Jenis-Jenis Variabel dan Skala Pengukuran

Variables in Research and Measurement Scales
📌 A variable is a phenomenon that needs to be measured or studied, characterized by having varied or diverse values.
🛑 Not every observable phenomenon qualifies as a research variable; it must have research relevance.
🧪 Variables can be classified by nature: dichotomous (having two mutually exclusive categories, e.g., gender) or continuous (having values within a specific interval, allowing for fractional values).

Types of Variables Based on Relationship
💡 Independent Variable (X): The variable that influences or causes the magnitude of the dependent variable (e.g., Motivation influencing employee performance).
📈 Dependent Variable (Y): The variable whose variation is influenced by the independent variable (e.g., Employee performance).
🔗 Moderator Variable: A variable that strengthens or weakens the relationship between the independent and dependent variables.

Measurement Scales Overview
📊 Likert Scale: Commonly used scale offering 5-7 response choices (usually 5), assigning numerical scores ($\text{5 to 1}$) to facilitate quantitative data processing.
✖️ Qualitative responses (e.g., "Strongly Agree") must be converted to numerical weights for mathematical operations, unlike raw categorical responses.
⭐ Guttman Scale: Presents mutually exclusive answer choices where options conflict (e.g., Yes/No, Good/Bad).

Semantic Differential Scale
🧠 Developed by Osgood, this scale measures values, resulting in interval scale data, allowing calculation of mean and standard deviation.
〰️ Respondents rate their response by placing a mark along a continuum line between bipolar adjectives (e.g., Very Bad to Very Good), which can be scored $\text{1}$ through $\text{5}$.

Rating Scales and Data Conversion
🔄 Rating Scale (Quantitative to Qualitative): Data obtained initially is quantitative (numerical ratings) and is then converted back into qualitative categories (the opposite of the Likert scale process).
📌 Rating scales are used to assign values to a measured attribute.

Four Levels of Measurement Scales
1️⃣ Nominal Scale: The lowest level, used only for categorization (e.g., gender: male/female). Once a category is assigned, the indicator cannot belong to another category.
2️⃣ Ordinal Scale: Used to indicate rank or level, but the distance or interval between ranks is not clearly defined or quantifiable (e.g., judging parking quality at two different locations as 'Good').
3️⃣ Interval Scale: Specifies rank and the interval between ranks is clear, but it lacks an absolute zero point (e.g., temperature in Celsius: $0^\circ \text{C}$ does not mean the absence of temperature). Time/clock scales also fit here.
4️⃣ Ratio Scale: The highest level, where rank, interval, and an absolute zero point are clearly defined (e.g., weight: $60 \text{ kg}$ is exactly twice $30 \text{ kg}$, and $0 \text{ kg}$ signifies no weight).

Key Points & Insights
➡️ Variables must possess research appeal to be suitable for study; research variables are a subset of all variables.
➡️ The Interval Scale has defined intervals but uses a non-absolute zero (e.g., $0^\circ \text{C}$ still represents a temperature).
➡️ The Ratio Scale is superior because it includes an absolute zero value, meaning zero truly represents the absence of the measured attribute (e.g., weight).

📸 Video summarized with SummaryTube.com on Dec 14, 2025, 11:26 UTC