1.6 THE FOUR VALIDITIES

validity

• validity: refers to how accurate it (claim) is given the data at hand

  • this is not a single measure or criteria

  • a claim from a single study is not simply valid or invalid

• every study balances four main types of validity

  • construct

  • external

  • internal

  • statistical

construct validity

• refers to how well the variables in a study measure what they intended to measure

• ask yourself

  • does the measure provide an accurate estimate of the theoretical construct?

  • is the operational definition appropriate?

• an invalid operational definition makes it impossible to evaluate claims

• evaluate construct validity of frequency claim by addressing measurement od the single variable in the claim

• evaluating construct validity for an association claim: address each variable

• in an experiment testing a causal claim the operational definition of an independent variable is how the variable will be manipulated in the study

  • evaluating construct validity of an experiment

    • does the independent variable and its manipulation reflect a change in the construct represented by the independent variable?

    • evaluate operational definition and measure for dependent variable

external validity

• refers to the degree to which a claim generalizes to a larger population or to other situations

  • ask yourself

    • about people- who does the result creating the claim most apply to?

      • if I conducted this study with different participants, would I expect the same results?

    • about situation- is this finding representative of other circumstances?

    • people- who are participants and who does the result apply to?

    • situation- what is the situation and would results apply to other contexts, times, or places?

internal validity

• refers to the extent to which the dependent variable changes due the independent variable, and not some alternative variable

  • ask yourself

    • is there an alternative explanation for change in the dependent variable?

      • if yes, you have a confound- something other than the intended manipulated variable also varies between conditions or levels of the independent variable

      • confounds weaken internal validity

    • were potential confounds in the manipulation controlled?

statistical validity

• refers to the degree to which the statistical results support the claim

  • ask yourself

    • did they do the right statistical test?

    • how big it the effect?

    • what is the margin for error in the outcome variable?

    • likelihood of occuring by chance- can the effect by replicated?