Chapter 3 Secondary Data, Literature Reviews, and Hypotheses McGraw-Hill/Irwin Copyright © 2013 by The McGraw-Hill Companies, Inc. All rights reserved. Nature, Scope, and Role of Secondary Data • Secondary data: Data not gathered for the immediate study at hand but for some other purpose – Internal secondary data: Data collected by the individual company for accounting purposes or marketing activity reports – External secondary data: Data collected by outside agencies such as the federal government, web properties, trade associations, or periodicals 3-2 Nature, Scope, and Role of Secondary Data • Secondary data research has gained substantial importance in marketing research with: – Increased emphasis on business and competitive intelligence – Ever-increasing availability of information from online / electronic sources • Used to examine marketing problems because of the relative speed and cost-effectiveness of obtaining the data 3-3 Some Sources of Internal Secondary Data • • • • • • Sales invoices Sales activity reports Online registrations Warranty cards Clickstream Data Online Analytics (i.e. Google Analytics, Tweetdeck, Facebook Insights, etc.) • Customer letters/ comments / e-mails • Social Media data (“found data”) • Past studies • Store Audits 3-4 Store Audits • Examination of how much of a particular product or brand has sold at retail – Product sales in relation to competition – Effectiveness of shelf space/POP displays – Sales at various price points – Effectiveness of POS and non-POS offers – Sales by store type, location, channel, etc. 3-5 External Secondary Data Sources • Government Studies & Reports (i.e. Census Data) • Business sources – Editors and Publishers Market Guide – Source Book of Demographics and Buying Power for Every Zip Code in the USA – IDS and Gartner Group (technology) – Neilsen/IRI (consumer products & media) 3-6 Consumer Panels • Benefits – Lower cost than other methods – Rapid availability and timeliness – Can track trends over time with same (similar) sample – High level of specificity • Risks – Sampling error (low minority representation) – Response bias – “inbreeding” effects 3-7 NPD Group 3-8 Neilsen Panels 3-9 What is a Literature Review? • A comprehensive examination of available information related to your research topic – Can help clarify and define the research problem and research questions – Can suggest research hypotheses to investigate – Can identify scales and research methodologies to use in your study – May provide an answer to your research question(s)! 3-10 Google Scholar 3-11 Criteria Used to Evaluate Secondary Data Sources • Purpose (i.e. why collected?) • Accuracy / Timeliness • Consistency over time and across units studied • Credibility • Methodology used • Bias in findings 3-12 Developing a Conceptual Model • Literature reviews can help you conceptualize a useful and empirically accurate model of reality • Elements required to conceptualize and test a model: – Variables – Constructs – Relationships – Hypotheses 3-13 Variable • An observable item that is able to be directly measured EXAMPLES: Gender, Age, Location, Intention to Purchase Construct • An unobservable concept that is measured by a group of related variables EXAMPLES: Brand Loyalty, Intelligence, Satisfaction Relationships • Associations between two or more variables or constructs Independent Variable • The variable or construct that predicts or explains the outcome (dependent) variable of interest Dependent Variable • The outcome variable or construct researchers are seeking to explain 3-14 Hypotheses and Models • Hypotheses: Theoretical statements about relationships between variables or constructs • Two hypotheses formally stated: – Hypothesis 1: Higher spending on advertising leads to higher sales. – Hypothesis 2: Higher prices lead to lower sales. • A set of related hypotheses form a Conceptual Model • Here we are trying to formulate a model of sales 3-15 Relationship Types • Positive relationship: An association between two variables in which they both increase or decrease together • Negative relationship: An association between two variables in which one increases while the other decreases (and vice-versa) • Non-Directional relationship: Does not specify a covariance pattern (i.e. no directional association specified) 3-16 Characteristics of Good Hypotheses • Follow from research questions • Clearly and simply stated • Must be testable (i.e. falsifiable) • Must lend themselves to variable identification and “operationalization” of variables • Data sources must be available to populate the operationalized model 3-17 Model Conceptualization • Development of a model that outlines all relevant variables/constructs and the hypothesized relationships between them • Involves: • Identifying the independent and dependent variables for your research • Specifying relationships between the variables – Preparing a diagram or other conceptual model that represents the relationships you will study – Developing or identifying theories that justify those relationships – Specifying “Boundary Conditions” for the relationships, if any – Identifying any control variables (pre-hoc) or confounding variables (post-hoc) that could affect results 3-18 A Model of New Technology Adoption 3-19 Boundary Conditions, Control Variables and Confounding Variables • Boundary Conditions: Conditions under which your hypothesized relationships and research conclusions will hold / not hold • Control Variables: Independent Variables (IVs) that affect the Dependent Variable (DV) but which are often not the main focus of the study • Confounding Variables: IVs that affect the DV which are not part of the conceptual model, not “controlled for” and not observable – BIG TROUBLE! 3-20 Hypothesis Testing • Hypothesis: An empirically testable though yet unproven statement developed in order to explain phenomena – Null hypothesis: No relationship between two variables/constructs. – Alternative hypothesis: Suggests that two variables/constructs are somehow related – Non-directional vs. Directional 3-21 Hypothesis Testing • A null or alternate hypothesis refers to a population parameter, not a sample statistic – Parameter: The true value of a variable – Sample statistic: The value of a variable that is estimated from a sample 3-22 Examples - Null Hypotheses • There is no significant difference between the preferences toward specific banking method exhibited by white-collar customers and blue-collar customers. • No significant differences exist in requests for specific medical treatments from emergency walk-in clinics between users and nonusers of annual preventive maintenance health care programs. 3-23 Examples – Alternate Hypotheses, Non-directional • There is a significant difference in satisfaction levels reported by Safeway and Lucky shoppers. • Significant differences exist between males and females in the number of hours spent online. 3-24 Examples – Alternate Hypotheses, Directional • We expect higher satisfaction levels to be reported by Safeway shoppers than Lucky shoppers. • We expect to find that males spend significantly more hours online than females. 3-25 Examples – Alternate Hypotheses (positive relationship) • More studying is related to higher GPAs. • Friendlier salespeople generate higher sales revenues. • More frequent advertising is associated with higher sales. 3-26 Examples – Alternate Hypotheses (negative relationship) • Students with higher GPAs consume less alcohol than those with lower GPAs. • The more pressure to close sales perceived by salespeople, the fewer the number of follow up, “relationship-building” sales calls made. 3-27 ACTIVITY: Formulating Research Objectives and Hypotheses • Develop a simple research question (example: What factors drive the adoption of a new technology?) • Formulate a simple model for your research question. Specify the following: – Independent Variables (several) vs. Dependent Variables (one) – The + or – relationship between each IV and the DV. – Theory behind the relationships – Any boundary conditions for the model – Any control variables/factors needed • Follow up: How to measure your IVs/DVs? 3-28