The descriptions of the types of items for your survey listed here in pdf and video format provide the overall structure of your survey, as well as an example of how to write your items. This structure serves as a general prototype for many marketing research surveys.
We use qualtrics to enter and administer your survey. Do not use the trial version of Qualtrics, rather use the official PSU Qualtrics link to access. Of course you must first design and write your survey, so focus on that task before learning how to use Qualtrics to implement.
This material is from our textbook, Basic Marketing Research, Vol 2, Building Your Survey
You need to choose the type of scale to define the possible responses to each of your survey items. Three basic scale types typically used for attribute and outcome items are Likert, Semantic Differential and Stapel, described on p55,56. Many examples of these scales follow on the successive pages. Chapter 8, The Survey Library, contains even more examples. You are welcome to use any of these on your own project.
Contact me this week sometime to show the progress of your survey, to provide feedback before you administer, and to answer any items you may have. Follow these two guidelines when forwarding your items.
Two ways to contact me:
One of the most important concepts in statistics and data analysis is the relation between two variables. In our study of the t-test for the mean difference, we saw how to analyze the relationship between a categorical variable (with two groups) and a numerical response variable. Here we examine the relationship between two categorical variables and between two continuous variables.
For a cross-tab table and associated bar graph of two categorical variables, X and Y, use BarChart(X,Y). For a scatter plot of categorical or continuous variables, ScatterPlot(X,Y). For just an analysis of the corresponding correlation coefficient, without the scatter plot, Correlation(X,Y).
Sec 7.2: a and b only, new videos as of Wed afternoon
Sec 8.1: a, b and c only
| 7.2 | Cross-Tab Table | pdf [21] |
7.2a [11:44] |
7.2b [5:00] |
|
| 8.1 | Correlation | pdf [48] |
8.1a [8:57] |
8.1b [8:16] |
8.1c (called d in the video) [9:35] |
Homework #7
Solutions #7
Regression analysis is the primary statistical technique that we use to explain the values of one variable, the response variable, from the predictor variables. In our case we will use regression analysis to attempt to explain customer satisfaction, or a related outcome variable, in terms of the product attributes. For this week we just have a single predictor variable.
Do regression with Regression(Y ~ X) or reg(Y ~ X), which regresses response variable Y on predictor variable X. Use reg.brief for this class, to provide simpler output (basically equivalent to that provided by Excel).
| 8.2 | Regression | pdf [38] |
8.2a [19:18] |
8.2b [16:01] |
8.2c [15:03] |
| 8.3 | Model Fit | pdf [a only] [24] |
8.3a1 [17:38] |
8.3a2 [23:00] |
Homework #8
Note: This homework is for Track ABC students only. Track C students continue to work on their projects with no new assigned homework.
Solutions #8
Multiple regression, multiple predictor variables in a regression equation, is the endpoint of our exploration of data analysis techniques. This technique, and the t-test for the difference between means, are the two most widely used inferential data analysis procedures in marketing research (descriptive histograms and bar charts are always done, for every analysis). The key advantage of multiple regression is that it provides the concept of statistical control. The level of control is not as sweeping as that provided by experimental control introduced in Week 6, but any level of control is better than none. In your readings for this week, understand the concept of control, and the benefits it provides regarding the analysis of causality.
Be aware that the interpretation of the slope coefficient from a multiple regression extends the interpretation from that of a slope coefficient from a regression with only one predictor variable. The extension is that of statistical control, that all the direct contribution of all other predictor variables in the model has been held constant. Always make sure to add that qualifying phrase to the interpretation of these slope coefficients from the model. For that reason slope coefficients from a multiple regression model are called partial slope coefficients.
Do multiple regression with Regression(Y ~ X1 + X2), which regresses response variable Y on predictor variables X1 and X2. Include as many predictors as needed. Multiple regression is the primary analysis you will do for your project, and, with the t-test for comparing group means, forms the foundation of statistical analysis of data.
| 9.1 | Multiple Regression [pdf] |
| 9.1a | Multiple Regression Model [10:02] |
| 9.1b | Statistical Control [13:08] |
| 9.1c-I | Input Into The Regression Analysis [9:02] |
| 9.1c-II | Interpreting The Model Estimates [10:07] |
| 9.1c-III | Assessing Model Fit [8:15] |
The difference between this homework and that of Week 8 is that here we have multiple regression, so the interpretations involve controlling for the other predictor variables in the model. Also there is a new question, k, which involves choosing just those predictor variables that seem to work the best.
Homework #9
Note: This homework is for Track ABC students only. Track C students continue to work on their projects with no new assigned homework.
Solutions #9
Project Week. This week is dedicated to answering questions about doing your project. No new material. ☺