DHR.

529

Quantitative Essay Part 1

I. Differences Between Descriptive and Inferential Statistics

Descriptive and inferential statistics serve distinct yet complementary roles in data analysis. Descriptive
statistics summarize and organize data, providing a clear overview of its key characteristics. Common
measures include the mean, median, and mode for central tendency, and the range, variance, and standard
deviation for variability. Graphical tools like histograms and boxplots further aid in visualizing data
distributions. These statistics are foundational, as they help researchers identify patterns and assess whether
data meets the assumptions required for further analysis, such as normality for parametric tests.

Inferential statistics, on the other hand, extend beyond the immediate dataset to make generalizations about
a larger population. Techniques like t-tests, Chi Square, ANOVA, and regression analysis are used to test
hypotheses and draw conclusions. For example, a t-test can compare mean differences between groups,
while regression analysis examines relationships between variables while controlling for confounders.
Descriptive statistics often inform the choice of inferential methods by revealing data properties, such as
skewness or outliers, which may necessitate non-parametric alternatives.
II. Null Hypothesis Testing: Theoretical Concept and Critical Appraisal

Null Hypothesis Significance Testing (NHST) is a cornerstone of statistical analysis, used to determine
whether observed effects are statistically significant or due to chance. The null hypothesis (H₀) posits no
effect, while the alternative hypothesis (H₁) suggests a significant effect. The p-value quantifies the
probability of obtaining the observed data under H₀, with a threshold (e.g., p < 0.05) guiding decisions to
reject H₀.

However, NHST has notable limitations. The arbitrary nature of the p-value threshold can lead to Type I
(false positives) and Type II (false negatives) errors. Additionally, p-values do not convey effect size or
practical significance. Critics argue that NHST oversimplifies complex data and encourages dichotomous
thinking (significant vs. non-significant). To address these issues, researchers increasingly supplement
NHST with effect size measures (e.g., Cohen’s d) and confidence intervals, which provide richer context for
interpreting results.

In practice, Null Hypothesis Significance Testing (NHST) is commonly used in research but is often
misapplied when its underlying assumptions such as normality of data or independence of observations—
are ignored. For example, a researcher might apply a t-test to compare the average test scores between two
groups of students, assuming the data are normally distributed. However, if the scores are actually heavily
skewed or the sample size is too small for the Central Limit Theorem to apply, the t-test may produce
inaccurate p-values. This misapplication can lead to misleading conclusions such as falsely claiming a
significant difference between groups when none exists ultimately compromising the integrity of the
research findings and leading to poor policy or practice decisions based on flawed evidence. For instance,
using a t-test on non-normal data without checks can yield misleading conclusions. Thus, while NHST
remains widely used, its limitations underscore the need for complementary methods and rigorous
adherence to statistical assumption.
III. Real-World Example: Psychosocial Interventions and Mental Health

Research Question: Do police officers who receive psychosocial interventions exhibit better mental health
outcomes compared to those who do not?

Variables:

The independent variable is the psychosocial intervention, which is categorized based on whether
participants have received the intervention or not. This variable is central to determining the impact of such
support mechanisms on mental health outcomes. The dependent variable is the mental health status of the
participants, which is quantified using standardized tools such as the General Health Questionnaire-12
(GHQ-12). This score reflects the psychological well-being of individuals and serves as an indicator of how
effective the intervention has been. Additionally, to ensure the relationship between psychosocial
intervention and mental health is not influenced by other factors, the study incorporates control variables
such as age, gender, and years of service. These demographic and professional attributes are important to
consider, as they may independently affect mental health and thus need to be accounted for in the analysis to
isolate the true effect of the psychosocial intervention.

Methods:
Descriptive statistics (mean, SD, boxplots) summarized baseline data and checked for normality. An
independent samples t-test compared mean mental health scores between intervention and control groups,
assuming equal variances after Levene’s test. Multiple regression controlled for confounders like age.

