Unit 2: Graphical Representations and Frequency Distribution - Subjective Questions

Graphical representation is the presentation of numerical data through diagrams, charts, and graphs. It is important in psychological research because:

• It simplifies large and complex data sets.
• It makes patterns, trends, and relationships easier to identify.
• It allows quick comparison between groups or variables.
• It helps researchers detect extreme values and unusual observations.
• It improves communication of research findings to both experts and general readers.
• It supports interpretation and decision-making by presenting data visually.

Thus, graphs provide an effective supplement to numerical and descriptive statistical analysis.

A good graph should have the following characteristics:

• Clarity: The information should be easy to understand.
• Accuracy: The graph must represent the data without distortion.
• Proper title: The title should clearly indicate the subject of the graph.
• Appropriate scale: The axes should use a suitable and uniform scale.
• Labels: Both axes, units, and categories should be clearly labeled.
• Legend or key: Different groups should be identified when necessary.
• Simplicity: Unnecessary decoration and complexity should be avoided.
• Proportion: The size of graphical elements should fairly reflect the values.

These features make a graph meaningful, reliable, and useful for interpretation.

Ungrouped data consists of individual observations listed separately. It is suitable when the number of observations is small and each value is important. Common representations include bar diagrams, dot plots, and simple line graphs.

Grouped data is arranged into class intervals along with their frequencies. It is useful when there are many observations or when the values cover a wide range. Common representations include histograms, frequency polygons, and frequency curves.

Grouping improves convenience but may result in some loss of detail.

A bar graph represents the frequencies or values of separate categories using rectangular bars.

Construction:

• Place the categories on the horizontal axis.
• Place frequencies or values on the vertical axis.
• Draw bars of equal width.
• Keep spaces between bars because the categories are discrete.
• Make the height of each bar proportional to its frequency or value.
• Add a title, labels, and scale.

Uses:

• Comparing frequencies of categorical variables such as diagnosis, gender, or preference.
• Presenting results from questionnaires.
• Showing differences between experimental and control groups.

Unlike a histogram, a bar graph is generally used for discrete or qualitative categories, and its bars do not touch.

A histogram is a graphical representation of a continuous frequency distribution. Class intervals are placed on the horizontal axis and frequencies on the vertical axis. Rectangles are drawn over the class intervals, with their areas representing the frequencies.

Differences between a histogram and a bar graph:

• A histogram represents continuous quantitative data, whereas a bar graph usually represents categorical or discrete data.
• The bars of a histogram touch each other, while bars in a bar graph are separated by spaces.
• In a histogram, the horizontal axis contains class intervals; in a bar graph, it contains separate categories.
• The order of class intervals in a histogram is fixed, whereas categories in a bar graph may often be rearranged.
• A histogram is useful for studying the shape of a distribution, including skewness and modality.

A frequency polygon is a line graph formed by joining points representing class midpoints and their frequencies.

Procedure:

1. Prepare the grouped frequency distribution.
2. Calculate the midpoint of each class using
   
3. Mark class midpoints on the horizontal axis.
4. Mark frequencies on the vertical axis.
5. Plot a point for each midpoint and its corresponding frequency.
6. Join adjacent points with straight lines.
7. Add one imaginary class of zero frequency at each end to bring the polygon back to the baseline.

A frequency polygon is particularly useful for comparing two or more frequency distributions on the same set of axes.

Both a frequency polygon and a frequency curve represent the distribution of quantitative data.

A frequency polygon is constructed by joining points with straight lines. A frequency curve is obtained by smoothing the polygon. The curve may show whether a distribution is symmetric, positively skewed, negatively skewed, unimodal, or bimodal.

A frequency distribution is a systematic arrangement of data values or class intervals along with the number of observations occurring in each value or interval.

Its major components are:

• Class limits: The smallest and largest values included in a class.
• Class boundaries: The actual boundaries separating adjacent classes, especially when continuous data are used.
• Class interval: The range covered by a class.
• Class width: The difference between the upper and lower boundaries of a class.
• Frequency: The number of observations in a class, represented by .
• Cumulative frequency: The running total of frequencies.
• Class midpoint: The central value of a class, calculated as .

A frequency distribution condenses raw data and makes further statistical analysis easier.

The following steps are used to construct a grouped frequency distribution:

1. Find the smallest and largest observations.
2. Calculate the range:
   
3. Decide the number of classes, often between 5 and 15 depending on the data size.
4. Determine the class width using approximately
   
   where is class width and is the number of classes.
5. Form mutually exclusive and exhaustive class intervals.
6. Tally each observation into the appropriate class.
7. Count the tally marks to obtain the frequency of every class.
8. Check that the sum of all class frequencies equals the total number of observations:
   

The intervals should be clear, equal where possible, and should not overlap.

The inclusive method includes both limits in a class interval. For example, intervals may be -, -, and -. This method is commonly used for discrete data.

The exclusive method includes the lower limit but excludes the upper limit. For example, intervals may be -, -, and -. This method is suitable for continuous data.

Main differences:

• Inclusive intervals have apparent gaps when data are treated as continuous; exclusive intervals have adjoining boundaries.
• Inclusive intervals are convenient for whole-number observations.
• Exclusive intervals are more appropriate for measurements such as reaction time, height, and test scores treated continuously.
• For inclusive intervals, true class boundaries may be obtained by applying a correction factor, such as for whole-number data.

