Comprehensive Notes on Sampling Bias, Frequency Distributions, and Excel Data Analysis

Sampling Error, Non-Sampling Error, and Bias

  • Random Selection & Non-Biased Sampling:

    • A sample selection is random when every individual or data point within the target population has an equal and fair chance of being selected.
    • True random selection ensures that the chosen sample is free from bias.
    • Choosing participants based on specific non-random criteria—such as selecting only students sitting in the front row under the assumption that they are the best students while assuming those in the back are not paying attention—constitutes a biased sample because not every population member has an equal probability of selection.

  • Bias and Sampling Bias:

    • Bias: Occurs when a systematic preference or influence sways the outcome or result of a study away from true randomness. It arises when data points or participants are chosen exclusively from a specific subset, depriving other members of the population of an equal chance of selection.
    • Sampling Bias: A specific form of bias that occurs during participant selection when certain members of the target population are systematically less likely, or unable, to be chosen or represented in the study.

  • Sampling Error:

    • Caused intrinsically by the process of sampling itself.
    • Natural sampling error dictates that a sample will almost never perfectly or exactly replicate the characteristics of the overall population.
    • Sampling error also occurs when a selected sample size is too small to accurately answer research questions or represent the population parameter under investigation.

  • Non-Sampling Error:

    • Errors unrelated to the sampling design or natural variation between a sample and a population.
    • Caused by data entry errors, manual calculation mistakes, incorrect counting of data points, or procedural errors occurring during data collection and processing.

Analysis of Sampling Scenarios and Case Studies

  • Scenario 1: High School Favorite Subject Study

    • Setup: A study determines the favorite subject of high school students. A random sample of 5050 students is collected from the entire high school population.
    • Sample Result: Math is identified as the most popular subject.
    • Population Result: When all students in the school are surveyed, Science is identified as the most popular subject.
    • Classification: Sampling Error.
    • Explanation: The sample collected was genuinely random and valid, but due to natural sampling variation, the sample result (Math) differed from the full population result (Science).

  • Scenario 2: European Travel Destination Study

    • Setup: A study evaluates top travel destinations between Europe and Asia by surveying individuals living in big cities throughout Europe.
    • Classification: Sampling Bias.
    • Explanation: Restricting the sample exclusively to residents of major European cities systematically excludes rural populations and non-European perspectives.

  • Scenario 3: Hospital Heart Monitoring Software

    • Setup: A hospital evaluates new heart monitoring software using all patients currently using the software.
    • Classification: Targeted sample evaluation.

  • Scenario 4: Cereal Brand Preference Survey

    • Setup: A survey asks consumers: "Do you prefer the delicious taste of brand X to the taste of brand Y?"
    • Classification: Biased Survey Design / Questioning.
    • Explanation: The phrasing uses loaded adjectives ("delicious taste") to influence or sway survey respondents toward brand X, introducing response bias.

  • Identifying Biased vs. Unbiased Sample Designs:

    • Nico's Study: Nico estimates the mean number of classes taken by university students by collecting data from a randomly selected, proportionate number of students from each grade level.
      • Evaluation: Unbiased / Valid Sampling (Proportional stratified random sampling giving fair representation to all grade levels).
    • One's Study: One estimates the mean weight of town residents by interviewing members of a local fitness club.
      • Evaluation: Biased Sample. Members of a fitness club are not representative of the overall town population's physical habits or average weight.
    • Sean's Study: Sean estimates voter preference for the Governor of Georgia by surveying a large group of randomly selected residents exclusively from Atlanta.
      • Evaluation: Biased Sample. As noted by Mr. Albright, residents of a single city (Atlanta) do not represent the political preferences of the entire state of Georgia.
    • Theresa's Study: Theresa estimates average weekly spending on clothes by mall shoppers by interviewing every 10th10\text{th} person entering a single clothing store at the mall.
      • Evaluation: Biased Sample. Although selecting every 10th10\text{th} person represents systematic sampling, restricting data collection to one clothing store excludes customers from all other mall stores who may exhibit different spending habits.

Frequency Distributions and Data Summarization

  • Frequency Tables:

    • A frequency table organizes raw data into two primary columns: the Data Value (or variable value) and its corresponding Frequency (the count of how many times that data value appears).

