Causal Relationships Between Variables - Academic Champion Tutors

Table of Contents

Threats to Internal Validity
Threats to External Validity
Tools and Tests
Variables
Statistical Techniques
References

The methodological basis of the proposed study is based on a quasi-experimental approach that allows the study of causal relationships between variables. It should be recalled that the independent variables in this study are the fact of cooperation between a general educator and a paraeducator and the socio-demographic characteristics of the specialists. Manipulations of these variables are expected to affect the dependent factors, which are overall classroom performance and student satisfaction in inclusive academic groups. The quasi-experimental design was not chosen by chance: the motivation is the absence of a control group and random assignment (McCombes, 2021). Conducting a quantitative study in a natural environment for teachers is also the reason behind the choice of this design, as it is assumed that placing an inclusive classroom into an artificial framework would be a confounding variable affecting the purity of the results.

Threats to Internal Validity

Internal validity in causal research conditions the possibility that relationships found in a quasi-experiment cannot be explained by other factors. To put it another way, if the project finds that teacher and paraeducator collaboration does have positive effects on student satisfaction and achievement, then high internal validity implies that no additional factors influence these relationships (Bhandari, 2022a). This makes the results credible and increases academic honesty. Primary conditions for meeting internal validity include simultaneous changes in both types of variables (influencing and responding), as well as the exclusion of any extraneous influencing factors. At first glance, in the proposed quasi-experiment, the confounding variables affecting student achievement and satisfaction in an inclusive classroom are age, learning fatigue, and classroom environment. In other words, it can be assumed that if the academic classroom environment is toxic and the child does not feel safe, then even with a highly effective teacher and paraeducator collaboration, this students satisfaction and achievement may be undervalued (Samara et al., 2021). It is recognized that children with disabilities may be exposed to episodes of school bullying and peer humiliation, which will affect research outcomes (DHHS, 2020). On the other hand, internal validity is also influenced by the degree of fatigue. For example, if the dependent variables are measured simultaneously for students who are heavily academically burdened and who do not pay enough attention to education, the scores may be different for these students.

The described threats to internal validity legitimately affect the reliability of the resulting data, so preventive measures must be taken to minimize undesirable effects. The elimination of these threats is addressed by forming a homogeneous sample of students who have confirmed friendly ties with each other and who do not show significant differences in academic loadings. To put it another way, it is necessary to select students within the class who can confirm friendships with each other mutual confirmation is necessary and who do not differ in their initial academic performance. This could be a group of classmates who have had similar grade point averages over the past few months. It is expected that the sample collected through such manipulations will have high homogeneity, thus eliminating any threats to internal validity.

Threats to External Validity

As opposed to internal validity, external validity should be understood as the researchers belief that the findings can be extrapolated to more general groups. For example, the findings from the proposed study could be generalized to all inclusive classes if the quasi-experiment had high validity. Classic threats to external validity include systematic sampling error, in which the group is unrepresentative, as well as unintended researcher effects on the outcome, the Hawthorne effect, and situation effects (Bhandari, 2022b). Each of these threats inhibits the ability to extrapolate results to more general groups effectively. At first glance, the strategy of collecting a homogeneous sample proposed in the last section is not appropriate for maximizing external validity. In reality, however, this is not the case: friendly ties and performance homogeneity are not threats to representativeness because student interactions during the quasi-experiment are not assumed. Excluding potentially toxic students from the sample made sense to increase student satisfaction in the initial conditions but did not contribute to decreased representativeness. On the other hand, an additional measure of the external validity threats implemented in this study is replication. Replication of the quasi-experiment for different but internally valid groups should counteract the external validity threats described and encourage the possibility of easy transfer of results.

Tools and Tests

As previously reported, a parametric one-way ANOVA test is used to conduct this quasi-experiment. The essence of such a test is to evaluate for statistical differences in means between cohorts. In this sense, ANOVA is an excellent solution to the research problem because the dependent variables are measured continuously and numerically, while the independent variables are distributed nominally. ANOVA analysis using IBM SPSS will be used to determine whether socio-demographic characteristics of teachers and the fact of collaboration with a paraeducator, as well as aspects of that collaboration (duration, frequency), have an effect on student achievement and satisfaction.

