Module 1 Handout - Basic Concepts & Intro to jamovi PDF

MODULE 1: BASIC CONCEPTS & INTRODUCTION TO JAMOVI© MODULE OVERVIEW In this module, we will give you a detailed discussion on the basic concepts of experimental designs. This includes definitions of experiment and how they fit in scientific research, and principles of exp...

MODULE 1: BASIC CONCEPTS & INTRODUCTION TO JAMOVI© MODULE OVERVIEW In this module, we will give you a detailed discussion on the basic concepts of experimental designs. This includes definitions of experiment and how they fit in scientific research, and principles of experimental designs. Also, we will introduce you to jamovi©, a license-free software, which we are going to use in this course. At the end of this module, you are expected to determine the role of Statistics in scientific research and understand the principles of experimental design. MODULE TOPICS I. Experiments in Scientific Research II. Principles of Experimental Design III. Introduction to and Navigation in jamovi© TOPIC 1: Experiments in Scientific Research Scientific research is an organized and systematic way of finding answers to questions by using the scientific method. It is organized because there is a structure or method in going about the research. It is systematic since there is a definite set of steps to follow. Whatever the discipline is, may it be social, biological or arts, the research process is done to produce accurate results. Researches that may be considered successful are those that uncovered patterns such as in exploratory analysis or found answers to specific research questions. Answers may not necessarily be the ones we expect but still they provide information surrounding the nature of the question posted at the beginning of the research. Statistics plays an important role in every essential stage of research. It provides tools and knowledge to carefully plan the procedures to take especially during data collection, analysis, and interpretation of results. The motivation for research is the research question which allows us to STAT 162 – Experimental Designs Page | 1 Institute of Statistics University of the Philippines Los Baños MODULE 1 [STAT 162: Experimental Designs] Basic Concepts and Introduction to jamovi© define successive steps within the process. Statisticians can help researchers to focus on the draft research question, to calibrate research questions, to convert questions into hypotheses, to devise appropriate statistical tools for every hypothesis and ultimately, to analyze data, and to interpret the results to provide answers. There are various research designs in literature- one of which is the experimental research design. This is applicable in all scientific disciplines and appropriate whenever practical and ethical. Experiment is a research design which provides the highest level of evidence of the cause-effect relationship between a treatment (or exposure) and response (or outcome). Experiments are characterized by the presence of treatments and experimental units to be used, by the manner of assigning the treatments to the units, and by the responses that are measured. We will define these in the next slides. In-depth discussion on these is available in the next modules. It should be emphasized that experiments may appear similar to another group of research designs called observational designs. However, there should be a clear distinction between experimental and observational studies. Observational studies are those wherein units are observed in the normal environment that they exist. There is no random allocation of experimental units to treatment levels or vice-versa. On the other hand, experimental studies are those wherein units are observed in an environment that is highly controlled by the researcher. There is random allocation of experimental units to treatment groups or random assignment of treatment levels to experimental units. Institute of Statistics Page | 2 University of the Philippines Los Baños MODULE 1 [STAT 162: Experimental Designs] Basic Concepts and Introduction to jamovi© Clearly, the difference lies in the presence of randomizing the treatment levels to experimental units in an experimental study and its absence in an observational study. Take note that random allocation or random assignment is different from random selection or random sampling which you might have learned in your basic statistics courses. To be more specific, whenever we say “experiment”, we actually mean “randomized experiment” especially in the context of this course. As mentioned, units are assigned to exposure or treatment levels or groups and this assignment is randomized, manipulated, or controlled by chance. The main goal is to see the effect of the treatment or exposure to the response while controlling for other factors. These extraneous factors are those that affect the response of the units but are not of interest to the current research. Thus, their effects are controlled. Since extraneous factors are ideally controlled in a randomized experiment, the experimental design presents the strongest evidence to causation among all research designs especially when the experiment is carefully planned and performed. This slide shows the schematic diagram for a randomized experiment. This illustration assumes four (4) treatment groups or levels A, B, C, and D. Notice that the levels should be randomly assigned to the experimental units. After applying the treatment level or after the units received their treatment, the units are observed