds of variables and potential threats to the experiment’s external and internal validity. Experiments are conducted using independent and dependent variables. A variable is dependent because it depends on the test subject’s reaction to the independent variable. The experiment also has control and random variables. Variables are controlled so that the independent variable is the only thing changing throughout an experiment. Random variables are used during selection to avoid a biased selection and so that the experimenters are able to make generalizations after the experiment is complete. Confounding variables are defined in the reading as “any circumstance that changes systematically as the independent variable is manipulated”. An experiment is internally valid if the change in the dependent variable is due only to the change in the independent variable. Threats to internal validity are history, maturation, selection, mortality, testing, statistical regression and interactions with selection.

Xiaoxuan Chen 20:00:53 2/26/2014

This article talked about how to do experiments. Experiments usually includes selecting variables, whether it's independent variables - the circumstance of major interest and is independent of participant's behavior, dependent variables - the behavior to measure, control variables - the circumstances that needs to be controlled, random variables - having some circumstances vary randomly, and confounding variables - any circumstance that changes systematically as the independent variable is manipulated. The major advantage of random selection, selecting randomly from a population in order to form a representative sample, it the generalizability of the results.This eliminates the bias from results. The article went more detail on random variables especially randomization with constraints. When we choose a random variable, we have to make sure it actually varies in a random way. Usually most of the circumstances that became random variables would be associated with participants. Those can be randomized by randomly assigning participants. A constraint on the random assignment or randomize within blocks can help have a circumstance in between of random and controlled. It then goes on to discussion some of the threats to internal validity including maturation - threat to internal validity caused by participant's more experienced, select - whenever participants are assigned nonrandomly and particularly self-selection, mortality - drop out of experiment, testing - can change behavior independently of any other manipulation due to testing, statistic regression - their score tends to move towards the mean when participants are chosen on the basis of having scored very high or very low, and interactions with selection - result from maturation and history. These experimental methods are of great help for doing experiment, considering how detailed it explained each possible variables and threats.

Guoyang Huang (Guh6) 20:40:04 2/26/2014

The chapter discussed two major points. The first point was about the types of variable that are in an experiment. It discussed independent, dependent, control, and random variable. I learned that the independent variable is the major interest of the experiment and the dependent variable depends on what the participant does. One should not always control everything because it would lead to a very specific case and not a generalize case. There is also external and internal validity. The external validity refers to how well a causal relationship can be generalized across people, settings and time. There is also randomization within constraints which dealt with finding a variable in the middle between the extreme of randomness. The second point discussed the ways to limit internal validity. This can cause confounding variables which are any circumstances that change systematically as the independent variable is manipulated. In order to avoid it, one must know the common causes. I learned about maturation which is the more experience the participant has, the problem becomes easier or can affect the outcome. Selection is choosing nonrandom. Testing can cause a change in awareness. Statistical regression can also affect the results because of choosing certain groups based on statistical means. Lastly, mortality and interaction with selection can also affect the result by the history of the results.

Zhanjie Zhang 21:20:07 2/26/2014

This chapter talks about how to do experiments. First we must look at the different variables that are at play. There is the independent variable and dependent variable. Once we know these two variables, we must make a hypothesis. However, other circumstances in an experiment will be needed to be accounted for in some way. We use the control variable to do this. This way, all circumstances other than the independent variable would say constant through the experiment. We must also be aware that it is impossible to control all of our variables. We also do not want to control all of our variables; otherwise we would create a unique set of circumstances that does not occur in the real world. The generalizability of an experimental finding has been referred to as external validity, how well a causal relationship can be generalized across people, settings, and time. Sometimes, we may want to have circumstances into more random. Falling between extremes leads to randomization within constraints. We can control one part of the experiment and randomize the other parts of the experiment. There are threats to internal validity. There is history of the project, maturation, selection, mortality, testing, and statistical regression. We must be aware of these variables that will affect our end result. Being aware of confounding variables that will cause low internal validity can reduce the difficulty of pinpointing which independent variable caused a change in the dependent variable.

