Identifying the Strengths and Weaknesses of a Scientific Investigation for the GED Science Test

By Murray Shukyn, Achim K. Krull

The GED Science test will expect that you are able to identify strengths and weaknesses in a scientific investigation. Scientific investigations should be empirical; that is, conclusions should be based on verifiable observation, experience, and experimental evidence. For a scientific investigation to produce reliable results, it must meet all the following criteria:

  • Participants/subjects must be chosen randomly.

  • Controls must be in place to reduce or eliminate variables not being tested.

  • Only one variable can be manipulated and tested. (More than one may be used, but that makes statistical analysis difficult.)

  • Results must be quantifiable — size, number, weight, or something else that can be measured. For example, the number of hairs on a cat is quantifiable, although they would be very difficult to count, whereas the softness of the cat would be qualitative — a judgment call, unless you could figure out some way to quantify it.

  • Participants/subjects must be assigned randomly to either the experimental or control group.

  • Participants/subjects may also be retested, so they’re tested in both the experimental and the control group. This “repeated measure” technique produces more uniform results. When repeated measures are done, counterbalancing may also be done to reduce the effects of the order in which participants are tested in either the experimental or control group.

  • All evidence must be reported, even if — and perhaps especially if — it doesn’t support the hypothesis. If evidence is excluded, the reason for the exclusion must be provided.

Controls are particularly important. When pharmaceutical companies conduct tests on medications, they commonly use the following three types of controls:

  • Control group: A group that establishes a baseline from which results are measured. The control group receives no treatment or a neutral treatment. Results from the treated and untreated (control) groups are compared to determine whether treatment had any effect.

  • Placebo: Untreated participants in a study often respond differently if they think they received treatment. To account for this placebo effect, researchers provide a neutral treatment (such as a “sugar pill”) that has no real effect.

  • Blinding: Those conducting the study hide the fact that some participants are receiving a placebo and some aren’t, because when people know they’re getting a placebo, they’re less likely to respond to it. In a double-blind test, neither the researcher nor the participant is aware of who’s getting the placebo, so nobody involved can inadvertently influence the results by what they say or do.

Another factor that determines the quality of a scientific study is its size. A study that involves a large number of experiments, participants, or measurements is likely to produce a more accurate body of data than does a smaller study. When testing products, researchers often refer to each experiment as a trial.

  1. The number of variables that should be changed in a properly designed and well-controlled experiment is

    • (A) 0

    • (B) 1

    • (C) 2

    • (D) depends on what is being tested

  2. When planning and conducting an experiment, one should

    • (A) carefully choose participants

    • (B) test as many variables as possible

    • (C) establish controls to eliminate as many variables as possible

    • (D) try to prove the hypothesis

  3. Which of the following evidence should be reported in the results of a scientific study?

    • (A) results from only one selected trial

    • (B) results from only those trials that prove the hypothesis

    • (C) results from only those trials that falsify the hypothesis

    • (D) results from all trials

  4. Which of the following is not quantifiable data?

    • (A) beauty

    • (B) height

    • (C) weight

    • (D) speed

Check your answers:

  1. Ideally, you test only one variable in any given experimental trial, Choice (B).

  2. When conducting an experiment, you should establish controls to limit variables so the only variable is the one you’re changing/testing, Choice (C).

  3. When publishing the results of a scientific study, you should report all results, Choice (D), even those that are excluded from your analysis.

  4. You can measure height, weight, and speed, but not beauty, Choice (A).