for their response. The observation usually takes time and responses of experimental units will be observed and recorded in the future. Since there are 4 treatment levels, there will also be 4 groups of responses. Ultimately, these 4 groups of responses indicating the effect of a particular treatment level will be compared with each other. Institute of Statistics Page | 3 University of the Philippines Los Baños MODULE 1 [STAT 162: Experimental Designs] Basic Concepts and Introduction to jamovi© Aside from providing evidence of causation, randomized experiment has the mechanism to layout the assignment of treatment levels to experimental units which allow researchers to set-up direct comparison between treatments or treatment levels of interest. Since there is randomization, the bias in the comparison of treatments levels is minimized. Another important aspect of randomized experiment is replication which minimizes the error in the comparison. Researchers can be in full control of the study since it allows researchers to manipulate treatment levels and control extraneous factors. Of course, these are easier said than done so careful planning of the experiment must be prioritized. The plan must be carefully followed and implemented. We will discuss more of these concepts when we go to principles of experimental designs within this module. Now, let us define some terms to guide you better for the next topics. Treatment is a set of experimental procedures or conditions whose effects are to be measured and compared. Usually, the mnemonic question for this concept is “Which are being compared?”. Treatment is the independent variable which is manipulated or controlled so that its effect can be measured or compared. Interchangeably, we can also refer to this term as factor or in some discipline, this is also called exposure. Treatment levels are the pre-set levels of a quantitative factor or categories of a qualitative factor under study. The mnemonic question for this concept is “Which are randomly assigned to experimental units?”. Yes, treatment levels are randomly assigned to experimental units and are applied during the actual conduct of the experiment. Institute of Statistics Page | 4 University of the Philippines Los Baños MODULE 1 [STAT 162: Experimental Designs] Basic Concepts and Introduction to jamovi© Response variable is the observed characteristic from which the effect of a treatment will be measured and eventually compared. This may also be referred to as the dependent variable. In most disciplines, this is also called outcome since we observe the response after applying a treatment to an experimental unit. The mnemonic question for response is “What is observed, collected and recorded for analysis?”. The unit that receives the treatment is what we call the experimental unit (eu). EUs are units or groups of units to which treatment levels are applied in a randomized fashion. Sampling units are those from which the response variable is observed or measured. In some cases, experimental units are also the sampling units. To avoid confusion, always remember that EU receives the treatment while SU gives the response. In some literature, sampling units are also called subsamples. However, we must not confuse this to “designs with subsampling” which will be introduced to us in the next modules. As a heads up, designs with subsampling are those experimental designs in which (1) each EU or SU is observed more than once, or (2) there are more than one SUs per EU. To better understand the terms defined in the previous slides, let us have the following examples. Institute of Statistics Page | 5 University of the Philippines Los Baños MODULE 1 [STAT 162: Experimental Designs] Basic Concepts and Introduction to jamovi© In this example, the experiment used three brands of iron supplements, and these were administered to mice such that a brand was randomly assigned to 3 out of 9 mice available. After 4 weeks, blood samples were taken twice from each mouse and were independently analyzed for RBC count. For instance, RBC counts from two blood samples of Mouse 1 given with Brand A yielded 12 and 8 as responses while Mouse 9 given with Brand C yielded 50 and 34 as responses. Based on raw data, we can see that RBC counts under Brand C were way higher than RBC counts in Brands A and B. In the example, iron supplement was the treatment, and its 3 brands were the treatment levels. The brands of the iron supplement were the ones compared and the experiment wanted to measure their effects on RBC count. Moreover, the response variable was the RBC count from the Institute of Statistics Page | 6 University of the Philippines Los Baños MODULE 1 [STAT 162: Experimental Designs] Basic Concepts and Introduction to jamovi© collected blood sample from each mouse. Experimental units were the mice since each mouse received a particular iron supplement while sampling units were the blood samples collected from each mouse. If there was only one blood sample collected from each mouse, then experimental units are similar to the sampling units since each mouse, or interchangeably each blood sample, will yield one RBC count. Let us have another example. In this example, the experiment used three amounts of zinc in fertilizers, and these were administered to 15 equally sized plots each with 3 seedlings. Each amount of zinc was randomly assigned to 5 plots. After a month, the total dry weight (g) of leaves collected in each plot was observed. Notice that it was not the dry weight of seedling leaves