Join 3,374 readers in helping fund MetaFilter (Hide)


Better synonyms needed
January 8, 2009 6:54 AM   Subscribe

Vocabularyfilter: I am looking for some near-synonyms for "binary" and "analog" but without the techie sheen those both have.

For a checklist of questions, I need a way to distinguish between two types of questions: some of the items are answerable by a simple "yes or no" while others have a spectrum of responses, each worth a certain value on a numerical scale. That is to say, if questions are worth five points each, some will be scored only zero or five (the "yes or no" items), while other items might get a two or a four or a one. The placeholder terms I am using right now are "yes or no" and "rateable," but I am sure there must be more elegant terms than these.
posted by ricochet biscuit to Writing & Language (23 answers total)
 
Dichotomous versus continuous (or ordinal, interval or ratio for slightly different meanings). Those are the standard terms for this in measurement/methodology.
posted by If only I had a penguin... at 7:01 AM on January 8, 2009


yes/no and gradated?
posted by delmoi at 7:02 AM on January 8, 2009


I'm not entirely clear on what you're doing. Do you want multiple choice questions with more than one right answer but one best answer? Or do you want short-answer type questions for which partial credit can be given?

A standard distinction in grading is objective/subjective. True/false and multiple choice are generally considered objective, in that there is only one correct answer. Short answer and essay questions are subjective, as their score depends on the impression they make on the grader. Fill-in-the-blank can call somewhere in between, depending on how they're constructed.

It sounds like what you're doing is pretty atypical, in that you seem to be wanting to offer a multiple-choice type question with partial credit. Most teachers avoid this because it makes grading a huge pain in the arse. I think there's a better way of doing it. Take the following, for instance.

1) 1 + 1 = 2
2) 2 + 2 = 4
3) 2 + 3 = 6

Choose the best answer:
A: 1 is true
B: 2 is true
C: 3 is true
D: 1 & 2 are true
E: all are true

Both A and B are "correct" answers, but D is the best one. This type of question is a good way of making students really pay attention to what they're doing rather than simply checking off the first answer that looks right.
posted by valkyryn at 7:08 AM on January 8, 2009


categorical vs. graded
posted by beniamino at 7:10 AM on January 8, 2009


continuous vs. discrete
posted by BaxterG4 at 7:21 AM on January 8, 2009 [2 favorites]


What's the audience? Who will have to understand these terms?
posted by amtho at 7:28 AM on January 8, 2009


"all-or-nothing" vs. "scaled"
posted by Jaltcoh at 7:35 AM on January 8, 2009 [1 favorite]


On tests I've always heard these described at True/False and Multiple Choice questions.
posted by vytae at 7:36 AM on January 8, 2009


"All or nothing" vs. "Partial Credit"
posted by toomuchpete at 7:38 AM on January 8, 2009 [1 favorite]


The usage is for site inspections of facilities around the country; the audience is the managers and inspectors. There are standards which are either all-or-nothing (they either have this service or they don't) and some which are graded on a scale (they get partial marks for doing this, but higher marks for doing THAT).
posted by ricochet biscuit at 7:38 AM on January 8, 2009


I think you need to pick up a textbook on the subject and look up levels of measurement as well as variable types. Then use that to classify the variables you are thinking about.

There are lots of ways of categorizing variables:

nominal/categorical(both mean the same) are those where there is no "order" involved. For example "Yes" is neither more nor less than "No". Religion, gender, race etc. are also nominal/categorical -- it's not a question of being/doing/having more or less of something, it's just which category you fit into.

ordinal variables: There's a clear order, but it's not clear how much bigger one value is than another. For example "very rarely, rarely, occasionally, often, very often" would be ordinal. You can order them, but it's not like you know that people who say "occasionally" have done something 3 times per week more than people who say "very rarely". Some variables that use ranges would also qualify as ordinal: "Less than high school, high school, some college, college graduate" or "Less than $20K,$20,001-$40K, More than $40,000K")

interval: There's an order and you know how far apart the values are. So for example year -- we know that 2008 is two years later than 2006. And temperature: 30 degrees is five degrees colder than 35. HOwever, note that 0 for both of these does not mean nothing -- Year 0, if it existed wouldn't mean "you have no year" or "there is no earlier". It's basically arbitrary. You cannot say that the year 1400 is twice as late as 700 or that 60 degrees is twice as hot as 30 degrees.

Ratio: Like interval, there's an order, you know the distance between answers. Now add a meaningful 0 point where 0 means you actually have "none" or whatever you're measuring. So temperature in Kelvin is ratio, so is age, dollars earned, network etc. are ratio.


continuous versus discrete:
This generally applies only to things that can be ordered. The question here is whether it's possible (in theory or practice) to have *any* value for this variable. For example, you can have any weight: You can weigh 200 lbs or 200.000038 pounds or 200.8883988 pounds. Therefore, weight is continuous. Number of jobs is discrete : You can have one job, or two jobs or three jobs. You can't have 2.499348 jobs.

