Market Research 412 Umass Final Exam

Term
1 / 33
Open-Ended Questions
Click the card to flip 👆
Terms in this set (33)
Open-Ended Questions
-Pose some topic and ask respondents to answer in their own words
-Used in exploratory research
Fixed Alternative Questions
-Most often used type of question
-Questions in which respondents are given specific, limited-alternative responses and asked to choose the one closest to their own viewpoint
-Used when researchers have some alternative in mind
Types of Fixed Alternative Questions
-Single-dichotomy (Y/N)
-Multiple choice alternative
Types of Multiple Choice Alternative Questions
-Determinant-choice question: Choose one response from multiple alternatives
-Frequency-determination question: Asks for an answer about general frequency of occurrence (always, sometimes, never)
-Checklist question: Provide multiple answers to a single question by checking off items
-Likert scales (1-5 or 1-7 ordinal scale) and semantic differential (measures connotative meaning, on a scale of low to high, clean to dirty, bright to dark, etc)
Avoiding Mistakes in Phrasing Questions
-Avoid complexity: the simpler the language, the better
-Avoid leading and loaded questions
-Avoid ambiguity: be as specific as possible
-Avoid double-barreled items
-Avoid making assumptions
-Avoid burdensome questions that may tax the respondents' memory
Sampling Approaches (2)
Probability and Non-Probability
Probability Sampling
-Known non-zero probability for every element
-Simple random sample
-Systematic
-Stratified sampling
-Cluster sampling
Non-Probability Sampling
-Probability for selecting any particular member is unknown
-Convenience
-Judgement
-Quota
-Snowball
Simple Random Sample
Each population member and each possible sample has equal probability of being selected (drawing names from a hat)
Systematic Sampling
Involves systematically spreading the sample through the list of population members
-A starting point is randomly selected and then every nth number on the list is selected
Stratified Sampling-The chosen sample is forced to contain units from each of the segments or strata of the population -The variable chosen for stratification should increase homogeneity within each stratum and increase the heterogeneity between strataProportionate and Disproportionate Stratified Sample-Proportionate: number of objects/sampling units chosen from each group is proportional to number in population -Disproportionate: sample size in each group is not proportional to the respective group sizes, used when multiple groups are compared and respective group sizes are smallCluster SamplingInvolves dividing population into subgroups (Universities --> majors --> students)Quota-Minimum number from each specified subgroup in the population -Non-probability "stratified" sampling -Include people who are easily found, willing to be interviewed -Introduce biasSnowball-Form of judgmental sampling (non-probability) -Appropriate when reaching small, specialized populations -Each respondent, after being interviewed, is asked to identify one or more others in the appropriate groupPros and Cons of Probability/Nonprobability Sampling-Nonprobability: reduces costs and trouble of developing sampling frame but results can contain hidden biases/uncertainties -Probability: higher costs and more difficult, but results are more accurateDescriptive AnalysisThe elementary transformation of raw data in a way that describes the basic characteristics such as central tendency, distribution, and variabilityTabulationThe orderly arrangement of data related to one categorical variable in a table or other summary format showing the number of responses to each response category -Would use frequency tableCross-Tabulation-Arrangement of data related to the relationships among multiple categorical variables -Contingency table -Marginals: row and column totals in a contingency table, which are shown in its marginsData Transformation-Process of changing the data from their original form to a format suitable for performing a data analysis addressing research objectivesCommon Ways of Data Transformation-Creating summated scales - avg or sum of scale items -Combining adjacent categories of a variable (reducing categories makes statistics more impactful and improves statistical analysis for cross-tabulation)(useful when bimodal distribution exists)Hypothesis Testing Procedure1) State hypothesis (derived from research objectives, null and alternative) 2) Choose your significance level (.05 means 95% confidence interval, .01 means 99% confidence interval 3) Choose an appropriate statistical test (calculate the p-value [sig. in SPSS] from your sample) 4) Compare p-value to your chosen significance level (if p-value > significance level, the null hypothesis is supported. if p-value < significance level, the null hypothesis is rejected and the alternative is supported- 95% confident that hypothesis is supported which implies differences)Hypothesis-An unproven proposition/supposition that tentatively explains certain facts of phenomena -Null and alternativeNull HypothesisThe DULL hypothesis - baseline or reference for a statistical test (ie there is no difference between x and y)Alternative HypothesisA change in dependent variable was caused by independent variable (scientific guess or answer to research objective)Significance LevelA critical probability associated with a statistical hypothesis test, indicates how likely it is that an inference supporting a difference between an observed value and some statistical expectation is true -Traditionally acceptable significance levels are .05 or .01 (95% or 99% confidence intervals)p-valueProbability value, or observed or computed significance level from your sample -If the p-value resulting from a statistical test is less than the pre-specified significance level (.05 or .01) the results support a hypothesis implying differencesWhat would a hypothesis test?Difference among groups or relationship among variablesChoosing appropriate test of difference-Test of differences: an investigation of a hypothesis that 2+ groups differ with respect to measures on a variable -2 questions to determine appropriate statistical test of difference: 1) how many independent variables and dependent variables are involved in the analysis? 2) what is the scale level of the independent and dependent variables involved in the analysis?Data/Measurement Scales-Nominal: labeling variables (mutually exclusive and not numerically significant) -Ordinal: order of the values is what's important but differences between each is not known (unhappy-very happy, unsatisfied-satisfied) -Interval: numerical scales in which we know the order and the exact difference between each value (time, temperature) -Ratio: numerical scale in which we know the order, the exact difference between values, and there is an absolute zero (height, weight, numbers) so you have the ability to calculate ratiosRule of thumb to choose statistical tests-Nominal and ordinal scales are categorical variables. Interval and ratio scales are continuous variables -If the independent variable is categorical (nominal and ordinal) and the dependent variable is categorical, use the chi-square test. If the independent variable is categorical and the dependent variable is continuous (interval and ratio), use t-test. -If the independent variable is continuous, use regression. *Interval scale with less than 5 categories can be treated as categorical variableInterpreting SPSS Output - Chi-SquarePearson Chi-Square x Asymp. Sig. (significance) (2-sided) = .xxx .xxx MUST be less than .05 (or otherwise specified confidence interval) to have 95% confidence in the resultInterpretation of SPSS Output - T-testEqual variances assumed x Sig. (2-tailed) = .xxx .xxx must be LESS than .05 or other confidence interval to be 95% confident in the result