that was measured but the TOTAL dry weight of leaves in a plot. In the example, the amount of zinc was the treatment and its 3 levels (10, 15, and 20) were the treatment levels. The amounts of zinc were the ones compared and the experiment wanted to measure their effects to dry weight. Moreover, the response variable was the total dry weight of the leaves from each plot. Experimental units were the plots since each plot received a particular amount of zinc. In this example, sampling units were the plots too. Seedlings were not sampling units (or subsamples) since the experiment was not particular to individual dry weight of leaves from each seedling. It was the total dry weight of leaves collected from the plot which concerned the researcher. Institute of Statistics Page | 7 University of the Philippines Los Baños MODULE 1 [STAT 162: Experimental Designs] Basic Concepts and Introduction to jamovi© TEST YOURSELF #1! Note: Answers can be found on Page 25. Data collected from randomized experiments are usually analyzed by accounting for statistical errors which may be a combination of random and systematic errors. Analysis of Variance or ANOVA, which will be the main statistical method that we will use in this course, partition the sources of variation into variation caused by the treatments applied and the variation caused by uncontrolled unavoidable factors. The latter can be measured by estimating the experimental and subsampling errors. These errors are caused by the (1) inherent variability of the experimental materials used, (2) procedural errors committed during experimentation and actual measurement, and (3) combined effects of all extraneous uncontrolled factors. Experimental error is the variation in the observed responses from experimental units treated alike. In an experimental setting where we try to eliminate the effects of factors other than the main treatment, EUs treated alike should ideally yield the same response. However, it does not happen almost all the time. EUs treated alike fail to yield the same response and these differences in their response is measured by the experimental error. Institute of Statistics Page | 8 University of the Philippines Los Baños MODULE 1 [STAT 162: Experimental Designs] Basic Concepts and Introduction to jamovi© If there are sampling units or subsamples, failure of subsamples within an EU to yield the same response gives the sampling or subsampling error. TOPIC 2: Principles of Experimental Design Let us now formally define what experimental design is. Experimental design involves the procedure of assigning treatments to the experimental units. Its structure concerns planning the experiment to obtain the maximum amount of information from available resources. Moreover, it involves the practice of deliberately changing or manipulating treatments in a specified manner to evaluate the effects of these changes in the response variable. When we say effect, we refer to the average measured change, increase or decrease, in a response variable associated with changes in a treatment. Institute of Statistics Page | 9 University of the Philippines Los Baños MODULE 1 [STAT 162: Experimental Designs] Basic Concepts and Introduction to jamovi© An experimental design consists of two structures. First is the design structure which refers to grouping of the experimental units into homogenous groups or blocks. The design structure is focused on laying-out the experiment and randomizing the treatment levels to EUs. Second is the treatment structure which refers to the set of treatments, treatment combinations or populations that the researcher has selected to study and compare. Modules 2 to 6 and 10 deal with design structures while Module 7 deals with treatment structures. Module 8 is a combination of both structures. These examples briefly illustrate one-way and factorial treatment structures. Simply put, one-way treatment structure means that there is only one factor or treatment in the experiment and the levels of this single treatment are being compared. On the other hand, factorial treatment structure means that there are more than one factor, and the focus now is to compare combinations of their levels. Let us now discuss details for consideration in creating an experimental design. To obtain the maximum amount of information while implementing the design, one should consider the concepts of accuracy and precision. Institute of Statistics Page | 10 University of the Philippines Los Baños MODULE 1 [STAT 162: Experimental Designs] Basic Concepts and Introduction to jamovi© Precision is the ability of a measurement to be consistently reproduced. This is estimated by the variance. In essence, precision is about the closeness of the observations with one another. This can be achieved by increasing the number of samples or replicates, by grouping the experimental units and proper selection of treatments and treatment levels. On the other hand, accuracy is the closeness of the observed values to the true value and is estimated by the bias. Accurate measurements can be achieved by refining the experimental technique and procedures, and by proper selection of treatment and treatment levels. To better understand these concepts, this illustration shows us different scenarios. Institute of Statistics Page | 11 University of the Philippines Los Baños MODULE 1 [STAT 162: Experimental Designs] Basic Concepts and Introduction to jamovi© For illustration, assume that the true value is the center of the circle. In the upper left circle, the