L5.pdf

Marketing Research Lecture 5: Data Collection I – Measurements PROFESSOR JIA LIU  From Last Class ▪ What are the objectives of doing descriptive research? Problem formulation Research design ▪ Wha...

Marketing Research Lecture 5: Data Collection I – Measurements PROFESSOR JIA LIU  From Last Class ▪ What are the objectives of doing descriptive research? Problem formulation Research design ▪ What is the main differences between descriptive and causal research? Data collection Analysis and interpretation Reporting results HKUST, MARKETING RESEARCH  Outline of Today Problem formulation Data collection Research design Measurements Data collection Questionnaire Analysis and interpretation Sampling Reporting results HKUST, MARKETING RESEARCH  Measurements Rules for assigning numbers to objects in such a way as to represent quantities of attributes Types Rules Applications Nominal Objects are either identical or Description different Classification Ordinal Objects are greater or smaller Rankings Interval Intervals between objects are equal Ratings (arbitrarily zero) Ratio Meaningful zero, for comparison of Amounts absolute magnitude HKUST, MARKETING RESEARCH  Instagram Red Tik Tok WeChat Nominal Scale Which of the following social media platforms do you use now? Check all that apply. Used for identification Numbers have no actual meaning HKUST, MARKETING RESEARCH  facebook Instagram Tinder Red Tik Tok WeChat Sample Data Collected from Nominal Scale Question Subject facebook Tinder Tik Tok WeChat 1 1 0 1 1 2 1 0 0 0 3 0 1 1 1 4 1 0 0 1 5 1 1 0 0 HKUST, MARKETING RESEARCH  facebook Instagram Tinder Red Tik Tok WeChat Ordinal Scale Please rank the following platforms according to the time you spend on the site (most = 1; least = 4) Used for rank ordering Numbers reflect relative standing The order matters but the differences do not HKUST, MARKETING RESEARCH  facebook Instagram Tinder Red Tik Tok WeChat Sample Data Collected from Ordinal Scale Question Subject facebook Tinder Tik Tok WeChat 1 1 4 2 2 2 2 3 1 4 3 4 2 3 1 4 3 4 1 2 5 1 4 2 3 HKUST, MARKETING RESEARCH  facebook Instagram Tinder Red Tik Tok WeChat Interval Scale Please indicate your liking for using each of these platforms (Extremely unfavorable = 1; Extremely favorable = 5) Used for relative comparison Numbers allow the comparison of the size of the differences among and between members HKUST, MARKETING RESEARCH  facebook Instagram Tinder Red Tik Tok WeChat Sample Data Collected from Interval Scale Question Subject facebook Tinder Tik Tok WeChat 1 5 1 5 3 2 4 3 5 4 3 4 4 4 4 4 3 2 4 1 5 4 4 3 2 HKUST, MARKETING RESEARCH  facebook Instagram Tinder Red Tik Tok WeChat Ratio Scale Over the past month, on average how many hours per week did you spend on each social media platform? Used to capture absolute magnitude Comparison of the absolute magnitude of the numbers is legitimate HKUST, MARKETING RESEARCH  facebook Instagram Tinder Red Tik Tok WeChat Sample Data Collected from Ratio Scale Question Subject facebook Tinder Tik Tok WeChat 1 3 0 3 8 2 2 1 1 0 3 10 4 5 3 4 4 2 3 5 5 3 1 7 1 HKUST, MARKETING RESEARCH  Measurement Scales of measurement Nominal (used for identification) Ordinal (used for rank ordering) Higher level of Interval (used for relative comparison) measurement Ratio (absolute magnitude) The type of scales dictates the analysis of the data Use higher level of measurement whenever possible HKUST, MARKETING RESEARCH  MEASURING UNOBSERVED CONCEPTS HKUST, MARKETING RESEARCH  Measuring Unobserved Concepts Attitudes Motivations Personality Purchase intentions Self-report is the most common approach Itemize-ratings scale (most popular) Graphic-ratings scales Comparative-rating scale HKUST, MARKETING RESEARCH  Itemized Rating Scale Summated-ratings (Likert) scale HKUST, MARKETING RESEARCH  