Dichotomous: Dichotomous variables are a special case of discrete variables where only two values are possible. Yes/No and gender are the most common. Also, any nominal variable can be recoded into a series of discrete variables: Instead of race as "choose one of the following" you would have one variable for each race coded yes or no. If you had race as "choose one or more of the following" you would have no choice but to recoded into a series of nominal variables for analysis.
posted by If only I had a penguin... at 7:40 AM on January 8, 2009 [5 favorites]


Black/White or Shades of Gray
posted by eagleapex at 7:48 AM on January 8, 2009 [2 favorites]


hit-or-miss questions vs. degree questions

The second group is harder to name; something like "additive" or "multi-value" also occurred to me. Also "scaled" questions.

One thing that might help is looking at the reason you're going to the trouble of dividing the two groups. That is, what distinguishes the groups functionally for the people reading the lists? Are they just separated to make adding scores easier, or is there a different time or phase of the inspection for which the yes/no questions are more appropriate?
posted by amtho at 8:09 AM on January 8, 2009


Again, I stress that this is for a general audience without particular interst in statistical theory. I need nothing more complex than a way to signal to facility managers that while many standards are gradated/scaled/continuous, some are not. "Either your fire extinguishers are charged or they are not. You don't get four out of five points for having most of them charged."

And for what it's worth, I live in Canada and will have to render these terms in both English and French.
posted by ricochet biscuit at 8:12 AM on January 8, 2009


For binary conditions: Pass/Fail
For analogue conditions: Graded out of a 5-point scale/100 percentage/scale of your choice
posted by Happy Dave at 8:35 AM on January 8, 2009 [1 favorite]


For binary it seems like Pass/Fail may work for you as the simplest option, in a safety setting. (I have also seen Compliance / Non-Compliance used for the binary safety stuff, but that is more jargony.)

When grading for safety reasons at my work place, there is a code assigned, 1 - 4 scale, called a risk assessment code (RAC).

See below for definitions:

RAC
LEVEL OF RISK
ACTION REQUIRED

1
Critical risk to health and safety.
(High severity / High probability of occurrence)
IMMEDIATELY CORRECT HAZARDOUS CONDITION (or reduce to LOW risk) WHEN IDENTIFIED.
In 30 day report, describe actions taken and programmatic measures in place to prevent recurrence.

2
Significant risk to health and safety
(High-to-moderate severity / High-to-moderate probability of occurrence)
Immediate action to mitigate completely or reduce risk to LOW. In 30 day report, describe actions taken, interim plans for completion, and program measures in place to prevent recurrence.

3
Moderate risk to health and safety (moderate-to-low severity/moderate-low probability of occurrence)
Within 60 days, action taken to mitigate completely or reduce risk to LOW. In 120 day progress report, describe actions taken, interim plans for completion, and program measures in place to prevent recurrence.

4
Low risk to health and safety
(low severity / low probability of occurrence); could include administrative deficiencies with minimal risk but high probability of regulatory citation.
Within 120 days, complete mitigation. In 120 day progress report, describe actions taken and plans to prevent recurrence.
posted by gudrun at 9:15 AM on January 8, 2009


and will have to render these terms in both English and French

One more reason to use "all or nothing," since that'll be fun to translate: "tout ou rien"!
posted by Jaltcoh at 9:23 AM on January 8, 2009


Dualistic vs. Non-Dualistic
posted by Roach at 9:53 AM on January 8, 2009


black and white
pass/fail
all or nothing

shades of grey
continuum, range, scale
posted by Good Brain at 10:31 AM on January 8, 2009


Gradual versus exclusive - [like a on-off button or whatever]
posted by Namlit at 10:44 AM on January 8, 2009


Pass/Fail puts a little fire under the issue.
Partial Credit for those where horseshoe rules apply.
posted by pointilist at 11:36 AM on January 8, 2009


Pass/Fail
Graded on a Scale

would be my two votes. Has a formal ring (which, to my eyes, "all or nothing" and "black and white" don't) while being pretty unambiguous. As an uncouth monolingual American, I have no idea how those sound in French.
posted by Tomorrowful at 12:12 PM on January 8, 2009 [1 favorite]


"Acceptable/Unacceptable"

and

"Performance"
posted by mrmojoflying at 3:13 PM on January 8, 2009


« Older I can't pronounce 'th' sounds ...   |  Pound for pound, megajoule for... Newer »
This thread is closed to new comments.