measurements are near the center. In fact, we can probably say that their average is close to the true value. However, they are not close with each other. So, this scenario shows accurate but not precise measurements. The circle in the lower left shows the reverse. Here, the measurements are very close with one another but all of them are far from the center. So, this scenario shows precise but not accurate measurements. The circle in the lower right is the worst scenario where measurements are not only not precise but also not accurate. What we want to achieve is the scenario shown in the upper right circle. The measurements are very close with each other, and they are clustered in the center which means that the measurements are both precise and accurate. The model which will be the basis for analyzing and interpreting the results is another important aspect of experimental designs. These models depend on the selection of treatment and treatment levels. One of these models is the Fixed Model. This model is appropriate when all factors under test are fixed factors. A factor is fixed when its levels are selected on purpose. In the example, the brand of racket is fixed since the researcher did not randomly choose brands of racket to be used as treatment levels. Automatically, the brands used were dictated by which brands are commonly used or purchased. The other model is called the Random Model which is appropriate when all factors under test are random factors. A factor is random when the levels of the factor tested are a random sample Institute of Statistics Page | 12 University of the Philippines Los Baños MODULE 1 [STAT 162: Experimental Designs] Basic Concepts and Introduction to jamovi© from a population of feasible levels. In the example, the varieties as treatment are random since the levels were randomly selected from a pool of experiment-feasible IRRI-bred varieties. Now let us focus more on the nature of experimental designs. A good experimental design must avoid systematic error. Systematic errors are reproducible inaccuracies that steadily appear in the experiment. Examples of these are wrong measurement procedure or wrong calibration of measuring instruments. The problem in systematic errors is that they cannot be analyzed. The observations with such errors will reach the analysis and interpretation stages without being detected so care must be included during planning. Another characteristic of a good experimental design is that it should allow estimation of random errors that we discussed in the previous topic. Random error is a statistical variability due to precision limitations. While estimating the random error, the experimental design should at the same time minimize such errors. Though we want to minimize random error, it should be noted that random error permits statistical inference. In fact, ANOVA uses the estimate of error to arrive at decisions. We will understand this more in the next modules. Let us take a look at this example. This experiment has a major flaw. It presumes that any difference between the yields of the two plots is caused by the varieties and nothing else which is certainly not true. If we incorporate the concept of experimental error here, we will be able to deduce that even if we plant the same variety on both plots, the yield will still differ since other factors such as soil fertility, moisture, and damage by insects and diseases also affects rice yield. The next slides will tell us how to improve this experiment. Institute of Statistics Page | 13 University of the Philippines Los Baños MODULE 1 [STAT 162: Experimental Designs] Basic Concepts and Introduction to jamovi© It was mentioned previously that a good experimental design should be able to estimate the experimental error since this will be the primary basis for deciding whether an observed difference is real or due to chance. If it is real, that is the time we reject the null hypothesis (Ho). If it is due to chance, we fail to reject Ho. The first principle of an experimental design called replication allows an experimenter to estimate the experimental error by replicating the treatment level or assigning it to more than one EU. Since a level is assigned to more than one EU, there will be more than one response for it allowing for the estimation to occur. Aside from providing an estimate of the experimental error, replication also increases the precision of the estimates of the parameter. Remember from your basic statistics courses that variance and standard error of estimates are a function of the number of observations usually denoted by n. As the number of observations increases, standard error decreases. Hence, precision is increased. In addition, replication also increases the scope of the experiment. For instance, we can replicate all treatment levels to another group or block of EUs which is the idea in an experimental design called Randomized Complete Block Design or RCBD. We can also replicate the treatment levels across time which is the idea of Repeated Measures ANOVA. In these scenarios, the experiment is extended to another group of EUs or to another time period with a different set of characteristics. We will discuss RCBD more in Module 5 and Repeated Measures in Module 10. There was no replication in the rice variety example above. In fact, rice varieties were applied to one plot each only. Ideally, you want to have many replicates. However, we should also consider cost and variability of resources and limited time for the experimenter to conduct the