Levels of Measurement in Structured Questions
Four lines on one slide carry the whole of this material's account of response choices. Three of the four terms are given no definition at all, the one worked scale is filed under a heading the slide's own wording contradicts, and the authority for the topic is deferred to a place that does not contain it.
The whole slide, in full
This is a short slide and the shortness is the point. It arrives two slides after the question-types comparison and five slides after the design question that asks "What response format can I use to collect the best information from the respondent?" — a question the material leaves open until this point and then answers in four lines.
Stated once, in your interest: this slide sits inside a week whose single learning objective is to evaluate methods of data collection. The delivery audit records questionnaires as delivered only in part and the objective's verb — evaluate — as not delivered at all. What follows is four labels, not an evaluation.
What is defined, and what is only illustrated
THE FOUR RESPONSE TYPES AND WHAT THE SLIDE SUPPLIES FOR EACH
| Response type | Definition given | Example given | Coding shown |
|---|---|---|---|
| Dichotomous | None — four instances only | yes/no; agree/disagree; true/false; male/female | None |
| Nominal | None | Language spoken at home | 1 = English, 2 = French, 3 = Chinese |
| Ordinal | "ranked in order" | Preferred travel destination | None |
| Interval | None | An agreement scale | 1 = strongly agree, 2 = agree, 3 = neutral, 4 = disagree, 5 = strongly disagree |
The five-point agreement scale is the only complete response scale printed anywhere in the supplied material. It is an example, not a recommended scale length: the source states no number of scale points, does not discuss the neutral midpoint, and never compares scale lengths.
The deferral that leads nowhere
The slide flags its own thinness with a parenthesis: "Measurement: (mentioned in topic 7)". A reader is being told that the substance sits in an earlier week.
Interval and ordinal cannot be told apart from what is printed
Read the two entries against each other. Ordinal is defined as "ranked in order". The interval entry gives a five-point scale running strongly agree, agree, neutral, disagree, strongly disagree — which is ranked in order, and therefore satisfies the slide's ordinal definition exactly.
What the slide states
- Ordinal is "ranked in order"
- The example of ordinal is a preferred travel destination, with no coding shown
- Interval is given no definition at all
- The example of interval is a five-point agreement scale, coded 1 to 5
What follows from that
- The interval example meets the stated definition of ordinal
- Nothing on the slide states any property that separates the two
- The two categories are therefore indistinguishable as presented
- The one worked scale in the material is filed under a heading its own slide cannot justify
Dichotomous sits beside the levels and is not one of them
The slide sets two co-ordinate headings: "Dichotomous" and "Measurement". That arrangement presents a two-option item as a fourth thing of the same kind as the three levels beneath it.
It is not the same kind of thing. A question with two options still has to sit somewhere on the measurement scheme — yes/no and true/false are unordered pairs, agree/disagree is an ordered one — and the slide never says which, never relates the two headings, and offers no rule for coding a two-option item. The relationship is simply absent.
What each level permits at analysis
This is the question a practitioner actually has when choosing a response format, and it is worth stating plainly what the material does with it.
WHAT THE SOURCE SAYS YOU MAY DO WITH EACH LEVEL
| Level | What the material states about analysing it |
|---|---|
| Dichotomous | Nothing |
| Nominal | One sentence, in the following week: central tendency is defined there as the "most common value for variables measured at a nominal level" |
| Ordinal | Nothing |
| Interval | Nothing |
The single statement in the nominal row is itself contested: the extraction records that it defines central tendency as the mode, in a table that separately defines the mode. See the note below.
The fourth level the source never names
Two vocabularies in one subject, never connected
There is a second classification of data elsewhere in the supplied material, from an earlier week, which divides data into attribute, ordinal and numerical. It is set out on Attribute, Ordinal and Numerical Data.