Itemized Rating Scale Sematic-differential scale HKUST, MARKETING RESEARCH  Itemized Rating Scale Other forms Extremely Extremely Unfavorable Favorable 1 2 3 4 5 HKUST, MARKETING RESEARCH  Graphical Rating Scale Please evaluate each attribute, in terms of how important the attribute is to you, by placing an “X” at the position on the horizontal bar that most reflects your feelings. Not Very Important Important Courteous service X Convenient location X Convenient hours X Low-interest-rate loans X HKUST, MARKETING RESEARCH  Comparative Rating Scale 1. Paired comparison: Which bank do you prefer? 2. Rank order: Please order the following banks according to your preference (1 = highest preference, 4= lowest preference) 3. Constant sum: Please rate the following banks on their preference (divide 100 points) ……… ……… Total: 100 points HKUST, MARKETING RESEARCH  Measuring Unobservable Concepts Number of items in a scale Single-item (global measure) Multiple-item (composite measure) HKUST, MARKETING RESEARCH  Measuring Unobservable Concepts Number of items in a scale Single-item (global measure) Multiple-item (composite measure) Number of scale positions 5-9 response categories Odd vs. even number of categories Very Very Unfavorable Favorable 1 2 3 4 5 Very Very Unfavorable Favorable 1 2 3 4 5 6 HKUST, MARKETING RESEARCH  VALIDITY AND RELIABILITY OF MEASURES HKUST, MARKETING RESEARCH  Measurement Errors Measuring unobservable constructs always involves errors Observed Systematic Random Truth response error error Sources: Sources: - Individual characteristics - Temporary aspects of the person - Measurement procedure - Or measurement situation HKUST, MARKETING RESEARCH  Validity The extent to which the measurement instrument indeed measures the construct that we intend to measure Observe System. Random d Truth error error response Decreasing systematic error can improve validity HKUST, MARKETING RESEARCH  Reliability The extent to which the measurement instrument gives the same results when the measurement is repeated (=consistency in measurement) Observed System. Random Truth response error error Decreasing random error can improve reliability HKUST, MARKETING RESEARCH  Unreliable & Unvalid Unreliable, but valid Reliable, but unvalid Reliable & Valid HKUST, MARKETING RESEARCH  How to Select Reliable and Valid Measures 1. Use existing scales (literature search) Book: Marketing Scales Handbook (available on Canvas) 2. Consider multi-item scales Example: General attitude toward a product/brand ◦ Good/bad ◦ Like/dislike ◦ Pleasant/unpleasant ◦ High quality/poor quality ◦ Satisfactory/dissatisfactory ◦ Favorable/unfavorable ◦ … HKUST, MARKETING RESEARCH  How to Select Reliable and Valid Measures 3. Use “common sense” ◦ If you cannot find existing scales, or existing scales do not make sense, come up with your own questions ◦ Discuss and check your choices with other experts 4. More rigorous procedure (new scales) i. Specify domain of construct (i.e. define the construct) ii. Generate sample of items (i.e. generate questions) iii. Collect data (i.e. survey with questions of step ii) iv. Purify scale (i.e. compute reliability) v. Assess validity HKUST, MARKETING RESEARCH  Summary Measurement and scales Four types of scales of measurement Measuring unobservable concepts Three different self-report scales Interpret relatively Validity and reliability of measures Components of measurement errors Validity and reliability HKUST, MARKETING RESEARCH  Checklist 1. We will help finalize all the groups by the end of today 2. Consultation over Zoom this Friday There will be breakup rooms for all groups Arrive your session on time Each session is 10 minutes, be prepared Sign up one, if you haven’t 3. Next class Questionnaire Design Readings: Chapter 13 HKUST, MARKETING RESEARCH