experiment. Institute of Statistics Page | 14 University of the Philippines Los Baños MODULE 1 [STAT 162: Experimental Designs] Basic Concepts and Introduction to jamovi© The second principle of experimental design is randomization. It is a process that ensures treatments will have equal chance of being assigned to an experimental unit. Randomization allows validity of the experimental error by eliminating systematic bias in assigning the treatments to the EUs especially when they are adjacent in space or in time. More importantly, randomization automatically satisfies the assumption of ANOVA that the errors should be independent. Assumptions of ANOVA will be discussed more in Module 2. Let us use the illustration above as a dummy field and consider the previous rice variety experiment as example. Here, we already replicated the rice varieties three times each. That is, each rice variety was planted into three plots. However, instead of randomly allocating or assigning the rice varieties across the six available plots, the experimenter planted them alternately not knowing that there is a fertility gradient in the experiment area going from left to right. In this case, the yellow variety will outgrow the blue variety and will yield more grain. If only the experimenter had randomized the rice varieties across plots, the impending effect of fertility could have been minimized if not eliminated. Last principle is local control or error control. Good experiments incorporate all possible means of minimizing the random error (experimental error and sampling error). With good local control, the design becomes efficient by making the tests more sensitive to detect differences among effects of treatment levels and powerful to reject a false hypothesis. Institute of Statistics Page | 15 University of the Philippines Los Baños MODULE 1 [STAT 162: Experimental Designs] Basic Concepts and Introduction to jamovi© Common techniques to improve local control of the experiment include the use of the most appropriate experimental design. Correct design leads to a correct model which will maximize the information offered by the dataset. Another technique is the use of proper shape and size of EU. There are designs called Split Designs that allow the experimenter to incorporate varying sizes of EUs which capture the inherent characteristic of the experimental area. More information on Split Designs can be learned in Module 8. Another technique is the use of a concomitant or confounding variable during data analysis. Concomitant variables are factors that are sometimes not part of the plan but existed during implementation of the experiment. These factors cannot be controlled and are not a focus of the experiment but there is a need to account for their effect since they influence the response. One method to analyze such an experiment is Analysis of Covariance or ANACOVA which will be discussed more in Module 9. Lastly, local control can also be improved by incorporating grouping, blocking, or balancing in the design. Grouping which is also called stratification in other disciplines uses homogenous EUs into groups and compares the treatments in each group. In this case, there will be separate conclusions for each group since each group has a separate unique set of EUs. There are techniques where conclusions and/or estimates from these groups are merged or combined. More of this in Module 10. Institute of Statistics Page | 16 University of the Philippines Los Baños MODULE 1 [STAT 162: Experimental Designs] Basic Concepts and Introduction to jamovi© Blocking, on the other hand, is the grouping of EUs into blocks such that the units within a block are relatively homogenous. As opposed to grouping, blocking factors are incorporated in the design and consequently in the model of the current experiment. More of this in Modules 5 and 6. Lastly, balancing is the assignment of the treatments to the EU to achieve a balanced configuration. When one treatment level is assigned to 3 EUs, the other levels should also be allocated to the same number of EUs. TEST YOURSELF #2! Note: Answers can be found on Page 26. TOPIC 3: Introduction to and Navigation in jamovi© For this offering of STAT 162, we will be using jamovi©, a free and open statistical software which is extremely user-friendly. We would like to emphasize from this point that you are not required to have jamovi© especially when you do not have the means to download and install the software. All modules will be utilizing the jamovi© system to aid you in the analysis. However, codes and outputs will be always given and ready for use. Also, module assessments will not require you to perform analysis in jamovi©. jamovi© is a desktop application so you will need to have Windows, Mac or Linux operating on your desktop. A beneficial feature of jamovi© is its close relation to the open-source programming language R which is a powerful statistical and data analysis software. To start, go to jamovi.org and download the latest version fit to your operating system. Samples in this module were run in Windows. Institute of Statistics Page | 17 University of the Philippines Los Baños MODULE 1 [STAT 162: Experimental Designs] Basic Concepts and Introduction to jamovi© When you open jamovi©, you will see a blank spreadsheet of rows and columns to the left. This is the Data Window. There is no data yet but jamovi© creates three variables by default. Like in common data structures, columns are variables while rows are observations or cases. On to the right is