Statistical Analysis
Statistical analysis is used to make informed, sound decisions that reduce bias and human error, thereby offering greater practical utility for applied purposes. One use of such analysis is to compare groups, which has vast practical potential. Research across a variety of fields uses data collection to examine multiple groups exposed to different interventions — comparing point estimates between groups allows differences to be determined at the level of significance.
Descriptive Statistics
It is important to emphasize that, in any statistical test, the first step is to obtain descriptive statistics. This type of result provides insight into the general distributions of variables, whether in terms of measures of central tendency or measures of variability. The outcomes of this type of analysis can lead to the primary insight that there are differences between group averages, for example. However, results from descriptive statistics alone are insufficient to assess the statistical significance of such differences; in other words, conclusions drawn from descriptive comparisons of averages are unreliable. In this context, inferential tests are an excellent technique for determining whether the difference between group means is statistically significant.
ANOVA
ANOVA is one of the most common examples of such tests that allow inferential comparison of point estimates between groups. Unlike classical t-tests, ANOVA compares means across two or more groups (Yu et al., 2022). In cases where there are three or more groups, however, ANOVA alone does not provide insight into which groups differ. Because ANOVA results report only the presence of statistically significant differences, identifying their location requires additional post hoc tests that perform pairwise comparisons between groups. The present analysis proposes evaluating the SPSS output from the directed ANOVA tests and discussing the results.
Statistical Analysis in Housing Research
Description
The research project examined differences in overall satisfaction and material well-being across groups, depending on the presence of housing problems. The project involved three groups based on the number of housing problems: No Housing Problem, One Housing Problem, and Two or More Housing Problems. It was hypothesized that groups with more problems would have lower levels of satisfaction, as this could lead to increased chronic stress and fewer opportunities to enjoy life.
Results
The results of the descriptive statistics provide a general understanding of the trends in the three groups: the overall satisfaction level for the No Housing Problem group was 12.71 (SD = 2.35), for the One Housing Problem group was 11.97 (SD = 2.588), and for the Two or More Housing Problems group was 10.57 (SD = 2.594). Although the averages differ, descriptive statistics alone, as stated earlier, do not provide insight into the significance of these differences. A parametric ANOVA test requires homogeneity of variance across groups, which is tested using Levene’s test (SL, 2021). This assumption was found to be satisfied (F(2, 932) = 2.109, p =.112). This implied that the null hypothesis of unequal variances among the three groups was rejected, and that the variances of the three groups were equal.
The SPSS results showed significant differences in overall satisfaction levels among the three groups (F(2, 932) = 61.674, p =.000). In other words, the groups’ means differ. However, as stated earlier, identifying the location of such differences is not possible with ANOVA; therefore, a Tukey post hoc test was conducted.
The test data provided insight into statistically significant differences among the averages, as each calculated p-value was below the alpha significance level. In more detail, differences were found between the No Housing Problem and One Housing Problem groups (MD = 0.739, p =.001). Statistically significant differences were also observed between the “No Housing Problem” and “Two or More Housing Problems” groups, (MD = 2.139, p =.000). Finally, the “One Housing Problem” and “Two or More Housing Problems” groups were also different from each other (MD = 1.401, p =.000). This implied that satisfaction was highest for the “No Housing Problem” group (M = 12.71, SD = 2.35), and lowest for the “Two or More Housing Problems” group (M = 10.57, SD =2.594), passing the “One Housing Problem” group (M =11.97, SD = 2.588). This indirectly supports the idea that increased housing problems lead to a drop in overall satisfaction.
Finally, an ANOVA was conducted in this project to determine whether group differences were statistically significant. Specifically, differences in overall satisfaction levels between the groups, as measured by the number of housing problems, were evaluated. The results showed that all three groups differed significantly in overall satisfaction. This implied that respondents with the highest number of housing problems had the lowest level of satisfaction, and vice versa.
References
SL. (2021). One-way ANOVA in SPSS statistics. Laerd Statistics.
Yu, Z., Guindani, M., Grieco, S. F., Chen, L., Holmes, T. C., & Xu, X. (2022). Beyond t test and ANOVA: Applications of mixed-effects models for more rigorous statistical analysis in neuroscience research. Neuron, 110(1), 21-35.