As an instrument to measure satisfaction, students will be asked to rate their overall level of satisfaction in the formed group under the guidance of a teacher and a paraeducator on a hundred-point scale. For the inclusive sample, each student provides a unique personal satisfaction value, and when this quasi-experiment is replicated in other schools, the procedure for measuring this indicator does not change. Meanwhile, the second dependent variable is the students academic performance in the classroom. However, because the sample is collected individually for the quasi-experiment, information about past academic performance is used only to generate the sample but not to analyze the results. In other words, in the newly assembled group of students, the achievement will be measured systematically through a test of the discipline being taught (math, reading, science). Test scores are measured at the beginning of the proposed quasi-experiment and at the conclusion of the experiment in order to compare them using the Pre/Post-tests methodology (LS, 2020). In this case, if the teacher and paraeducator collaboration is not practical, students will either not change or decrease their performance over time.

Variables

The proposed study uses variables on two scales at once, namely categorical and quantitative. The categorical variables include information about the socio-demographic characteristics of the teacher as well as information about the teachers collaboration with the paraeducator. In contrast, the quantitative variables in this quasi-experiment are student achievement and satisfaction in the sample collected. Table 1 and Table 2 below provide a more detailed overview of which categories are included in the independent variables of the project.

Independent Variables
Name	Type	Values
Teachers age	Nominal	18 to 24 25 to 34 35 to 44 45 to 54 e55
Teachers ethnicity	Nominal	White/Caucasian Black or African American Asians Latino/Hispanic Others (not listed)
Teachers academic degree	Nominal	Bachelor Masters Postgraduate (PhD)
Experience (years of teaching)	Nominal	0 to 5 6 to 10 11 to 15 16 to 20 e21
Experience with IEPs classes (years)	Nominal	0 to 2 3 to 5 6 to 8 9 to 12 e13

Table 1 Description of socio-demographic categories of independent variables.

Independent Variables
Name	Type	Values
Facts of cooperation with a paraeducator	Nominal	Yes No
Experience of cooperation with the paraeducator (years)	Nominal	0 to 5 6 to 10 11 to 15 16 to 20 e21

Table 2 Description of the independent variables of cooperation.

Manipulation of these independent factors would be expected to affect student achievement and classroom satisfaction. The block of socio-demographic variables is used to identify potentially confounding effects; it is not unlikely that the impact of teacher-paraeducator collaboration may be mediated by additional factors, so their study is of applied value to the quasi-experiment. At the same time, in the formed samples, students under the guidance of a teacher and a paraeducator undergo a one-week study, the result of which is collected information on achievement and satisfaction, which is analyzed through a Pre/Post-test.

Statistical Techniques

In this paper, a one-way ANOVA test conducted in SPSS will be used to evaluate the significance of differences in student achievement and satisfaction in inclusive classrooms with paraeducators. However, performing this test requires adherence to a number of statistical assumptions, without which the reliability of the analysis is questionable (LS, 2021). The first of these assumptions is the condition of normality of the distribution. For this purpose, the continuous variables (achievement and satisfaction) will be evaluated using the Kolmogorov-Smirnov and Shapiro-Wilk tests as well as Q-Q plots. These techniques will determine the normality of the distribution of the variables and their ability to satisfy the assumptions for the ANOVA test. The second condition is the homogeneity of variance between groups. Since the proposed subgroups in the ANOVA are quite large, it is necessary to run Levenes test for each of them to evaluate the assumption of homogeneity of the sample.

The second part of the statistical experiments used in the paper is the Pre/Post-test. A parametric Dependent t-Test can be used to determine statistical differences in student achievement and satisfaction at the beginning and end of the experiment. This test also has a number of assumptions that must be evaluated before analysis (LS, 2020). In particular, the condition for this test suggests that there are no abnormal outliers, which can be evaluated by constructing boxplots, as well as the normality of the distribution. As for the ANOVA, in the case of the Pre/Post test, the normality of the distribution is checked using the same statistical techniques.

An important note should be made about the assumptions stated above. In the present studies, some of them, which do not concern the nature of the variables, may not be observed this is a consequence of non-ideal quasi-experimental conditions. In this case, violation of the assumptions prevents the use of statistical test data and creates the need to search for alternatives. Specific alternatives can be transformations of variables into another form, followed by a test of compliance with assumptions or a search for a similar test that does not have such strict input conditions.