Good morning everyone. In this lecture we will start covering the basic background information of this course. Going over the basic types of studies for which will be teaching you methods for conducting power and sample size analysis. This lecture we'll go over longitudinal studies and some introductory concepts surrounding research design. The learning objectives cover the basic information that characterizes the studies we will be focusing on, including defining important aspects of longitudinal studies, and then defining longitudinal studies themselves. We'll also see if you can recognize examples and discuss how this type of study affects correlation. A term that you will hear a lot of in this course is independent sampling unit. This is an important term to understand because you will need to be able to recognize the independent sampling unit in order to correctly run a power or sample size analysis. The independent sampling unit is the unit of study that is independent from any and all other units. This can be people, or even larger groups. It will make more sense once we discuss correlation and take a look at an example. Another important term is to understand, is unit of observation. This is just the measurement of interests for the study which comes from within each independent sampling unit. In a simple example, if you were running a study to determine the height of each person in your class, each classmate would be an independent sampling unit, as the height of individuals in your class are all statistically independent of one another. Then from each individual, his or her actual height would be the unit of observation. We will be discussing power and sample size analysis pertaining to longitudinal studies in this course. Time is key to what you need to remember about longitudinal studies. Longitudinal studies always include multiple measurements of the same person, group, or other form of independent sampling unit being taken over time. Correlation is an important aspect of running a powered sample size analysis. Here, you can read the definition of correlation according to Rosner. Correlation tells us two things about an association or relationship, it tells us the strength and direction. On the right side of this graphic, you can see a strong positive association. These are the values close to positive one. On the left side of the graphic you can see this strong negative association. Values of these kind are close to negative one. Towards the middle of the graphic around either side of zero are the weak associations. Then a correlation of zero means two variables are unrelated. When we say direction of correlation, we are talking about how two variables are changing. For example, height and weight are typically positive correlations, as height increases, weight typically increases as well. This would be a correlation somewhere between zero and positive one. Height and comfort level on airplanes however may have a negative correlation. As height increases, comfort level in the airplane seat may decrease. This would be a correlation somewhat between zero and negative one. Correlations range between negative one to one, and perfect correlations occur at these two exact values. A perfect negative correlation of negative one, means two variables are changing at the exact same rate in opposite directions, such as one increasing, the other one decreasing at constant rates. While a perfect positive correlation of one means two variables are changing at the exact same rate in the exact same direction, such as both increasing or decreasing at constant rates. As stated earlier, a correlation equal to zero means there is no correlation at all, or the pattern or strength of the direction of change between the two variables is not related at all. This complete lack of correlation between two variables means that the variables are independent of one another. While this may not always be exactly true mathematically speaking, for purposes of this class we will accept correlation of zero as independent. The whole reason we are discussing correlation and longitudinal studies induced correlation. Remember that longitudinal studies include repeated measurements of the same independent sampling unit over time. The multiple measurements from one independent sampling unit is correlated. In this course, we will use the term factor to describe a dimension of interest within a study. For example, time can be a factor within a study. Treatment groups versus control groups can also be a factor within a study. Longitudinal studies have what we call within-independent sampling unit factors, or within factors for short. This just means that it has a factor in which the same independent sampling unit offers different values or measurements. Longitudinal studies always have a within factor because this is time, and time is a within factor subject. In the example here, each independent sampling unit or person provides a different measurement in distance walked, each day for five days. The within factor here is time. While longitudinal studies always have a within factor, they may also have, but do not have to have, a between factor as well. Between-independent sampling unit factors are those that take on only one value or measurement per independent sampling unit. Between factors are looking at differences between different independent sampling units, while within factors are looking at different values within the same independent sampling unit. In this example, drug assignment would be a between subject factor because some independent sampling units would be given a drug while others would be given a placebo. Using within and between factors together in a study allows you to look at patterns. For example, how does using this drug differ from not using this drug? How does this difference look over time? We'll now discuss all of these concepts we have gone over using the example. This example will be used many times throughout the course, so please pay attention. Please read it carefully now as it will not be explained as much in the future lectures. This particular study explores dental patients in their long-term memory pertaining to the following, a root canal procedure. Some of the patients were instructed to focus on the sensations of their mouth during the procedure. These patients made up the treatment or intervention group. While others were not instructed to focus on anything in particular, to control or intervention, non-intervention group. These groups were assigned randomly. The intervention group was instructed to focus on physical sensations on their mouths by listening to audio instructions during the procedure. The non-intervention group on the other hand listened to audio on a neutral topic during the procedure. This flowchart provides a synopsis of the procedures of this study. Patients are randomly assigned to either the intervention or sensory focus, or the non-intervention or standard care on this graphic. It's worth noting that throughout this course we'll be using many different terms to describe control group of studies. We'll use terms like standard of care of control and non-intervention. These terms are all synonymous and used in different contexts. But, just know that we are essentially meaning control group. After patients are randomly assigned to the intervention group or the control group, each patient has their memory of pain measured directly following the procedure, six months after the procedure, and 12 months after the procedure. Here, you can see the null hypothesis. That there will be no difference in the pattern of pain over time between two groups. The alternative hypothesis would be that the participants receiving the intervention will have different patterns of pain over time as participants who do not receive the intervention. Each patient is an independent sampling unit, as they are assumed to be independent of one another. The unit of observation is the memory of pain, as this is the measurement of interest in the study. This study includes both within and between-independent sampling unit factors. The within factor at a play is time. As each patient has their memory of pain measure right after the procedure, six months after the procedure, and 12 months after the procedure. Remember that longitudinal design induces correlation in these measurements for each patient. The between subject factor at play was intervention, or sensory focus versus non-intervention, or the standard of care. The combination of within and between factors allows us to see the pattern of memory of pain over time based on different treatments. The combination of within and between factors allows us to see the pattern of memory of pain over time based on these different treatments. As you can see here, it appears that one group's memory of pain decreased more intensely than the other over the 12 month period. The longitudinal design allows us to look at this pattern over time. Let's do a quick review summary. We know that correlation measures associations between variables. Longitudinal designs, which have repeated measures taken over time, induce correlation between the measurements taken from the independent sampling unit. Finally, we went over between and within subject unit factors. A between subject factor has only one value per independent sampling unit, such as a patient being either in the treatment or a control group. They can only be in one of those conditions. An independent sampling unit takes multiple different values for a within factor, such as a patient taking four tests over a span of a year. These are different test scores that are taken from the same patient. Remember, this is a within factor of time is a characteristic of longitudinal studies, and this can create correlation patterns that we will be talking about a lot in this course. That's all for this lecture, thanks for now.