TWO SCHEMES, ONE SHARED WORD
| The earlier week's scheme | This slide's scheme | Relationship stated by the source |
|---|---|---|
| Attribute (categorical) data | Nominal | None |
| Ordinal data | Ordinal | None — the same word appears in both schemes, in different company |
| Numerical data | Interval | None |
| Not addressed | Dichotomous | None |
The two schemes are never mentioned together, neither refers to the other, and no mapping between them appears anywhere in the supplied material.
Both schemes are the source's. Neither is presented here as the correct one, and a reader should be aware that using them interchangeably in a write-up would be assuming a correspondence the material never states. Pick one vocabulary, say where you took it from, and use it consistently.
Recording your response formats defensibly
The material gives you four labels, one definition and two coded examples. That is enough to document an instrument and not enough to justify an analysis. The decision aid below is this page's construction; it uses the source's own labels and adds nothing to them.
Deciding and recording the format of each question
Correction: the fourth level does appear
What to carry forward
- Four response types are named on one slide and only ordinal is defined — "ranked in order".
- The five-point agreement scale is the only complete response scale in the material. It is an example, not a recommended scale length.
- The interval example satisfies the slide's own definition of ordinal, and nothing on the slide separates the two categories.
- The slide defers its definitional content to an earlier week that does not contain any of these terms.
- Dichotomous is presented as parallel to the levels and is never related to them; a two-option item is left unplaced.
- The material states nothing about what any level permits at analysis, while the following week builds a coding rule on exactly that knowledge.
- A fourth level exists in standard practice, is never named here, and is where most counts, durations and amounts belong. Get it from a second source.
Frequently asked questions
How does this material define nominal, ordinal and interval?
It defines only ordinal, as "ranked in order". Nominal and interval are given through examples alone — a language-spoken-at-home item coded 1, 2, 3 for nominal, and a five-point agreement scale for interval. Dichotomous is likewise given as four instances with no definition.
Why does the slide say the topic was covered in an earlier week?
It carries the parenthesis "(mentioned in topic 7)", and the material of that week contains none of these terms — not nominal, not ordinal, not interval, not level of measurement. The cross-reference is dead, and the slide gives no warning that it is.
Is a five-point agreement scale interval or ordinal?
The slide labels it interval; the slide's own definition of ordinal — ranked in order — also covers it, and nothing on the slide distinguishes the two. The material cannot settle the question, so if your analysis depends on the answer, settle it from a statistics or research methods text before you collect.
Are there only three levels of measurement?
No. The source gives three and does not say its list is partial; a fourth is recognised in standard practice and is not named anywhere in this material. This page deliberately does not teach it, because inventing a definition and attributing it to a source that lacks one would mislead you. Take it from a second source.
What analysis does each level allow?
The material never says. Its only statement connecting a level to an analytical operation appears in the following week, where central tendency is defined as the most common value for variables measured at a nominal level — a definition the same table contradicts by separately defining the mode. Treat the question as unanswered by this source.
How does this relate to the attribute, ordinal and numerical scheme used earlier in the subject?
The supplied material never connects them. Two classifications appear in different weeks, sharing the word ordinal and nothing else, with no mapping stated between them. Choose one vocabulary for your write-up, say where you took it from, and do not assume the two sets of terms line up.
References and source attribution
- Bryman, A. 2016, Social Research Methods, 5th ed., Oxford University Press, Oxford.
- Veal, A. J. 2005, Business Research Methods: A Managerial Approach, Longman.
- O'Leary, Z. 2017, The Essential Guide to Doing Your Research Project, 3rd ed., Sage Publications, London.
- Trochim, W. M. K. 2006, Research Methods Knowledge Base.
- The supplied teaching source: the week 10 slide deck on collecting data, the slide on choices of responses in structured questionnaires, read with the week 11 slides on coding and on terms used in quantitative data analysis.
Suggested questions for Ask KEVOS
- Classify each question in my instrument by the response format this material names.
- Write the code list for my questionnaire, with a note on what each set of numbers means.
- Which of my measures fall outside the three levels this slide names?
- What do I need from a second source before I can average my agreement scale?
- Draft the measurement paragraph for my methodology chapter using only what this material supports.