the Output Window where outputs or results of analyses will be printed and ready for copying and exporting. Located in the upper left area of the interface are the command tabs under Analyses menu. Click Exploration then choose Descriptives to bring up the Descriptives dialogue. Selecting variables and placing them in the Variables window will prompt jamovi© to output descriptive statistics for that variable. Institute of Statistics Page | 18 University of the Philippines Los Baños MODULE 1 [STAT 162: Experimental Designs] Basic Concepts and Introduction to jamovi© You may also choose various plots and add more summary statistics to the output by clicking Statistics and Plots menu at the lower left area. To go back to the data window, click the arrow at the upper right corner of the Descriptives dialogue. The other command tabs provide more suites of analyses which you can explore on your own. You will be introduced to these tabs more in the next modules. Since jamovi© was built on R, the system can be extended to more suites of analytical tools. Click the cross icon located in the upper right corner to open the jamovi© library for more collection of modules. For the meantime, it is important you install Rj Editor library. This will enable us to run any R code in jamovi©. After installing Rj Editor, an R icon will be added as another tab. Click on it to bring up the Rj Editor Window. We will get back to this shortly. Let us explore first other menus in jamovi©. Institute of Statistics Page | 19 University of the Philippines Los Baños MODULE 1 [STAT 162: Experimental Designs] Basic Concepts and Introduction to jamovi© Next to the Analyses menu is the Data Menu which opens a new set of command tabs. This menu allows us to import data into jamovi© and set up the characteristics of each variable through the Setup command tab. Clicking Setup opens a dropdown dialogue which allows you to change and validate variable names, descriptions and variable types which are very important before starting the analysis. The rest of the command tabs under data are simply straightforward and resemble some functions in Microsoft Excel. These will be left for you to explore. The last two menus that you need to be familiar with are the three-line menu in the far left and the three-dot menu in the far right. The three-line menu gives us the basic commands like open a file, save and export while the three-dot menu gives us the preference options. Institute of Statistics Page | 20 University of the Philippines Los Baños MODULE 1 [STAT 162: Experimental Designs] Basic Concepts and Introduction to jamovi© An important option in the three-dot menu is the Syntax mode which allows us to transform the output with an R command. It will add or attach the R code that resulted to the output. This is greatly useful especially if you want to export the outputs together with the commands and reproduce them at another session for a different data set. Also, it will allow beneficial interchange between jamovi© and R if you are using R too. Now, let us open and work on Tooth Growth dataset from the three-line menu. Opening a data set will start a new jamovi© session. The data contains three variables. From the Setup command tab, we can see that the first variable named len is a continuous variable, supp Institute of Statistics Page | 21 University of the Philippines Los Baños MODULE 1 [STAT 162: Experimental Designs] Basic Concepts and Introduction to jamovi© is nominal text variable with levels OJ and VC while dose is a nominal integer variable with levels 500, 1000, and 2000. Take note of the icons particular to a certain variable or data type. Ruler icon is for continuous or numerical variables, bar chart is for ordinal variables, blue tag is for ID variables while the three circles that look like a Venn diagram are for nominal variables. Nominal variables can either be text or integer. Changing the nominal variable to text will change one of the circles of the Venn diagram to “a” to denote the variable as a nominal text variable. Always make sure to assign the correct type for each variable in your data sets. TEST YOURSELF #3! Note: Answers can be found on Page 26. Now let us discuss how to enter data to jamovi©. Institute of Statistics Page | 22 University of the Philippines Los Baños MODULE 1 [STAT 162: Experimental Designs] Basic Concepts and Introduction to jamovi© Though we can manually encode data to the built-in spreadsheet, jamovi© allows us to easily import data sets by navigating to the three-line menu then select Import. Supported data formats can be seen from this menu. TEST YOURSELF #4! Note: Iris.csv is included as a separate file in this module. Answers can be found on Page 26. After importing a dataset, it is very important that you check the variable types of the imported data before using them for analyses. Most of the time, jamovi© can automatically recognize the variable type after import but checking their validity can free you from errors during analysis and most especially during interpretation of results. Institute of Statistics Page | 23 University of the Philippines Los Baños MODULE 1 [STAT 162: Experimental Designs] Basic Concepts and Introduction to jamovi© Now, let us go back to Rj Editor. As mentioned previously, jamovi© is based on R. Implementing R commands to jamovi © will appear as if you are working in R environment through the help of Rj Editor. Though there are built-in commands in the point-and-click menu of jamovi©, the Rj Editor library extends the capabilities