Primary Data Collection Methods Overview | AskSia

您好，您刚刚上传了文件，但没有附加具体问题或说明。请告诉我：

• 您希望我帮您做什么？（比如：总结内容、提炼重点、解答具体问题等）
• 是否需要可视化辅助或图片讲解？
• 如果是针对考试复习或某章节的知识点，也可以直接告诉我具体需求。

温馨提示：我可以记住您的学习重点、错题或易混淆知识点，随时帮助您快速回顾。如果需要，可随时告诉我“记住这个知识点”或“以后不要重复提这个内容”，我会为您记录！

Primary Data Collection Methods: An Overview

This summary outlines various methods for collecting primary data, focusing on communication-based approaches, their administration, and associated advantages and disadvantages.

Primary Data: Definition and Types

Primary data refers to information collected directly by the researcher for a specific purpose. It can be categorized into:

• Demographic / Socioeconomic Characteristics: Objective attributes of individuals.
  • Examples: Age, education, occupation, marital status, income.
• Psychological / Lifestyle Characteristics: Subjective traits and behaviors.
  • Examples: Activities, interests, risk-taking levels.
• Attitudes / Opinions: Feelings or judgments about something.
  • Example: "Do you like that bookcase?"
• Awareness / Knowledge: Familiarity with or understanding of a subject.
  • Examples: "Have you heard of this cell phone?"
  • Unaided Recall: Remembering information without prompts (e.g., "What ads do you remember from yesterday's newspaper?").
  • Aided Recall: Remembering information from a given list (e.g., "Among the list given to you, what ads do you remember from yesterday's newspaper?").
  • Recognition: Identifying previously encountered information (e.g., "Do you remember this ad?").
• Behavioral Intentions: Plans or likelihood of future actions.
  • Example: "Are you planning to buy this toothpaste?"
• Actual Behavior: Observed actions or past events.
  • Examples: "Why did you purchase this car?" or "What did the customer do after clicking on a link?"

Qualities of Primary Data

When collecting primary data, researchers consider:

• Business Logistics: Speed and cost of data collection.
• Data Quality: Objectivity and accuracy of the information gathered.

Communication Methods of Data Collection

Primary data can be collected through communication methods, primarily surveys. These surveys can be administered in various ways, differing in structure and disguise.

Structure and Disguise in Communication Methods

• Disguised: The true purpose of the study is hidden from the respondent.
  • Techniques include: Motivation Research (Word Association, Sentence Completion, Story Telling).
• Undisguised: The purpose of the study is clear to the respondent.
  • Typical Questionnaire: A common tool used in undisguised surveys.
  • Interviews: Can be open-ended, allowing for deeper exploration.

Methods of Survey Administration

Surveys can be administered through different channels:

1. Mail Surveys:

   • Advantages:
     • Wide distribution possible.
     • Not subject to interviewer bias.
     • Respondents work at their own pace.
     • Assures anonymity.
     • Generally least expensive.
     • May be the only method to reach certain respondents.
     • Sampling frames easily developed with mailing lists.
     • Best for personal, sensitive questions.
   • Disadvantages:
     • Very little control in securing responses from specific individuals.
     • Cannot secure responses from illiterates.
     • Cannot control the speed of response; long response time.
     • Researcher cannot explain ambiguous questions.
     • Does not allow probing with open-ended questions.
     • Difficult to change the sequence of questions.
     • Sequence bias as respondents can view the entire questionnaire.

2. Telephone Interviews:

   • Advantages:
     • Wide distribution possible.
     • Interviewer supervision is strong, leading to less interviewer bias.
     • Relatively strong response rates (higher than mail).
     • One of the quickest methods of data collection.
     • Less difficulty and cost in handling "call backs" than in-home interviews.
     • Allows easy use of computer support.
     • Sequence of questions is easily changed.
   • Disadvantages:
     • More difficult than personal interviews to determine if the appropriate respondent is being interviewed.
     • Difficult to establish a representative sampling frame due to unlisted numbers.
     • Cannot use visual aids.
     • More difficult to establish rapport than in person.
     • Does not handle long interviews well in most cases.
     • Subject to some degree of interviewer bias (though less than personal interviews).