Max Campolo 21:25:17 2/26/2014

This reading was on how to do experiments. The first topic that was covered was the types of variables in experiments. Variables are independent variables, control variables, dependent variables, and random variables. Independent variables are the circumstances of major interest in an experiment and are independent of the participants behavior. The independent variable is changed during the experiment but is not measured. Dependent variables are what is being measured in the experiment. The measurements are in response to manipulations of the independent variable. A hypothesis is a statement about the expected behavior of the dependent variable in response to the independent variable. Control variables do not vary and are kept constant throughout an experiment so that measurements only depend on changes in the independent variable. Random variables are variables that are allowed to vary during the experiment. This allows the experiment to be generalized to different situations or environments. This is called external validity. Another topic that was covered were variables randomized within constraints which is when a variable is allowed to be random but only within a set of limits. Another type of variables are confounding variables. These should try to be eliminated because they change based on the independent variable and can confuse results of the experiment. Internal validity is the concept that all of the criteria of the experiment is correct. Confounding variables can decrease the internal validity because they may cause changes in the results. When an uncontrolled event in an experiment happens it is called history and also negatively effects the validity. Other concepts covered which effect the validity is maturation when the experiment changes with age, selection which is the participants in the experiment, and morality which is the loss of individuals from the experiment. Lastly, the reading covered the importance of testing, statistical regression, and interactions with selection which all also affect the internal validity of an experiment. In order to have a good experiment, these are all important issues to consider.

Cory Savit 23:34:30 2/26/2014

Today's reading, How To Do Experiments, focused on the experimental method and more specifically the variables that exist in experiments and research. The first three are commonly known (at least they seem to be; independent variables, dependent variables, and control variables. The independent variable cannot be controlled by the subject but is manipulated by the experimenters to determine if it elicits some sort of reaction or effect which will be measured. This measured response is the dependent variable. Control Variables are used to limit external influence on the study by keeping certain things consistent. Random variables are circumstances that are permitted to vary randomly, although they often have to be “constrained” so as not to bias the results. Finally, Confounding variables are factors which could have had a significant effect on the dependent variable, but are not the independent variable(s), and in doing so were threats to internal validity. While it succeed in explaining the variables that exist in experiments, I believe the main point of this chapter was to demonstrate how to structure an experiment. It went beyond just providing a rubric and explained that there are many ways to structure experiments based on the purpose of your experiment. While having many control variables limits the number of random variables and decreases the likelihood of confounding variables, and so strengthens internal validity, it also means the results can't be generalized since they only apply to very specific circumstances. On the other hand, allowing for too many confounding variables diminishes internal validity (Who wants to apply findings to a large population when those results are likely wrong?). The correct balance is somewhere in between, but where that lies depends on the purpose of the study.

