When making a statement from a survey...

hihihi9993

Given the survey population is a representative of the whole, let's say that a survey result showed 50% of people polled believed that Sally is silly, whereas 30% believe that she is not.

Is it valid to conclude that more people believe that Sally is silly than believe she is not?

Achen165

There is no straight answer here...it is definitely contingent upon context. A rule of thumb is to be skeptical when the LSAT presents statistical data such as percentages to numbers, as it is a flaw commonly tested upon. It is less likely that the survey is actually representative of the whole than is not.

Depending on the stimulus, you need to be positive nothing was missed or overlooked in your read through to know that the survey is FULLY representative of the same set, through the premises and the conclusion. The amount of those polled and those that represent an entire population/relevant subset of the population must be on par with one another. You'd likely need some indication that all of the necessary participants were actually participants in the survey.

There is a bit of a loophole here, for the remaining 20%...what if they had no opinion/ alternate opinion? That would disprove a conclusion that more people believe that Sally is silly...a 50/50 split.

This sort of conclusions that can be drawn are also contingent upon question type...paying close attention to language of the relevant answer choices to see how strong of a conclusion can be drawn.

BinghamtonDave

is it the same survey? is that explicitly stated?

hihihi9993

Thank you @achen013 and @BinghamtonDave !

The reason why I asked this question was because of PT22 S2 Q25 AC (A). It is a flaw question, and I initially chose (A - draws a conclusion about the population in general based only on a sample of that population) for the stimulus not explicitly stating that the survey population is a representative of the whole, and for it making statements about #s when only % are given. At the time of taking the practice test, I didn't notice that the stimulus was talking about different sets, which is (B - suff nec confusion), I chose (A). After acknowledging that (B) is the correct answer, I tried to eliminate (A) by the reasoning that one should not doubt the representativeness of a sample unless we are given a reason to doubt.

So to rephrase my questions...
(1) Is it valid to think that one shouldn't doubt the representativeness of a sample unless we are given a reason to doubt? But if we say that we can conclude that a sample is a good indicator of a whole merely based on the fact that [reasons to doubt] is absent, I feel like we could also conclude that a sample is not a good indicator of a whole for [reasons to accept such representativeness] is absent.
(2) Given that [we know the distribution (%) of each part of the pie] + [the survey population FULLY represents the general population], are we allowed to draw conclusions about #s about the general population just based on %s of the sample population?

Thank you for your help!

Achen165

The lingual shift in that question is indeed very subtle. 'A' is incorrect because the argument does not actually draw any conclusion about the "general population," the words 'more' and 'most' are not interchangeable here. The argument says that 'more' people vote a certain way (indicating that the the percentages and numbers are directly proportional...regardless of whether the number 100 or 100 billion...more will still indicate a higher numerical value)...if the argument said 'most' then it could potentially be making a general claim about the population and support an answer choice like 'A,' especially in the absence of other flaws.

To answer your questions (2) we do not know if the survey in this case is representative of the population...it is also irrelevant here...the conclusion drawn pertains strictly to the data that there is 'more' evidence to support the idea that indictment should lead to resignation than believe that convictions should, based on solid percentage disparities. The strength of the inference you can draw is always context dependent. (1) Doubt representativeness when representativeness is an issue...the flaw in this Q can be overlooked with quickness for this reason because the percentages seem to contain flaw, which distract from s.c/n.c. distortion that occurs between premises and conclusion as it is far more subtle and seemingly overshadowed by percents.

hihihi9993

@achen013 said:
The lingual shift in that question is indeed very subtle. 'A' is incorrect because the argument does not actually draw any conclusion about the "general population," the words 'more' and 'most' are not interchangeable here. The argument says that 'more' people vote a certain way (indicating that the the percentages and numbers are directly proportional...regardless of whether the number 100 or 100 billion...more will still indicate a higher numerical value)...if the argument said 'most' then it could potentially be making a general claim about the population and support an answer choice like 'A,' especially in the absence of other flaws.

To answer your questions (2) we do not know if the survey in this case is representative of the population...it is also irrelevant here...the conclusion drawn pertains strictly to the data that there is 'more' evidence to support the idea that indictment should lead to resignation than believe that convictions should, based on solid percentage disparities. The strength of the inference you can draw is always context dependent. (1) Doubt representativeness when representativeness is an issue...the flaw in this Q can be overlooked with quickness for this reason because the percentages seem to contain flaw, which distract from s.c/n.c. distortion that occurs between premises and conclusion as it is far more subtle and seemingly overshadowed by percents.

Thank you so much for helping out! Things got clear!