3. Personal Interviews:

   • In-Home Interviews:
     • Advantages:
       • Probably the highest response rate.
       • Best for getting responses from specific, identified persons.
       • Allows use of any type of question/questionnaire.
       • Sequencing of questions is easily changed.
       • Allows probing of open-ended questions.
       • Allows clarification of ambiguous questions.
       • Permits easy use of visuals.
     • Disadvantages:
       • Generally most expensive method of administration.
       • Costly to revisit "not-at-homes."
       • Relatively slow method of administration.
       • Subject to interviewer bias.
       • Sample control is more difficult than with mall intercepts in terms of identifying a representative sample.
   • Mall Intercept Interviews:
     • Advantages:
       • Same advantages as in-home interviews.
       • Relatively short project completion time.
       • Less expensive than in-home interviews.
       • Much better interviewer supervision and control than in-home interviews.
     • Disadvantages:
       • Generally narrow distribution.
       • Interviewer supervision and control difficult to maintain.
       • Often difficult to identify individuals to include in the sampling frame.
       • Interviews typically need to be shorter than in-home interviews.

4. Internet/Web/Email Surveys:

   • Advantages:
     • Fairly versatile (can show print ads, product visuals, play music, video).
     • Survey responses automatically entered into a data file.
     • International samples are possible.
   • Disadvantages:
     • Sample is still not representative of general consumer markets.
     • Response rates are dropping as novelty declines.
     • Respondents may have concerns with privacy.

---

---

---

Chapter 9: Questionnaires and Data Collection Forms

This chapter outlines the procedure for developing effective questionnaires and data collection forms, detailing each step and providing guidelines for question wording, sequencing, and physical presentation. It also discusses various question types, response formats, and administration methods, including a specific technique for handling sensitive questions.

Procedure for Developing a Questionnaire

The development of a questionnaire follows a systematic, nine-step process:

1. Specify What Information Will Be Sought: Clearly define the objectives and the specific data needed.
2. Determine Type of Questionnaire and Method of Administration: Decide on the overall structure (structured vs. unstructured, disguised vs. undisguised) and how it will be delivered (mail, phone, interview, electronic).
3. Determine Content of Individual Questions: For each question, assess its necessity, whether the answer can be inferred from others, if multiple questions are needed, and if the respondent possesses the required information and is willing to provide it.
4. Determine Form of Response to Each Question: Decide how respondents will answer (e.g., open-ended, fixed-alternative, rating scales).
5. Wording of Each Question: Craft clear, unbiased, and easy-to-understand questions.
6. Sequence of Questions: Arrange questions logically to guide the respondent.
7. Physical Characteristics of Questionnaire: Ensure the questionnaire is well-presented and professional.
8. Re-examine Steps 1-7 and Revise if Necessary: Review the entire questionnaire for clarity, consistency, and effectiveness.
9. Pre-test the Survey, Revise Where Needed: Conduct a pilot test to identify any issues and make necessary adjustments.

Key Steps and Considerations in Questionnaire Development

Step 2: Types of Questions and Methods of Administration

• Structure:
  • Structured: Pre-defined questions and response options.
  • Unstructured: Open-ended questions allowing for free-form responses.
• Disguise:
  • Disguised: The purpose of the survey is not immediately obvious.
  • Undisguised: The purpose of the survey is clear to the respondent.
• Administration Methods:
  • Mail
  • Phone
  • Interview (in-person)
  • Electronic

Step 3: Individual Question Content

• Necessity: Is the question essential for the research objectives?
• Redundancy: Can the answer be derived from other questions?
• Efficiency: Could one question replace multiple?
• Respondent Capability: Does the respondent have the necessary information?
• Respondent Willingness: Will the respondent provide the information?

Step 4: Form of Response

• Open-ended: Allows respondents to answer in their own words (e.g., "Can you name three sponsors of last night's football show?").
• Fixed-alternative: Provides pre-defined response options.
  • Dichotomous: Two options (e.g., Yes/No).
  • Multiple Choice: Several options.
  • Rating Scales: (e.g., Likert scales: Agree, Neutral, Disagree).

Weaknesses of Fixed-Alternative Questions:

• The list of options may not be exhaustive.
• Does not allow for elaboration by the respondent.
• Clarity on whether alternatives are mutually exclusive.
• Potential susceptibility to order bias.

Step 5: Wording of Each Question

Guidelines for Question Wording:

• Simplicity: Use simple words and construct straightforward questions.
• Clarity: Avoid ambiguous words and questions.
• Neutrality: Avoid leading questions that suggest a preferred answer.
• Explicitness: Avoid implicit alternatives (e.g., "Do you favor or oppose...?").
• Assumptions: Avoid implicit assumptions (e.g., assuming the respondent has a certain experience).
• Specificity: Avoid generalizations and estimates where precise answers are needed.
• Single Focus: Avoid double-barreled questions that ask about two things at once (e.g., "How was our price and service?").