MJ McLaughlin 23:51:43 2/26/2014

This chapter of Martin’s “Doing Psychology Experiments” provides a great overview of the many different aspects of experiments and the results we can obtain from them. Following the experimental method is a powerful tool in that it allows us to make causal statements about the factors we take into consideration in our experiment. We can say that one thing causes another, which is incredibly useful, but also requires that strict standards be met in order for that statement to be correct. To make sure our experiment lives up to these standards, it is necessary to understand the many different aspects of an experiment. Variables are those factors that change in an experiment. An independent variable is one that is not affected by a participant’s behavior-it is independent of their actions and is rather controlled only by the experimenter. It is the variable that the experimenters manipulate in order to see how that affects results gathered from the participants in the experiment, and as such must have at least two different conditions, or levels, that will be tested to see the effect had on the experiment’s results. These differences are found in the dependent variables, those that are dependent on the participants’ actions and that may change in response to changes in the independent variable. Predictions about this relationship, specific or general, can be said to be the hypothesis being tested in the experiment. To make sure that the results of an experiment can be validly used to support or disprove a hypothesis, one must make sure that as few variables as possible affect the cause and effect relationship being tested. One way to do this is by controlling any outside variables that could affect the results of the experiment, to ensure that the change in the independent variable is the only variable that caused a change in the resulting dependent variable scores. But it is impossible to control all variables, and doing so would make the results of the experiment not generalizable, or applicable, to real life, where variables also cannot be controlled. This external validity, or generalizability of experimental results to real people in real life, can be strengthened through randomization, as life itself is random. Participants can be randomly selected to ensure that the experimental sample is representative of the target sample, and participants can also be randomly assigned to different experimental levels to make sure that, for example, participants with similar backgrounds or characteristics that may affect experimental results aren’t all assigned to the same independent variable level. But one must make sure that the randomization process is really random, and that what is being randomized can actually be randomized! It is also useful to understand that control and randomization lie on a continuum, and you can randomize within constraints. You can control certain aspects of an experiment, and then randomize within those controlled aspects by, for example, randomly assigning controlled blocks of tasks rather than randomly assigning all tasks, which can actually skew results. It must also be understood that no experiment is perfect and that outside interference can always occur. One typical form of such interference is confounding variables, which systematically change with the independent variable and make the relationship between the independent and dependent variables unclear. They cause a threat to internal validity, which measures whether the independent variable being tested was actually responsible for the experiment’s results, or if there was something else affecting and confounding the results. Such threats to internal validity include historical threats, which consist of historical events or changes that occur between trials and could affect results, maturation of a participant between trials, selection threats that can produce bias when participants assign themselves to experimental conditions usually because of a bias that could affect results, mortality, or participants dropping out of the experiment, which is not too troublesome when it occurs evenly across participants but can be very much so when participants differentially drop out of certain conditions, usually for confounding reasons that could affect results, testing,which can make participants unnaturally aware of experiment-related content and hypotheses, and statistical regression, which describes how extreme measures tend to trend towards the mean or less extreme measures on retesting, because of extreme random error usually being responsible for the extreme results in the first place, error which is likely to be less extreme in subsequent tests and affect the “true” measure less. If groups are selected based on such measures, it is likely that these measures are due to error, and the validity of results is threatened. Finally, interaction with selection describes how threats can have differential effects on nonequivalent groups, which can also bias results. The experimental method is a very powerful and useful tool and is applicable to all kinds of processes, including interface design. By gaining a deeper understanding of all the intricacies of the method, we are better equipped to carry out experiments that produce valuable, useful results that make our endeavors that much more fruitful.

Ariana Farshchi 0:08:16 2/27/2014

This weeks reading, Doing Psychology Experiments outlined the different components that make up an experiment. The reading first outlines the different kinds of variables needed to do an experiment, such as independent variable which is a variable independent of the participants behavior, dependent variable which is dependent on what the participant does, control variables which are variables that can be controlled in such circumstances, random variables which are variables that vary randomly, and confounding variables. After outlining the variables, it talks about threats to internal validity such as maturation, selection, morality, testing and statistical regression.

Matthew O' Hanlon 0:18:05 2/27/2014

The reading described the variables that play into the results of an experiment. The author described each of the variables in detail giving their definition followed by an example for further clarification. He discussed independent and dependent variables first as those are the two most interesting variable in any given experiment. The independent variable is controlled by the person conducting the experiment to see if it causes certain behavior in a test subject. The testing participant’s behavior in this case would be the dependent variable which depends on the independent variable for stimulus. The expected relationship between the independent and dependent variable can be described in the hypothesis. The next section of the reading talked about external validity and how to achieve greater external validity. This mean that you can have control variables in an experiment, but not too many because having too many would make the conditions of the experiment more specific than would occur naturally. If others are to take advantage of the results of an experiment, the the conditions should be generic enough such that the independent variables tested would also affect others in the same way.This is why it is important to have random selection and assignment of participants to your experiment. The more random the selection of people, the more it reflects the general population which in turn lets others take advantage of it. The last section had to do with problems related to Internal validity. The author pointed out already that external validity is of great concern to the tester, but so is internal validity. The types of problems that show up in experiments are non-random selection, mortality, pretesting, and statistical regression. Essentially, non-random selection can allow for bias in experiments. The mortality rate can change the characteristics of the group.Pretesting can change what your participants are looking for, so their behavior might be different had they not known what you were looking for. Statistical regression can be troubling because one test doesn’t say enough about a group of people.

