Turner, J. Using statistics in small-scale language education research: Focus on non-parametric data Chapter Three: Practice Problem Key Practicing the Concepts—Variable Roles For each of the brief problems that follow, identify the variables and indicate which have been assigned the role of independent variable and which have been assigned the role of dependent variable. 1. A researcher plans to investigate whether a teacher’s amount of experience has an impact on the students’ success in learning a foreign language. Independent variable(s): teacher’s amount of experience Dependent variables(s): students’ success in learning 2. Jean (that’s me) plans to do a research problem to investigate whether students who are enrolled in morning classes have the same degree of success in learning as students who are enrolled in late afternoon classes. Independent variable(s): time of class meeting Dependent variables(s): students’ success in learning 3. I wonder whether learners whose first language is Spanish or French find learning Japanese more challenging than learners whose first language is Thai or Vietnamese. Independent variable(s): learner first language (Though really, the variable is a characteristic of the first language—whether the learner’s first language is tonal or not (+/- tonality). Spanish and French are not tonal languages; Thai and Vietnamese are tonal languages. Japanese, by the way, is not a tonal language.) Dependent variables(s): degree of challenge in learning Japanese 4. Do you think that extroverts are more likely to develop a high degree of oral fluency than introverts? Independent variable(s): degree of extroversion or if the variable is nominal, introversion/extroversion Dependent variables(s): oral fluency 5. Can you think of any moderator variables that might play a role in Studies 1 – 4? In the first study, I might include teacher’s type of experience as a moderator variable. In Study 2, I think I’d include as a moderator variable how learners characterize themselves—so the variable could be labeled as morning people/night owl. In Study 3, I might include number of languages spoken as a moderator variable. The sex of the learners might be an important consideration in Study 4 so I believe I’d include it as a moderator variable. 1 Turner, J. Using statistics in small-scale language education research: Focus on non-parametric data 6. Can you think of any control variables that might play a role in Studies 1 – 4? For Study 1, I think it would be a good idea to focus on students in just one type of school—I think I’d first study learners in high school, so the control variable could be labeled type of school, or student’s level of education. In Study 2, I think I’d introduce student’s program of study as a control variable, and include as participants only those who are in the applied linguistics program. For Study 3, I’d only include learners who are between the ages of 20 – 25, so age would be a control variable. In Study 4 I’d have two control variables. This first is language of study—I’d only include people who are learning Spanish. The second control variable that I’d introduce is age. As in Study 3, I’d only include participants who are between the ages of 20 and 25. 7. Write two brief research problems of your own. No answer key for this one! 8a. What is the independent variable and how many levels does it have? type of curriculum; it has two levels (new and existing) 8b. What is the dependent variable? reading ability 8c. What type of scale do you think “measures” the dependent variable? I think the researcher defined reading ability as an interval scale variable, because it appears that the variable has mathematical properties. (The researcher calculated the “average” as part of the analysis of the data). 8d. Is there an explicit control variable? The researcher doesn’t say that there is a control variable, so there isn’t an explicit one. However, I believe the researcher had access to students in grades K – 6 and chose to include only first-graders as participants in this study, so perhaps learner’s grade level is a control variable. 8e. Is there an explicit moderator variable? The researcher doesn’t say that she compared the outcomes for the native speakers and nonnative speakers of English assigned to the two different types of curriculum, but I think she should. If she does, learner’s first language, would be a moderator variable. 8f. Identify two major threats to the internal validity of the study and what might be done to attempt to reduce each. Here are a few that might threaten the soundness of the study. 1) teachers’ familiarity with the curriculum Teachers who participate in the study should be trained to use the curriculum that they teach—whether it is the new curriculum of the existing one. 2) teachers’ attitude toward the curriculum What teachers think of the curriculum they teach could have an impact on their teaching—I think I’d exclude teachers from participating in the study if they reported having a negative attitude toward the curriculum they use. 2 Turner, J. Using statistics in small-scale language education research: Focus on non-parametric data 8g. Identify two major threats to the external validity of the study and what might be done to attempt to reduce each. 1) The fact that only first-graders participated in the study threatens the external validity of the study. The only way this threat could be reduced is by including students from other grade levels as participants; however, the researcher may not have sufficient resources to do this. 2) If the researcher didn’t make learner first language a moderator variable, I think one would have to be very careful about generalizing the findings of the study to both native speakers of English and non-native speakers of English. To reduce this threat, I recommend that this factor, learner first language (with two levels, native speaker of English and nonnative speaker of English), be studied as a moderator variable. 8h. Into which of the four categories of research design does this study best fit? I used the three ordered questions on page 80 in the textbook as a guide. 1) Is there an experimental treatment? Yes, there is an innovation; the students who use the new curriculum represent an experimental condition. The students who use the existing curriculum are in the nonexperimental condition. 2) Are there legitimate comparison groups? If the number of first-graders in the district is sufficiently large to be considered a population, it appears that the random selection of the 60 participants from the population would result in their being comparable to the population of first-graders, thus there are legitimate comparison groups. 3) Are the legitimate comparison groups formed through random selection and random assignment? If one believes that the district is sufficiently large to be considered a population, random selection of the participants indicates that the design of the study is true experimental. Note: If the school district is so small that the first-graders cannot be considered a population, the design must be considered pre-experimental. 9a. What is the independent variable and how many levels does it have? travel to Spanish-speaking country (immersion in Spanish-speaking country); it has two levels, yes and no. 9b. What is the dependent variable? Spanish ability (or Spanish proficiency) 9c. What type of scale do you think “measures” the dependent variable? I think it’s an interval scale variable, though it could be an ordinal scale that has mathematical properties (see pp. 18-19 in the textbook). 3 Turner, J. Using statistics in small-scale language education research: Focus on non-parametric data 9d. Is there an explicit control variable? There doesn’t seem to be any explicit control variables, but there are several implicit ones, including teacher (she only included her students!) 9e. Is there an explicit moderator variable? I think she treated sex as a moderator variable. 9f. Identify two major threats to the internal validity of the study and what might be done to attempt to reduce each. There are a lot of serious threats to the soundness of this study! Here are a few: 1) duration of time the learner spent in a Spanish-speaking country I think the nominal independent variable, travel to Spanish-speaking country, which has only two levels, is inadequate. It should have been defined to include several more categories reflecting length of time spent in a Spanish-speaking country. 2) purpose of travel to a Spanish-speaking country The researcher doesn’t take into account why the learners travelled to a Spanish-speaking country or what they did while they were there. The researcher should create a moderator variable to categorize the participants according to their purpose for the travel. 9g. Identify two major threats to the external validity of the study and what might be done to attempt to reduce each. 1) a very limited sampling of learners of Spanish All of the participants in this study were the former students of one teacher. I assume there were other teachers of Spanish at the school—their students should have been included too! 2) There are so many threats to the soundness of this study that it simply doesn’t have any external validity! I suggest the teacher create a new research design and start over! 9h. Into which of the four categories of research design does this study best fit? To answer this question, I used the ordered three questions on page 80 as a guide, but needed to answer only the first one to identify the research design. 1) Is there an experimental treatment? No, there is no innovation or change. The teacher simply reviewed the data that she had; there is no experimental treatment so the design is ex post facto. 4