Descriptive vs. Inferential Statistics for Intervention Data

1. The Foundation: Population vs. Sample

Before we can differentiate between the two types of statistics, we must understand the core concepts of population and sample. The distinction between these two terms is the primary reason we need two different statistical approaches.

To get a clear picture of this, let's watch a short segment from a tutorial on quantitative data analysis.

To put this in a Swasthawritta context:

• Population: If you want to study the effect of Ashwagandha on stress in IT professionals, the population is all IT professionals. This group is vast and practically impossible to study in its entirety.
• Sample: You might recruit 200 IT professionals from a few companies in Bengaluru for your study. This group of 200 is your sample.

The central goal of research is to learn something about the population by studying the sample. This is precisely where descriptive and inferential statistics play their distinct roles.

2. Descriptive Statistics: Describing Your Slice of the Cake

Descriptive statistics do exactly what the name implies: they describe or summarize the data from your sample. They don't make any conclusions beyond the data you have in front of you. Their purpose is to organize the data into a manageable and understandable format.

Key functions of descriptive statistics in an intervention study:

• To summarize the characteristics of your participants (e.g., average age, gender distribution).
• To present baseline values of your outcome measures (e.g., the average stress score before the intervention).
• To check if your intervention and control groups were comparable at the start of the study.

The following video section details the most common types of descriptive statistics.

Application in an Ayurvedic Study

Let's see how this looks in a real research paper. The following resource is a study investigating the effectiveness of Swasthavritta Aahar (Ayurvedic dietary principles).

As you saw in the paper, Table 1 is a classic example of descriptive statistics at work.

• Mean: The authors report that the mean "Overall Health Assessment" score was 8.6 for the Ayurvedic diet group and 6.8 for the modern diet group. This gives us a single number that represents the central tendency of each group's health score within the sample.
• Standard Deviation (SD): The SD for the Ayurvedic group was 1.2, while for the modern diet group it was 1.6. This tells us that the health scores in the modern diet group were slightly more spread out or variable than in the Ayurvedic group.

At this stage, we have only described the 100 people in the study. We cannot yet claim that an Ayurvedic diet is definitively better for everyone. We are simply painting a picture of our sample.

3. Inferential Statistics: Generalizing to the Whole Cake

While descriptive statistics describe the sample, inferential statistics help us make inferences or generalize our findings from the sample to the larger population. This is where we perform hypothesis testing to determine if the effects we see in our sample are "real" and not just due to random chance.

This is the part of the analysis that answers the primary research question, such as: "Does my intervention work?"

The Logic of Inferential Statistics

To make these inferences, we use a formal process of hypothesis testing. Let's explore the core components of this process.

Key Concepts in Hypothesis Testing:

• Null Hypothesis ($H_0$): A statement of no effect. In our example study, the null hypothesis would be: "Swasthavritta Aahar has no effect on health scores."
• Alternative Hypothesis ($H_1$): The statement that there is an effect. "Swasthavritta Aahar has a significant positive effect on health scores."
• p-value: The probability of observing the data you collected (or something more extreme) if the null hypothesis were true.
  • A small p-value (typically < 0.05) suggests that your observed data is unlikely if there's truly no effect. This gives you evidence to reject the null hypothesis in favor of the alternative. The popular mnemonic is: "If the P is low, the null must go."

Common inferential tests include the t-test (to compare means of two groups), ANOVA (to compare means of more than two groups), and the chi-square test (to compare categorical data).

Application in the Ayurvedic Study

Now, let's return to the Swasthavritta Aahar paper and see how the authors used inferential statistics.

In Table 2, the authors use a t-test to compare the health scores of the two groups.

• They found a p-value of 0.0001.
• Since this p-value is much smaller than the standard cutoff of 0.05, they rejected the null hypothesis.
• This allowed them to conclude that the higher health score in the Ayurvedic diet group was not just a fluke in their sample but represents a statistically significant difference. They can now infer that Swasthavritta Aahar is likely to have a positive effect on health in the broader population from which the sample was drawn.

4. Bringing It All Together: The Roles in an Intervention Study

Let's summarize the complementary roles of descriptive and inferential statistics in the context of a Swasthawritta clinical trial.

Scenario: You are testing if a specific Pranayama technique (intervention) reduces anxiety scores compared to a simple breathing awareness exercise (control) in 80 university students.

1. Role of Descriptive Statistics:

   • Describe Participants: You first use descriptive statistics to create a "Table 1" that summarizes your sample. You would report the mean age, the percentage of male/female participants, and, crucially, the mean baseline anxiety scores for both the Pranayama group and the control group.
   • Purpose: This demonstrates who was in your study and checks for baseline comparability. If the Pranayama group happened to have much lower anxiety to start with, it would bias your results. Descriptive statistics help you confirm that the two groups were similar before the intervention began.

2. Role of Inferential Statistics:

   • Test the Hypothesis: After the 8-week intervention, you compare the final anxiety scores (or the change in scores) between the two groups using an inferential test (like an independent t-test).
   • Make a Conclusion: The test yields a p-value. If p < 0.05, you reject the null hypothesis ("Pranayama has no effect on anxiety"). You can then conclude that the Pranayama technique was significantly more effective at reducing anxiety than the control exercise.
   • Purpose: This allows you to generalize your finding beyond the 80 students in your study, making a scientific claim about the potential benefit of this Pranayama technique for the wider population of university students.

Conclusion

Today we've demystified the two fundamental branches of data analysis. We've seen that they are not opposing methods, but rather partners in the scientific process, each with a distinct and vital role.

Key Takeaways:

• Descriptive Statistics summarize and describe the characteristics of your immediate sample. They answer the question, "What does my data look like?" Common tools include mean, median, and standard deviation.
• Inferential Statistics use data from your sample to make generalizations or predictions about the broader population. They answer the question, "What do my results mean for the real world?" Common tools include t-tests, ANOVA, and the interpretation of p-values.
• In an intervention study, descriptive statistics are used to characterize the sample and ensure baseline comparability, while inferential statistics are used to test the hypothesis and determine if the intervention had a statistically significant effect.

Preview of the Next Lesson:

We now understand why we need inferential statistics to test our hypotheses. However, there are many different inferential tests available. How do you know whether to use a t-test, a chi-square test, or another method? In our next lesson, we will address the learning outcome: "Choose appropriate basic statistical tests (e.g., t-test, chi-square) to compare groups based on the type of data collected." This will equip you with the practical knowledge to select the right statistical tool for your specific research question.