Michael Mai 0:42:24 2/27/2014

This reading had a lot to do with psychology and scientific experiments. It went over a lot of stuff that I learned in high school science courses. The author first went through and discussed different types of variables. He goes over independent, dependent, control, and random variables. Independent variables are those that aren't dependent on anything else. The dependent variables are those that are dependent on the independent variables. The control variables are things that the person carrying out the experiment controls. An example of this is making every participant of an experiment do it in the same location so there is no other factors in the results. The random variable refers to random assignment of circumstances to levels of the independent variable. The author then goes on to talk about threats to internal validity. He goes over a few of these threats including maturation, selection, and mortality. Maturation is a threat because as participants grow older or more experienced, results can be different. Selection is also a problem because the example he showed was in his psychology class, a lot of people in the first semester class would be freshmen and the second semester class would contain a lot of seniors that held off the class until their last year. Obviously results would be different for these 2 groups of people. Mortality is another smaller issue because sometimes participants die unfortunately. In the end, the author goes on to give some examples of these variables in an experiment that he carried out.

Chris Solis 0:53:21 2/27/2014

The opening of the reading seemed to be a simple introduction to basic psychology terminology such as different variables and assignments. It talked about different types of random terms such as random variables, random assignment, and randomization within constraints. It didn't seem to focus much on the benefit of each tool but more on what to do to make sure each tool is used effectively in an experiment. The next section talked about possible negative occurences that should be taken into consideration when creating an experiment. These items include maturation, selection, mortality, testing, statistical regression, and many more. Each has its own drawback to gaining reliable results and because of this, they need to be dealt with to ensure reliable results. The final section provides an example of what a well structured experiment would look like by breaking down each component such as the different variables. Overall the reading was easy follow and concise.

Nicholas Amoscato 0:55:21 2/27/2014

In the second chapter of his book “Doing Psychology Experiments”, psychology professor David Martin describes how to execute an experiment. Martin begins by defining several relevant terms. An independent variable is the circumstance of major interest that does not depend on the participant’s behavior but rather is manipulated by the experimenters. A dependent variable is directly affected by the participant’s behavior. A hypothesis is a statement about the expected nature of the relationship between the independent and dependent variables. Control variables remain the same across different levels of the experiment, and the effectiveness of an experiment is largely determined by the attention given to control variables. Circumstances that are permitted to vary randomly are called random variables. There are several different contexts of random worth describing: random selection involves selecting a representative population; random assignment ensures that the participants on each experimental level are representative. Randomization within constraints involves controlling part of the event assignment while randomizing the other part. A confounding variable is a circumstance that unintentionally changes as the independent variable is manipulated. External validity refers to the generalizability of an experiment in regards to people, settings and time; it is important to note that the more highly controlled an experiment, the less realistically applicable it is. Internal validity refers to whether or not the manipulation of the independent variable solely influenced the dependent variable, or if the process was influenced by some other confounding variable. Martin goes on to explain different threats to internal validity. He begins with history, or the occurrence of some event between two experiments that inadvertently affects the results. Maturation refers the side effects of growing older. Selection involves a bias when participants are assigned nonrandomly. Differential mortality in which particular kinds of participants drop out of an experiment, can also prove to be a threat. Testing can influence the response of participants by prompting them with things to think about. Statistical regression refers to the fact that aggregate scores of participants tend to move toward a mean. Martin concludes his chapter by defining a schematic framework that incorporates the concepts he introduced. Overall, I enjoyed the chapter. Some of the beginning terminology was a review from a statistics course I took a few years ago; however, the details involving external and internal validity were new.