  • Example 1: Student Work Hours Dataset:

    • Data Value 22 hours: Frequency = 33 students
    • Data Value 33 hours: Frequency = 55 students
    • Data Value 44 hours: Frequency = 33 students
    • Data Value 55 hours: Frequency = 66 students
    • Data Value 66 hours: Frequency = 22 students
    • Data Value 77 hours: Frequency = 11 student
    • Total sample size calculation:         Total Students=3+5+3+6+2+1=20\text{Total Students} = 3 + 5 + 3 + 6 + 2 + 1 = 20

  • Grouped Frequency Distributions (Class Intervals):

    • When dealing with continuous or wide-ranging quantitative data, data values are grouped into class intervals or range bins.
    • Example class intervals:
      • 4.5 to 54.5 \text{ to } 5
      • 5 to 5.55 \text{ to } 5.5
      • 5.5 to 65.5 \text{ to } 6
      • 6 to 6.56 \text{ to } 6.5

  • Example 2: Daily Emergency Doctor Visits:

    • Dataset recording average daily doctor visits in a small town: 2,1,1,3,3,3,9,9,…2, 1, 1, 3, 3, 3, 9, 9, \dots (total of 1111 data points).
    • Frequency of visit count 33: 33 occurrences.
    • Frequency of visit count 99: 22 occurrences.
    • Sum of frequencies: 1111 total recorded data values.

  • Example 3: Recreation Center Visitor Age Bins:

    • Age Bin 0 to 90 \text{ to } 9 years: Frequency = 00
    • Age Bin 10 to 1910 \text{ to } 19 years: Frequency = 66
    • Age Bin 20 to 2920 \text{ to } 29 years: Frequency = 66
    • Age Bin 50 to 5950 \text{ to } 59 years: Frequency = 22
    • Total number of data values calculation:         Total Data Points=0+6+6+2=14\text{Total Data Points} = 0 + 6 + 6 + 2 = 14

Data Organization and Analysis Using Excel Spreadsheets

  • Overview of Spreadsheet Utility:

    • Spreadsheet software such as Microsoft Excel streamlines data organization, processing, and retrieval, particularly when handling extensive datasets.

  • Step-by-Step Procedure for Extracting Unique Data Values in Excel:

    1. Paste raw dataset into the first column of an Excel worksheet (e.g., column A, starting at cell A2).
    2. Select the entire column containing the raw data values.
    3. Navigate to the Data tab on the top menu bar.
    4. Click Advanced within the Sort & Filter group.
    5. Select the option Copy to another location.
    6. Specify the target destination cell (e.g., cell E5).
    7. Check the checkbox labeled Unique records only.
    8. Click OK to generate a list containing only non-repeating entries.

  • Step-by-Step Procedure for Calculating Frequency using COUNTIF:

    1. Click on the destination cell adjacent to the first unique record (e.g., cell F6).
    2. Enter the formula structure:        =COUNTIF(range, criteria)
    3. Specify the full cell range of the original raw dataset (e.g., A2:A27).
    4. Insert a comma and reference the cell containing the specific unique value being counted (e.g., cell E6).
      • Complete example formula: =COUNTIF(A2:A27, E6)
    5. Press Enter to calculate the frequency count for that item.
    6. Hover the mouse over the bottom-right corner of the formula cell until the cursor transforms into a fill handle crosshair.
    7. Click and drag the fill handle down across the remaining unique value cells to apply the formula dynamically.

  • Case Study: Employee Sales Performance:

    • Dataset: Employee sales records containing repeated entries for sales reps including Pearson, Mark, Jeremy, Simon, Lee, Christian, and Kirsten in cells A2 through A27.
    • Unique Names Extracted: Pearson, Mark, Jeremy, Simon, Lee, Christian, Kirsten.
    • Formula Applied: =COUNTIF(A2:A27, E6) dragged down column F.
    • Analysis Question: How many employees had exactly 33 total sales?
    • Result: Exactly 22 employees achieved 33 total sales.

Questions & Discussion

  • Instructor Office Hours:

    • Office hours are available until 3:003:00 for additional assistance with spreadsheet applications or conceptual statistics questions.

  • Classroom Procedures:

    • In-class worksheets must be completed and submitted prior to departure.
    • An official attendance sign-in sheet must be signed by all attending students before leaving.