Questionnaire
A formalized set of questions for obtaining information from respondents to generate the data necessary to accomplish the objectives of the research process
Questionnaire Objectives
1. Provide the necessary information to aid the decision- making
2. Translate needed information into a set of specific questions that respondents can and will answer
3. Should 'speak' to the respondent
4. Should facilitate data processing. Should be easy
Questionnaire Design Process
1. Planning what to measure
2. Formatting the questionnaire
3. Question wording
4. Sequencing and layout decisions
5. Protesting and correcting problems
Planning what to measure - Step 1 DP
Step 1 in Design process. What are your research objectives? What information is needed to address the research objectives?
What information is needed about Buying process? (Step 1 in design process)
Brand awareness & knowledge, attribute importance, brand preference, choice behavior (past, intention)
What information is needed for segmentation variables? (Step 1 in design process)
Consumer lifestyle & demographics
Formatting the Questionnaire - Step 2 DP
1. Determine the content of each question
2. Determine the form of each question
3. Response order Bias
Determine the content of each question - Step 2 DP
1. How many questions - capture the needed data using as few questions as possible
2. A single piece of information from one question
3. Question should be answerable - ability to answer and willingness to answer
Determine the form of each question - Step 2 DP
open ended questions - versatility, difficult to code, analyze, interpret the answers, good for explanatory research, gives rich array of information (probing is possible)
close ended questions - easy to answer, code, analyze, and interpret, not good for
How to get rid of response order bias?
split- ballot techniques which are response options are reordered or randomized to create different versions of the survey
Question Wording - Step 3 DP
avoid complexity- use simple words, avoid ambiguity, avoid double-barreled questions, avoid leading/loading questions, avoid burdensome questions
Burdensome questions
questions that depend too much on an excellent memory
ex: do you recall any commercials during Super Bowl 2017?
should be - "do you recall a Budweiser commercial during Super Bowl 2017?
Double barred questions
do you think Coca Cola is a tasty and a refreshing soft drink?" - need to split this question into two questions
Leading questions
clearly provide clues to the answer
loaded questions
emotionally charged or suggest socially desirable answers
Sequencing and Layout Decisions - Step 4 DP
Use funnel approach, question order bias
What is funnel approach?
1. start broad and progressively narrow down the scope
2. establish rapport - easy to answer general questions first
3. Tough and important questions in the middle
4. sensitive/demographic questions at the end
Pretesting and correcting problems - Step 4 DP
pretest is the most inexpensive insurance you can buy to ensure the success of your questionnaire and research project. 2 pretests are recommended:
1. pilot run on a small samples ( 25 is "rule of thumb")
2. personal interview pretest
What is sampling
The process of obtaining information from a subset ( a sample ) of a larger group ( the universe or population )
What is an inference?
an inference about the population are based upon observations of a selected portion of the population
Reasons for sampling
1. less costly than census
2. less time consuming than census
3. May be more accurate than census if population is difficult to reach
4. Sampling can produce very accurate estimates about population
What is the sampling process?
1. Determine the sampling frame
2. Determine the sampling procedure
3. Determine sample size (Ch. 15)
Determine the sample frame
a list of all or random selection of population member used to obtain a sample
commonly used sample frames include:
customer database
telephone directories
list developed by data compliers
Determine the sampling procedure
non probability samples/ probability samples
Non-probability sample
1. a sample that relies on personal judgement in process of selecting population
2. impossible to assess the degree of sampling error
3. cannot evaluate how much it resembles the population
4. techniques include convenience samples, judgement samples, sno
Probability samples
1. A sample in which each population member has a known, nonzero chance of being included in the sample
2. The amount of sampling error can be estimated and the sample statistics be used to infer about the population
3. Techniques include simple random, s
Simple Random Sample
1. Each population member has a known and equal probability of selection
2. May over/under represent small subgroups - depends on the amount of variation in each group
stratified random sample
1. Divide population into homogeneous groups. Then, randomly sample from each group.
2. Subgroups are homogenous within but heterogenous between with respect to key variables
3. will not miss important subgroups
4. estimates will be more accurate than ran
Systematic Sampling
Select every k-th element on a list
accuracy compared to random sample
1. web sampling frame us arranged in a random order
2. when sampling frame is arranged in a monotonic order
3. when sampling frame is arranged in a cyclical order
Cluster Sampling
Two step process:
1. Draw a random sample of clusters
2. Do a census within each
cost effective compared to simple random sampling
Determine the sample size
sample size is a financial issue, statistical issue, and it is a managerial issue
Sampling error
1. Error due to the act of sampling
2. If the researcher uses probability sampling, the magnitude of this error is known
Statistical theory of sample size
1. use statistical theory to guarantee a desired level of accuracy
2. Sample size is the number of respondents required to produce a certain level of accuracy
3. Optimal sample size calculations ONLY make sense when probability sampling is undertaken
Parameter
population mean and standard deviation u and o
Statistic
sample mean and standard deviation X and s
sampling distribution
1. frequency distribution (histogram) of sample elements
2. is generally not so smooth but look like the population distribution
interval estimation for population mean (confidence interval)
because the sample mean X varies from sample to sample, we want to construct interval estimate around the sample mean that includes population mean.
Formula for sample confidence interval
���� ?�?�����?������/?������
X + sampling error (E)
What is data coding?
The process of transforming raw data into numbers that can be utilized for analysis
Strategy for Data Analysis
1. Tabulation
2. Cross- Tabulation
Tabulation
Counting the number of cases that fall into the various categories
Goals:
1. determine frequency distribution
2. Calculate descriptive statistics
Cross-tabulation
Access if any association is present between two nominal variables. Try to see whether one variable (independent variable or predictor variable) has an influence on another variable (the dependent variable or outcome variable). Hypothesis testing helps us
Null Hypothesis Ho
1. the hypothesis of status quo, no difference, no effect
2. The hypothesis that a proposed result is not true for the population
3. we attempt to reject the null hypothesis in favor of alternative hypothesis
Alternative hypothesis Ha
1. The hypothesis that a proposed result is true for the population
2. Never has equal sign
3. Direction of tests
Example for hypothesis test
The manager asks... "Are you sure that we have unequal number of males and females in our customer base?'
Ho: The number of females is equal to that of males
Ha: The number of females is different from that of males
Type 1 error
rejection of null hypothesis when, in fact, it is true
Type 2 error
Failure to reject the null hypothesis when, in fact, it is false
Rejection region
1. Reject Ho if test statistic falls into rejection region and then that means Ha is TRUE.
2. Fail to reject Ho if test statistic DOES NOT fall into rejection region and that means Ha is FALSE
Testing frequency distributions
one group - chi square of goodness of fit
two groups - chi square tests of statistical independence = observed - expected ^ 2 / expected
X^2 used for frequency distributions
Hypothesis test of the mean
1. Z distribution if population standard deviation is known
2. T distribution if population standard deviation is NOT known
Proportion testing
1. Z distribution if population standard deviation is known
2. T distribution is population standard deviation is NOT known
Comparing multiple sample means
Two or more = F tests
ANOVA tests used for
1. when there are only two groups :
unrelated (independent) - t test
related (dependent) - t test
are used to determine if the mean score on the dependent variable for one group is significantly different from that of the second group
2. if there are more
When can t tests, z tests, and anova tests be used?
When dependent variable is continuous
What is chi squared test appropriate for?
When dependent variables are categorical
Likertz scale
Considered continuous variable and anova is used to analyze differences in means across groups
Correlation (R)
correlation coefficient captures both direction and strength of the linear relationship. -1<r<1
Underlying assumption of Correlation
the variables are continuous variables measured by the interval scales
properties of correlation
1. independent of sample size and unit of measurement
2. -1<r<1
3. r<0 = x and y are negatively correlated
4. r > 0 = x and y are positively correlated
5. r=0 = x and y are uncorrelated
6. as r is closer to either -1 or 1 = stronger correlation
Regression Analysis
a statistical technique to relate the dependent variable to one or more independent variables
dependent variable - response variable
(Y) = the variable of interest
independent variables - predictor variables
(X's) - the variables influencing Y
Simple regression
there is only one independent variable
Multiple regression
there are multiple independent variables