Step 6: Sequence of Questions

Guidelines for Question Sequencing:

• Engaging Start: Begin with simple, interesting questions.
• Funnel Approach: Start with broad questions and narrow down to specific ones.
• Branching Logic: Carefully design questions that direct respondents to different parts of the questionnaire based on their answers.
• Demographics Last: Place classification information (e.g., age, income, political affiliation) towards the end.
• Sensitive Topics: Position difficult or sensitive questions near the end, after rapport has been established.

Step 7: Physical Characteristics of Questionnaire

• Professional Appearance: Ensure the questionnaire looks neat and professional to avoid appearing sloppy.

Step 8: Re-examination and Revision

• Thoroughly check and revise the questionnaire based on the previous steps.
• Always perform pre-testing to identify and rectify any issues.

Step 9: Pre-test the Survey, Revise Where Needed

• Conduct a pilot test with a small group similar to the target audience.
• Analyze the results of the pre-test to identify problems with clarity, wording, flow, or response options.
• Revise the questionnaire based on the pre-test findings before full deployment.

Handling Sensitive Questions: Randomized Response Model

• This model is used to increase the accuracy of responses to sensitive questions by introducing an element of chance.
• Respondents are randomly assigned to answer one of two questions:
  • Question A (Sensitive): A direct question about the sensitive topic (e.g., "Have you shoplifted?").
  • Question B (Innocuous): A non-sensitive question (e.g., "Were you born in January?").
• The respondent answers "Yes" or "No" to the question they are randomly assigned.
• The interviewer or data analyst knows the probability (p) of being assigned Question A and the proportion of "Yes" answers to Question B in the general population.
• This allows for the estimation of the true proportion of "Yes" answers to the sensitive question without knowing which question each individual answered.

Sample Cover Letter for Mail Questionnaire

A cover letter accompanying a mail questionnaire should include:

1. Personal Communication: Address the recipient personally if possible.
2. Importance of Research: Explain the project's purpose and significance.
3. Recipient's Importance: Justify why this specific recipient was chosen (e.g., part of a sample).
4. Recipient Benefit: Outline how the respondent or their community might benefit.
5. Time Commitment: State that completing the questionnaire will take a short time.
6. Ease of Completion: Emphasize that the questions are simple.
7. Return Envelope: Include a stamped, self-addressed envelope for easy return.
8. Confidentiality/Anonymity: Assure respondents that their answers are anonymous or confidential.
9. Offer of Results: Offer to send a report on the survey findings.
10. Appreciation: Express gratitude for their participation.
11. Sender's Credibility: Mention the sender or their organization to establish legitimacy.
12. Incentive: Describe any incentive offered (e.g., a dollar bill, as in the example).
13. Style, Format, and Appearance: The overall presentation should be professional.

Example Snippet from a Cover Letter:

"We are conducting a nationwide survey among executives and managers in the metalworking industry. The purpose of this research is to find out the opinion of experts like you on the advantages and disadvantages of using three new steel products... Your name appeared in a scientifically selected random sample. Your answers are very important to the accuracy of our research... It will take only a short time to answer the simple questions on the enclosed questionnaire and to return it in the stamped reply envelope. Of course, all answers are confidential... The enclosed dollar bill is just a token of appreciation."

---

---

---

Methods for Assessing Attitude

This document outlines various methods for assessing a person's attitude towards a specific object or idea. Attitudes are defined as a person's ideas, convictions, or liking regarding a particular subject. Measuring attitudes requires scales of measurement, and the document details several techniques, primarily focusing on self-report methods.

I. Direct Observation and Performance-Based Methods

• Observation of Behavior: Observing how individuals act in relevant situations.
• Performance of Objective Tasks: Assigning a specific task to an individual and evaluating their performance.
• Physiological Reactions: Measuring biological responses such as heart rate to gauge attitude.

II. Self-Report Techniques

These are the most common methods for attitude assessment.

A. Self-Report Scales: Equal-Appearing Intervals (Thurstone Scale)

This method involves developing a scale with statements that have pre-determined scale values representing different degrees of favorability.