Limitations:

The use of a cross-sectional research design presents a notable limitation in that it restricts the ability to
make causal inferences. Since data are collected at a single point in time, it becomes challenging to establish
the directionality or temporal order of relationships between variables. Consequently, while associations
may be identified, it is not possible to determine whether one variable actually causes changes in another.
Additionally, even when statistical controls are applied to account for known variables, the possibility of
unobserved confounders still exists. These are hidden or unmeasured factors that may influence both the
independent and dependent variables, potentially biasing the results. The presence of such confounders
undermines the internal validity of the study and complicates the interpretation of findings, as observed
relationships may be attributed to factors not captured in the analysis.
Justification of Descriptive and Inferential Statistics

Descriptive Statistics

Measures of Central Tendency and Variability

The mean and standard deviation are used to summarize mental health scores for both the intervention group
(psychosocial support) and the control group. The mean provides the average score, offering insight into the
general well-being of participants, while the standard deviation indicates how much individual scores vary
from the mean. A low standard deviation suggests that most participants' scores cluster closely around the
mean, whereas a high standard deviation reflects greater variability in mental health outcomes. These
statistics help assess the consistency of the psychosocial intervention's effects compared to the control
condition.

Frequency Distributions for Categorical and Continuous Variables

Frequency distributions are employed to analyze demographic data (e.g., age, gender, occupation) by
showing how often each category occurs within the sample. This helps identify trends or imbalances in
participant characteristics.

For continuous variables, such as mental health scale scores, frequency distributions can be constructed by
grouping scores into intervals (bins) and counting how many observations fall into each range. This
approach visualizes the spread of scores—whether they cluster in certain ranges (e.g., moderate symptoms)
or are dispersed widely (e.g., from mild to severe). Histograms or frequency tables can then display these
distributions, aiding in the detection of common patterns or outliers in mental health outcomes across
groups.

Boxplots provide a visual representation of the distribution of mental health scores and help identify
outliers. Each boxplot displays the median (middle score), interquartile range (middle 50% of scores), and
possible extreme values. Outliers, which appear as individual points outside the main box, indicate
participants whose mental health scores differ significantly from the rest. By comparing boxplots for the
peer counseling group and the control group, researchers can quickly assess differences in score distribution,
variability, and any unusual data points that may require further investigation.
Inferential Statistics

To evaluate the effectiveness of peer counseling on police officers’ mental health, this study employs
inferential statistical methods, including the Independent Samples *t*-test and multiple regression analysis.
However, the research question focuses on improvements in mental health rather than differences between
groups. Since assessing improvement requires repeated measurements (e.g., pre- and post-intervention
scores), future iterations of this study should incorporate longitudinal data to better capture changes over
time.

The Independent Samples t-test compares the mean mental health scores of two distinct groups: officers
who received peer counseling and those who did not. This test assumes that the dependent variable (mental
health scores) is normally distributed and that variances between groups are either equal or unequal. Before
conducting the test, the plausibility of these assumptions must be verified through diagnostic checks, such as
normality tests (e.g., Shapiro-Wilk) and homogeneity of variance tests (e.g., Levene’s test). If these
assumptions are violated, non-parametric alternatives, such as the Mann-Whitney U test, should be
considered.

While the T-test helps identify differences between groups, multiple regression analysis isolates the effect of
peer counseling by controlling for confounding variables, such as age and years of service. However, even
when observable factors are accounted for, unobserved variables (e.g., baseline resilience, personal
stressors) may still influence outcomes. Without random assignments to intervention and control groups,
there remains a risk of selection bias, where unmeasured differences between participants could drive the
observed effects.

Descriptive statistics, including measures of central tendency (mean, median) and dispersion (standard
deviation, skewness), provide an initial overview of the data distribution. These analyses ensure that the
assumptions of subsequent inferential tests are met. If violations are detected—such as non-normality or
heteroscedasticity, appropriate adjustments such as data transformations or robust statistical methods should
be applied.

The application of these statistical techniques offers evidence-based insights into the effectiveness of peer
counseling. However, since this analysis focuses solely on methodology without presenting results no
conclusions can yet be drawn about the intervention’s impact. Future research should include effect sizes
and confidence intervals to quantify the magnitude and precision of observed effects, further strengthening
the validity of the findings.