The choice depends on the nature and precision of the observations.

A cumulative frequency distribution shows the successive totals of frequencies up to or beyond specified class boundaries.

Less-than cumulative frequency: Frequencies are added from the lowest class upward. It indicates the number of observations less than a given upper boundary.

More-than cumulative frequency: Frequencies are added from the highest class downward. It indicates the number of observations greater than a given lower boundary.

For example, if class frequencies are , , and , the less-than cumulative frequencies are , , and . The more-than cumulative frequencies are , , and .

Cumulative distributions are used to construct ogives and to locate the median, quartiles, deciles, and percentiles.

An ogive is a cumulative frequency curve. It is constructed by plotting cumulative frequencies against class boundaries.

Types:

• Less-than ogive: plots less-than cumulative frequencies against upper boundaries.
• More-than ogive: plots more-than cumulative frequencies against lower boundaries.

To locate positional values:

1. Find the total frequency .
2. Mark , , and on the cumulative-frequency axis for the median, first quartile, and third quartile respectively.
3. Draw horizontal lines from these values to the ogive.
4. Drop perpendiculars to the horizontal axis.
5. The corresponding values are , the median, and .

The intersection of a less-than and a more-than ogive provides an approximate graphical location of the median.

Quartiles are positional measures that divide an ordered data set into four equal parts.

• First quartile (): Approximately 25% of observations lie below it.
• Second quartile (): The median; approximately 50% of observations lie below it.
• Third quartile (): Approximately 75% of observations lie below it.

For grouped data, the quartile formula is:

where , is the lower boundary of the quartile class, is total frequency, is cumulative frequency before that class, is class frequency, and is class width.

Quartiles help psychologists describe the distribution of test scores, identify the central half of observations, and assess variability through the interquartile range:

Quartiles, deciles, and percentiles are positional measures that divide an ordered distribution into equal parts.

• Quartiles: Divide the data into four equal parts and are denoted by , , and .
• Deciles: Divide the data into ten equal parts and are denoted by through .
• Percentiles: Divide the data into one hundred equal parts and are denoted by through .

Their positions in grouped data are calculated as:

These measures are useful for interpreting an individual's relative standing in psychological tests. For example, a score at the 80th percentile is higher than approximately 80% of the reference group.

The th percentile is the value below which approximately percent of the observations fall. Its position in a distribution of observations is:

Suppose this position lies in a particular percentile class. Let:

• = lower boundary of the percentile class
• = cumulative frequency before the percentile class
• = frequency of the percentile class
• = class width
• = total frequency

The number of observations still required after the preceding classes is . Assuming observations are uniformly distributed within the class, the required fraction of the class is:

Multiplying this fraction by the class width and adding it to the lower boundary gives:

This formula estimates the percentile by interpolation within the appropriate class interval.

Deciles divide an ordered distribution into ten equal parts. There are nine deciles: through . The decile has approximately percent of observations below it.

For grouped data, use:

where:

• is the lower boundary of the decile class.
• is the total number of observations.
• is the cumulative frequency before the decile class.
• is the frequency of the decile class.
• is the class width.

To calculate a decile, first locate the class containing the position , then substitute the relevant values in the formula. Deciles are useful for reporting relative performance and dividing respondents into ten groups.

A percentile indicates the relative position of a score within a reference group. The th percentile is the score below which approximately percent of observations fall.

Applications in psychological testing:

• Comparing an individual's performance with a norm group.
• Interpreting aptitude, achievement, intelligence, and personality test scores.
• Identifying individuals who fall in unusually low or high ranges.
• Selecting candidates for educational or occupational programs.
• Communicating test performance in an easily understandable form.

A percentile rank is not the same as a percentage score. A score of 70% means 70 correct units out of 100, whereas the 70th percentile means the individual performed better than approximately 70% of the norm group.

For grouped data, the percentile is calculated using:

A percentage score expresses the proportion of items answered correctly or the proportion of a total obtained by an individual. It is calculated as:

A percentile rank expresses the percentage of people in a comparison group who scored below a particular individual or score.

The two values may be numerically different and should not be interpreted as equivalent.

Class boundaries and class midpoints are essential when representing grouped continuous data.

• Class boundaries are the exact points separating adjacent classes. They remove apparent gaps between intervals and are used for constructing histograms and calculating positional measures.
• Class midpoint is the central value of a class and is calculated as:
  
  where and are the lower and upper class limits.

Importance:

• Histograms require continuous class boundaries so that rectangles touch correctly.
• Frequency polygons use class midpoints on the horizontal axis.
• Mean estimation from grouped data uses midpoints as representative values.
• Quartiles, deciles, and percentiles use class boundaries in interpolation formulas.
• Correct boundaries and midpoints improve the accuracy of graphical and statistical interpretation.

Advantages:

• Graphs summarize large amounts of data efficiently.
• They make comparisons quick and clear.
• Trends, variability, and distributional patterns can be recognized easily.
• They help communicate findings to audiences with limited statistical training.
• They can reveal unusual values, gaps, clusters, and skewness.

Limitations:

• Important numerical details may be lost through summarization.
• Poor scales or distorted axes can mislead readers.
• Some graphs require technical knowledge for accurate interpretation.
• Graphs may oversimplify complex relationships.
• Different graphical choices can produce different impressions of the same data.

Therefore, graphs should be prepared accurately and interpreted along with tables and numerical summaries.