• Step 1: Statement Generation: Create a large number of opinion-based statements about the subject (e.g., banks).
  • Example Statements: "Bank offers convenient hours," "Bank offers low interest rate on loans," "Bank has convenient location."
• Step 2: Judge Sorting: A sample of judges sorts these statements into categories based on their favorability towards the subject.
• Step 3: Scale Value Computation: For each statement, calculate its scale value (median) and dispersion (Q value, the interquartile range). The Q value indicates how consistently the statement was interpreted.
• Step 4: Statement Selection: Choose statements that span a range of favorability and have low dispersion (meaning they were interpreted consistently by judges).
• Step 5: Survey Placement: Randomly arrange the selected statements in a survey.
• Step 6: Respondent Agreement: Respondents indicate whether they Agree or Disagree with each statement.
• Attitude Score Calculation: The respondent's attitude score is the average of the scale scores for the statements they agree with. A score above the neutral point (e.g., 6 on a scale where 6 is neutral) indicates a positive attitude.

B. Self-Report Scales: Summated Ratings (Likert Method)

This method allows for the expression of the intensity of feeling.

• Step 1: Statement Development: Create numerous statements about the subject, classifying them a priori as favorable or unfavorable.
  • Example Statements: "The bank offers courteous service," "The bank has a convenient location."
• Step 2: Judge Agreement: Judges indicate their level of agreement or disagreement with each statement.
• Step 3: Scale Assignment: Assign numerical values to the levels of agreement (e.g., 1-5, where 1 is "disagree" and 5 is "agree").
  • Correction for Direction: For unfavorable statements, the scale is reversed during analysis (e.g., 5 for "disagree" and 1 for "agree") to ensure consistency.
• Step 4: Total Attitude Score Calculation: Calculate a total attitude score for each respondent by summing their scores across all statements (after direction correction).
• Step 5: Statement Consistency Check:
  • Assumption: A person with a favorable attitude will consistently agree with favorable statements and disagree with unfavorable ones.
  • Method 1 (Correlation): Compute the correlation between each statement's score and the total attitude score. Statements with high correlations are considered good.
  • Method 2 (Group Comparison): Compare the mean scores of statements for the top 25% and bottom 25% of respondents (based on total score). Statements where the top group's mean score is significantly higher than the bottom group's mean score are considered consistent. Statements with high differences between these group scores are preferred.
• Step 6: Survey Inclusion: Randomly include selected consistent statements (both favorable and unfavorable) in the survey.
• Respondent Score: The sum (or mean) of the scores across the selected statements (after direction correction) represents the respondent's attitude score.

III. Scales of Measurement

The document briefly touches upon different scales of measurement and their applications in assessing attitudes:

• Nominal Scales: Categorical data (e.g., Male-female, User-nonuser).
• Ordinal Scales: Ranked data (e.g., Preference for brands, Social class).
• Interval Scales: Data with equal intervals but no true zero (e.g., Temperature scale, Attitude toward brands).
• Ratio Scales: Data with equal intervals and a true zero (e.g., Units sold, Number of purchasers).

Examples are provided for assessing soft drink liking using these scales, including:

• Checking all that apply (Nominal).
• Indicating liking on a scale (Ordinal/Interval).
• Ranking drinks (Ordinal).
• Dividing points to represent liking (Ratio).

IV. Other Self-Report Scale Types

• Semantic Differential Scaling: Uses bipolar adjective scales (e.g., discourteous-courteous, inconvenient-convenient) to measure attitudes. Respondents rate a subject on these scales. Average responses are plotted to create profiles.
• Comparative Rating Scale: Asks respondents to divide a fixed number of points (e.g., 100) among different attributes to indicate their relative importance or preference.

V. Visual Scales

• Sad-to-Happy Faces: A visual scale often used with children (and adults) to express feelings or attitudes.

The document also includes mathematical formulas for covariance and correlation, which are relevant to statistical analysis in attitude research, though their direct application within the described scaling methods is not fully elaborated.

---

---

---

Sampling and Population Summary

This document outlines the fundamental concepts of sampling, its procedures, and various techniques used in statistical analysis. It explains how information is gathered from a portion of a larger group to make inferences about the entire group.

Main Idea: Sampling

• Target Population: The complete set of cases that meet specific criteria for a study.
• Sampling: The process of collecting data from a subset of the population to draw conclusions about the entire population.
• Reasons for Sampling:
  • A complete count (census) may not be feasible.
  • Sampling can sometimes be more accurate by minimizing staffing errors that might occur in a census.

