Educational Research On-Line
Practice Exercise for Inferential Statistics
Determine which statistical test(s) should or could be used for each example. Possible tests include: Correlation, t-test, ANOVA, ANCOVA, or chi-square.
1. A researcher wishes to determine whether cooperative learning, computer assisted learning, or self-paced learning provides the most academic achievement for students of educational research.
The IV, type of instruction, is qualitative, and the DV, achievement, which most likely be measured by scores on a test, so it is quantitative. Since the IV has more than two categories (cooperative learning, computer assisted learning, and self-paced), and since there is no clear covariate to control with, ANOVA will be used to analyze the results of this study.
2. A researcher decides to test whether cooperative group testing (CGT) makes a increase in learning among college students. The researcher uses his two classes of educational research students to test whether CGT increases learning. Class A is the control group and they receive no special treatment. Class B will use CGT.
t-test or ANOVA:
The IV, type of testing allowed, is qualitative, and the DV is quantitative (achievement test scores will be used to measure learning). Since the IV has only two categories, either a t-test or ANOVA could be used for this study. Recall that t-test and ANOVA are appropriate for analysis of two groups (two means). Mathematically, both the t-test and ANOVA give identical results.
3. A researcher wishes to know whether dynamic or static assessment will increase teachers expectations for mentally handicapped (intellectually challenged) children. The researcher randomly assigns the teachers to two groups. The first group of teachers views the child via a dynamic assessment procedure. Following this viewing, both groups of teachers view the child via a static assessment. After this viewing, both sets of teachers complete an instrument designed to measure the teachers expectations of the childs ability to learn.
t-test or ANOVA:
The IV, type of assessment, is qualitative, and the DV is most likely quantitative (expectations of a child's ability is probably measured with categories that can be ranked from high to low expectations, so it will be quantitative). Since the IV has only two categories, dynamic plus static assessment, or static assessment only, either a t-test or ANOVA could be used for this study. Recall that t-test and ANOVA are appropriate for analysis of two groups (two means). Mathematically, both the t-test and ANOVA give identical results.
4. A researcher wants to determine whether elaborative summaries or writing-to-learn provides the most academic achievement for students of educational research. The researcher uses his two classes to perform the experiment. The researcher believes that intelligence could be confounded with achievement, so the researcher uses students intelligence scores as a control variable.
The IV, type of writing activity, is qualitative, and the DV, achievement, will be quantitative. The IV has only two categories, so a t-test or ANOVA appears to be appropriate, but the researcher wishes to control for a possible confounding variable, intelligence. ANCOVA, not t-test or ANOVA, enables one to statistically control for confounding effects by adjusting the DV means to take into account any differences among the groups in intellectual ability.
5. The state department of education funds a teenage pregnancy prevention program in a local middle school. The effectiveness of the program is evaluated at the end of the third year of the programs existence. The question of interest is whether pregnancy (pregnant vs. not pregnant) is less prevalent among girls who attended the program as compared with similar girls who did not attend the program.
The IV, attendance in the program (with 2 categories: attend or not attend), is qualitative, and the DV, pregnancy (with 2 categories: pregnant or not pregnant), is qualitative. Chi-square is appropriate for determining whether two qualitative variables are related.
6. A researcher wishes to determine whether need for dominance and need for achievement are related to academic achievement in college classrooms. Academic achievement will be measured using a test administered at the end of the semester and scores will range from 0 to 100 with 100 indicating the highest level of achievement. Both need for dominance and need for achievement will be measured in such a way that scores can be sorted from high to low needs in both areas.
All three variables (need for dominance, need for achievement, and academic achievement) are measured in such a way that enables them to be quantitative. To learn whether these variables are related, Pearson's r may be used to determine the degree of relationship among the variables.
7. A researcher wishes to determine whether cooperative group testing, computer assisted learning, or traditional lecture provides the most academic achievement for educational research students. The researcher realizes, however, that students do not come into educational research with equal backgrounds (i.e., understanding of research methods). The researcher administers a pretest on the first night of class and uses scores on the pretest as a control variable in the analysis to control for initial understanding of research methods.
The IV, type of instruction with three categories (cooperative group testing, computer assisted learning, and traditional lecture), is qualitative, and the DV, achievement, is quantitative. Note that the researcher thinks prior experience or knowledge of educational research could confound the outcome of the study, so the researcher wishes to control for prior knowledge using pretest scores as a covariate in the study. ANCOVA enables the researcher to make this statistical control.
8. Does a relationship exist between ones intelligence score and ones level of academic self-efficacy?
Both variables, intelligence and academic self-efficacy, are quantitative. For those not familiar with academic self-efficacy, it refers to one's belief in their ability to perform academic tasks well. This construct is usually measured in such a way that allows one to rank people from high to low self-efficacy--that is, some have more self-efficacy than others. Since both variables are quantitative, Pearson's r may be used to determine the relationship between these two variables.
9. Students in a biology class are asked to participate in an experiment with textbooks. On the first night of class a test was administered to all students to determine their level of knowledge of biology content to be covered in this particular course. One-third of the class is asked to use textbook A, one-third to use textbook B, and one-third textbook C. At the end of the semester students are administered an achievement test to measure their content knowledge.
The IV is book read (A, B, or C), the DV is achievement test, and the covariate are scores from the test administered on the first night of class. ANCOVA allows one to statistically control for any possible group differences among the three groups of students in regard to their initial understanding of biology content covered in this course.
Copyright 2000, Bryan W. Griffin
Last revised on 02 March, 2018 09:09 AM