MKB2705 Chap.8 Descriptive Data Analysis
Descriptive Data Analysis
Descriptive analysis establishes what the dataset contains and what it can represent before an inferential method compresses the evidence. Begin with the case ledger, eligible population and missingness, then inspect each variable using summaries suited to its scale. The denominator is part of every result.
A percentage of all eligible cases, valid item respondents or people routed to a question can answer different questions, so publish the base and explain exclusions. Nominal variables need counts and proportions; ordered responses need a visible distribution and defensible positional summaries; quantitative measures may support means and standard deviations when their interpretation is justified.
Centre and spread answer different questions. The mean uses every value and is useful for additive planning, while the median describes the middle ordered case and resists extreme observations. The mode identifies the most common category or value. Range uses endpoints, interquartile range describes the middle half, and standard deviation expresses variation around the mean in original units.
No single statistic captures shape, clusters, gaps, skew or a decision-relevant tail. Extreme values need provenance rather than automatic deletion. They may be coding errors, genuine heavy users or rare cases that reveal operational risk. Compare conventional and robust summaries when conclusions are sensitive. Cross-tabs show joint categorical distributions, but row, column and total percentages use different denominators.
Choose the direction that answers the research question and show counts so a large percentage from a tiny cell is visible. Charts should make one comparison readable without distorting scale. Use bars for categorical magnitude, histograms for distributions, points and intervals for estimates, and scatterplots for paired quantitative measures. Titles and captions should state population, measure, period, base and weighting.
The independent perceived-access example reports mean, median, SD, IQR, valid base, missingness and a lower-access tail. It also notes an evening-day pattern without converting the cross-sectional association into a staffing effect. Description should end with a decision inventory: which patterns are robust enough to act on, which require inference, which need design follow-up and which remain uninterpretable.
This chapter teaches standard descriptive-analysis canon aligned to the official sequence and SPSS report requirement. All numbers and practice items are independent.
What this chapter covers
- 01
Decision focus
- 02
Evidence control
- 03
Error control
- 04
Claim boundary
- 05
Decision use
AskSia-authored practice weighting (not an official mark scheme): Original worked model: descriptive data analysis
- +1State the management or evidence question, population and decision use before naming a method.
- +1Choose and defend the relevant design, measure, sample or analysis, preserving the correct valid base.
- +1Report the result or planned output with magnitude, uncertainty and a precise evidence noun.
- +1State the main limitation and a conditional decision or follow-up that does not exceed the evidence.
Key terms
- valid base
- A key concept in Descriptive Data Analysis: define it in the population, context and decision for this study rather than relying on a label alone.
- interquartile range
- A control term used to keep descriptive data analysis technically and conceptually traceable across the project.
- conditional percentage
- A boundary or diagnostic that should appear beside the relevant result, not only in a generic limitations paragraph.
Descriptive Data Analysis FAQ
How does descriptive data analysis support MKB2705 assessment?
It supplies a specific research decision, evidence control and claim boundary that can support the proposal, class-test reasoning, SPSS report or reflection. Apply it to your own project and current Moodle task rather than copying the worked model.
What is the most common descriptive data analysis mistake?
The common failure is letting a convenient method, software output or confident phrase replace the decision question and represented evidence. Keep population, construct, valid base and design visible.
Can AI complete this descriptive data analysis work?
No. Follow the current task-specific rule. Use only permitted support, verify every source and number, author submitted prose yourself and complete the required declaration. Quizzes say AI should not be used.
Assessment move
Use a fixed description order for every primary variable: eligible and valid base, missingness, labels and range, full distribution, centre, spread, extremes and decision meaning. Produce a frequency table before collapsing any category and explain whether row, column or total percentages answer the question. For a quantitative measure, compare mean with median and SD with IQR, then inspect the plot to explain any difference.
Trace extreme observations to source before choosing a treatment. Create one honest chart from a named analysis table, include population, period and base, and test whether it remains readable without colour. End with a decision inventory separating patterns ready for action, patterns needing inference and patterns blocked by measurement or coverage.
Confirm current task conventions and reproduce all outputs from controlled syntax.
Working through Descriptive Data Analysis in MKB2705? Sia is AskSia’s AI Marketing tutor — ask any MKB2705 Descriptive Data Analysis question and get a clear, step-by-step explanation grounded in how MKB2705 is taught and assessed. Read this chapter free, then take your hardest questions to Sia.