6-Step Procedure for Drawing a Sample

1. Define the Target Population: Clearly identify the group you want to study.
2. Identify the Sampling Frame: Create a list or map of all accessible elements from which the sample will be drawn (e.g., a voter registration list).
3. Select a Sampling Procedure: Choose the method for selecting elements from the sampling frame.
4. Determine the Sample Size: Decide how many elements will be included in the sample.
5. Select the Sample Elements: Choose the specific individuals or items for the sample.
6. Collect the Data: Gather information from the selected sample elements.

Types of Sampling Procedures

• Fixed Sample: The sample size is determined before data collection begins.
• Sequential Sample: Data is collected in stages, and the sample size is adjusted based on interim results. If the initial sample doesn't provide a clear answer, more data is collected.

Classification of Sampling Techniques

1. Nonprobability Samples

• Characteristics:
  • Sampling error cannot be statistically estimated.
  • Elements are selected based on convenience or judgment, not random chance.
• Types:
  • Convenience Sampling: Selecting easily accessible participants (e.g., asking friends).
  • Judgement Sampling: Researchers select participants based on their expertise or judgment (e.g., choosing "swing" communities in an election).
  • Snowball Sampling: An initial group of participants identifies and refers other potential participants.
  • Quota Sampling: Ensuring the sample reflects certain characteristics of the population by setting quotas for different subgroups (e.g., ensuring specific majors are represented in a business school sample).

2. Probability Samples

• Characteristics:
  • Elements are selected using an objective process where the likelihood of each member being selected is known (though not necessarily equal).
  • Allows for the estimation of sampling error.
• Types:
  • Simple Random Sampling: Every element in the population has a known and equal chance of being selected.
    • A subset of members is chosen randomly from the population.
    • Used to estimate population parameters (like the mean) when a complete census is not possible.
  • Stratified Sampling:
    1. The population is divided into mutually exclusive and exhaustive subgroups (strata) based on specific criteria.
    2. A random sample is then drawn from each stratum.
    • Benefits: Produces more accurate estimates and allows for the analysis of specific subgroups.
  • Cluster Sampling:
    1. The population is divided into subgroups (clusters), often geographically based.
    2. A random sample of clusters is selected.
    3. One-stage: All elements within the selected clusters are included.
    4. Two-stage: A random sample of elements is selected from within the chosen clusters.
    • Trade-offs: Less statistically efficient (higher standard error) but often more economically efficient.
  • Systematic Sampling: A type of cluster sampling where every k-th element is selected after a random start.
  • Area Cluster Sampling: Cluster sampling based on geographical areas (e.g., zip codes).

Parameters and Distributions

• Parameters: Characteristics of a population (e.g., population mean $\mu$).
• Statistics: Characteristics of a sample (e.g., sample mean $\bar{X}$).
• Normal Distribution: A common probability distribution characterized by its bell shape. The formula provided describes the probability density function of a normal distribution.
  • A linear combination of normally and independently distributed random variables is also normally distributed.
• Central Limit Theorem (CLT):
  • States that if the sample size ($n$) is sufficiently large, the distribution of sample means ($\bar{X}$) will be approximately normal, regardless of the population's distribution.
  • The mean of the sample means will be close to the population mean ($\mu$).
  • The variance of the sample means is $\sigma^2/n$, where $\sigma^2$ is the population variance.
  • Formula: $\bar{X} \sim N(\mu, \sigma^2/n)$

Sample Mean and Standard Deviation

• Population Mean ($\mu$) and Variance ($\sigma^2$): Calculated using formulas involving all population elements.
• Sample Mean ($\bar{x}$) and Sample Variance ($s^2$): Calculated from a sample.
  • Note the difference in the denominator for variance calculation: $N$ for population variance vs. $n-1$ for sample variance (Bessel's correction).

Confidence Intervals

• A range of values within which the true population parameter is likely to lie, with a certain level of confidence.
• Case A: Population variance ($\sigma^2$) known: Uses the Z-distribution.
  • Formula: $\bar{X} \pm Z \sqrt{\sigma^2/n}$
• Case B: Population variance ($\sigma^2$) unknown: Uses the t-distribution (or Z if $n$ is very large).
  • Formula: $\bar{X} \pm t \sqrt{s^2/n}$ (where $s^2$ is the sample variance)
• Example Calculation: Demonstrates how to calculate a 95% confidence interval for the population mean using sample data, population standard deviation, and a Z-score. The result indicates a range within which the true population mean is estimated to fall.