Aamir Nayeem (aan14) 2:05:18 2/27/2014

Today's reading was all about how to run experiments and different types of variables for which an experimenter must account. It mostly dealt with different kinds of confounding variables which could cause massive errors and therefore must be avoided. It's a lot of stuff I've learned before in psych classes and classes on research methods, but it's nice to see it all together in one neat place as a good reference. It definitely seems like there are at least a few big mistakes that a researcher could potentially make, so it was important to brush up on this, especially as we run experiments with subjects as our projects progress. Some of these errors seem easier to avoid than others, while some of them seem downright inevitable, unfortunately, and just have to be minimized and reported with one's findings. Overall, though, it seems as though there are a few critical things that one must avoid, and an experiment can hopefully go well if those things are avoided and those guidelines followed. It shows the importance of taking time and designing a good study, as, just like everything, good planning ahead of time can save a lot of problems later.

Bret Gourdie 2:25:02 2/27/2014

When performing an experiment, much care must be done in making sure that your results remain meaningful. This reading would have been good to review before our group used a paper-prototype, as we could make sure to follow these instructions as well as the ones in the paper-prototyping book. The first concern is that of variables. While independent variables are directly manipulable by the experimenters, it is the participant's reaction to the dependent variables that we wish to harvest. To narrow our focus as to what dependent variables to record reactions of, we form a hypothesis to mark a central idea for these experiments. However, we need to not control all the variables in order to generalize our results, which is called the external validity. There is also much discussion on randomness; discussions based around double-blind studies are noted but not referenced outright. Randomization within constrains is also important: we don't want to present a particularly stressful situation to the participant based solely on the grounds of randomness! Finally, there is the notion of the confounding variable, which affects results as a hidden sort of external force, such as a learned behavior being repeated slowly at first. It is here that internal validity is quite important. Internal validity is the measured change that the independent variable makes to the dependent variable. Just a few things that can affect internal validity are history, maturation, selection, statistical regression, testing, and interactions with selection. These concepts are able to discredit the precious internal validity in one fell swoop! It is important to make sure these forces remain away from the experiment.

Buck Young 2:43:57 2/27/2014

Todays reading was interesting. It was about psychological experiments and considered different kind of testing variables. Independent variables are set by the experimenter while the measured behavior is known as the dependent variable. Invariable variables (as ridiculous as that sounds) are called control variables -- they are set and not allowed to changed over the course of the testing. Random variables, however, are allowed to change over the course of testing. Random variables can be bound to a set of constraints if needed. Confounding variables are detrimental to the relationship between independent and dependent variables because they change in relation to the independent variables. This can be bad over the course of multiple tests and confuse the results.

Robert McDermot 6:13:39 2/27/2014

Today's reading was regarding the scientific method and how to design an experiment. There are many types of variables that need to be either controlled or accounted for when designing and experiment. If you are not careful, the experiment may not be valid and your work would be for not.
I did not personally get much new information out of this reading having heard it all before in an introductory psychology course.

Sara Provost 7:55:44 2/27/2014

This week’s article is about how to perform experiments. First the article discusses the different types of variables that exist in an experiment. The independent variable is the subject of major interest. The dependent variable is the behavior that is under experimentation. The control variables are the behaviors that are not under experiment and are controlled. Random variables are the behaviors that are allowed to vary randomly, such as members of the sample. Any variable that changes systematically with the independent variable is a confounding variable. It also describes different types of validity, one of which is internal validity, which refers to whether the manipulated change in the independent variable caused the change in the dependent variable. Maturation is a threat to internal validity, because of the growth of participants’ experience. Selection is a threat to the testing sample, because it can affect the experiences of participants as well. Additionally, testing itself can be a threat to internal validity, because the tester can unintentionally change behaviors. The final threat to internal validity is mortality, which refers to when a participant leaves a study. All of the previously mentioned concepts are combined into the testing schema, which allows for experiments to take place in a more controlled way. I think that this article has useful information for when our groups due our own experiments. Using this information we can be more aware of how our actions will affect the experiment and the behavior of the subjects.

