Test design question

@joefollette, dont’ understand what you mean.

Pls show some example.

1. I love to eat grapes. False (3), Slightly True(2), Mainly True(1), Very True(0) [Category Fruit]
1. I hate grapes. False(0), Slightly True(1), Mainly True(2), Very True (3) [Category Fruit]

Suppose I answered False to both questions – Total Fruit category would be 3

Ok, so:

* add “Category Fruit” scale
* add “false”, “sl true”, “mly true”, “true” global answers
* link those scale/answers with test
* add those 2 questions and save
* fill-up scores in cells

You should get something like this then:

https://gyazo.com/0f9ad62fb0bebd4041f9b8394b280650

In the example you provided love and hate had the same score (3). Added together will equal 6. In my example they have different scores 3 and 0. For a total of 3 in that category. The same answer False – is a assigned a different score in each question.

@joefollette, it’s a same as in your example:

love+no = 3
love+yes = 0

and

hate+no=3
hate+yes=0

Your example also can’t have 3 as a total as if respondent will select both:
* I love to eat grapes. False (3)
* I hate grapes. Very True (3)

It will be total 6 as 3+3 is 6.

I think I dont’ fully understand your model, let’s try another example with questions/answers/scores AND with some results, that respondent will choose — so what should respondent get in different situations.

[Scale=Fruit] Trying to see how much I like fruit. The higher the score the more I like fruit. Answers in parenthesis
1. I love to eat grapes. “False(0)”, Slightly True(1), Mainly True(2), Very True (3)
2. I hate grapes. “False (3)”, Slightly True(2), Mainly True(1), Very True(0)
3. I love apples “False(0)”, Slightly True(1), Mainly True(2), Very True (3)
4. I hate apples False (3), “Slightly True(2)”, Mainly True(1), Very True(0)

So with the above answers my total score for fruit scale is 5.

The test needs two answer templates that are used based on how the question is asked.

False(0), Slightly True(1), Mainly True(2), Very True (3) and
False (3), Slightly True(2), Mainly True(1), Very True(0)

The reason this is done is to ensure the respondent is answering questions consistently. It asks the same question in a negative way and flips the scoring to be sure they match. So if I hate apples 0 I shouldn’t then also love them 0. It should be I love apples 3,2 or 1 and I hate apples 0

Yes, I see, that you use here control questions.

For this case in tests typically used lie scales.

See example of test: https://gyazo.com/96bd51e38213230ab1f71345f3b26913

It’s not ideal, but something like this.

So q1 is main question. In lie score we mirror it’s scores. But the main requirement for them is to be different and positive.
q2 is our control question — it does not have anything in fruit scale. Only inverted scores in lie scale.

So when all answered truly:

https://gyazo.com/aa921f92f10bc16793c808bff24b3193

And two cases of lie:

https://gyazo.com/0fbfe59e42ba2c93d4aecc5b423b686b
https://gyazo.com/4910350296e9f576223f0469f75f3886

The scales for this test are already set so I can’t add a lie scale. So are you saying your test is unable to handle different answers for each question? And add up the answer scores from the answers they collected? They must be the same answer points for each question?

test is unable to handle different answers for each question?

No it can handle global/individual answers and also any scores that you put into it.

And add up the answer scores from the answers they collected? They must be the same answer points for each question?

I dont’ understand these questions. Pls rephrase.

The scales for this test are already set so I can’t add a lie scale.

Then your model with only single scale and mirroring questions becomes unusable. For example how you will differ these two cases:

* respondent tells always true and gets low scores
* respondent lies and gets low scores

Somthing like this is possible.

I think you should step back and check your test’s model, not it’s implementation.

Just sent you a copy of the test. Maybe you can show me how t implement this.

Ok, so as I see there are 22 scales, and if we look on it as subsccales — there are 42 scales.

If you need only scales, then add 22 scales.
If you need only sub scales in results — then add only sub scales as scales.
If you need both — then add scales with sub scales.

So for example you have scale BLA and sub scale BLA-A and sub scale BLA-B.
You will add both of these scales and then will fill scores for both scale and sub scale:

* q1 scores will be filed same in BLA and BLA-A
* q2 scores will be filed same in BLA and BLA-B

In result BLA will have scores from BLA-A and BLA-B.

In your concrete situation you will have 347 questions multiplied on 22 scales multiplied on 4 answers will give 30536 score inputs in text editor. You will need to increase max_post_vars limits to 40000.

Current plugin version will handle this test, but I’m not sure regarding UI and your server settings. In future versions I will plan to improve it for this many-questions-scales cases.

Ok thanks.