MKTG3001 Chap.8 Preparing Survey Data and Basic Analysis
Preparing Survey Data and Basic Analysis
Week 8 turns a raw survey export into something you can analyse and introduces the first of three analysis levels. You learn to recognise the six closed-ended response types (single-response, multiple-response, demographic, Likert, semantic differential and comparative), then place every technique the course teaches into basic, intermediate or advanced analysis.
Data validation removes incomplete responses, metadata and screening answers while keeping the Respondent ID. One-way tabulation reports a count and percentage for each option of a single question, with the twist that select-all items need not total 100%. Cross-tabulation sets a single-response question against a demographic one to compare groups.
Finally, data coding turns answers into numbers, the dataset is laid out in Excel with one row per respondent, and the mean summarises a coded Likert-type item. Expect exam questions asking whether a proposed method suits a given question.
What this chapter covers
- 01
Closed-ended response types: predefined categories and measurement scales
- 02
Basic, intermediate and advanced levels of quantitative analysis
- 03
Data validation: incomplete responses and unnecessary columns
- 04
One-way tabulation across five question types
- 05
Cross-tabulation of a single-response and a demographic question
- 06
Data coding, Excel layout and the mean of a Likert-type item
Worked example · free
Validating bike-share survey rows and building a one-way tabulation with percentages
- 2Validation drops the 11 incomplete rows, so 171 minus 11 leaves 160 usable riders; the name, email and start-time columns and both screening answers are deleted, while the Respondent ID column is kept so every rider stays distinguishable.
- 3With 160 as the base, each share is count / 160 × 100: Commuting 64 / 160 = 40%, Exercise 36 / 160 = 22.5%, Avoiding parking 28 / 160 = 17.5%, Leisure rides 20 / 160 = 12.5% and Cost savings 12 / 160 = 7.5%, which together make 100% because the question allowed one answer.
- 2Commuting is the dominant motive, named by two riders in every five and almost twice as often as exercise, while cost savings ranks last; the operator could therefore promote the scheme chiefly as a commuting option, noting that this result describes the sample rather than testing any group difference.
Key terms
- Data validation
- A completeness and suitability check on collected survey data that removes incomplete responses, metadata columns and screening answers while retaining each Respondent ID.
- One-way tabulation
- A count of how often each response option to one question was chosen, reported as a frequency and a percentage for every option.
- Cross-tabulation
- A table of two categorical variables, typically a single-response question against a demographic question, that reports frequencies and percentages within each group.
- Data coding
- The assignment of numerical values to survey answers so that statistical analysis can use them, with codes acting as labels rather than quantities.
Preparing Survey Data and Basic Analysis FAQ
Why don't the percentages add up to 100% on a select-all-that-apply question?
Each respondent can tick several options, so the same person is counted under more than one category. Every percentage still uses the number of respondents as its base, which means the column can total well over 100%. Compare options by how many people selected each one rather than reading the shares as slices of a single whole.
Can I calculate a mean for a single-response question once it has been coded?
No. Codes for categories such as store type or favourite brand only name the groups; brand 3 is not three times brand 1. An average of those numbers changes whenever someone renumbers the list, so it carries no meaning. Use frequencies and percentages instead, and keep the mean for ordered scale items like Likert-type questions.
What should I remove from a survey export before analysing it?
Delete columns that carry no answers you will analyse: timestamps, names, email addresses and similar platform metadata, along with the responses to screening questions that only filtered who could continue. Remove respondents who skipped required questions. The Respondent ID stays, because without it you cannot tell individuals apart or trace a row back if something looks wrong.
What is the difference between one-way tabulation and cross-tabulation?
One-way tabulation examines a single variable and shows how its responses are distributed. Cross-tabulation examines two variables at once, usually a single-response question against age group or gender, and shows frequencies and percentages within each group. The second reveals whether response patterns vary between groups, although it describes that variation rather than testing it statistically.
Exam move
Start by drilling the six response types until you can label any closed-ended question on sight, since every method question in this chapter depends on that label. Next, write the three analysis levels from memory with their techniques and weeks, and practise sorting short research aims into the right level.
For tabulation, invent a small set of counts, convert each to a percentage by hand and write two sentences of interpretation; do one select-all version so you remember the totals can exceed 100%. Rehearse the coding rule with one example where a mean is valid (a Likert-type item) and one where it is meaningless (a coded category list).
Finally, answer each practice prompt in about 220 words under timed conditions: state whether the proposed method fits, name the response type, justify with the rule and suggest the better alternative.
Working through Preparing Survey Data and Basic Analysis in MKTG3001? Sia is AskSia’s AI Marketing tutor — ask any MKTG3001 Preparing Survey Data and Basic Analysis question and get a clear, step-by-step explanation grounded in how MKTG3001 is taught and assessed. Read this chapter free, then take your hardest questions to Sia.