Derrick Ward 8:20:01 2/27/2014

In today's reading we discuss how to properly conduct an experiment. In this discussion we talk about a myriad of elements that help us establish external validity. Some of the elements we discuss are independent variables, dependent variables, control variables, random variables, random variables with a constraint, and confounding variables. To define a few a theses, a independent variable is one in which it is not dependent on user input of behavior. An example of this in the reading was light. In an example the reading gave, a participant of the experiment was be triggered by a light to press a button. A dependent variable is one in which it is dependent on the user input or behavior. Using the same example, the dependent variable would be the time it takes the participant to press the button. A control variable is one in which it does not depend on user input or behavior but controls the independent variable. Using the same example stated earlier, our control variable would be light intensity. In the experiment the researcher would have a way to control how bright and how dim the light is. This was an attempt to see if there were changes in the time it takes a participant to press a button, depending on light intensity. The section on random variables and random variables with a constraint was most likely the most interesting for me. I found defining this part of the experiment to vital to the experiments attempt to be externally valid and generalizable for real world applications. Via a statistics course, I knew ahead of this reading that an experiment needs a random sample size. I also learned that it is imperative to keep the population exposed to two or more stimulants equal. From personal experience, I found it difficult to identify and control confounding variables, that were present in my psychological experiments from a previous class. All in all I enjoyed the reading. It was a little redundant in some cases, but it is always important to restate important information.

Megan Ziegler 8:37:47 2/27/2014

This article covers basic experimentation methods and their application to psychological testing. First, it covers variables: the variables being tested, independent and dependent, and those that must be accounted for, controlled and random. Most variables in this case will be randomized on constraints, with a balance of randomness and control, so that the data is more valid. There are many other threats to a test's validity, however, such as changes that have happened to a subject outside of the test, selection bias that occurs due to practical necessity, and regression of extreme scores towards a mean value. All of this must be accounted for when constructing an experiment.

MJ (Mary Letera) 8:49:48 2/27/2014

Today's reading goes over the basics of doing experiments, which are common to all fields. It discusses hypotheses, control variables, independent and dependent variables, etc. This is also familiar from statistics. Of the whole paper, the thing that I found most valuable to take away was the importance of selection. I think it would be easy to pick poor test subjects without even realizing it, and this can invalidate a whole study. Even as a consumer of information, I think these are important things to be aware of. Given how easy it is to stumble upon information, both good and bad (such as through facebook), everyone should be aware of at least the basics of good experiment methodology. Too often people are whipped up into hysteria based on the conclusions drawn from a study that was done poorly and thus can't be relied upon.

Hao Zhang 11:24:30 2/27/2014

This chapter is talking about the importance of experiments. It shows us to learn how to design, execute, interpret, and report on simple psychology experiments and guides you through the experimentation process in a step-by-step manner. It emphasizes the decision-making aspects of research, as well as the logic behind research procedures. It also devotes many of the ethical questions that confront new experimenters - giving us a complete introduction to the psychology laboratory. Variables are important elements of an experiments. There are independent variables, dependent variables, control variables, random variables and confounding variables. Threats to internal validity compromise our confidence in saying that a relationship exists between the independent and dependent variables. History is a threat for the one group design but not for the two group design. In the one group pre-post test design, the effect of the treatment is the difference in the pre-test and post-test scores. This difference may be due to the treatment or to history. Is not a threat for the two group (treatment/experimental and comparison/control) design because the comparison is between the treatment group and the comparison group. If the history threat occurs for both groups, the difference between the two groups will not be due to the history event. : In a short experiment designed to investigate the effect of computer-based instruction, Ss missed some instruction because of a power failure at the school. : In an experiment involving reading instruction, subjects grouped because of poor pre-test reading scores show considerably greater gain than do the groups who scored average and high on the pre-test. Experimental Mortality is differential loss of participants across groups. After reading this chapter, I have more sense to design a good experiment.