of jamovi© by accommodating more functions. After installing the module, click the R icon in the commands tab to show the Rj Editor Window. Open Anderson's Iris Data then type summary(data) in the editor. Put the cursor in the line or highlight the code then press Ctrl + Shift + Enter to run the program. You may also click the green arrow located at the upper right corner of the editor. Institute of Statistics Page | 24 University of the Philippines Los Baños MODULE 1 [STAT 162: Experimental Designs] Basic Concepts and Introduction to jamovi© After running the program, the output window should show you this output. summary command outputs the minimum, maximum, mean, median, first quartile, and third quartile for numerical data while frequencies for categorical data. Whenever you want to edit your R code (or commonly called script), click the output in the Output Window to show the editor for that output again. Do not click the R icon in the commands tab again as it will create a new blank script. Also, notice that the Iris data is simply called “data” in the Rj Editor of jamovi©. This should not be a problem since every data set requires a separate jamovi © session. Calling “data” will include the whole dataset or all variables. If you want to produce outputs for a specific variable or if you want to point out a subset of variables, use $ or column number indicators as in this example. Both codes will produce descriptive statistics for variable Sepal Length but with different orientations. Note that “1” in summary(data) refers to the first column. Institute of Statistics Page | 25 University of the Philippines Los Baños MODULE 1 [STAT 162: Experimental Designs] Basic Concepts and Introduction to jamovi© Last detail that we need to mention in this module is on saving your jamovi© files. You may save your outputs to an external file by right clicking the output and exporting it as html or pdf. You may also turn on the syntax mode in the three-dot menu to attach the R code for each output. If you want to save the whole jamovi© session including the outputs, data and R codes, click the three-line menu and choose Save or Save As. The jamov © session will be saved as.omv in your chosen directory. You may also save the current progress in the dataset by clicking Export and saving it to your desired data filetype (e.g..csv). Saving the session as.omv will save the data, outputs and codes in the Rj Editor that were already executed. All codes that were not run previously will not be saved in the.omv file. Make sure to run the code first before saving the session so you can retrieve the codes again by double-clicking the output. Copying and pasting the code to another file (.txt or.docx) can also help you save the current progress of your codes. GOOD JOB! YOU FINISHED THE FIRST MODULE! 1/10 COMPLETED Institute of Statistics Page | 26 University of the Philippines Los Baños MODULE 1 [STAT 162: Experimental Designs] Basic Concepts and Introduction to jamovi© REFERENCES 1. Datalabcc at https://www.youtube.com/user/datalabcc 2. Gomez, K. A., & Gomez, A. A. (1984). Statistical procedures for agricultural research. John Wiley & Sons. 3. Hinkelmann, K., & Kempthorne, O. (1994). Design and analysis of experiments (Vol. 1). New York: Wiley. 4. Montgomery, D. C. (2014). Textbook: Design and analysis of experiments (No. 575-672). Pg. 5. The jamovi project (2020). jamovi. (Version 1.2) [Computer Software]. Retrieved from https://www.jamovi.og ACKNOWLEDGEMENT This module was written and prepared by: PROF. MARK JAY P. DATING PROF. MARIE JOY F. LOPEZ-RELENTE MR. JAN ANDREW P. REFORSADO TEST YOURSELF ANSWERS 1. Whiteness Rating Experiment Treatment: Brands of detergent Levels: Brand A, Brand B, Brand C, Brand D or 4 Brands Response Variable: Whiteness rating Experimental Unit: Shirts Sampling Unit: Shirts 2. Drying Seeds Experiment Replication happened when each method was assigned to three glass dishes, dishes serving as replicates while seeds serving as subsamples. Randomization happened when each method was randomly assigned to dishes. There is no blocking. 3. jamovi© 1 a. Open Tooth Growth data via three-line menu. b. Click Exploration → Descriptives c. Place len in the Variables window and dose to Split by window. d. Check the Q-Q plot under Plots menu. e. Check N, Mean, Sum, Standard Deviation and Variance in Statistics Menu. Uncheck the other information checked by default. 4. jamovi© 2 a. Open a new jamovi© session. This is important since importing a new dataset might append the new data to the current data in the session. b. Import iris.csv via three-line menu. Institute of Statistics Page | 27 University of the Philippines Los Baños MODULE 1 [STAT 162: Experimental Designs] Basic Concepts and Introduction to jamovi© c. Place Sepal Length in the Variables window and Species to Split by window. d. Check the Box plot and Data Jittered under Plots menu. e. Check N, Mean, and Standard Deviation in Statistics Menu. Uncheck the other information checked by default. NOTICE Note that the course pack provided to you in any form is intended only for your use in connection with the course that you are enrolled in. This covers all materials such as handouts, lecture slides, and other materials uploaded in Canvas, sent to you through flash drive, or sent to you as hardcopies. The course pack is not for distribution or sale. Permission should be obtained from your instructor for any use other than for what it is intended. Institute of Statistics Page | 28 University of the Philippines Los Baños